Comparing FS 530 and TP 550.

[hitechjb1]

How is a Fortron Source 530W compared to a TP 550 since their rated wattage is about the same. The FS 530W has a cap of 510W on the 3.3, 5, 12 V together, whereas the Antec TP 550 has a 530W cap.

Other than just the spec comparison, anyone has first hand experience with either PSU under full load. Typical config: 1 DVD, 1 CDRW, 2 HD, CPU fan + 4 case fan, 2 memory sticks, a 9700/9800 video card, a 2.5+ GHz XP or a 3.4+ GHz P4, ... The higher CPU freq the better for comparison.

Running XP under 2.5+ GHz won't stress either PSU enough.

How good
1. each holds the 5V/12V rail from no load to full CPU load (% voltage drop),
2. % voltage drop when running CPU from 2 GHz to 2.5+ GHz (or even higher),
3. voltage fluctuation during full load (like running Prime95, ...).

If possible, please quantify the comparison than just comment like "its great PSU", ...

[larva]

Originally posted by hitechjb1

Running XP under 2.5+ GHz won't stress either PSU enough.



I agree there, but rasing the clock rate to 3GHz or introducing a 3.6GHz P4 doesn't change the situation. These things can be run easily with a top notch 400W supply, there is no realistic load one can create that tests a power supply that truly makes over 500W.

Both are great supplies, and both are overkill. Exactly which is more overkill isn't a question easily answered without the means to create a stiffer load than any mb/cpu/ram/drive/video card combination you can currently assemble.

[hitechjb1]

The other day, AntecRep pointed to these test results for TP.

Power Supply Specifications, Testing, and Antec TruePower (http://www.antec-inc.com/info_DIYArticle1.html#)

Test Results for True480 (http://www.antec-inc.com/info_DIYArticle_true480spec.html)

Test Results for True550 (http://www.antec-inc.com/info_DIYArticle_true550spec.html)

Have you come across some test results like that for Fortron, and the 530W in particular.

[hitechjb1]

I also have some questions about the testing results:

It is an attempt to show how the TP series has some hidden juice. About 16% for the TP430, and about 15% for the TP480, esp for the TP480 which is like a TP550 when overdrawn. The test did not show any overdrawn for the TP550.

I hope I can find these detailed testing for some Fortron PSU, which few people on this forum say Fortron PSU can be overdrawn by 30%. I'd like to see some detailed testing results for Fortron, like what Antec TP did.

I looked at the Fortron 530W (high watt series, non EPS). It also has a limit of 510W for 3.3V + 5V + 12V.

Due to the Fortron 510W limit, and the fan noise and location, I am leaning towards the TP480/550 for upgrade if needed. Also the FS 530W has lower 12V current than the TP 550 by 6A (18A compared to 24A). And 12V is important for most CPU oc. Also Antec TP has 3% line regulation, compared to 5% for FS. Other than this, rest of the spec are very similar.

Some comments/questions:

1. In the TP430 test, I don't know why when they overdraw the PSU by 10%, they cap the 5V line current to 36A (the 5V rated current). Also the 12V and 3.3V current are overdrawn by 10% at the expense of the 5V line (from 36A to 17A !!!). I am having problem with the 5V line in the TP430. Is there a hard limit of 36A for 5V ???

2. Same for the TP480 test, the 12V current is overdrawn at the expense of the 5V current (from 38A to 22A !!!!). The TP480 12V line is overdrawn as heavy as the TP550 12V rated current.

3. I am surprise that they did not show the hidden juice of a TP550. Is it because of legal ATX limit of 530 W, and they don't want to show it ???

4. The test shown the TP480 can be overdrawn to 540W (actual) which is as high as shown for the TP550 549W (actual). Both theoretical output are the same (555W vs 556W). Also the TP480 12V current is overdrawn to 24A as high as the TP550 (which is 24A).

I also don't know why they didn't overdrawn the 550. Does this mean the TP550 cannot be "overclock" as much as the TP480???

If TP550 cannot be overdrawn, the TP480 when overdrawn is like a TP550. So the TP480 is a better buy.

[flapperhead]

If u decide to get the antec, go with the channewell 550. they are antec,s oem manufacturer. They can be purchased for as low a 80 bucks... I paid well over a 100 for my antec tp550

[hitechjb1]

I did a serch on pricewatch.

There is only one supplier for the Channel Well 550, asking $85, but adding USP ground will be $94.

For Antec True Power, there are many suppliers. Cheapest is around $103 (including shipping).

The price diff is $9, if I had to buy, I would buy the Antec TP550 in case of problem and needed RMA down the road.

Also one concern for the True Power is that the fan is in auto mode. There are some complaints/questions about how to run the fans at higher speed if needed.

I am not aware of any mod to do that, unless getting the True Control instead of True Power, and Antec will get $20 more from the sale.

[mattspalace]

I just installed my TP550 this morning, and I'm happy with it so far. I got mine from Directron for 94.99 plus $8 ground shipping.
Pretty good, and when I found it, I posted to the cyberdeal section.

-I thought about going with Fortron, but since my last PS was an Antec, and I was happy with it, I stuck with them..

[hitechjb1]

Originally posted by mattspalace
I just installed my TP550 this morning, and I'm happy with it so far. I got mine from Directron for 94.99 plus $8 ground shipping.
Pretty good, and when I found it, I posted to the cyberdeal section.

-I thought about going with Fortron, but since my last PS was an Antec, and I was happy with it, I stuck with them..

Thanks for the info.

Do you find the auto fan speed control sufficient? Some complain that it is too slow under some situation.

They should give the user an option to disable the auto fan mode. Anyone knows such option if hidden or mod.

[mattspalace]

I think the auto fan speed control works fine really. Most of us have cases with several fans, so temps don't ever really get that high. I did notice that the fan speed would automatically increase as I added FSB.

My case temp is 32c and my CPU diode is 37c. The extra money for the true control just wasn't worth it to me, but $94.99 versus $109 isn't a huge price difference. If I had temp problems, then I probably would have gone with the true control though, or if I only had a single fan in the rear of the case.

[cV]

I have a true550. Very stable lines.

Everything I have is high-powered, look in the sig. Several 15K drives will test its ability.

[hitechjb1]

I posted on the AMD CPU section for testing the Fortron 530W with a 1700+. Repost here, maybe some experts here can help to see whether I got a bad PSU, or I did something wrong, or the Fortron 530W is indeed not that good.

There are pics about the snapshots of the voltages at idle and during prime95. It is on page 5 of this link: Tbred B 1700+ DLT3C 2.5+ GHz air results (http://forum.oc-forums.com/vb/showthread.php?s=&threadid=187887&pagenumber=5). If needed I can post them here too

Testing a Fortron 530W PSU

I have been thinking about upgrading to an Antec TP 550 or a Fortron 530W PSU, hoping to get better voltage stability for oc. Many ppl have highly recommend the Fortron 350W and 530W. Some even say the Fortron 350W is enough to do most job and can deliver 454W (I do not agree on this number).

I usually like to look at spec and testing, rather than just based on word of mouth. The Antec TP550 or TC550 has better specification, +-3% line tolerance instead of 5% of Fortron, and also higher current in the 12V line of 24A compared to 18A of Fortron. In additional Antec has better fan control and fan location to vent the CPU.

Anyway, for its cheaper price ($40-50 less), I got a new Fortron 530W PSU and tested on the Tbred B 1700+ setup, with a SLK-800U + TT SFII.

Here is some measurement of the voltage lines at idle and running prime95, @1.92V, 2544 MHz.

1. Voltage lines at idle

The 5V line dropped to 4.703 V, and flutuated between 4.703 - 4.977 V.

The 12V line was at 12.42 V, this was too high and might be too much for hard drives, CDRW, DVD, ...

For the Antec TP430, 12V line is around 11.93 V, +- 0.04 V.

See the two figures in next post for snapshot of voltages, ....

2. Voltage lines with Prime95

I saw the 5V fluctuated to a high of 4.977 V from 4.703 V, which is 0.274V. This fluctuation is more than the 5% line tolerence of the PSU. If taking potential measuring uncertainty of 5% of the ASUS probe, that fluctuation would still be 0.274/5 * 0.95 = 5.2% (almost in line with the spec tolerance).

Also I saw the Vcore bounds around between 1.85 to 2.0 V (in the snapshot it bounds between 1.888 V and 2.0 V), also this is outside of the AMD Vcore tolerence of +- 50 mV.

See the three figures in next post for snapshot of voltages, ....


To my surprise and many's recommendatation, it is disappointed to see the Fortron 530W performs like this. The 12V too high (12.42 V), and % of tolerence is more than 5% of spec, or right at the margin if 5% measurement uncertainty is taken into account.

My system does not have excessive hardware (many may even think a Fortron 350W would do it): 1 mb, 1 CPU, 1 TT SFII, 4 80mm fan, 1 DVD, 1 HD, 1 video card.

The Antec TP430W that I have been using has smaller tolerance on the various lines and within spec, for powering exactly the same system.

Due to the voltage fluctuation of the Fortron 530, I got BSOD even running at lower CPU clock, compared to that when using the Antec TP430.

I don't know whether it is a bad PSU or others have looked at the Fortron 530W carefully. I hope I have not done anything wrong with the setup ???

I am going to return this PSU and ask for full refund. It seems there is no other choice except an Antec TP/TC 550.

[hitechjb1]

Tbred B 1700+ DLT3C 2.54 GHz voltage idle (Fortron 530W PSU testing)

http://hitechjb1.dynu.com:8081/images/TbredB_OC/dlt3c_1700_2545_fsp530_idle_c.JPG

http://hitechjb1.dynu.com:8081/images/TbredB_OC/dlt3c_1700_2545_fsp530_vgraph_idle_c.JPG

[hitechjb1]

Tbred B 1700+ DLT3C 2.54 GHz voltage under Prime95 (Fortron 530W PSU testing)

http://hitechjb1.dynu.com:8081/images/TbredB_OC/dlt3c_1700_2545_fsp530_prime1_c.JPG

http://hitechjb1.dynu.com:8081/images/TbredB_OC/dlt3c_1700_2545_fsp530_prime2_c.JPG

http://hitechjb1.dynu.com:8081/images/TbredB_OC/dlt3c_1700_2545_fsp530_prime3_c.JPG

http://hitechjb1.dynu.com:8081/images/TbredB_OC/dlt3c_1700_2545_fsp530_vgraph_prime_c.JPG

[ayoung]

Due to the Fortron 510W limit, and the fan noise and location, I am leaning towards the TP480/550 for upgrade if needed


Fan noise? What are yo talking about, the F530 is the quietest fan(s) in my whole system.

[Oklahoma Wolf]

FWIW the only place I can find the CWT 550 in Canada is at www.computerboulevard.ca for $139... and the only place in Canada I know of for the FSP 530 is the Aopen model at www.ntcw.com for $159. I doubt the Antec version of the 550 is any cheaper, but for the $20 difference I'd still grab the FSP 530 since they have the better build quality anyway.

Edit: I take that back, the Antec Truecontrol at www.ncix.com is $166, and $152 for the normal version. I'd personally still get the FSP though ;)

[hitechjb1]

But I just tested a FSP530W, it does not work according to spec. at all if you see my previous post and the pics:

The Antec TP430 I have been using gives less line fluctuations on both 5V and Vcore, for powering the same system.


Here is some measurement of the voltage lines at idle and running prime95, @1.92V, 2544 MHz.

1. Voltage lines at idle

The 5V line dropped to 4.703 V, and flutuated between 4.703 - 4.977 V.

The 12V line was at 12.42 V, this was too high and might be too much for hard drives, CDRW, DVD, ...

For the Antec TP430, 12V line is around 11.93 V, +- 0.04 V.

2. Voltage lines with Prime95

I saw the 5V fluctuated to a high of 4.977 V from 4.703 V, which is 0.274V. This fluctuation is more than the 5% line tolerence of the PSU. If taking potential measuring uncertainty of 5% of the ASUS probe, that fluctuation would still be 0.274/5 * 0.95 = 5.2% (almost in line with the spec tolerance).

Also I saw the Vcore bounds around between 1.85 to 2.0 V (in the snapshot it bounds between 1.888 V and 2.0 V), also this is outside of the AMD Vcore tolerence of +- 50 mV.

[Oklahoma Wolf]

Knew I was missing something... you tested the voltage lines with a multimeter I trust? I don't put a lot of faith in BIOS/MBM type readings when it comes to verifying output from a PSU. The fluctuations could be caused entirely by the board itself, or magnifying a somewhat minor fluctuation from the PSU in some way.

At any rate, if you're more comfortable with the Antec units you could always sell the FSP... if your board fluctuates less with the Antec, then it's worth keeping. Both PSU's are more than decent units :)

[hitechjb1]

I saw similar fluctuation with a voltmeter. I did not measure Vcore using voltmeter, avoiding accidental shorting.

Every thing being equal, the fluctuations reported by the ASUS probe on 5V and Vcore are much smaller when using an Antec TP430 than a FSP 530W.

The Vcore bounce when using Antec TP430 is within the AMD CPU DC tolerance of +- 50mV, even running at 2.59 GHz.

Whereas it is outside the +- 50 mV AMD CPU DC tolerance using the FSP 530W at 2.54 GHz.

All these points to the Antec TP430 has better line regulation when the CPU is drawing heavy active current and power at high frequency. So I suppose the Antec TP/TC 550 would be even better. From spec, Antec is 3% and Fortron is 5%.

I don't know whether it is a bad PSU I got (out of spec on both line voltage and line tolerance under heavy load) or someone has actually looked at the FSP 530W closely with even detailed measurement and numbers than what I did, .... than just saying it is good, ....

As shown in previous post, AntecRep pointed to the links with detail line measurement on overdrawing the TP 430, 480, and 530. And I had asked for similar information for FSP 530, it seems no one has come across something like that for Fortron PSU.

[Oklahoma Wolf]

I have no idea how that 530 performs myself, since I don't have one and haven't seen a detailed review like that of the Antecs, but the evidence does seem to show better regulation from the Antec... the only other thing I can think of is maybe some of the individual pins on the ATX connector of the FSP are a bit loose in comparison to the Antec. The results of the FSP are a surprise to me as well.

[dustybyrd]

i have the 530 and the 350 watt fortrons and dual p3's clocked at 1.62 ghz...with 1 gig of RAM, 3 hard drives, cdrw, zip, sound, tvcard, 6 case fans, floppy, 8500le o/c to 275/265....

and the 12 volt never budges always 12.1

the 5 volt only goes from 4.91 to 4.88 under full load with 2 prime 95 instances and CPU burn running all at once...

the 3.3 goes from 3.29-3.3 all the time regardless of load....


i'd say that your powersupply is bad...

[hitechjb1]

Originally posted by Oklahoma Wolf
Knew I was missing something... you tested the voltage lines with a multimeter I trust? I don't put a lot of faith in BIOS/MBM type readings when it comes to verifying output from a PSU. The fluctuations could be caused entirely by the board itself, or magnifying a somewhat minor fluctuation from the PSU in some way.

At any rate, if you're more comfortable with the Antec units you could always sell the FSP... if your board fluctuates less with the Antec, then it's worth keeping. Both PSU's are more than decent units :)

Originally, I was planning to get a TP/TC 550, since I went through its spec, some detailed test results, 8A higher on the 12V line, 3% line regulation and fan location, and found it to be better in my opinion. But somehow I tried to save $40-50, and also many are recommending the FSP 350/530, so I went to buy a FSP 530W. Apparently, it did not work out for me and I end up wasting time and money (rma and two shippings).

I am not going to keep it, or exchange, risking wasting more time and money. I am going to return it and get an Antec TP, which I found much better (I have a TP 430), more stable line and regulation for overclocking at high Vcore and high active current.

[Lithan]

Fortron 530, According to mbm5 on my nf7-s

+3.33v line = 3.23v-3.33v with an average of 3.29v
+5v line = 5.00v-5.05v with an average of 5.03v
+12v line = 11.86v-11.98v with an average of 11.93v

[hitechjb1]

Originally posted by Lithan
Fortron 530, According to mbm5 on my nf7-s

+3.33v line = 3.23v-3.33v with an average of 3.29v
+5v line = 5.00v-5.05v with an average of 5.03v
+12v line = 11.86v-11.98v with an average of 11.93v

At what Vcore and CPU clock are these measured?

How is the Vcore fluctuation?

Are these measurement done during running Prime95? If not, do you have these fluctuations on the various lines and Vcore when running Prime95?

[Gautam]

For some strange reason, I've heard of others having issues with the 530w and 550w Fortron units. They actually seem to perform significantly worse than their 400w and 350w counterparts. I don't think yours is an isolated example, and I don't know why. My voltage lines on my 300 watter are rock solid, and thats at similar speeds to which you're running. In fact, they're better than Lithan's results, too. My +3.3 fluctuates from 3.34v-3.36v, never any higher or lower. My +5v never seems to budge from 5.05v, +12v ranges from 11.95 to 12.01, averaging around 11.98v. And this is with a 300w unit. Something seems wrong with their higher rated PSU's.

[hitechjb1]

Originally posted by Gautam
For some strange reason, I've heard of others having issues with the 530w and 550w Fortron units. They actually seem to perform significantly worse than their 400w and 350w counterparts. I don't think yours is an isolated example, and I don't know why. My voltage lines on my 300 watter are rock solid, and thats at similar speeds to which you're running. In fact, they're better than Lithan's results, too. My +3.3 fluctuates from 3.34v-3.36v, never any higher or lower. My +5v never seems to budge from 5.05v, +12v ranges from 11.95 to 12.01, averaging around 11.98v. And this is with a 300w unit. Something seems wrong with their higher rated PSU's.

Are your numbers measured during Prime95? It makes a big difference between lightly load and heavy load.

Will see what Lithan has to say about Vcore and during Prime95, since he has a FSP 530W.

[Gautam]

Yeah, this is with prime95 running. Without it voltages don't change at all. VCore fluctuates from 1.93 to 1.95.

[hitechjb1]

Another bad thing about my FSP 530W is that:

My mb uses 5V to derive Vcore, when prime95 is run, the 12V fluctuates. I don't see this happens with the Antec TP430 which hold the 12V line solid.

It looks like the Fortron 530W 5V and 12V lines are not completely decoupled, this is really bordering, since I heard ppl are saying/implying the Fortron 3.3, 5, 12V are independent. Antec TP PSU claims to have independent circuits for the three lines, which seems to be the case from what I have seen. But I cannot find any info about the Fortron at all about the independence of the 3.3, 5, and 12V lines, except ppl talking about it.

I like to see the detailed spec from Fortron, but up to now, I cannot find any except sketchy few lines on their web site.

[hitechjb1]

Originally posted by Gautam
Yeah, this is with prime95 running. Without it voltages don't change at all. VCore fluctuates from 1.93 to 1.95.

That is good Vcore regulation.

Is it from a Fortron 300W. If so, what exact model is your Fortron 300W?


PS: Your mb nf7-s derives Vcore from the 12 V line which uually gives better Vcore regulation.

[dustybyrd]

Another bad thing about my FSP 530W is that:

My mb uses 5V to derive Vcore, when prime95 is run, the 12V fluctuates. I don't see this happens with the Antec TP430 which hold the 12V line solid.

It looks like the Fortron 530W 5V and 12V lines are not completely decoupled, this is really bordering, since I heard ppl are saying/implying the Fortron 3.3, 5, 12V are independent. Antec TP PSU claims to have independent circuits for the three lines, which seems to be the case from what I have seen. But I cannot find any info about the Fortron at all about the independence of the 3.3, 5, and 12V lines, except ppl talking about it.

I like to see the detailed spec from Fortron, but up to now, I cannot find any except sketchy few lines on their web site.


i think you just got a bad 530 watt fortron...and i would RMA it...


my system easily requires at least as much power as your's (dual p3's at 1.62 ghz require about 100 watts total on the 5v line)...

and my rails hardly change at all even when running passmark burnin test (that test all components at the same time...and i mean everything)...2 instances of prime95 and cpu burn...all at the same time...

with all those programs running at once the software read voltages are:

my 3.3 volt was solid 3.29-3.3 (same as no load)....
the 5v was 4.86-4.88 (it's 4.91 with no load)...
and the 12v was 12.1 (same as no load)

[larva]

Originally posted by hitechjb1


That is good Vcore regulation.

Is it from a Fortron 300W. If so, what exact model is your Fortron 300W?


PS: Your mb nf7-s derives Vcore from the 12 V line which uually gives better Vcore regulation.

Which kind of shows the 12V output on the Fortrons is great, all spec sheet bs aside.

[larva]

Originally posted by hitechjb1


It looks like the Fortron 530W 5V and 12V lines are not completely decoupled, this is really bordering, since I heard ppl are saying/implying the Fortron 3.3, 5, 12V are independent. Antec TP PSU claims to have independent circuits for the three lines, which seems to be the case from what I have seen. But I cannot find any info about the Fortron at all about the independence of the 3.3, 5, and 12V lines, except ppl talking about it.



http://forum.oc-forums.com/vb/showthread.php?s=&action=showpost&postid=1659735#1659735

[larva]

Originally posted by hitechjb1

Some even say the Fortron 350W is enough to do most job and can deliver 454W (I do not agree on this number).



Well, now you know why that outlandish bit of advice comes from "some" of us.

Originally posted by Gautam

For some strange reason, I've heard of others having issues with the 530w and 550w Fortron units. They actually seem to perform significantly worse than their 400w and 350w counterparts. I don't think yours is an isolated example, and I don't know why. My voltage lines on my 300 watter are rock solid, and thats at similar speeds to which you're running. In fact, they're better than Lithan's results, too. My +3.3 fluctuates from 3.34v-3.36v, never any higher or lower. My +5v never seems to budge from 5.05v, +12v ranges from 11.95 to 12.01, averaging around 11.98v. And this is with a 300w unit. Something seems wrong with their higher rated PSU's.

[dustybyrd]

Well, now you know why that outlandish bit of advice comes from "some" of us.

[/B]


larva, i don't follow what you mean here

[larva]

I was pointing out why I recommend the 350W model as being adequate. The 300W model is adequate, but the 350 carries such a small price premium over it I don't recommend the 300 unless you already have it.

[dustybyrd]

I was pointing out why I recommend the 350W model as being adequate. The 300W model is adequate, but the 350 carries such a small price premium over it I don't recommend the 300 unless you already have it.


i see...but he thinks the 350 watt can't hit 454...


anyway...for anyone who is interested i am replacing my p3's and board for two AMD 2100xp's and i will overclock them to 2+Ghz...

they will require more power than a 2.5+ghz amd or a 3.5+ghz p4....probably 150 watts just for the cpu's...

i will use the 530 watt supply...so we'll see how good it is...

[hitechjb1]

Originally posted by larva
I was pointing out why I recommend the 350W model as being adequate. The 300W model is adequate, but the 350 carries such a small price premium over it I don't recommend the 300 unless you already have it.

I have an old FSP 300 60GT and it was able to power the 1700+ up to around 2.4 GHz @ 1.8 V, beyond that the system became unstable and the 5V line fluctuated beyond 5%, and the PSU was getting very warm.

So I got a Antec TP430, and I can get the oc further to 2.59 GHz and 1.9+V with system stable, and the various line flutuation within 3-5%.

But then the FSP 530W I got performs much much worst than the TP430 as described earlier.

I know you have been recommending Fortron 350W and sometime also 530W. But since the Fortron 530W I got does not do well at all, and also this comment:

Originally posted by Gautam
For some strange reason, I've heard of others having issues with the 530w and 550w Fortron units. They actually seem to perform significantly worse than their 400w and 350w counterparts. I don't think yours is an isolated example, and I don't know why. My voltage lines on my 300 watter are rock solid, and thats at similar speeds to which you're running. In fact, they're better than Lithan's results, too. My +3.3 fluctuates from 3.34v-3.36v, never any higher or lower. My +5v never seems to budge from 5.05v, +12v ranges from 11.95 to 12.01, averaging around 11.98v. And this is with a 300w unit. Something seems wrong with their higher rated PSU's.

At this point, I'd like to hear what other's opinion on the Fortron 530W, in terms of line stability during heavy load, ....

Between the FSP 530 and the TP 550 which are what I have been trying to compare, there are many people are recommending the TP550. At this point, I am going to return the FSP 530 and get a TP/TC 550 and get over with it.

[larva]

Originally posted by hitechjb1

I have an old FSP 300 60GT and it was able to power the 1700+ up to around 2.4 GHz @ 1.8 V, beyond that the system became unstable and the 5V line fluctuated beyond 5%, and the PSU was getting very warm.


The GT series is a cheaper unit with less 3.3+5V output (175W). The other 300W FSPs are rated at 200W 3.3+5V and are more capable.

Originally posted by hitechjb1

I know you have been recommending Fortron 350W and sometime also 530W. But since the Fortron 530W I got does not do well at all


I find it very hard to believe the 530W model is somehow less capable than the 350 (or 300, for that matter). Dusty's 530 performs better for him than does his 350, and I doubt this is an isolated incident. I'm sorry you aren't happy with your supply but FSP does indeed know what they are doing and I strongly doubt there is any negative issue inherent in the design or performance of the 530.

[star882]

I'd go with:
http://www.zmoz.com/pics/AntecTruepower.JPG
My friend Christina Mahoney has a dual processor(2x 2.2GHz Britney IIRC)box powered by an Antec Truepower 550.
She says the machine draws 480w, so it's starting to get close to the limits.
She also has lots of HDs(including one seagate cheetah), many Delta fans(for the CPUs and PSU(she modded the PSU with Deltas)), and a Radeon 9700 Pro AIW.

[hitechjb1]

Originally posted by Gautam
For some strange reason, I've heard of others having issues with the 530w and 550w Fortron units. They actually seem to perform significantly worse than their 400w and 350w counterparts. I don't think yours is an isolated example, and I don't know why. My voltage lines on my 300 watter are rock solid, and thats at similar speeds to which you're running. In fact, they're better than Lithan's results, too. My +3.3 fluctuates from 3.34v-3.36v, never any higher or lower. My +5v never seems to budge from 5.05v, +12v ranges from 11.95 to 12.01, averaging around 11.98v. And this is with a 300w unit. Something seems wrong with their higher rated PSU's.

Originally posted by Gautam
Yeah, this is with prime95 running. Without it voltages don't change at all. VCore fluctuates from 1.93 to 1.95.

Originally posted by hitechjb1

That is good Vcore regulation.

Is it from a Fortron 300W. If so, what exact model is your Fortron 300W?

PS: Your mb nf7-s derives Vcore from the 12 V line which uually gives better Vcore regulation.

1. If we trace the posts, Gautam was saying others are having issues with the Fortron 530W (in responding to my line regulation issue with the Fortron 530W), he was referring to his Fortron 300W PSU being able to power the system with Vcore fluctuation from 1.93 to 1.95V, with 12V lines between 11.95 and 12.01V, ...


Originally posted by larva

Which kind of shows the 12V output on the Fortrons is great, all spec sheet bs aside.

Since we have been discussing Fortron 530W. But then coming to the conclusion from a Fortron 300W to "the 12V output on the Fortrons is great", without saying which models (300W, 350W, 530W, ...) becomes not clear to me which Fortrons (plural) were referred to. Whether it was the 300W that he tested or it is being extended to other Fortrons (other models), ... Since we have been discussing the Fortron 530W that I had been having line fluctuation problem with.


2. Also the words like "all spec sheet bs aside" is not very useful for us. Due to your high reputation in this forum, many new comers may misinterpret that we can ignore detailed specifications, .... I guess probably you referred to those label on some PSU's that might have border line ratings.

But if we oversimplify the situation that we have to treat those specification or commercial spec with care, and extend them to ignore reading specifications such as % of line regulation, temperature sensitivity, ripple, ... for PSU in particular, Vcore, temperature, current, power, ... for AMD, Intel CPU, timing, Vdimm, ... for memory modules, etc, etc, and the secifications for many other components in PC, or that for enginerring fields in general, I think that is oversimplication and wrong. Without or ignoring specification, large projects that involves thousands of people, or multi-industry cannot be carried out effectively. Without specification, we would not have that much fun in putting different parts together in building our PC, we would not have space shuffle flying in space and robots landing on mars, ...

I know most of us and you know about all these, but just to point out that using these phrase to describe "specification" may not be helpful, especially for new comers for doing technical works (overclocking, building PC, ...). I read and rely on detailed specification all the time, I based on both specification, actual experiment and measurement results, and people recommendations, ...

[dustybyrd]

I based on both specification, actual experiment and measurement results, and people recommendations, ... [/B]


but, you seem to have ignored my posting on my real world experiences with the 530 and 350watt....

i think it's likely that either your 530 watt was a bad apple or your tests are flawed...because of software voltage reading problems..

also...i noticed in your posts with screen shots of the 530watt on your 2.5+ghz AMD that during prime95 the 5v went from 4.7something to 4.9something....that doesn't make sense does it?

if the power supply was weak wouldn't the 5v go DOWN under a prime 95 load?

[hitechjb1]

I know about your results on the 530W, that is why I have asked others to post some actual experience with the Fortron 530W under heavy load.

I am still trying to figure out why the 5V line fluctuated that much between a TP430 and a FSP530. I indeed see the Fortron 5V line went from 4.703 V to 4.977 V after Prime97 started.


PS:
I looked at the 5V fluctuation of the Fortron 530W many times to make sure there is no handling errors or flaws as suggested, with different settings of CPU clock (2.5+ GHz) and Vcore (1.9+ V). It is bouncing between 4.703 and 4.977 V, during both idle and during heavy load (Lithan also found line fluctuation is independent of load), sampled over a period of time. That is a fluctuation of 274 mV (5.5% of the 5 V line). Taking into account the possibility of voltage sampling error, assuming as high as +-5%, the line regulation is slightly above the 5% line regulation spec of the Fortron 530W. Had the sampling error be +-10%, the line regulation would be at the margin of the 5% line regulation specification.

[dustybyrd]

I know about your results on the 530W, that is why I have asked others to post some actual experience with the Fortron 530W under heavy load.


ok...i am curious about more real world experiences too



I am still trying to figure out why the 5V line fluctuated that much between a TP430 and a FSP530. I indeed see the Fortron 5V line went from 4.703 V to 4.977 V after Prime97 started.


i agree that that is weird....don't you think it's possible that it is a software reading problem...it just doesn't make any sense that the 5v would go up under extreme load...

therefore i would say that it is more likely that the software readings are wrong...

[hitechjb1]

I am aware of the accuracy issues of voltage reported by software probing.

Pros:
1. They are readily avaiable for sampling the voltage lines and stored and taking pics.
2. They may be even faster than voltmeter in response time and sampling time.

Cons:
1. Most believe that it is inaccurate (I don't know whether this is true or false). It does not matter in my experiment (see below).

BUT, here I am not talking about measuring ABSOLUTE voltage accuracy. I am using the same software setup and probing to measure and compare two different PSU situations with the remaining system being the same.

For my Antec TP430: I got 5V line fluctuation between 4.73 - 4.892 V, which is 162 mV fluctuation.

For my Fortron 530W: I got line fluctuation between 4.703 - 4.977 V, which i s 274 mV fluctuation.

Due to the higher 5V line fluctuation which results in higher Vcore fluctuation 150 mV, the same sytem is less stable than that setup with the TP430 (around 30 MHz lower in oc frequency using SLK-800U and TT SFII).


Regarding to absolute/relative measurement analogy:

If you really want to measure the distance between two points, any rulers found in the house, school probably is not reliable to measure to certain accuracy. We may have to resort to use atomic measurement for "absolute" accuracy to atomic scale.

But we can use a house ruler to compare the length of two objects, say if one measured 31.0 mm and another one measured 31.8 mm, we would know which is longer.

So I think using the same software probing to report voltages and temperatures for two similar setups with one changing element (iin this case being the PSU) is a fair experiment.

[Oklahoma Wolf]

It's not inconceivable to me the Fortrons might do better with fluctuations under extreme loads where all rails are being hit hard, but it's hard to know where its limits are I guess with so few people owning one... don't know about you guys, but I'd like to see one really pushed like in the Tom's review or using the methodology at SilentPC Review and see how it does with regulation when it's having to run full power all the time ;)

[dustybyrd]

BUT, here I am not talking about measuring ABSOLUTE voltage accuracy. I am using the same software setup and probing to measure and compare two different PSU situations with the remaining system being the same.

For my Antec TP430: I got 5V line fluctuation between 4.73 - 4.892 V, which is 160 mV fluctuation.

For my Fortron 530W: I got line fluctuation between 4.703 - 4.977 V, which i s 274 mV fluctuation.

Due to the higher 5V line fluctuation which results in higher Vcore fluctuation 150 mV, the same sytem is less stable than that setup with the TP430 (around 30 MHz lower in oc frequency using SLK-800U and TT SFII).



i would agree with you...but i believe it was oklahoma wolf who said something i hadn't thought of that might cause the voltage monitoring software to still be off (even when just changing the Power supply and nothing else in the system)...

and that possibility is that the software magnifies changes in the voltages that might be much smaller if actually measured with a voltmeter....in other words maybe there is a change under different load conditions (however backwards your changes seem to be), but the changes are actually .05 volts instead of .25 volts and the software is magnifying the voltage changes...

that would also imply that the antec 430 has tighter rails...but they might only be 0.000001 tighter because once there is a slight real voltage fluctuation the software magnifies it 10x or by some other large factor....

[Since87]

FWIW Prime95 does not cause a constant load on the power supply.

If you go about halfway down the page of this ProCooling thread (http://forums.procooling.com/vbb/showthread.php?s=&threadid=6442&perpage=25&pagenumber=3), you will see a graph of CPU temperature vs time taken while Prime95 is running.

Temperature varies substantially, which implies the CPU's power dissipation varies substantially.

[dustybyrd]

FWIW Prime95 does not cause a constant load on the power supply.

If you go about halfway down the page of this ProCooling thread (http://forums.procooling.com/vbb/showthread.php?s=&threadid=6442&perpage=25&pagenumber=3), you will see a graph of CPU temperature vs time taken while Prime95 is running.

Temperature varies substantially, which implies the CPU's power dissipation varies substantially.


that might be true (not sure though as when i have run prime95 the cpu usage is 100% any time i have looked), but that still doesn't explain why the 5volt rail would go from 4.7something at IDLE to 4.9something at ANYTIME during prime95....

[Since87]

Originally posted by dustybyrd

that would also imply that the antec 430 has tighter rails...but they might only be 0.000001 tighter because once there is a slight real voltage fluctuation the software magnifies it 10x or by some other large factor....

I think you are reaching here. It's easy enough to test though. Through the BIOS set Vcore a bit higher and see how much the Motherboard reading changes using both supplies. If it is the same amount of change with both PSUs, then it can't be what you suggest.

[Since87]

Originally posted by dustybyrd



that might be true (not sure though as when i have run prime95 the cpu usage is 100% any time i have looked), but that still doesn't explain why the 5volt rail would go from 4.7something at IDLE to 4.9something at ANYTIME during prime95....

Ah, I didn't realize the rail was 4.7 at idle.

[dustybyrd]

I think you are reaching here. It's easy enough to test though. Through the BIOS set Vcore a bit higher and see how much the Motherboard reading changes using both supplies. If it is the same amount of change with both PSUs, then it can't be what you suggest.


i think he has already done this...and the antec didn't change and the fortron did...therefore you can't tell from that test if the readings from the power supply are actually changed in both circumstances....

[Lithan]

Originally posted by hitechjb1


At what Vcore and CPU clock are these measured?

How is the Vcore fluctuation?

Are these measurement done during running Prime95? If not, do you have these fluctuations on the various lines and Vcore when running Prime95?




1. vcore was at 1.80v cpu at 230x10=2305

2. Vcore varied from 1.725-1.775 but that is the motherboard. NF7's have low/fluctuating vcores.

3. It was high/low readings after 2 days running... about 8hrs of prime, a couple hours of wc3, alot of surfing and some idling.



In my experience load has NOTHING to do with variances. 8 hrs of prime gets the exact same high/low as 8 hours in standby. This PSU is beefy enough that 'I' sure can't load it. But it's regulation isn't the best. I'd say it's probably on par or slightly worst than antecs regulation... but in the end I'd wager it is a stronger brute strength PSU, just off of Fortrons reputation and the weight difference between the 350 and the 530, I doubt 2.5 extra pounds is there for no reason.

[hitechjb1]

In order to measure DC fluctuation of a voltage line, we have to use instruments with high frrequency repsonse such as an oscilloscope or high speed sampling devices that can sample the voltage of a line over a period of time. Otherwise if the voltage of a DC line fluctuates faster enough the DC voltmeter will not be able to SEE it.

E.g. if the CPU Vcore changes 50 mV within 1 ms (this 1 ms = 2000000+ clock cycles, is just a guessing number at this point, since I have to look into the decoupling RC of the Vcore line etc), due to current surge, the DC voltmeter would not be able to record it and SEE it. It requires some high speed sampling device to monitor the line over a period of time to measure it.

Those programs such as MBM or ASUS proble may be able to do this. I have to go deeper into finding what are their response time, .... Or the best thing is to use is a high speed sampling oscilloscope to capture the Vcore or 5V or 12V over a period of time.

I think response time is more important than absolute accuracy in comparing two setups. Even the probing program is off by say -5%. That is reading 5V as 4.75V. But we are measuring the differential result of two setup, the absolute error should cancel each other. Just like the example I used earlier, an inaccurate ruler can tell which object, between two objects of 31.0 mm and 31.8 mm, is longer by measuring their length in relative terms.

So if you take that 5% (or even 10%) error into account for the probing software and multiply the voltage fluctuation in the two cases, the result of which has less fluctuation would still be the same, except until the error of the probing program is as high as 50% (which is not the case), then both measurement would become meaningless.


It is correct that in the FSP 530W case that at light load, the 5V line is at 4.703 V lower than during heavy load. I am still trying to figure out why.

I have some theories under mind, one is due to the non-independent circuit of the Fortron when under heavy load, more current or lower impedance is sourced to the line(s) that are more heavier drawn. Others are that something related to regulation response time, inductance and dI/dt on the line. These are just premature thinkings, so maybe someone can throw some more ideas onto them.

[dustybyrd]

But we are measuring the differential result of two setup, the absolute error should cancel each other. Just like the example I used earlier, an inaccurate ruler can tell which object, between two objects of 31 mm and 38 mm, is longer by measuring their length in relative terms.



yes, but only assuming, the factor that the software is off is a LINEAR factor...suppose that if one power supply is wavering between 5.01 and 4.99 and the other is wavering between 5.05 and 4.95....but the software program voltage reports are close to accurate for the "barely discernible" difference of 0.02, but the software program reports very large differences if there is a 0.1 fluctuation...

the difference between what the software program reports and the actual voltages don't have to be linear over all voltage differences...

[hitechjb1]

It is possible that non-linear error property of the two measurment can skew the results.

It only happens when the two measurements have disjoint intervals or non-completely overlapping intervals.

But in my measuring case:

For my Antec TP430: I got 5V line fluctuation between 4.73 - 4.892 V, which is 162 mV fluctuation.

For my Fortron 530W: I got line fluctuation between 4.703 - 4.977 V, which is 274 mV fluctuation.

The Antec 430W fluctuation interval is completely within the Fortron 530W fluctuation interval, so regardless of non-linear error concern, it is a non-issue, I think.


I looked at the 5V fluctuation of the Fortron 530W many times to make sure there is no handling errors or flaws as suggested, with different settings of CPU clock (2.5+ GHz) and Vcore (1.9+ V). It is bouncing between 4.703 and 4.977 V, during both idle and during heavy load (Lithan also found line fluctuation is independent of load), sampled over a period of time. That is a fluctuation of 274 mV (5.5% of the 5 V line). Taking into account the possibility of voltage sampling error, assuming as high as +-5%, the line regulation is slightly above the 5% line regulation spec of the Fortron 530W. Had the sampling error be +-10%, the line regulation would be at the margin of the 5% line regulation specification.

[dustybyrd]

For my Antec TP430: I got 5V line fluctuation between 4.73 - 4.892 V, which is 162 mV fluctuation.

For my Fortron 530W: I got line fluctuation between 4.703 - 4.977 V, which is 274 mV fluctuation.

The Antec 430W fluctuation interval is completely within the Fortron 530W fluctuation interval, so regardless of non-linear error concern, it is a non-issue, I think.


ok, so let's say you are right on that issue...

then, i forgot have you done measurements of the 5v on both supplies with a voltmeter?

if the software voltage program is right then the voltmeter directly on the fortron should read 4.703 on idle, right? can a voltmeter be used to test voltages under load (i don't know)?

if the fortron's voltage on the 5v rail at idle is between 4.95 and 5.05 (with the voltmeter) then i think the whole thing is a nonissue since the software reading program would clearly be WAY off...

[hitechjb1]

As said in earlier post, I think in order to measure line fluctuation, we have to use instruments such as a sampling oscilloscope, or approximation using software monitors sampling over a long period, taking into account of the accuracy of those programs. Or the monitoring program should be able to measure the relative voltage fluctuation difference of two similar PSU setups.

I think DC voltmeter cannot see these fluctuations due to response time. For DC line regulation, there are some measured result as shown in these links for Antec TP. I hope we can find something similar for Fortron 530. The spec for Antec TP and Fortron 530W are respectively 3% and 5%.

Originally posted by hitechjb1

...
Power Supply Specifications, Testing, and Antec TruePower (http://www.antec-inc.com/info_DIYArticle1.html#)

Test Results for True480 (http://www.antec-inc.com/info_DIYArticle_true480spec.html)

Test Results for True550 (http://www.antec-inc.com/info_DIYArticle_true550spec.html)
...


I don't want to get into endless debate of which PSU to get, which is not the original intent of this thread which is for using actual measured data between these PSU's. It is a personal choice to get which PSU, like which PC3500 CAS2 memory module to get is a personal choice, the memory specification already speaks for itself.

I hope others can post some more actual results and experience on either PSU, so it may help us to make future decisons.

[dustybyrd]

I don't want to get into endless debate of which PSU to get,

i agree...


It is a personal choice to get which PSU, like which PC3500 CAS2 memory module to get is a personal choice, the memory specification already speaks for itself.


i agree it's a personal choice...but i disagree on the fact that the specification speaks for itself....as can be seen by toms hardware power supply test found here http://www.tomshardware.com/howto/20021021/index.html

that you CAN NOT BELIEVE the specs...


I hope others can post some more actual results and experience on either PSU, so it may help us to make future decisons.


which brings me to a final thought that i agree with you that actual results and experience with a PSU are the most important factor...

all the theoretical specs and software reported voltages don't mean anything if they are incorrect...

so real world experience with a PSU in tough conditions for extended periods of time is the real test of whether a PSU is good or not...


i'm not knocking antec at all....1st i've never used one...2nd many users swear by them with very few negatives about them except price (great evidence that they are great ps's)....and 3rd i picked a 480watt true power up at bestbuy and they weigh a ton (which in itself doesn't mean anything...but combined with other factors is assuring)

[hitechjb1]

Originally posted by Since87

Ah, I didn't realize the rail was 4.7 at idle.

Since87,

It was also fluctuating between 4.703 and 4.977 V during idle/lightly load/heavy load.

I looked at the 5V fluctuation of the Fortron 530W many times to make sure there is no handling errors or flaws as suggested, with different settings of CPU clock (2.5+ GHz) and Vcore (1.9+ V). It is bouncing between 4.703 and 4.977 V, during both idle and during heavy load (Lithan also found line fluctuation is independent of load), sampled over a period of time. That is a fluctuation of 274 mV (5.5% of the 5 V line). Taking into account the possibility of voltage sampling error, assuming as high as +-5%, the line regulation is slightly above the 5% line regulation spec of the Fortron 530W. Had the sampling error be +-10%, the line regulation would be at the margin of the 5% line regulation specification.

1. There were concerns about the software probing error. But I account that even with 5-10-20% error, the measurement should still track between the two setup with one changing elment being the PSU.

2. There were concerns about non-linear calibration characteristics of the software probing to skew the line regulation results. But I noted that the fluctuation interval of the Antec 430 measurement was completely within (complete interval inclusion) the Fortron 430W fluctuation interval, so the result was not skewed, and Antec 430 had a much smaller line fluctuation than the counterpart.

3. I think I have taken good measure to do the measurement and I found that, even with 8% software probing error (that is as much as 5V being read as 4.6V, very conservative error allowance):

For my Antec TP430 measurement: I got 5V line fluctuation between 4.73 - 4.892 V, which is 162 mV fluctuation +-13 mV. This is in line with the Antec 3% line regulation spec on the 5V (not defeative).

For my Fortron 530W measurement: I got line fluctuation between 4.703 - 4.977 V, which is 274 mV fluctuation +-22 mV. This is in line with the Fortron 5% line regulation spec on the 5V (not defeative).

Since87: Would appreciate you take a look and any comment are welcome.


BTW, that link about the tempearture and Vcore fluctuation of a CPU under load is interesting. I have some question about the time constant of temp vs Vcore, and its impact on stability.

So it looks like for programs that require smaller cache, the CPU temp can fluctuate close to 3C, most of the time are CPU intensive since there is less cache miss.

For programs that require larger cache, the CPU temp can fluctuate as much as 5C, there is marked segments of idle and temp drops further lower during cache misses.

Do you know of any number/estimate of the time constant of CPU temp over Vcore?

[Since87]

Originally posted by hitechjb1

Do you know of any number/estimate of the time constant of CPU temp over Vcore?

In this procooling article (http://www.procooling.com/articles/html/amd_thermal_diode_testing_cali.php) there is a graph showing the temperature of the die as the power is cut. The temp of the die drops very quickly and then trails off. (Tracking the temperature drop of the larger thermal mass of the heatsink.) The time scale of the graph is too coarse to determine an accurate time constant.

As far as your PSU issues; I think you would really need to get a scope on it to know what is happening.

One think I suspect, is that there may be an interaction between the switching of the PSU and the switching of the Vcore regulator. I've run into this before when developing a switching supply similar to a Vcore regulator. The ripple in the current draw of the Vcore regulator will cause some fluctuation in its input rail. The PSU (which is trying to regulate that input rail) may overshoot and undershoot the current it supplies on alternate switching cycles depending on the relative switching frequencies of the the two supplies.

In other words, I would guess that your PSU can handle a high power output just fine if the load was a simple resistor, but it is unstable when powering the ripple current load of your particular mobo's Vcore regulator.

My solution to this, was to decouple the input supply's sense lines, so that the input supply did not respond to the high frequency load fluctuations caused by the lower voltage supply. For this to work acceptably you would need to increase the capacitance at the input to the Vcore regulator. (The extra capacitance makes up for the slowed response of the PSU.)

You could do this to your system if you wanted to, but I'm doing a lot of speculating here, and what I suspect may be totally irrelevant. You are right. Getting a scope on it is the only way to know what's going on.

One thing that might be worth a try is upping the capacitance at the input side of the Vcore regulator. Using something like the 680uF 6.3V capacitors shown on this page of the Digi-Key catalog (http://dkc3.digikey.com/PDF/T032/0681.pdf) might help. Adding one of these, per phase of the Vcore supply, to the input side of the Vcore regulator might decrease any interaction between the PSU and the Vcore regulator enough to allow the system to operate stably.

Investigating this is a lot of trouble to go to. Unless you really want to get the system working with the Fortron supply, trading the Fortron in for an Antec TP may be the best way to go.

[star882]

"One thing that might be worth a try is upping the capacitance at the input side of the Vcore regulator. Using something like the 680uF 6.3V capacitors shown on this page of the Digi-Key catalog might help."
The input to the Vcore regulator is 12v on all Pentium 4 and Britney systems, you need at least 16v caps.

[hitechjb1]

I think Since87 referred to the 5V line from which my mb Vcore regulator derives Vcore, hence 6.3V capacitor is sufficient.

Since87, thanks for the links, you are good source for info.

CPU Power Model

The CPU is a non-linear device, and to the first order, I would model it as an inductor in series with a resistor and a capacitor in parallel. The inductor accounts for the inductance in the package (VL = L dI/dt), and the resistor accounts for leakage and biasing current in the CPU (V^2/R heat up the CPU), and the capacitance accounts for charging and discharging the internal 100 million of capacitors for logic switching (C V^2 f accounts for computation at clock f). Vcore = VL + V.

So there is a time constant for Vcore and Icore switching between different level of loads. And this time constant can be different than the time constant of Vcore regulator (2 or 3 phase). And also can be different than the time constant resulted from the PSU switching cycle. I think that is what you refer to?

Does line regulation of PSU (e.g. 3% for Antec TP, 5% for Fortron and most of the rest) depends on the load, whether it is a linear resistor, or a linear network of R, C, L, or a non-linear component like a CPU. Is that 3% defined for pure resistive load, does it cover linear RLC and non-linear devices? I suppose the answer is yes, otherwise how can they use for motherboard.

Is line regulation a function of the switching cycle of a PSU? Does smaller line regulation (tighter rail) requires smaller PSU switching cycle?

What are the typical time constant of PSU and Vcore regulator?

And pointers and links to this area, thanks in advance.


Regarding to the PSU, I have been suspecting that is relating to the regulation response time of the PSU to the non-resistive current switching of the CPU and hence the Vcore regulator (as you pointed out).

Apparently the Antec TP I have regulates better the Fortron 530W I have. I am not going to mess around the Vcore regulator with adding capacitors, I would just return the Fortron and get a TP/TC 550W.

[Since87]

Originally posted by star882

The input to the Vcore regulator is 12v on all Pentium 4 and Britney systems, you need at least 16v caps.

I think hitechjb1 is using an AMD processor though.

Edit: And, what's a Britney system?

[dustybyrd]

this is all interesting, confusing, and maybe enlightening...

but i think a good real world test would be to overclock your machine as much as possible (on each power supply), notice if there is a difference in max overclock between the power supplies....then...perform these stability tests at the max overclock.....

all at once run:

passmark burnin test
prime95
seti
3dmark2001

and do that for 24 hours, heck a week....

and see which, if either, psu fails...

also check out the rails...

now that's a test....

[Since87]

Originally posted by hitechjb1
So there is a time constant for Vcore and Icore switching between different level of loads. And this time constant can be different than the time constant of Vcore regulator (2 or 3 phase). And also can be different than the time constant resulted from the PSU switching cycle. I think that is what you refer to?

I'd simplify the CPU model to a resistor and capacitor in parallel. (Where the resistor can change value substantially and suddenly depending on how loaded the CPU is.) Because current travels from the Vcore regulator to the CPU through copper planes and large numbers of pins, the inductance can be reasonably neglected.

What I was referring to when I mentioned the different frequencies of the PSU and the Vcore regulator was the frequency at which power the regulator controller switches the MOSFETs driving the output inductor. You might be interested in taking a look at the datasheet for the Vcore regulator on an Epox 8K7A+. (http://www.semtech.com/pdf/sc2422a.pdf)

The PSU probably switches at around a 100kHz rate, and the Vcore regulator probably switches at a rate from 500KHz to 1 MHz. Non-integer ratios between these two frequencies, mean that for each switching period of the PSU, the amount of charge that the PSU must transfer to the input cap of the Vcore regulator is different than during the last period. This is where instability can arise.

Originally posted by hitechjb1
Does line regulation of PSU (e.g. 3% for Antec TP, 5% for Fortron and most of the rest) depends on the load, whether it is a linear resistor, or a linear network of R, C, L, or a non-linear component like a CPU. Is that 3% defined for pure resistive load, does it cover linear RLC and non-linear devices? I suppose the answer is yes, otherwise how can they use for motherboard.

Line regulation refers to how well the supply maintains its output voltage with changes in the input voltage. Line regulation is frequently 1% and is usually not a problem.

Load regulation refers to how well the supply maintains its output voltage with different loads. To know for sure what the manufacturer means you'd have to ask them how they test, but I would guess that the specs are generally based on loads drawing a constant DC current. I doubt any PSU is designed to work with a significantly inductive load.

What I suspect is an issue in your case is transient response. I've never seen any specifications for transient response for PSU's. If you look at Figure 2 on that datasheet I linked, you'll see the transient response curve to a sudden load change of 44A. The output voltage of the regulator swings about 6.5% above and below the nominal output voltage. However, because the output voltage settles back to the same level before and after the transients, the load regulation could be considered to be better than 1%.

Originally posted by hitechjb1

Is line regulation a function of the switching cycle of a PSU? Does smaller line regulation (tighter rail) requires smaller PSU switching cycle?

Faster switching rates allows the regulator to respond to changes in output load (transients) more quickly, but I don't think PSU manufacturers consider the transient response to be part of the load regulation. I would guess Intel and AMD have transient response requirements as part of their PSU approval process, but I haven't really looked into it.

If you want to know more, Intel and AMD have probably published info on PSU issues. You can get a lot of info on switching supplies by searching for application notes at places like:

http://www.ti.com
http://www.linear-tech.com
http://www.maxim-ic.com

[hitechjb1]

Thanks for the info.

I used the wrong terminology. The line regulation I referred to is the load regulation you use.

CPU Power Model (cont)

The CPU model (ignoring L) can be like this: the resistance R accounts for the total current through all the leaking and biasing paths through the dozen million of transistors. It is roughly constant for a given Vcore and temperature. The power dissipated on the R = Vcore^2/R heats up the chip (background heat). The capacitance C varies during different level of load, lightly load smaller C, and vice versa. So the active power on C is used for computation (charging and discharging all the internal capacitors) and is given by C Vcore^2 f. At higher CPU activity/load, the C is bigger to account for more switching power. The faster the clock, the higher the active power. Typically, the steady current through R is 10-20% of the max active current through C. Active current = C Vcore f.

If the CPU is running at f (e.g. 2 GHz), and assume the task switching or load change (of a small intensive program of a few lines of C code) can be represent as N CPU cycles (e.g. N = 1000). So the time in switching between idel to heavy load = N/f (e.g. 500 ns), or at a switching rate of f/N (e.g. 2 MHz).

In order for the PSU and Vcore regulator to switch faster enough to regulate for the active current Icore between task switching, does that mean they need to have a switching time shorter than N/f or a faster switching rate than f/N (e.g 2 MHz). There are decoupling capacitors at the regulator output and input to hold the lines steady (much longer time constant).

So if Vcore regulator switches at 1 MHz, so seems in line with the above scenario. And PSU switches at 1/10 of the regulator rate, since the Vcore regulator does most of the regulation on Icore between load switching.

After that, I looked at your link for the current mode controller, it turned out that it operates at roughly the same switching rate as the above scenario. Make sense?

[star882]

"the Vcore regulator probably switches at a rate from 500KHz to 1 MHz"
I know, Christina's dual Britney machine makes AM radio completely useless when the machine is on and TVs display diagonal interference lines(the distance from one line to the next would even vary depending on the CPU load).

[hitechjb1]

Originally posted by Since87

...

Edit: And, what's a Britney system?


Star882, What is a Britney system, I'd like to know too.

[Since87]

Originally posted by hitechjb1
So the active power on C is used for computation (charging and discharging all the internal capacitors) and is given by C Vcore^2 f. At higher CPU activity/load, the C is bigger to account for more switching power. The faster the clock, the higher the active power. Typically, the steady current through R is 10-20% of the max active current through C. Active current = C Vcore f.

That's an okay way of looking at it.

No power is dissipated in an ideal capacitor. In the case of a CPU, the switching power is dissipated in resistances in series with the capacitance being charged. I would leave C as a fixed value and consider the influence of CPU loading as affecting f. Suppose we add a factor Lcpu to represent the fraction of maximum load the CPU is operating at. When the CPU is loaded Lcpu = 1, when the CPU is truly idle Lcpu = 0. Then the equation becomes:

Active current = C * Vcore * ( f * Lcpu )

There's no mathematical difference between this and varying C, but it reflects the actual situation a little better maybe.

Originally posted by hitechjb1
In order for the PSU and Vcore regulator to switch faster enough to regulate for the active current Icore between task switching, does that mean they need to have a switching time shorter than N/f or a faster switching rate than f/N (e.g 2 MHz). There are decoupling capacitors at the regulator output and input to hold the lines steady (much longer time constant).

I'm not sure that N is of much relevance. To the best of my knowledge, the current draw of the processor can go from minimum to maximum and back to minimum in three clock cycles. The faster the Vcore regulator switches, (and responds accurately to the Vcore rail's need for current) the lower the amount of decoupling capacitance required.

The MOSFET's switched by the Vcore regulator have a lot of gate capacitance as well. Although faster Vcore regulator switching will improve the regulator's transient response, and lower the requirement for output decoupling, it will also reduce the efficiency of the regulator. (Because of the power dissipated in charging and discharging the gate capacitance of the MOSFET's and overlap in the turn on and turn off of the two MOSFET's connected to the switching inductor.) The 500 KHz to 1 GHz range is probably chosen as the point of optimum balance for these issues.

There are a whole bunch of issues traded off in the design of the Vcore regulator including: cost, component size, power efficiency, and quality of regulation.

[hitechjb1]

Originally posted by Oklahoma Wolf
It's not inconceivable to me the Fortrons might do better with fluctuations under extreme loads where all rails are being hit hard, but it's hard to know where its limits are I guess with so few people owning one... don't know about you guys, but I'd like to see one really pushed like in the Tom's review or using the methodology at SilentPC Review and see how it does with regulation when it's having to run full power all the time ;)

I think there are two important factors in PSU at high level of overlcoking (high Vcore and CPU active current): one is the reserved wattage which has been paid much attention, another one is the load regulation (1% for PC Power&Cooling, 3% for Antec TP, 5% for most others).

I think the PC Power&Cooling and Antec TP when under heavy current drawn such as the CPU active current (Icore to ~60+A/CPU, biggest component in most system) can hold the PSU output lines much better. And the small line fluctutation should give advantage for the last 50-100 MHz of overclocking (this is for testing and not for price/performance).

When I tested by pushing the CPU to the edge of its Vcore stable limit, a larger current surge resulted in voltage fluctuation on Vcore would result in system crash/instability which could otherwise be avoided if the Vcore could be held tighter by a smaller PSU line fluctuation. The TP430 indeed can hold the line within 3% compared to Fortron 530 of 5% in my tests.

[hitechjb1]

Originally posted by Since87

...
I'm not sure that N is of much relevance. To the best of my knowledge, the current draw of the processor can go from minimum to maximum and back to minimum in three clock cycles. The faster the Vcore regulator switches, (and responds accurately to the Vcore rail's need for current) the lower the amount of decoupling capacitance required.

The MOSFET's switched by the Vcore regulator have a lot of gate capacitance as well. Although faster Vcore regulator switching will improve the regulator's transient response, and lower the requirement for output decoupling, it will also reduce the efficiency of the regulator. (Because of the power dissipated in charging and discharging the gate capacitance of the MOSFET's and overlap in the turn on and turn off of the two MOSFET's connected to the switching inductor.) The 500 KHz to 1 GHz range is probably chosen as the point of optimum balance for these issues.
...



CPU Power Model (cont)

So if CPU current as you described can swing from min to max within few (3) clocks, is it feasible for the Vcore regulator to switch at 500KHz-1GHz (or even higher to track for faster CPU) due to its large cap components (for optimum balance you suggested) in order to track the 3 CPU clock swing? If I understand correctly, you mentioned the Vcore regulator switches around 500KHz-1MHz earlier. At some point for higher CPU clock (2-3 GHz and beyond), the regulator switching alone cannot track the CPU clock to compensate for the Icore few-cycle swing any more. What can be done beyond that?


Back to the CPU model, not trying to get into pure academic and regardless of the mathematical eqivalence. I think a varying lossy capacitance in parallel with a resistor model better refects the internal oepration during load changes.

Since CPU clock is constant at frequency f (e.g. 2-3+ GHz) regardless of light or heavy load. When charging and discharging all the internal capacitors to do logic switching (computation), the lossy part of the capacitors (series resistive part from FET channels, plus metal and via resistance) will generate heat, while the C V^2 f part for active power and C V f for active current accounts for the power/current surge/change between load levels. During heavy load, i.e. more capacitive charging/discharging takes place due to more active functional units and cirucits are working/switching, hence bigger total effective capacitance (with lossy resistive part in series). And during light load, that lossy capacitor is smaller. As mentioned earlier, there is always a resistor in parallel (from Vcore to GND) to account for the constant heating (in background) of the CPU due to leakage and biasing current through all the FET transistor paths.

[Since87]

Originally posted by hitechjb1
So if CPU current as you described can swing from min to max within few (3) clocks, is it feasible for the Vcore regulator to switch at 500KHz-1GHz (or even higher to track for faster CPU) due to its large cap components (for optimum balance you suggested) in order to track the 3 CPU clock swing? If I understand correctly, you mentioned the Vcore regulator switches around 500KHz-1MHz earlier. At some point for higher CPU clock (2-3 GHz and beyond), the regulator switching alone cannot track the CPU clock to compensate for the Icore few-cycle swing any more. What can be done beyond that?

I'm not sure I understand the question. No modern motherboard has a Vcore regulator that tracks the CPU load 'closely'. Decoupling/output capacitors provide the stored charge to make up for the 'slow' response of the Vcore regulator. If in the future, the mismatch between CPU frequency and Vcore frequency becomes greater, the amount of decoupling capacitance will need to increase. Power MOSFET's are continually improving, so by the time CPU's are running at 20GHz, Vcore regulators may be running with acceptable efficiency at 10MHz.

Originally posted by hitechjb1
Back to the CPU model, not trying to get into pure academic and regardless of the mathematical eqivalence. I think a varying lossy capacitance in parallel with a resistor model better refects the internal oepration during load changes.

I'm not much interested in arguing this. If varying the capacitance seems a more useful model to you then use it.

No capacitance actually changes though. What does change is the number of capacitors which are charged to a different state during a given clock cycle, and the power dissipation associated with charging those capacitors to a different state through a resistance.

In any case, we are talking about gross simplifications of what is actually going on. Who's gross simplification is better? I don't really care.

[hitechjb1]

CPU Power Model (cont)

So is it a fair assumption/model that Vcore regulator switches (at a max rate) about 1/1000 times (order of magnitude) that of a CPU clock, with the decoupling capactiance circuit to regulate Vcore given Icore can changing from min to max within a few CPU cycles. And PSU switches (at a max rate) about 1/10 that of a Vcore regulator, with decoupling capacitance at the PSU output (regulator input).

BTW, are the switching rate of a PSU and a Vcore regulator constant through out, or they vary according to demand (load fluctuation)? Or is it an implemention option, constant switching rate vs varying switching rate based on load demand?


Yes, the changing capacitance to account for active switching power and current between load level is a result of the number of capacitors (in the order of 10-100 millions) in a CPU (each with series resistor) have to be charged/discharged. Under heavy load, many more of these capacitors have to be charged/discharged, the voltage of the remaining capacitors remain constant. Under light load, less have to be switched, and most of these capacitors remain unchanged in voltage. The number is larger during heavy load (hence larger effective capacitance) and smaller during light load. Hence there are larger active power and active current during heavy load and smaller during light load. Or if you like, you can look at it as

C_active = C * activity_factor

where C is the total lossy capacitance to represent logical switching in CPU, activity_factor = 1 for 100% load, and 0 for no switching. So active power = C_active Vcore^2 f, and active current = C_active Vcore f. Anyway, I think there is no disagreement here, and just for making the things clearer.

[star882]

In all Pentium 4 and Britney voltage regulators, the regulators are designed with muliple phases.
For example, a Britney mobo has a 6 phase voltage regulator switching at 1MHz(right in the middle of the AM radio band), for an effective switching frequency of 6MHz.
In the Britney mobos, the voltage regulator has a small passive heatsink(those Britney CPUs draw over 100w(130w for the 3.0GHz version!), and thus require a powerful regulator), and the south bridge also has a small heatsink(the north bridge is integrated into the CPU).
A Britney CPU has an effective resistance of about 0.019 ohms(110(Vcore power usage)/1.45(Vcore)= ~75.8A , 1.45/75.8=~0.019).

[TidyBowl]

I know this is a little late in the game for throwing in 2 cents of opinion but....

I ran into this before where I used to work, we had to test two power supplies to see which one was better. The guys doing it mounted one inside the system and tested it -- then proceded to hook up the other PSU from outside the case and running the wires into the open case. The second PSU test not as well.

After they presented thier findings to me I made them go back and reverse the test and put the second PSU inside the case properly mounted. Needless to say the second PSU got much better scores the second time around.

Not that this has any bearings on osciliscope testing but let's make sure the test is fair and equal.

Just my .02 cents

T

[hitechjb1]

Originally posted by TidyBowl
I know this is a little late in the game for throwing in 2 cents of opinion but....

I ran into this before where I used to work, we had to test two power supplies to see which one was better. The guys doing it mounted one inside the system and tested it -- then proceded to hook up the other PSU from outside the case and running the wires into the open case. The second PSU test not as well.

After they presented thier findings to me I made them go back and reverse the test and put the second PSU inside the case properly mounted. Needless to say the second PSU got much better scores the second time around.

Not that this has any bearings on osciliscope testing but let's make sure the test is fair and equal.

Just my .02 cents

T

I have taken good measures, cares on possible measurement errors (including absolute error uncertainty, relative error, potential non-linear error characteristic, ...), if you trace back to all the previous posts in this thread, identical setup (mb, cpu, voltage, frequency, temperature, ...), including the physical location of the two different PSU (Antec TP430, Fortron 530) during the two tests.

The two PSU's were respectively screwed and secured into the same position in an ATX case during the two different tests. So noise shielding and grounding should be as equal as it should be possible.

[Since87]

Originally posted by hitechjb1
So is it a fair assumption/model that Vcore regulator switches (at a max rate) about 1/1000 times (order of magnitude) that of a CPU clock, with the decoupling capactiance circuit to regulate Vcore given Icore can changing from min to max within a few CPU cycles. And PSU switches (at a max rate) about 1/10 that of a Vcore regulator, with decoupling capacitance at the PSU output (regulator input).

In general yes. Actual frequencies for PSU's and Vcore regulators may well vary more than 2 to 1 from those numbers though.

Originally posted by hitechjb1

BTW, are the switching rate of a PSU and a Vcore regulator constant through out, or they vary according to demand (load fluctuation)? Or is it an implemention option, constant switching rate vs varying switching rate based on load demand?

The most common way of controlling the power output of a switching regulator is with Pulse Width Modulation. Switching frequency is usually fixed. The output MOSFET is turned on for a larger fraction of the switching period to deliver more power to the output. Some switching controllers are designed so that they will actually skip outputting during some switching periods when the load is really low, but I doubt that any Vcore controllers do this. (The quiescent current draw of a CPU is high enough that this is not needed.)

[dustybyrd]

check this post out....i think this should let you realize that you likely can't trust any of the voltage readings from software programs....this post from a user with a enermax 465 and an antec truepower at 550....

i don't believe for a second that that 465 enermax is a better power supply than the antec 550...but that's what you would think looking at the numbers...

http://forum.oc-forums.com/vb/showthread.php?s=&threadid=203450

"Help Antec TruePower 550W Low Rail!
I just recently upgrade my power supply from Enermax 465P-VE 431W power to Antec TruePower Supply and notice one strange problem. The problem I'm seeing is that the TruePower 550W shows low rails compare to my Enermax 465P-VE 431W.

The wierdest part is that the TruePower is suppose to provide better rail than my Enermax because of more watts then the Enermax. The chart below is the comparison between the two with the same setup.

Enermax 465P-VE 431W
Idle Load
+5 = 5.03 +5 = 4.89-5.00
+3.3 = 3.35 +3.3 = 3.30-3.37
+12 = 11.90-12.09 +12 = 11.78-11.9

Antec TruePower 550W
Idle Load
+5 = 4.80 +5 = 4.67-4.70
+3.3 = 3.10-3.20 +3.3 = 3.00-3.2
+12 = 11.7-11.8 +12 = 11.50-1.60

See what I mean between the two, the Enermax seem to be performing better then the bigger Antec TruePower 550W. Both were purchase as Brand New.

Question, what does the 230V or 115V voltage output mean in the back of the power supply. Can I safely adjust those two setting or what?

Any suggestion is highly appreciated."

[hitechjb1]

This seems to concur with what I have seen and required for the overclocking of my TB B 1700+ DLT3C, at 1.9+ V Vcore, 2.5+ GHz on air.

If looking back the posts in this thread, I found that besides sufficient wattage reserve which has been paid much attention and talked about, tight line regulation (1% PC P&C, vs 3% fo Antec TP, vs 5% for the rest) is important for system and Vcore stability for the last 50-100 MHz of CPU overclocking.

It has been pointed out by Since87 that CPU huge load change (min to max) and current surge can occur in a short time of 3 CPU cycles. And I think such requires fast transient response of the switching regulator and PSU to regulate the Vcore line (+- 50 mV max from AMD spec) and the underlying PSU line to within small fluctuation for system and CPU stability.

Quote:

"In the power-use class, the 450- to 520-Watt league, the THG lab technicians only arrived at one recommendation: the Antec True 480P, which produced the most power in the test. This power supply also delivers stable voltages on all lines and comes with a good equipment package. The Antec True 480P is the power supply we recommend for high demands, and is especially suited to extreme overclockers."
Reference: June 9, 2003 High Power For Power Users - 13 Power Supplies In The Spotlight (http://www6.tomshardware.com/howto/20030609/power_supplies-18.html)

[dustybyrd]

Quote:
"In the power-use class, the 450- to 520-Watt league, the THG lab technicians only arrived at one recommendation: the Antec True 480P, which produced the most power in the test. This power supply also delivers stable voltages on all lines and comes with a good equipment package. The Antec True 480P is the power supply we recommend for high demands, and is especially suited to extreme overclockers."
Reference: June 9, 2003 High Power For Power Users - 13 Power Supplies In The Spotlight (http://www6.tomshardware.com/howto/20030609/power_supplies-18.html)


thanks for pointing this review out to me...didn't see it before..

i agree that the antec true480 seems to be a great supply...484 watts

but they didn't test anything higher than a 400watter from fortron and it was the most efficient performer out of all the supplies they tested and it was the most underrated supply in that it produced the most power to rated power ratio...452 watts (which ironically is about the same as the 350 watt fortron supply)

however, they have not tested the newer 530 watt fortron or 550 watt fortron or the 550 watt antec for that matter...

i would bet that they would both be great...

notice that the 350 watt fortron and the 400 watt fortron produce more power than the 465 enermax (rated 431 watts and produced that much as well)

[hitechjb1]

CPU at exterme overclocking (much higher Vcore and frequency than rated) can draw twice the active current and almost three time the active power compared to that at rated Vcore. For details, there are some threads discussing this in the AMD CPU section.

These are links about how to estimate CPU power, especially for high level of overclocking.

Question: How much power is increased ... (http://forum.oc-forums.com/vb/showthread.php?s=&threadid=188850)

how many watts does your cpu take up? (http://forum.oc-forums.com/vb/showthread.php?s=&threadid=190084)

Those current changes and hence Vcore fluctuation can occur within a very short period of time (As in earlier post, Since87 suggested to be of the order of CPU cycles). If so, these fluctuations cannot be detected or SEEN by the commonly used DC multimeter. They have to be measured by special instrument such as a sampling oscilloscope over a peroid of time (which are not readily available at home), or possibly using software probing over a long period of time to estimate the line fluctuation (%). I think the fast fluctuation is more an issue for the last 100 MHz of overclocking than the absolute DC voltage values, provided that they can held within tight range, say 3%. Hence I think it requires Vcore regulator and PSU to have fast response time to regulate those voltages due to the fast CPU/GPU current changes, than just good DC regulation.

[hitechjb1]

Form the PSU spec (whether you want to rely on it or not).
For 12 V,
TP 480 is 22 A
TP 550 is 24 A
Fortron 530 is 18 A

Many Intel and AMD mb use 12 V to generate Vcore and for video card, both of which consume a large % of the 12 V current. I have an estimate for active current for an AMD CPU during high level of CPU overclocking, the 12V line current is close to 18 A for both CPU and video alone.

Further, the intake fan position of the Antec PSU is right above the CPU area (I saw an AMD techsheet recommends such fan position). Also I think the 3% line regulation of Antec TP (vs Fortron 5%) is crucial for high level CPU overclocking stability.

If one does not heavily overclock the CPU (e.g. under 1.8V, 2.4 GHz for AMD XP CPU), then the above concerns do not matter.

[dustybyrd]

Form the PSU spec (whether you want to rely on it or not).
For 12 V,
TP 480 is 22 A
TP 550 is 24 A
Fortron 530 is 18 A

Many Intel and AMD mb use 12 V to generate Vcore and for video card, both of which consume a large % of the 12 V current. I have an estimate for active current for an AMD CPU during high level of CPU overclocking, the 12V line current is close to 18 A for both CPU and video alone.

Further, the intake fan position of the Antec PSU is right above the CPU area (I saw an AMD techsheet recommends such fan position). Also I think the 3% line regulation of Antec TP (vs Fortron 5%) is crucial for high level CPU overclocking stability.

If one does not heavily overclock the CPU (e.g. under 1.8V, 2.4 GHz for AMD XP CPU), then the above concerns do not matter.

I am using the fortron 530 watter WITH the 4 pin 12v connector for the two AMD 2100xp's in my dual system...

here are my results on a DUAL AMD machine with 2100xp's at 2 ghz and 4 hard drives (2 in RAID 0), 1 GIG of memory, case fans, tv card, 275/275 overclocked 8500le, sound, cdwriter, zip, floppy, etc....see sig...

so i ran Passmark's Burn-in test at 100% on both cpus (this program stresses cpu, 3d and 2d graphics, hard drives, cdrom, floppy, memory, sound ALL AT ONCE...i also ran two instances Prime 95 separately and got about the same results...

and here are the software voltage rail results:

no load-versus-load...

12v=12.16-12.22
5v=5.02-5.06
3.3v=3.34-3.36


I'd say that we can safely ignore the "18A" spec of the fortron 530 power supply...because by my calculations my cpus ALONE use about 16A and obviously i am running much more off the 12v rail in this system....

the 12v rail barely fluctuates at all

[hitechjb1]

I do not agree that we can "safely ignore" spec, as you suggested. Without even specifying what the implications are, e.g. on line regulation, on temperature tolerance, effect on other lines, ... of the unofficial rating (if any). Reputable manufacturers put up specification for some reasons. Specifications are needed for an industry, and large scientific and engineering projects.

"Safely ignore spec" is different than showing an individual good result. A test case or two or few cannot be generalized to replace and ignore specification which have gone through extensive engineering tests.

I don't know what hidden rating you have found out and how many units of that model you have been tested. You suggested to "safely ignore" spec without giving a substituted number for that specification, whether it would be 20A or 22A or 24A or even more. Also it is not clear what is the chance of achieving the unofficial rating (if there is any). It will not help people looking for a new PSU, since we don't even know what the unofficial current rating is, and what to expect, and what is the chance of getting that?

You once recommended me to get a Fortron 350W claiming it as 454W without knowing what my system and requirements were, and you said
for $48 you get 454 watts tested power...that is completely silent with a super large 120mm intake fan (that removes internal hot air from the case and shoots it out the back)---i have a dual system with 2 hd's in RAID, etc...on this power supply and it is rock solid and pushes only cool air out the back...i personally guarentee that this will be all you need and more for super low money and low noise
I am glad I did not take your suggestion, and I did my homework/calculation way back, otherwise I would have problem powering my CPU/system to 2.5-2.6 GHz, which I estimated to consume twice the power compared to running at 2GHz at 1.5V. Recently, I saw some posts saying that Fortron 350W has problem powering P4 beyond certain frequency.

Further, if CPU is not overclocked as expected, the result may be just that the system will not be running as fast. For PSU, if it is not working according to the unknown/unofficial spec, it would affect the entire system and stability.

E.g. even if I can put 2.2 V on a CPU myself, I would NOT recommend others to ignore the CPU voltage rating from AMD or Intel.

My point is that we cannot ignore, replacing manufacturing spec based on one or two tested cases, and recommend parts based on unofficial numbers (if any), and ignoring spec. New comers may not know the implications. If the parts are not meeting the expectation, will waste time and money.

I am not disputing your result on the Fortron 530W, and recommending your PSU, but I disagree going from a test result to saying "safely ignore" the spec, ... for reasons mentioned above.

[dustybyrd]

Originally posted by hitechjb1
[B]Reputable manufacturers put up specification for some reasons.

i'm not sure why they do...since very, very few of the specifications are correct....most are over-rated (like deer,etc..) and some are under-rated (like antec, fortron)

I only know of one that is usually right on their provided specs and that is enermax...

i get this data from tmg's review


"Safely ignore spec" is different than showing an individual good result. A test case or two or few cannot be generalized to replace and ignore specification which have gone through extensive engineering tests.


why not ignore them since almost all specs are wrong (some under and some over rated)

heck...by your calculations that i saw in some other thread, the antec true 550 is also likely to be underrated in the 12v...since it must be putting out more than what it is rated in your system...


You once recommended me to get a Fortron 350W claiming it as 454W without knowing what my system and requirements were,


and i would recommend the fortron 350w again...without knowing what your system specs are...assuming you are building a normally overclocked system...


am glad I did not take your suggestion, and I did my homework/calculation way back, otherwise I would have problem powering my CPU/system to 2.5-2.6 GHz, which I estimated to consume twice the power compared to running at 2GHz at 1.5V.

i would bet the fortron 350watter would power your system---try it and see


My point is that we cannot ignore, replacing manufacturing spec based on one or two tested cases, and recommend parts based on unofficial numbers (if any), and ignoring spec. New comers may not know the implications. If the parts are not meeting the expectation, will waste time and money.


almost all psu manufactors inflate their specs...therefore if a newcomer believes the specs then they will likely get an overrated generic psu that will be crap

i am not disputing if the antec true 550 is great...i'm sure it is....what i am saying is that it is overpriced compared to a fortron 530 watter...that i would be willing to bet would do just as well as the antec 550 and when someone tests these supplies fully I'd bet that will be the case...

furthermore, i still contend that a fortron 350 watter would power your 2.6 ghz AMD system....i KNOW of several people using the fortron 350 watter on DUAL AMD systems at 2ghz (which i am SURE these systems require more power than your 2.6ghz AMD---in addition to the extra CPU...the motherboard requires more power---with all the other power consuming goodies...

[hitechjb1]

Originally posted by dustybyrd

and i would recommend the fortron 350w again...without knowing what your system specs are...assuming you are building a normally overclocked system...
...
i would bet the fortron 350watter would power your system---try it and see
...
i am not disputing if the antec true 550 is great...i'm sure it is....what i am saying is that it is overpriced compared to a fortron 530 watter...that i would be willing to bet would do just as well as the antec 550 and when someone tests these supplies fully I'd bet that will be the case...


From what you wrote, it is clear to me that you recommended PSU, to me at least, based on assumptions, not knowing what I wanted and what my system was.

Also you still recommend fortron 350W to me and asked me to "try and see", given that in other posts of this thread that I already returned a Forton 530W since I found that it did not work out for me and for my system due to higher line fluctuation, fan locaton, ...

Some facts: a 2.6 GHz 1.95 V CPU consumes an active power more than twice that of a CPU running at 2 GHz 1.5 V (using Tbred B 1700+ DLT3C as example).

I am not going to spend more time on this issue, I know what to get and how to calculate my system requirements and to obtain sound recommendations, rather than from recommendations based on assumptions, "try and see", and bets.

[dustybyrd]

Also you still recommend fortron 350W to me and asked me to "try and see", given that in other posts of this thread that I already returned a Forton 530W since I found that it did not work out for me and for my system due to higher line fluctuation, fan locaton, ...

Some facts: a 2.6 GHz 1.95 V CPU consumes an active power more than twice that of a CPU running at 2 GHz 1.5 V (using Tbred B 1700+ DLT3C as example).




you said that your software read voltage rails were fluctuating with the 530 watt fortron...but did your system overclock the same? was it stable? did it reboot? who cares about software voltages...what matters is REAL WORLD performance!---

the fact is someone using the antec true 550 got horrible software read voltage fluctuations compared to a wimpier enermax 465:

Antec TruePower 550W
Idle Load
+5 = 4.80 +5 = 4.67-4.70
+3.3 = 3.10-3.20 +3.3 = 3.00-3.2
+12 = 11.7-11.8 +12 = 11.50-1.60

Enermax 465P-VE 431W
Idle Load
+5 = 5.03 +5 = 4.89-5.00
+3.3 = 3.35 +3.3 = 3.30-3.37
+12 = 11.90-12.09 +12 = 11.78-11.9

( see this post http://forum.oc-forums.com/vb/showthread.php?s=&threadid=203450)

do YOU believe those software read voltages? if you do, that means YOU should have gotten the enermax 465 instead of the antec true power 550...and I would say that the antec is the better supply...

here is a quote from larva (i assume you know his credentials):

"It doesn't matter how many software programs you try, they are all using the innacurate motherboard sensors to read voltage. These are not accurate enough to blame the supply based on their readings."

lastly, my dual 2ghz AMD's are at 1.63 volts, not 1.5....therefore, my 2 AMD processors consume more 12 volt current than your 2.6ghz at 1.95v....

and yet my fortron "software" voltages on my dual AMD 2ghz were really, really good...(from mbm5 for over 2 days and sisoft sandra 2003)

no load-versus-load...

12v=12.16-12.22
5v=5.02-5.06
3.3v=3.34-3.36

(BUT big deal...it doesn't matter what the software voltages display...all that matters is that it's rock stable)

[hitechjb1]

Testing Antec True Control 550

Got an Antec True Control 550 PSU. Its 3.3 V, 5 V, 12 V lines are very stable at high level of overclocking of CPU and GPU on air, compared to my previous test results of a Fortron 530, Antec True Power 430.

With the 1700+ overclocked to 2.545 GHz 1.925V, for the three PSU tested, for the 5V line that generates Vcore,
- Antec TC 550 has a fluctuation of 1.6% compared to
- Fortron 530W fluctuation 5.5% and
- Antec TP 430 fluctuation ~3%.
The Antec TC 550 12V line flutuation is less than 0.5%, but this may due to lack of heavy 12V current until I test it with a mothboard that uses 12V ffor Vcore.

System:
A7N8X-DLX, 512 MB PC3500 C2 memory
CPU: Tbred B 1700+ DLT3C 2545 MHz (204x12.5), 1.925 V, SLK-800U, TT SFII
Video card: 9500np/9700 (softmod) at 360/311
1 HD, 4 case fans, 1 CPU fan, ...

The line fluctuations of the True Control 550 and Fortron 530 from ASUS probe when running Prime95 are shown below. From the PSU's that I tested, it is clear (also shown from pics) that the True Control 550 has much better line regulation than the Forton 530 and Antec TP 430 under similar conditions of high level of CPU/GPU overclocking.

Fortron was tested with a Radeon 8500 only (not the 9500/9700).
5V fluctuated 274 mV (5.5%), if 5% measurement uncertainty is taken into account, fluctuation would be 5.2% .
Vcore bounds between 1.85 to 2.0 V (150 mV fluctuation), this was outside of the AMD Vcore tolerence of +- 50 mV.
12V idle is at 12.4 V (high)
Got BSOD when running Prime95 at 2545 MHz.

Antec True Control 550 (spec 3.3V 32A, 5V 40A, 12V 24A, 3% line regulation)
3.3 V: adjustable 3.10 - 3.35 V, idle = 3.232 V
5 V: adjustable 4.73 - 4.97 V, idle = 4.838 V
12 V: adjustable 11.60 - 12.40 V, idle = 12.032 V
The PSU fan speed can also be adjusted from ~1700 to ~2800 rpm.

When running Prime95
3.3 V: 3.152 - 3.248, fluctuation = 96 mV (2.9 %) <- with 9500np//9700 (softmod), Vagp uses 3.3V. Less 3.3 V fluctuation if R300 overclocking is down to 333/300.
5 V: 4.784 - 4.865 V, fluctuation = 81 mV (1.6 %) <- line that generates Vcore
12 V: 11.968 - 12.032 V, fluctuation = 64 mV (0.5 %)
Vcore: 1.888 - 1.952 V, fluctuation = 64 mV (within +- 50 mV of AMD tolerance)

Fortron 530 (spec 3.3V 28A, 5V 40A, 12V 18A, 5% line regulation)
3.3 V: idle = 3.312 V
5 V: idle = 4.703 - 4.977 V
12 V: idle = 12.416 V (high)

When running Prime95
3.3 V: 3.248 - 3.296 V, fluctuation = 48 mV (1.5 %) or more <- Radeon 8500 only, without 9500np/9700 (softmod), Vagp uses 3.3V
5 V: 4.703 - 4.977 V, fluctuation = 274 mV (5.5 %) <- line that generates Vcore
12 V: 12.416 - 12.608 V, fluctuation = 192 mV (1.6%) or more
Vcore: 1.850 - 2.000 V, fluctuation = 150 mV (outside AMD Vcore tolerence of +- 50 mV)

For the Antec True Power 430, its 3.3V, 5V and 12V were around 3% from specification and around 3% line regulation when running Prime95. The TP 430 PSU was able to power the above CPU to 2.60 GHz Prime95 stable. Later, it was also tested OK to run with a 9500np/9700 (softmod) overclocked to 360/311.

It would be interesting to see how the TC 550 performs with a NF7-S (that uses 12V for Vcore) with the 9500np/9700 (softmod). I expect that the CPU, the video card, HD and fans, ... will draw about 24A from the 12V.


Next two pictures show the line fluctuations of an Antec TC 550 and an Fortron 530 running Prime95. Note the difference in the 5V line (for generating Vcore) between the two.

Antec True Control 550 line fluctuation when running Prime95 at 2545 MHz

http://hitechjb1.dynu.com:8081/images/TbredB_OC/dlt3c_1700_2545_tc550_prime95_vmon3_c.JPG

Fortron 530 line fluctuation when running Prime95 at 2545 MHz (not stable)

http://hitechjb1.dynu.com:8081/images/TbredB_OC/dlt3c_1700_2545_fsp530_vgraph_prime_c.JPG

[dustybyrd]

Originally posted by hitechjb1
[b]Testing Antec True Control 550

Got an Antec True Control 550 PSU. Its 3.3 V, 5 V, 12 V lines are very stable at high level of overclocking of CPU and GPU on air, compared to my previous test results of a Fortron 530, Antec True Power 430.

With the 1700+ overclocked to 2.545 GHz 1.925V, for the three PSU tested, for the 5V line that generates Vcore,



it looks like from your software voltage read probe results that the fortron 530 had higher rails for all three lines...and from your picture it looks like the 5v line of the fortron fluctuated very little around 4.95v and the antec was fluctuating (even less) around 4.85v....i would think that slight fluctuation around 4.95 is better than less fluctuation around 4.85v, no?

and it looks like from the Asus probe picture that the 3.3v line on the Antec fluctuated MORE than the Fortron's....isn't the voltage supplying memory important?

furthermore, you said in the post that you set the vcore at 1.925...and the fortron 530 seems to be at 1.92 volts (with almost no fluctuations) while the antec is at 1.936 volts (also with almost no fluctuations)---isn't the antec maybe a little too high? wouldn't want to accidently fry a chip...

maybe i should post the same Asus probe picture of my fortron 530's results powering TWO 2ghz AMD cpu's at 2Ghz....

no load-versus two prime 95 instances-load:

12v=12.16-12.22
5v=5.02-5.06
3.3v=3.34-3.36

and remember this is what someone else posted about the antec true 550:

Antec TruePower 550W
Idle Load
+5 = 4.80 +5 = 4.67-4.70
+3.3 = 3.10-3.20 +3.3 = 3.00-3.2
+12 = 11.7-11.8 +12 = 11.50-1.60


if your forton 530 was a lemon...(which it seems to be fine to me)...then the antec true 550 above is also a lemon (which i also doubt)....too much emphasis on software voltages maybe?

should I have paid $40-50 more (about 70% more) for an antec true 550? i don't think so...



Also your calculations on 12v current consumption of the AMP XP processors are incorrect...this is from the AMD site:

http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/26003.pdf

http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/25175.pdf

2. To calculate the processor current at 12V VRM source: ((Processor Core Voltage X Processor Core Current) / 12) X 1.25

therefore the current requirement of a 2600+ AMD at 1.65volts would be 41.4X1.65/12X1.25=7.11A's

and at 1.925volts:

41.4X1.925/12X1/25=8.3A's

and...since processor working state current (Icc) doesn't increase that much with increases in processor speed (1.73ghz=38.8A while 2.133ghz=41.4A)...then even at 2.6ghz the Icc is likely to be at the most 45A....

so an AMD 2.6ghz processor at 1.925 volts requires this much 12v current:

9A!!!

not ~11A...like you had said in previous posts...

Also my dual 2100's at 2ghz (1.63 vcore) require 14.1A's...

[cache]

wait wait, britney system=amd=cheap?

[hitechjb1]

I just got a TC 550 and its line regulation under CPU/GPU overclocking and Prime95 are 2.9% on 3.3V, 1.6% on 5V (Vcore line), 0.5% on 12V (detailed in previous post).

By adjusting the variable resistors in the control panel, the voltages can be adjusted to almost (+-) 3.3, 5 and 12V. Usually for PSU's, I do not look for perfect absolute voltage value, since the various regulators can regulate them, as long as the lines are and fluctuate within spec.

Question:

- Can the True Control 550 be used without the control panel?

- What are the worst case high and low voltages on the 3.3, 5 and 12V lines without hooking up the control panel.