***Volt Mod's For AsRock SataII***

[RollingThunder]

I dont think I understand the question. but if you are implying if there is a direct relation between vcore/vdimm increase and overclocking? there is but not a set formula or rule.
more voltage helps w/ overclocking that's about it and as far as i understand it.
From my own experience, i have learned that for every 50 mhz increase in cpu speed, you need ~ 25mv(one notch higher) vcore increase.
another word, if you increase fsb from 200x10 to 205x10 you need 25mv increase. But even this is not a set rule. Most cpu's will overclock about 20% without any vcore increase. When cpu is unstable at an overclock, then "may be then" you should increase vcore one notch and keep increasing speed until it is unstable again, and repeat the process. The 25mv increase in vcore for every 50 mhz speed increase, should be applied when highest cpu overclock is found @ stock vcore (you ran into a wall w/ stock vcore).
I hope I am making sense.

OC,

Yes, your answer was what I was implying and it's pretty much what I was seeing overclocking my old Asrock before it went out. When I was stock testing the CPU for an overclock my old Asrock board hit the wall at about 11x234 - 236 MHz and probably the vcore limit output for the board or over. 11x230 ran ok for a while.

A few RAM brands state what the vdimm specs increase with overclocks or they give you something like 2.6v - 2.9v and you figure it out from there.

 

[Hipcrostino]

Hey, whats the calculation to figure out the vDimm after you have modifide the value of #2 resisitor?

 

[ochungry]

Hey, whats the calculation to figure out the vDimm after you have modifide the value of #2 resisitor?

I think I wrote it somewhere in this thread. the way you calculate the final resistor value when connected in parallel to #2, is:
Rnew= (R1*R2)/(R1+R2).
the existing (#2) is 1.2k. if you solder a 1K on top of it, you get:
Rnew= (1.2*1)/(1.2+1)= 1.2/2.2=.545k=545 ohm

From experiments i have found:
800 ohm gives you 2.82v
545 ohm gives you 2.90v
300 ohm give you 2.95v
200 ohm gives 3.v
150v give 3.05v
those can vary +/- .05v depending on your system.
If you want the final value to be a "known" and need to know what resistor to attach to #2, you use same formula, but move things around:
If R1 is existing (#2), and Rnew is also known value (what you want final value to be)
what to buy/solder:
R2= (Rnew*R1)/(R1-Rnew)
R2 = what to buy/solder to #2 in parallel.

 

[Hipcrostino]

thanks. currently i have the resistor with the graphite mxture on top and im measureint 470ohms. going to try that. this is TCCD and i know it cn take 2.9. not going to test it with my good ram first though. I'll let you know hoe it gose, should give me just over 2.9v


UPDATE: Success! 2.870v meassured as the vdimm!! all right!

UPDATE 2: Well, my ram hasn't reponded to the voltage at all . Im stable at 225mhz 2.5-3-2-7 where as before i was 2.5-3-3-8. Its a Pitty, but arr well. I've had one random reset so far too, but at the time the vcore was below the stock levels so im hopeing it was that. Just primeing at 2.8ghz now and have yet to use the extra vcore i should now have. Load temp was 30, but its just failed. im at 1.425 in the bios right now so its to be expected. Anyway ill update how it gose.

 

[chinobino]

Hipcrostino, I noticed in your sig that you have your RAM's command rate set at 1T. This board often has problems with 1T and high HTT frequencies. The best way I have found to test 1T stability on this motherboard is to browse on the internet and look for a page that has a lot of flash animations. For some reason flash accesses memory in a certain way that causes the CPU's memory controller to fail with 1T at high HTT frequencies on this board. I have found the Silkroad Online website to be good for testing 1T stabiilty.

I have run up to 264 MHz HTT and RAM (1:1) at 1T on my board, but others have reported not even being able to run at more than 225 MHz.

Just thought you should be aware of this potential problem .

 

[RollingThunder]

Chinos' Sig >

AMD ATHLON 64 3700+ KACAE @ 2913MHz (11 x 264) @ v1.472
Asrock 939Dual-SATA2 Mobo
PCB Rev. 1.04 BIOS 2.20ADV (ULi chipset) socket/wire mod for v1.55 vcore
2GB Corsair XMS PC4400 (DDR550, 4 x 512M) Samsung TCCD
RAM 1:1 @ 2.5-4-4-8, 2T, @ 2.7v
BFG 7800GS OC (400/1250 Default) @ 469/1380
Western Digital 250GB PATA HDD (8MB, 7200 RPM)
Western Digital 80GB PATA HDD (8MB, 7200 RPM)
Scythe Ninja Heatpipe (SCNJ-1000)
Thermaltake Toughpower 750W

Chino,

Your sig is interesting in that you are at 2913 MHz with a vcore mod with some voltage to spare if necessary to go along with a nice set of RAMs. I'm particularly interested in your vcore of 1.472 at your overclock. Since I am electronically challenged I hesitate to put Multimeter probes to check this while it's running........ That's a pretty nice overclock all around!

I got my RMA Asrock board back and will go to work on it with OC's help.

The new Asrock board is up to Rev. 1.05 but I don't know what the changes are. I still can't find out what chips these Kingston HyperX use but they have a good repuation so I'm not concerned. I'll be tickled to death to get 11x230 or 11x240 with some moderate vcore and vdimm mods.

I did run my old board at 1T for a short time with Crucial Ballistix 2x512 @ 11x230 but with my inexperience at the time I was seriously pushing my vdimm limits and was over it from the beginning because the Ballistix needed 2.8v stock. The board was never right after that.

When the system finally failed I dropped it all back to stock, sent back the Ballistix and bought Patriot 2.6v XBLKs PC3200-PC4200 . I could never run 1T or any overclock after that, not even with an underclock. An IC finally burned out on the board so I RMAd it back. Who knows what vdimm those Crucials were pulling at that overclock. I hate to even think about what I was trying to do with them without mods.

When these new Kingstons failed individually and in pairs in every slot locking up Memtest86 I knew the problem wasn't the RAMs so I looked around the board and found the burned IC. Not sure what it controlled but OC seems to think it was the USB IC, not sure. The only peripheral I use a USB for is my APC UPS and the occasional digital camera plug-in.

Needless to say I made some mistakes from inexperience, cost me a few bucks but didn't break the bank. If I knew back them what I have learned now.............. I RMAd those Patriots back because I thought they were bad when apparently they were fine all along. I was wrong and now they are discontinued and out of stock. Too bad, I never got a chance to work them

It's been a experience to say the least!

 

[Hipcrostino]

Hipcrostino, I noticed in your sig that you have your RAM's command rate set at 1T. This board often has problems with 1T and high HTT frequencies. The best way I have found to test 1T stability on this motherboard is to browse on the internet and look for a page that has a lot of flash animations. For some reason flash accesses memory in a certain way that causes the CPU's memory controller to fail with 1T at high HTT frequencies on this board. I have found the Silkroad Online website to be good for testing 1T stabiilty.

I have run up to 264 MHz HTT and RAM (1:1) at 1T on my board, but others have reported not even being able to run at more than 225 MHz.

Just thought you should be aware of this potential problem .

Cheers man. Yeah it is at 1T, have never run the mem at 2T on this board so far. im like that though. Best way to test is find the max 1t and 2t freq's and see what the bandwidth differance is. I'd like to get close to my HTT (i.e be 1:1) but ive now amanged to get windows 22mins prime stable @ 2.9Ghz with some voltage to spare (but getting close to the 1.6v max). im confident at getting 2.9ghz stable, so that means 290mhz for 1:1. this ram it rated at 2 2 2 6 200Mhz, so that would be good going.

Anyway, thanks for the heads up

EDIT: just a note. this ram HATEs runnign at CAS 3. Ive never seen it not error at that speed(as in not even at stock 200mhz). cas 2 or 2.5 only it seems

 

[RollingThunder]

EDIT: just a note. this ram HATEs runnign at CAS 3. Ive never seen it not error at that speed(as in not even at stock 200mhz). cas 2 or 2.5 only it seems

Hip,

This board really seems to have a lot of variance from board to board and without any pattern to it.

I did run my first board @1T for a short time with the Ballistix CAS2.5, then it errored and that was it, 2T and CAS 2.5 or nothing. All the RAM sets I used defaulted to CAS 2.5 and that was all it would ever run them at, default, overclocked or underclocked. CAS 2 or CAS 3 crashed every time. Anything else errored right away and I was locked in to 2.5.

This is why I am so interested in everyone's CPU performance and RAMs they are using. When I get my new board running with the latest revision, I'll let it set its default timings and see what happens. These Kingston HyperX are 2-3-2-6 but I'm betting they default to 2.5. I'd be surprised if they didn't.

 

[chinobino]

Chino,

Your sig is interesting in that you are at 2913 MHz with a vcore mod with some voltage to spare if necessary to go along with a nice set of RAMs. I'm particularly interested in your vcore of 1.472 at your overclock. Since I am electronically challenged I hesitate to put Multimeter probes to check this while it's running........ That's a pretty nice overclock all around!

Thanks RT , I very carfully selected all of my components to make sure I would get a good oc. The v1.472 vcore is only possible because of my CPU stepping (KACAE). I read up a lot on all the different AMD socket 939 CPU models and steppings before I decided to get this CPU (I bought the Asrock board back in January but couldn't decide on a CPU until July lol).

I got my RMA Asrock board back and will go to work on it with OC's help.

The new Asrock board is up to Rev. 1.05 but I don't know what the changes are.

The v1.05 boards seem to have more stability at 1T with more types of RAM. Also, it is rumoured that the v1.05 boards do not have the cold boot issue when using an HTT above 274MHz - perhaps you can confirm this?

I still can't find out what chips these Kingston HyperX use...

That's quite a story! Your Kingston RAM could be Hynix D4 or D5, Samsung TCCD or even Winbond BH-5 (although less likey). You really should try to find out what chips are under those heat spreaders.

I don't think you should take the heat spreaders off though becuase sometimes they are attached like super glue and you can damage the IC's (not that I would ever do something silly like that and stuff up one of my Corsair XMS sticks ).

The best way to find out is to look for some reviews of your RAM on the net and see if you can find a mention of the IC's or maybe a photo. The date of the review can give you a good idea of whether your RAM is likely to be the same (you can also check serial numbers).

Different types of RAM respond to different voltage ranges. For example, my TCCD RAM performs well with v2.7-2.9. As you can see in this screenie, I am running my RAM @ 624MHZ DDR with only v2.7 (RAM set to High) but with crap latenices. When I get around to doing the vdimm volt mod I should be able to do ~275MHz or more with tight timings.

Take your time if you do the mods and print out all the pictures and info you need before you start (unless you've got another PC with internet access). That should reduce any chances of error. Let us know how you go!

EDIT: just a note. this ram HATEs runnign at CAS 3. Ive never seen it not error at that speed(as in not even at stock 200mhz). cas 2 or 2.5 only it seems

That is very odd . I can run my RAM fully relaxed (3-6-6-15, 2T) from 200MHz right up to 312MHz... 2.9GHz with RAM 1:1 and tight timings sounds like a good goal, lots of memory bandwidth, bring it on! I am aiming to do the same. I know the RAM is capable, it just needs more juice!

 

[chinobino]

Oh, I forgot why I came here lol.

3GHz stable - that Opteron 156 better watch its back!



Chino's to do list;

1) Get to 3.1GHz stable... VID5 mod?

2) Vdimm mod - RAM 1:1 at either 282 x 11 or 310 x 10 with tight timings.

3) Heatsink mod for 7800GS - Arctic Cooling Accelero S2

4) Make that Opty 156 wish it was never invented! Mwhahaha!

 

[Hipcrostino]

haha nice CB. With the 1.55 vcore mod I can get a much better clock than 2.7Ghz. This weekend im going to get 2.9ghz stable (note the 'going to'). did a little testing last weekend and managed 20mins prime stable at 2.9 with voltage to spare. this board overvolts 0.033v more than what its set too (and reports higher) so at 1.55 im at around 1.58. Im on air (although good air) so its about my limit i think. I think it'll get into windows at 3ghz now, but now way it will be stable. 2.9 makes me friggen happy though. im going for 24/7 clock btw.

im pritty sure my rev is 1.05. havn't had any cold boot issues yet and ive only ever run my ram at 1T. anyway ill let you know how it all gose on the weekend

 

[RollingThunder]

Chino,

I am assuming the Kingston RAMs are CH5 form the stickie provided in the Memory section:

Kingston HyperX PC 3200AK2 - Winbond CH-5

http://www.ocforums.com/showpost.php?p=4005092&postcount=3

But I don't know for sure because that post is 11 months old. And no, I won't take the heat spreaders off to find out. For what I want to do, most all performance RAMs will work for me. I'm not in the same league with you guys so an oc of 10-15% makes me happy! I'll search around the Internet for a more recent evaluation of these Kingstons.

Maybe I'll be able to run 1T with the new board with a CAS of 2. I'll find out soon enough.

 

[chinobino]

To get some idea of how CH-5 perform;

CH-5 – This is a newer ram chip than BH-5 and on average cannot achieve the results BH-5 can. It is known to do around 220-230 MHz at cas2-3-2-X. This ram is a little different than BH-5 as it does not respond as well to voltage. Usually, anything over 3 volts does not yield any better results. This could be due to the manufacturer's PCB design, or simply the ram chips themselves. In some cases, however, this ram can achieve similar results to BH-5, it just does not happen very often. This is still great ram for the money and can most likely achieve cas2-2-2-X at 200 MHz with the right voltage applied. This ram is easier to find than BH-5 and usually costs a fraction of what BH-5 costs. It can be identified by the last 3 characters on the chip reading CH-5. It can be found in Corsair XMS, Kingston Hyper X, and many other offerings.

What the Winbond UTT CH-5 chips can do (eg, OCZ GOLD VX PC4000)
-------------------------------------------------------------------------------
2-2-2-8 @3.3v @250Mhz <Very good
2-3-3-8 @2.6v @200Mhz
2-2-2-5 @3.0v-3.2v @200Mhz <Not so good

They don’t call it Voltage extreme for nothing.

Winbond UTT CH-5
- Requires 3.0v-3.2v @200Mhz, more than UTT BH-5
- When overclocking it needs less than UTT BH-5 to achieve the same results.
- Can maintain low timings at high frequencies
- Gets very hot, must be cooled

So like BH-5 it requires high voltages to run at higher HTT frequencies with no RAM divider (1:1) or to run at stock HTT (200MHz) with tight timings (2-2-2-5). It unfortunately is not quite as good and may not reach as high HTT frequencies as the BH-5 can with high volts (~v3.3). The good news is that you can do the simple vdimm mod (no vr, just solder a wire) to put v3.13-3.20 through your RAM for a moderate oc. You need to 100% sure you don't have TCCD RAM because this mod will kill your RAM;



Hopefully that gives you some idea of what to expect.

 

[RollingThunder]

CHino,

Some searching found this:

2G set= AK2/2G Infineon BE-5 (July 2006)

1G set UK2/1G = TCCD

I don't know whether you can compare (guess) chips in a 2G pair with a 1G pair. I have AK2/1G so I really don't know what I have......

 

[chinobino]

You have to be 100% sure which chips are used.

KHX3200AK2/1G = BH-5

KHX3200AK2/1GR = CH-5 (R = refurbished)

Apparently Kingston changed the chips it used from Winbond UTT (CH-5) to (cheaper) Nanya chips around November 2005. They confused their customers by advertising the Nanyo product with the same name (Hyper-X PC3200 2-3-2-6-1) but the cheaper chips they used didn't oc well at all, barely reaching 225MHz.

I am wondering what CPU-Z might tell you about them. Use CPU-Z v1.33 if you can and not v1.35 or v1.36 as they are really buggy and you might get something like this;



LMAO! DDR2 PC40000 2500MHz 768MB - Man these TCCD sticks are smokin'

 

[RollingThunder]

Chino,

Do these photos I posted tell you anything?

http://www.ocforums.com/showpost.php?p=4633232&postcount=8

 

[chinobino]

RT - good photo's. I think you should try going to the source.

Here is a link to get Technical support. Just enter the model number (KHX3200AK2/1G) from your photo's. They also have a registration page for owners of Kingston RAM here. You should request that they tell you what chips are on it as you need to know the voltage specifications. They should be able to tell you and if they can't then I'll never buy Kingtson RAM again! lol.

 

[RollingThunder]

RT - good photo's. I think you should try going to the source.

Here is a link to get Technical support. Just enter the model number (KHX3200AK2/1G) from your photo's. They also have a registration page for owners of Kingston RAM here. You should request that they tell you what chips are on it as you need to know the voltage specifications. They should be able to tell you and if they can't then I'll never buy Kingtson RAM again! lol.

Thanks Chino, done on both. I picked at the heat spreader retainer clips with a dental pick to pry them open. They are tighter than a bull's a$$ so I'll forget doing that.

 

[RollingThunder]

Kingston HyperX Tech Support Reply

Chino,

Got a reply from Kingston>
_________________________________________________

Hello,

These particular modules have Infineon C die. But not all of this part number will have the same chips. The chips used should not matter. The modules are tested to the speed and timings listed on our website no matter what chips are used.

If you have any other questions or concerns, please feel free to reply to this e-mail. Thank you for using Kingston on-line technical support.

Regards,

Glenn Neumeister
Kingston Technology
Technical Support
______________________________________________

From what I read here, they are not the best OC chips but probably fine for my wants and needs. I should have a spare set anyhow so I probably will get another pair of better ones.

 

[chinobino]

So, news to me, Kingston just buys whatever RAM chips are cheapest at the time, just so long as it will meet the specs - and then slaps their name on it and rate it. This inconsistency in production and lack of details for the consumer has put me off Kingston. On the Corsair forums a mod has listed all the Corsair products and specified which chips were used with each 'revision' of each product - a much more professional approach.

Anyways, from what I have been reading, Infineon die C RAM can oc quite nicely in the 220-250MHz range with quite reasonable timings. Apparently it prefers low voltages similar to the Samsung TCCD chips - about v2.6-2.9. This means you'll need to solder a variable or fixed resistor in place to get the required vdimm. This is definitely the safer approach as we do not know for sure what the chips are.

I wish they wouldn't attach the heatspreaders so well! It would be even better if companies would label their products correctly, including the chip manufacturer, so that consumers will know what they are buying is what they are actually want! I wonder how many people are using mixed RAM chips that are labelled as the same product...