PL-iP3/T Voltages

[thecondor]

I read the thread on the iP3/T with great interest. I have just received one of these and it is a poor overclocker at best. I can only boot into w2k at 103 fsb. At 105 it's bsod. This is a shame
since so much potential is there. So, I would like to help with knowledgable folks to get this straightened out for the OC community. I have the bare adapter installed and can reach the pins while it's running. So far I have measured 1.3 volts on the
vss-vcc pins and 1.45 volts on the vss-vtt pins. So where do we go from here? Ol'man, are you around?

[Pinky]

Yeah, I think Ol'man might be one of the members best equiped to field this. I'm replying to bump it back to the top of the list.

Welcome to the forums BTW :)

[ol' man]

Originally posted by thecondor
I read the thread on the iP3/T with great interest. I have just received one of these and it is a poor overclocker at best. I can only boot into w2k at 103 fsb. At 105 it's bsod. This is a shame
since so much potential is there. So, I would like to help with knowledgable folks to get this straightened out for the OC community. I have the bare adapter installed and can reach the pins while it's running. So far I have measured 1.3 volts on the
vss-vcc pins and 1.45 volts on the vss-vtt pins. So where do we go from here? Ol'man, are you around?

So the adapter is running the vtt out of spec? Well that may explain them wanting to run the vcore lower then so the heat would not increase. I am almost positive the vtt supplies the cache bus with its juice. At least it is not running the vtt at vcore but none the less it is still too high. There should be a thread around here about what one guy did to increase the vcores for that chip. It was wild but doable.

[thecondor]

I read through that thread and couldn't really follow what he was doing. He said he was going to document and post some pic's but that didn't happen. Did you catch completely what he was doing?

[ol' man]

Originally posted by thecondor
I read through that thread and couldn't really follow what he was doing. He said he was going to document and post some pic's but that didn't happen. Did you catch completely what he was doing?

He had to cut some traces and then add dip switches to get all the vcores cause at 1.32v it had to have all the vid pins at 1 I think. This woiuld make it impossible to use the wire trick to get any vcores higher than 1.32v.

[thecondor]

Well, he actually states later on after his dips are installed that he
has determined the stock voltage is 1.425. That would indicate all
pins are open except VID1. If that is the case, you now have something to work with, and the VID1 to VID2 wire wrap gives you 1.625 V. I'm going to wrap it and just see if that is correct.

[thecondor]

OK. Here's some facts. I went ahead and wire wrapped VID1 to VID2 to VID25m. This should equate to 1.60 V. On the first boot
I went for 112 FSB for 1344 Mhz on WCPUID,and it booted clean right in. Before mods it was 103 fsb max. I'm going to try and get a good Vcc reading, but I'm working with an el cheapo analog meter that I don't trust to much at this point.

[ol' man]

Originally posted by thecondor
OK. Here's some facts. I went ahead and wire wrapped VID1 to VID2 to VID25m. This should equate to 1.60 V. On the first boot
I went for 112 FSB for 1344 Mhz on WCPUID,and it booted clean right in. Before mods it was 103 fsb max. I'm going to try and get a good Vcc reading, but I'm working with an el cheapo analog meter that I don't trust to much at this point.

Well It seems some of the adapters are different. Youy may have one that doesn't have all the traces connected as I heard. It is just that at 1.425v you should be able to hit higher than 103fsb. My old PIII used to do 1.4GHz on 1.4v from the 1.2GHz @ default.

[thecondor]

I agree. There are some things here not making a lot of consistent sense. I'm going to continue to experiment and will post facts as I can verify them. My adapter may be different. The Mfg. date is 11/01, pretty recent. One thing that I don't like about
this adapter is there is no way to get a temp reading. My BIOS reports it as 210 deg( which is obviously wrong ) and the Powerleap utility is pretty has no temp monitor.

[doer]

check this mod by Alderan out:
http://koti.mbnet.fi/~alderan/mod/

It has not any processor pins cut but instead the VID traces on Powerleap adapter have been cut. (2nd picture)

I've followed that example and made me a similar dip-switch (6 dips, 5 used). And it's working fine. Now one can use all the V-cores from 1.05 to 1.825.

I even got Prime running with cel1.2@1.6Ghz,1.825V, 1h22min (after which I stopped it myself) case open, but case closed it crashed after 5min. With the normal voltage I couldn't even get my cpu booting @1.6.

[thecondor]

That looks great. I guess I don't understand if the vid traces are cut, how the signals get back to the processor. Then again, I don't need to understand if it works! What do you think about
ol'man's theory that the processor cache voltage is also being
supplied by the same Vcc voltage thus leading to premature
cache failure? AS soon as I get a better meter, I'm going to measure Vcc and Vtt. Right now with my wire mod, I'm booting real stable at 1393 Mhz( 116 fsb ). Haven't tried higher.

Great info.
Thanks

[ol' man]

Originally posted by doer
check this mod by Alderan out:
http://koti.mbnet.fi/~alderan/mod/

It has not any processor pins cut but instead the VID traces on Powerleap adapter have been cut. (2nd picture)

I've followed that example and made me a similar dip-switch (6 dips, 5 used). And it's working fine. Now one can use all the V-cores from 1.05 to 1.825.

I even got Prime running with cel1.2@1.6Ghz,1.825V, 1h22min (after which I stopped it myself) case open, but case closed it crashed after 5min. With the normal voltage I couldn't even get my cpu booting @1.6.

I know for a fact with these high of vcores that the chip itself is putting out a large amount of heat. If you do not have a decent HSF it will overheat every time. I know many here will dispute this but after three of these .13u chips I still stand by the theory that if the chip gets into the 50 deg. C area it will fail. This of course with everything maxed(ie. vcore and hurtz) Now if you were to take that same chip and run it back down to stock vcore hurtz it may be able to run hotter than 50 deg. C but I think in my experience anything much higher than 50 deg. C will cause chip failure too at most operating speeds. Personally I think 1.825v for the phillipine chip is too much but if you have success with it so be it. I think the adapters always run the chips a little under default vcore so it would probably be at like 1.775 to 1.8v but still that to me seems a little high. This cause some here have been hitting 1.6GHz on defualt vcore or a little above. This of course with the philli chips though. I have yet to see the malay's do that well.

[doer]

This is Malaysian chip... I ran it ~1603Mhz, 1.825V(?), 11h 55min Prime, case open. And after that Prime reported an error. :/

Case closed I managed to run Prime 49min. Now I'm just hoping that I'd have an unconnected wire on the dip-switch, cause it looks like it's difficult to get that chip running stable @1600.

At 1.725 I can't get Prime running even a minute case open. At 1.675 I couldn't get into Windows. I tried 1.55 just to test, it didn't boot at all @1600.

Cooling is Global Win WBK-38, 6800rpm, 37CFM and silver thermal paste.

Now I'm trying it @ 1.75V case open.. hmm.

[Ruiner]

Time for water cooling, dude!

[ol' man]

Originally posted by doer
This is Malaysian chip... I ran it ~1603Mhz, 1.825V(?), 11h 55min Prime, case open. And after that Prime reported an error. :/

Case closed I managed to run Prime 49min. Now I'm just hoping that I'd have an unconnected wire on the dip-switch, cause it looks like it's difficult to get that chip running stable @1600.

At 1.725 I can't get Prime running even a minute case open. At 1.675 I couldn't get into Windows. I tried 1.55 just to test, it didn't boot at all @1600.

Cooling is Global Win WBK-38, 6800rpm, 37CFM and silver thermal paste.

Now I'm trying it @ 1.75V case open.. hmm.

Sounds very very familiar. Those malay chips are different for sure. You never tried lower vcores before ramming the vcore to 1.825v??????

Why don't you give some more specs about your system. If you don't have 1/4 dividers that may pose somewhat of a problem.

[doer]

abit vt6x4, got 1/4pci-divider starting from 133Mhz, 250Wpsu, 384Mb, win2000

@1.6Ghz @ Voltage
1.55 No boot
1.60 boot, can't get into bios
1.675 problem getting into windows
1,725 windows ok, can't run Prime (crash after 30sec)
1,75 can run Prime sometime
1,775 Prime did still crash
1,825 Prime ran almost 12 hours, but then crashed

@1.2Ghz @ Voltage
1.325 just tested and gets fine in win... I tried fast Prime and it worked.

Global WIn WBK38... aluminum alloy, not good enough. I consider getting a heatsink made of copper. The 1,6Ghz isn't far away...

[Ruiner]

By the sound of those results, its definitely time to go water cooled!

[Lyron]

Yes, sorry about not documenting what I did...
I've got a lot going on right now... From taking the last 2 tests of the ged, having a family member have surgery, and now I have a really bad cold (got it from some jerk the ged testing "lab").

When am I going to get started?
I will try tonight...

[thecondor]

Lyron, some pics would be nice, but don't knock yourself out. Sounds like you have a handfull right now. My VID1, VID2, VID25m
wrap seems to be working fine. However, ol'man is right about the temperature. I had some spurious lockups yesterday and found the backside of the chip piping hot. Picked up a Thermaltake
Volcano and I have now been running solid at 124 fsb( 1489 mhz ) all day. Chip is barely warm to the touch.
We need to figure out how to get the internal thermistor to read right, and why Powerleap don't include temp reading on their CPU
utility is beyound me.

[Lyron]

Originally posted by thecondor
Lyron, some pics would be nice, but don't knock yourself out. Sounds like you have a handfull right now. My VID1, VID2, VID25m
wrap seems to be working fine. However, ol'man is right about the temperature. I had some spurious lockups yesterday and found the backside of the chip piping hot. Picked up a Thermaltake
Volcano and I have now been running solid at 124 fsb( 1489 mhz ) all day. Chip is barely warm to the touch.
We need to figure out how to get the internal thermistor to read right, and why Powerleap don't include temp reading on their CPU
utility is beyound me.

Pics you say?

http://www.speakeasy.org/~norton/all1.jpg
http://www.speakeasy.org/~norton/cutleads.jpg
http://www.speakeasy.org/~norton/pins3.jpg
http://www.speakeasy.org/~norton/vttreadout.jpg
http://www.speakeasy.org/~norton/voltagereg.jpg
http://www.speakeasy.org/~norton/dipswitches.jpg
http://www.speakeasy.org/~norton/pins2.jpg
http://www.speakeasy.org/~norton/pins1.jpg
http://www.speakeasy.org/~norton/installed.jpg

[mastsablaster]

Doer ,where did you get the dip switchs,i'm thinking of trying that out..looks pretty cool.:rolleyes:

[Lyron]

Originally posted by mastsablaster
Doer ,where did you get the dip switchs,i'm thinking of trying that out..looks pretty cool.:rolleyes:


Not that you asked me or anything.
I got mine off an old 386 server mobo.
You know the ones with the 12, 32pin simm slots? hehe

[doer]

mastsablaster, well I bought the dipswitch from a local electronic store, in USD it cost under 1 dollar. you of course need some wire too... rather very thin. I had about 30cm for each 6 connections.

I found it quite difficult to solder those wires on the Powerleap adapter, cause there is only about 1millimeter space between the pins. You better practise with the dip-switch soldering 1st, if you don't master soldering... In an ideal case, you could get a long enough wire to put that dipswitch outside of your computer case.

***

my stupid cele1200 CAN NOT do 1600 stable even case open. I got a new HSF of copper, but it didn't help it. I even put a temporal duct with a 9,2cm fan pushing air into the intake of my cpufan... but no use. DAMN. :/

Though I got it starting to load win2000 @ 1680+Mhz/140Mhz, but it froze at the start. (the f8-key pressing). It could be my memory or my raid-1 setup causing stop too.

BUT I'm sure that a new CeleTualatin 1.1 WILL do 1466 with no problem. Cause I now got a voltage-selectable adapter and much better case cooling. AND that cel1200 almost did 1600... and many have reported success under 1500Mhz...

[bearulez]

Check this place out for a semi-smooth mod on your adapter, I guess y'all know the dude already, but anyway....

http://koti.mbnet.fi/~alderan/mod/:D

[thecondor]

Unless you are experienced at soldering small circuits, I wouldn't risk runining the adapter. What I have actually done is wrap the VRM contacts just as you would processor Vid pins. You also have to run a small wire from the wrap to Vss(ground) and sever the 5 leads coming from the cpu socket.
Specifically, I wraped Vid0, Vid1, and Vid2. Vid3 and Vid25m stay open. On a high quality digital VOM, this gives me a Vcc reading
of 1.605 volts. This allows me to run 1.489 GHz(124 fsb) on a P3V4X and 1.501 GHz(125 fsb) on a P3C-D. I'm happy with this and stability is solid with air cooling. I also measured 1.278 Vcc
on the unmoded adapter and is probably why these adapters don't overclock worth a hoot stock. Vtt is reading 1.482.
My adapters use the Phillipine chip. These are great little setups after moding. These are 2 year old motherboards posting benches better that 1.6 P4's!

[ol' man]

Originally posted by thecondor
Unless you are experienced at soldering small circuits, I wouldn't risk runining the adapter. What I have actually done is wrap the VRM contacts just as you would processor Vid pins. You also have to run a small wire from the wrap to Vss(ground) and sever the 5 leads coming from the cpu socket.
Specifically, I wraped Vid0, Vid1, and Vid2. Vid3 and Vid25m stay open. On a high quality digital VOM, this gives me a Vcc reading
of 1.605 volts. This allows me to run 1.489 GHz(124 fsb) on a P3V4X and 1.501 GHz(125 fsb) on a P3C-D. I'm happy with this and stability is solid with air cooling. I also measured 1.278 Vcc
on the unmoded adapter and is probably why these adapters don't overclock worth a hoot stock. Vtt is reading 1.482.
My adapters use the Phillipine chip. These are great little setups after moding. These are 2 year old motherboards posting benches better that 1.6 P4's!

1.482 vtt is abit high and the access heat is most likely the reason they do not OC well on these adapters. A 1.1GHz chip would be better maybe! It seams to me that the vtt supplies the cache with its juice so overheating the mem may not be a good thing. Run prime to see what happens please.

[BlueThunder]

Can somebody make some Pics where I can see the FRONT and the BACK of the PL-iP3/T? I need two PICS:

One of the Back and one of the Front!

Thx!:cool: