• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

running Tualatin on CuMine MB w/o Powerleap

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
Re: Re: One last question...

jmsandrsn said:


I don't think it will mater.
However, it may not be a good assumption to assume that all slockets with that TVC16222 will work.

Agreed, also the other way around -- I think the Soltek and MSI parts don't use a TVC16222 and are modable.

BTW, I've taken a look at the data sheet and from what I can understand it's called a "Translation Voltage Clamp" (hence TVC16222).

I'm curious and would like to know how it is being used on the slotkets, so If anyone's got a scanner and some webspace, you could do me a big favour by scanning a slotket with that TVC and upload it somewhere. TIA!
 
Re: Re: Re: Re: There's a lot talking for this!

Celemine1Gig said:


Ebenfalls Grüße nach Nordrhein-Westfalen!

I'll keep you updated. The Slotket I'm going to try the mod with is a NoName Card. No manufacturer is listed on the package. It only says 370 PPGA/FC-PGA Card and FC-PGA Coppermine. As I already said, I'll do te mod for a friend of mine, but he doesn't have his new Tualatin Celeron CPU yet. I'll have to wait until he gets his new CPU and perhaps I'm able to get my hands on a Gigabyte Slotket until then.

Ah, thanks. Those no name things seem to be the only stuff more or less easy to find. Asuscom.de has some PPGA only S370 left and the prices on eBay are, well eBay.

Reichelt.de still have a Soyo card, but the part doesn't look more advanced than a generic slotket.
 
jmsandrsn said:


I believe it does but I'll have to double check when I get home.

I know that the Gigabyte slocket did not have this chip...it seems to have 2 smaller but similar chips on it.

The Asus S370-133 slocket that I have DOES have the TVC16222 chip on it. Too bad the mod does not work with it(assuming I had all of the pins adequately insulated). I believe they were insulated properly because I took the CPU right out of the Asus slocket and put it into the MSI slocket and it booted right up.

I just tried the slocket with a PPGA celeron and it booted right up so I at least know that there is nothing wrong with the slocket.
 
info

@ Brosken:
TVC16222 w/o A chip: you do not have to worry; it works (I have got my friends MSI one slotket with that chip on, but this now mine slotket is IMHO a bit damaged-sometimes work, sometimes not-was making problems even before! Yesterday my Tuallie-1.0A worked on it after a lot of trying for few hours @1,35G with ak4-ak26 bridged across 100 ohms resistor with no problems, but now I lapped the IHS & put it again with direct wire bridge in I have again problem not to work! Mine Cpu is not a good one & doesn´t like to go higher :-( even with higher voltage than 1,7V... I am trying to get an Asus slotket...

@ jmsanddrsn:
If your ASUS s370 slotket is a DL version should work if you DO NOT make ak4-ak26 bridge-there are reports that it does NOT need that bridge=does not boot with it!!!

@ DSTA:
that chip is 22bit bidirectional DataBus (agtl i/o) signal voltage mirror-translator (clamp) - I studied a little bit its datasheet... The other similar 48pin but smaller & only 16bit is IDT 74FST, which is found on a Tekram slotket, which reported also works perfectly. Still no other reports with other chips; my friend also tryed Abit Slotket!!! ver.0.83 with other same looking chip but other markings & did not work! I will try on this again myself, maybe will work somehow!

@ all:
looks still no "juice" with generic cheap slotkets.. :-(
 
Re: info

Spajky2 said:
@ jmsanddrsn:
If your ASUS s370 slotket is a DL version should work if you DO NOT make ak4-ak26 bridge-there are reports that it does NOT need that bridge=does not boot with it!!!

The Asus slocket I'm using is the S370-133 rather than S370-DL. It does have the TVC16222 chip but I can't get it to work with or without the AK4/AK26 connection.
 
Hey i just have an idea. Lets make list with modding results. It would be easier to other people to know if their mobo can work with Tualatin. Send me messages with :

1)your mobo(revision,bios version). if slot1 what slotket.
2)your cpu
3)what had you done
4)result(works/doesn't work, overclocking speed)
5)any other comments

I'll make the table and publish it.

my e-mail: [email protected]
 
I've tried the mod with a Celeron 1.0A, a P2B and a soltek sl02-a++. (This adapter doesn't have the tvc chip, it just has two lvc07a "hex buffer with open-drain output"). I just insulated the three pins (using insulation from a thin wire).
First it didn't boot at all, but after some resets, suddenly it did. And sometimes it still fails to post. But if it does boot, it is way too unstable - I could never load windows or linux, it always crashes (usually just reboots, sometimes freezes). Tried 1.3V, 1.5V, 1.7V and 666Mhz/1000Mhz/1333Mhz (not all combinations of course) - always the same. Works more stable under dos (played quake timedemo for about 10 min. before locking up).
Is there any chance the connection between ak4/ak26 will fix things (I just couldn't get that connection correctly using conductive silver paint, everything is so small!) or is the adapter just not going to cut it?
 
mczak said:
I've tried the mod with a Celeron 1.0A, a P2B and a soltek sl02-a++. (This adapter doesn't have the tvc chip, it just has two lvc07a "hex buffer with open-drain output"). I just insulated the three pins (using insulation from a thin wire).
First it didn't boot at all, but after some resets, suddenly it did. And sometimes it still fails to post. But if it does boot, it is way too unstable - I could never load windows or linux, it always crashes (usually just reboots, sometimes freezes). Tried 1.3V, 1.5V, 1.7V and 666Mhz/1000Mhz/1333Mhz (not all combinations of course) - always the same. Works more stable under dos (played quake timedemo for about 10 min. before locking up).
Is there any chance the connection between ak4/ak26 will fix things (I just couldn't get that connection correctly using conductive silver paint, everything is so small!) or is the adapter just not going to cut it?

i have the same problems before doing the ak4-ak26 bridge
i did it with silver paint and i'm glad with my 1000@1500 now ;)

i suggest to u to do the same thing, it's a good new that a Tually
works on a P2B without a modified BIOS!
 
@ jmsandrsn:
It looks that your Asus Slotket is DAMAGED or oxidated (symptom-probably works with coppermines but not a Tuallie mod)! Check all soldering points, connections etc..! Mine is a little bit too; but IMHO I solved the problem: I cleaned it very well with a brush & put it in a WD-40 "bath" :))), than dryed it & non-booting problem looks like resolved; mine runs perfectly now for few hours already; knock on wood...

@ mczak:
I did a wire bridge for ak4-ak26 on S370 socket: carefully with non-isolated thin wire plugged into ak26 hole & isolated ak4 hole traced ski slalom like & than covered with transparent sellotape on top side to isolate it to prevent shortcutting pins; but you better use isolated one wire...no need soldering, painting or breaking pins; the CPU stays still intacted & not voided warranty when you pull it out of socket..oh yeah, for those 3 isolated pins I widened holes in the socket with 1mm drill & isolated holes with thin plastic wire cover.. After using a sellotape you have to use a needlle to make holes again for the cpu pins at that area; do NOT try to force CPU in without doing that before, you can brake pins!
 
Spajky2 said:
@ jmsandrsn:
It looks that your Asus Slotket is DAMAGED or oxidated (symptom-probably works with coppermines but not a Tuallie mod)! Check all soldering points, connections etc..! Mine is a little bit too; but IMHO I solved the problem: I cleaned it very well with a brush & put it in a WD-40 "bath" :))), than dryed it & non-booting problem looks like resolved; mine runs perfectly now for few hours already; knock on wood...

Did you get it to work using the S370-DL or the S370-133. I actually tried it on another S370-133 and had no luck with this adaptor either.
 
Yes! it works stable now! Thanks for all your help.
I didn't connect ak4 and ak26 however, after studying the datasheets (especially the timing diagrams for the startup-sequence) I thought that connecting ak4 to aj5 (a vcc pin) might just work the same - and it did. This is far easier to do with conductive silver paint.
If you want to try this yourself, one warning however: Make really sure you have isolated ak4, if not you will short Vss to Vcc!
Now running at 1.5V, 1330Mhz, 512MB ram with 3 modules (one of them a pc100 - had to lower ram timings for this one to get it to work...).
I was really sceptical about this mod, especially since I have a bios which is known to have problems with faster cpus (cpus with a multiplier higher than 10 will not work, one of the reasons I got a 1.0A) and an almost noname slotket.
(btw I think the reason why you need to connect ak4 is because if you don't the cpu will not read / or will read at the improper time (before the signals have stabilized) the configuration of some configuration pins during reset - this isn't mentioned in the datasheet, but that's my theory. It could be wrong of course, just a speculation.)

mczak
 
(btw I think the reason why you need to connect ak4 is because if you don't the cpu will not read / or will read at the improper time (before the signals have stabilized) the configuration of some configuration pins during reset - this isn't mentioned in the datasheet, but that's my theory. It could be wrong of course, just a speculation.)

Super idea to try that!

I think the original thought was that a Tualatin would not boot at all if there's no connection to vtt_powergood, so the mod "pioneers" took something seemingly close in function and connected to that.

Turns out it does boot without AK4 connection though, but often at the penalty of being unstable -- so I'm not sure there's a problem reading the config pins. I'd figure it would not boot at all or at the wrong VCore or whatever if there was a problem there. BTW there is a part in the spec sheet saying it's there to make sure the VID signals are ok and ready to sample -- question is why does the CPU have to know (by means of the input pin vtt_pwrgood), since the VID signals are sampled by the VRM.

Can't recall where I read that, but I think the PWRGOOD signal on the CPU (AK26) is there for the CPU itsell to know about VCC being stable (without spikes, swings etc) enough to turn some CPU internal stuff on. Perhaps this is the same for vtt_pwrgood -- and that's not mentioned in the datasheet...

@Spajky2

Thanks for the input!

I'm wondering what they do exactly with these TVCs since it seems to have only room for 22 "bits" or data lines (so it's not enough to host all AGTL I/O pins). On a PPGA / FCPGA adapter that could just mean they put the few signals that changed going from PPGA to FCPGA thru the TVC -- but Powerleap seem to use the thing as well? Can you check to what signals the TVC is actually connected?

Perhaps the problem with some slotkets (including the generics) is something else. Here's a bit I that I found in "Intel® 815E Chipset Platform
For Use with Universal Socket 370 Design Guide " (Intel doc # 298350-001":

"Processor pin AG1 requires additional attention since it is a ground pin on a Pentium III processor
(CPUID=068xh) / Celeron processor (CPUID=068xh) and a VTT pin on a future 0.13 micron
socket 370 processor. A separate switch controlled by the TUAL5 reference schematic signal
determines whether pin AG1 is pulled to ground or VTT. Refer to Figure 11 for an example
implementation. "

That's only one of the differences. There are a few more of this kind, and perhaps one of them is the reason for the "no show" condition on some slotkets.
 
@ jmsandrsn:
I am still searching for a propper Asus slotket...

@ mczak:
...an almost noname slotket ???? brand, model, revision & what kind of biggest chip is on board ? Please!!!
about your new discovery: bridge ak4 isolated to any VCC pin: do it thrue resistor, especially if highering Vcore to prevent that input pin to kill it inside; it is NOT much more than 1,5V tolerant !!! according to datasheets! You could end with non functional CPU if that signal in line dies!!!

@ DSTA:
>Processor pin AG1 requires additional attention ..
Doesn´t matter, according to PIII datasheet/p.79 it is used only for backward compatibility /ppga´s/!

IMHO chip TVC is actualy connected to CPU i/o (agtl) buses; it is bidirectional voltage mirroring/translating Cpu in/out signals (example level 0-1,5V translates to chipsets 0-3,3V (or 2,5V) & viceversa! For old ppga IMHO even to TTL levels (5V). Bit rate here in bites defines only how fine in steps is set this mirrored voltage to outputs.Tekram slot adapter uses 16-bit voltage clamp chip instead & reported works perfectly! IMHO would be enough even 12 or 10 bits AD conversion. Have in mind that this chips are ultraHigh speed A-D-A converters; for this purpose that are used, you can not simply use simple analog ones, because of strong spurious signals there & digitals are quite imune to that! & yes PLs370-2-fcpga2 uses the same chip according to picture I found of it... :)

The datasheet for newer chipsets design for s370 differs a bit to prevent easily running Tuallies on non-compatible boards-just to Intels sake to earn more money!

@ ALL:
there are NO diferences according to datasheets between CellyTuallies (or PIII-D now renamed & cripled L2 latency) & PIII-S pins; this refers about my very first post on this forum: does anybody thinking to mod 2 slotkets & 2 Tuallies & try them on doubleCPU Slot1 board if someone has it, just to TRY IF it works SMP; IMHO i believe it SHOULD (even if Intel /normally/ says it won´t!) I think it is just another just Bios issue /like AMD XP story/ ! Maybe some additional wire on slotkets like for PPGA Cellys mod for SMP! Guys, think also about that too...
 
@Spajky2
> ...an almost noname slotket ???? brand, model, revision & what kind of biggest chip is on board ? Please!!!

I posted it some posts earlier - soltek sl02-a++, it has two lvc07a hex buffers.

> about your new discovery: bridge ak4 isolated to any VCC pin: do it thrue resistor, especially if highering Vcore to prevent that input pin to kill it inside; it is NOT much more than 1,5V tolerant !!! according to datasheets! You could end with non functional CPU if that signal in line dies!!!

I don't think so. It's true this signal should be only 1.25V, but if you connect it to pwrgood, it will be even 1.8V! And since I have absolutely no idea how much current will flow through that Vtt_pwrgood pin, it would be difficult to calculate a reasonable resistor value. The reason I used a Vcc pin is I could just connect two adjacent pins with conductive silver paint which is very easy to do - using a resistor will be very complicated compared to that. Since I have only 1.5V Vcc, my solution shouldn't do much harm (and if you use 1.8V ore more Vcc, you will kill your cpu anyway in the long run).

btw about dual tualatin celerons - I'm sure it won't work. I've seen tests (sorry, can't remember the URL) with two normal (non-S) PIII Tualatins, it simply didn't work - same as with the Celerons FCPGA, some pins are just not connected internally. Of course you could still try it with Celerons, but why should intel enable it with those if they don't even enable it with the regular PIII?

mczak
 
@ mczak:
Sorry about that stupid req : Soyo slotket is just not an almost noname slotket! :)

about dual tualatin celerons :
you have seen tests on Tualatin compatible mobos /they have already bios block not too run SMP anything than PIII-S, imposed by Intel/ & maybe in reality smp capability was not really disabled in a hurry /strong AMD Duron competition!/. I have checked with multimeter a bunch of critical pins on my Tuallie; there are not non-connected pins except al1, which is also on PIII-S marked as reserved, while on coppermines is for GND-no need isolating it so..No other differences probably exept L2 size & latency!
But non-Tually boards Bioses normally recognize CellyTuallie as PIII Cu-Mine (like mine MoBo), IMHO with maybe some additional rewiring we could trick the mobo to run it so in SMP. Here were some rumours aboth that: http://forums.2cpu.com/showthread.php?threadid=16028&perpage=20&pagenumber=2 or here: http://www.vp6-board.com/cgi-bin/ubbcgi/ultimatebb.cgi?ubb=get_topic&f=5&t=000217&p=
or here: http://forum.oc-forums.com/vb/showthread.php?threadid=76926
For about FcPga Celly is your statement true, because there are a bunch of differences regarding normal PIII Cu-mines! (not the case for any Tualatins!!!) If I would have a dualCpu slot1 mobo & 2 moded slotkets working & 2 Tuallies I would first simply try on slotkets this mod first: http://www.hardware-one.com/reviews.asp?aid=92&page=4 , maybe would even work straight; if not maybe some additional rewiring? The problem is that none of that I have.. :-( & buying just for that material ex. VP6 just to try is nonsense & much complicated than just modifying slotkets...Just still nobody tried it on some BX dual mobo! /no bios block!!! here/
Thats only my thoughts, IMHO I could be even right about that issue but if Intel statement is true about SMP /could be one of their lies too, not to spoil their server market profits/, I could also be wrong & declared "Only a Dreamer".. :) ..But IMHO there is still a bit of chance that I am even right! :) ..
 
1.0A at SMP

Spajky2:

I did get my 2. celeron back from rma and have been playing with them at Asus P2B-DS. I have tested them with various slotkets and no luck to get them to work in smp. It seems that BR1 pin is missing (or is located somewhere else than old celeron or P3). They work fine at single configuration with modded slotket!!! or PL-iP3/T. With 2 tualatin I don't get any post at all, if I try tualatin and P3 side by side then machine does post and freezes where 2. cpu shoud be found. I'll continue to work with them.
 
Re: 1.0A at SMP

Jarmo said:
Spajky2:

I did get my 2. celeron back from rma and have been playing with them at Asus P2B-DS. I have tested them with various slotkets and no luck to get them to work in smp. It seems that BR1 pin is missing (or is located somewhere else than old celeron or P3). They work fine at single configuration with modded slotket!!! or PL-iP3/T. With 2 tualatin I don't get any post at all, if I try tualatin and P3 side by side then machine does post and freezes where 2. cpu shoud be found. I'll continue to work with them.

Is there a way you can test that your slotkets work at all in SMP? That is, do you have two P3 coppermines and a pair of slotkets that are proven to work in your P2B-DS? Don't know if you are aware, but depending on slotket BR1 (pin X2) can be mapped to the strangest of places.

One more question :) : in your Coppermine P3/Tualatin experiment, did you try swapping the CPUs in the mobo slots? Perhaps P3 Coppermine was sitting in CPU0 slot.
 
Back