Need P2B-S mod help

When I first built my current main PC, I intended on NOT bein
obsolete in two years. All the best components. I must say, thi
machine has done much better than anticipated. Seven years, averagin
14 hours of use. Three keyboards, two track balls, two monitors, tw
chairs, two tables and two houses. Amazing

As much as I could glorify Asus products, the time has come to upgrad
my beast. I thought I was going to build a new PC, until I came acros
Slot-T. Oh joy of joys! I get to keep my baby

I had wanted to use a Pentium 1.4, but was talked out of it by th
Slot-T rep. The rep said I might or might not get it to work. Well,
want it to work, so, I got a 1.3 Celeron (because it's 3/5 to cost o
a 1.4)

Then, I came across the pages of P2B's site that details modifying th
P2B-DS board with an ICS 9250CF-08

Ok, I've got a 9150AF-08..

I did some more searching, but apparently this clockgen is so far ou
of date, even google can't reveal anything usefull about it. And it'
not in Allied's Catalog.

Then I read that adding the missing jumper header and resistor onl
gives you a paltry 107Mhz. Hey, there's already a setting fo
that..

But then, on another of P2B's pages, where he tests various cpu's
it's mentioned that a P2B-S board is run at 150Mhz, but no detail i
provided on how this was accomplished

So................................

Mr P2B... how did you get the P2B-S to run at 150Mhz FSB? Solder in
different clock gen? Find a magic pin setting? Place a magnet on th
FSB

Please enlighten me as to the way to get the FSB to become mor
compatible with the chip I'd really like to use, a P3-S. I'll tak
the Celeron, over my SL3CC, but there's a reason the machine sports
Katmai and not a Mendocino

I thank you in advance

David -AKA- Senior Flamingtac

 

You can underclock the Tualatin. (I think I have tried 75MHz [66MHz FSB
is not working], me it is working at 103MHz FSB (3MHZ OC - 82443),
gives fantastic 1083MHz hypercached i686)



Best Regards,

Daniel Mandic

 

A Celeron 1.4 works just as well as a 1.3, if you have the latest 1014
beta3 bios (otherwise the 1.4 will not boot at all IIRC, not sure about
the 1.3, older bios have a bug with high multipliers). In any case
though it's not really a big win, those chips are usually highly
bandwidth-starved due to the slow fsb (so, a celeron 1.0a overclocked to
1.33Ghz will be quite a bit faster than a celeron 1.4 unless your
workload fits into cache like with synthetic cpu benchmarks), that 8%
clock speed advantage doesn't really translate into that much of
real-world performance advantage. Tualatin P3 as opposed to Tualatin
Celeron doesn't really provide any benefits, other than the (quite big)
advantage due to higher fsb, which you can easily get with the
lower-clocked celerons by overclocking (1.0A and 1.1 sould overclock to
133Mhz fsb). P3-S has two times the cache, which is nice but may be
overpriced for the modest performance advantage (for single core systems
that is). Since you can't get 133Mhz FSB this is all only interesting in
theory only though anyway...

I think I have seen a datasheet a long time ago but couldn't find it
neither recently.

Well, new revistion P2B-S/LS/L boards (those are all the same pcb) use
newer clock chips, as well as vrm which go down to 1.3V (which you
really want for using slot-t + tualatin).
Board rev. 1.04 pcba D02 and newer are guaranteed to have new vrm (there
are apparently even 1.03 boards out with new vrm), and for the new clock
chip you need even a newer rev. (unless you really want to try your
soldering skills...)
(shameless plug
http://homepage.hispeed.ch/rscheidegger/p2b_procupgrade_faq.html)

Roland

 

I started by buying a lot of new-old-stock P2B-S revision 1.04
motherboards, some Tualatin Celeron 1A processors, Corsair PC150 SDRAM,
and Slot-T adapters.

The P2B-S 1.04 has a Tualatin-capable voltage regulator, an ICS9250-08
clock chip, and vacant spots on the board for a 4th FSB jumper. After
soldering in the 4th jumper, the 150Mhz FSB setting is available.

My success rate at 150Mhz was 50%. The Celeron 1A processors all run
stable at 1500Mhz on default voltage, but only half the P2B-S 1.04
boards are completely stable at 150Mhz - the others can only manage 140Mhz.

I have successfully transplanted ICS9250-08 clock chips from dead
motherboards onto P2B-DS boards which originally had 9150s, but have
never tried it on a P2B-S - should be possible, but perhaps not worth
the trouble given the result may not be stable at 150Mhz.

P2B

 

I'd take 133 if I can get it. Anything to feed the CPU a littl
better.

Do you have any boards for sale? Or chips? Would I need a VRM as well
or does the slot-t card handle that

Also, have you looked for better cooling options, or are the stoc
HSF's up to the task? I'm considering fabricating a clip to attac
something like this

http://www.frozencpu.com/images/products/detail_secondary_hires/ex-blc-212_3.jp

or this

http://www.scythe-usa.com/product/cpu/009/scktncu1000_detail.htm

I also think this might fit without impinging upon the first RAM slot
I'm going to measure shortly. Won't do much for overclocking, though

http://www.pcabusers.com/reviews/zalman/p1.htm

Thanks
Davi

 

That would require changing the clock chip - not a trivial task,
especially since new ones aren't available so you'd have to transplant.

Your Celeron 1.3 is unlikely to be stable on 133Mhz FSB anyway - most
Tualatin cores become unstable between 1.5 and 1.6Ghz, very few can
manage 1.73Ghz. The safe choice for 133Mhz FSB is a Celeron 1.1, while a
1.2 carries some risk but usually works.

I have one brand new P2B-S 1.04 on the shelf, but it's the spare for my
file server - not for sale, sorry. There's a couple of known-good
9250-08 clock chips on otherwise dead boards, but it would be cheaper
and easier to buy a P2B-S 1.04 on eBay than to buy a used chip and
install it.

If you went the eBay route, you could have it shipped to me to add the
4th FSB jumper on it's way to you.

Slot-T does not have a VRM. You have not posted your board revision or
VRM chip details, so I don't know if you need one. If it's rev 1.03 it
should have a Tualatin-capable VRM, otherwise not.

Tualatin processors run nice and cool. The stock HSFs are overkill at
stock speeds and perfectly adequate for overclocking since Tualatins
usually don't respond to Vcore increases. There may be cases where a
Tualatin overclock is *almost* stable and would benefit from improved
cooling, but I have not encountered one.

P2B

 

FYI datasheets for most (all?) ICS clock generators are still available
here:

http://www.icst.com/datasheets/

Nice of them

P2B

 

Considering how much ICS charges to do a clock chip design, posting their
datasheets is the least they could do...though I'd trade that for them getting
it right on the first pass...

 

Yeah, upgrading older P2B-series boards would be a lot simpler if the
9150-08 133Mhz setting actually worked, instead of producing an ugly
signal the BX interprets as 109Mhz.

Maybe next time you are beating on them about the cost of deficient
designs, you could suggest they atone for past sins by making 9250-08
samples available to those of us who still care...

P2B

 

.. it would be cheaper and easier to buy a P2B-S 1.04 on eBay tha
to buy a used chip and

The question is... how long until a 1.04 shows up? There were 5 las
night, all were 1.03. I am truly impatient ;

Board work is no problem for me since Uncle Sam was generous enough t
train me to on multi-layer board repair. My dis-advantage is a lack o
experience with home PC mobo's, and no schematics :

or VRM chip details, so I don't know if you need one. If it's re
1.03 it should have a Tualatin-capable VRM, otherwise not
I'm sorry, it's a 1.03 board. I thought you'd know what revision sinc
I'd mentioned the 9150AF-08 clockgen

Yes, no VRM, but my card (upgradeware rev 1.1) does have a voltag
clamp, I found the specs via digikey..

http://rocky.digikey.com/WebLib/Texas Instruments/Web data/SN74TVC16222A Series.pd

...and presumed it's purpose was to bring the voltage down to th
range needed by the tualatin chips (admittedly, I've poured over th
board but not the information that came with it ~ for shame)

Excellent! Now that I've spent many hours browsing o/c sites for hsf'
and planned for the worst (pocketbook wise), it's good to know that
can just slap a small one on there and call it a day. I ordered a
oem chip with no sink as I've planned to o/c... oh well

So... The upgradeware rev 1.1 with tualatin 1.3 celeron should b
plug-and play (unless the vc on the board has a different use...).
need beta bios 1014 from asus, and I need to do a code patch as well
Sound correct

I thank you for the wealth of information you've provided me, BTW

 

Another item while I've got people's attention..

Anyone have an idea of what generation are the best video cards fo
these boards? So far I've worked my way up to an nVidea MX serie
card (second generation GPU?). It's been a long while since I'v
looked at the cards.

I don't want to pay an extra $100 for that last 1/10th of a percent o
performance (that kind of spending is for the car!), but if my curren
card would be significantly dusted by a newer generation with th
tualatin chip and limitations of the P2B AGP (2x?), I'm good for a
upgrade. The computer sees some serious gaming periodically (Falco
4.0 - has two "game engines")

Thanks

 

As I've already told you, some 1.03 (and older) boards could have an old
vrm. You should just look at the chip...

NO. The clamp is there for some i/o signals (from the top of my head,
some of these are 2.5V with the old spec (for old p2), and 1.8V (or was
that 1.5V?) with the new spec (coppermines/tualatins). It neither
corrects the bus signal voltage (just runs out of spec, no problems),
nor the vcore - and if you really have an old vrm (unlikely), you'd need
to set the slot-t to 1.8V to work at all (not recommended for tualatins).

Roland

 

9150AF-08 only tells us it's *not* revision 1.04 D03. I don't know which
revision 1.03 PCBA level introduced the new VRM, you'll have to check
the chip as Roland said.

The TVC16222A voltage clamp is used to meet the Coppermine/Tualatin CMOS
I/O signal specification of 1.5V vs. 2.5V used by P2 processors. The
Slot-T does nothing to address AGTL voltages (Tualatin 1.25V,
P2/Coppermine 1.5V) or Vcore voltages. AGTL voltages can be adjusted on
the motherboard as described here:

http://tipperlinne.com/p2b-dsvtt.htm

.... but most people don't bother since the discrepancy doesn't appear to
cause problems. Tualatin processors do not last long on 1.8V Vcore
(minimum available from old VRMs).

Only *if* your board has the new VRM - check the chip!

I can email you the patched BIOS.

P2B

 

Ok, where is the VRM? I see two power fet's close to the slot marke
32N03. Sorry for the short post, got to run

 

Between the USB/PS2 connectors and the CPU slot, adjacent to the PWR_FAN
header. I expect either:

HIP6004ACB :-( (not Tualatin capable)
HIP6004BCB
HIP6004CB

P2B

 

Jeez, this is going to be my FIFTH time attempting to get this posted
Fisrt four three had a screwy keyboard problem, then I made th
mistake of allowing the site to time me out

I confused the VRM and clockgen. I've got an HIP6004BCB. Core voltag
adjusts to 1.3, correct? What's a good starting voltage for a 1.
celery

Any ideas on doing the agtl/agtl+ compromise like you did with th
6019 on your dual

The slot-t board has fsb adjustment. Does it take over the mob
control? I'd like to get the 112 the mobo allows if I cannot modif
the board for the faster clock rates

All the celery tualatin images I see online show the chip with no hea
speader. Did they have a production change at some point?

I need to know the thread pitch and size for the cpu retentio
mechnism studs in the asus boards. I did not get one originally, bu
need to add it

I ordered the AMP universal SEC/SECC/SECC2 retainers. They do not fi
the slot-t card! What gives there? The card looks like it fit'
intel's SECC only mechanism

http://support.intel.com/support/processors/sb/CS-017374.ht

but these are supposed to fit the SECC, too. Here they are

http://www.pccables.com/07054.ht

Am I supposed to remove the top-half of the connector and let th
pressure from the "sprung" piece in the side hold it in
Upradeware's card won't slide all the way to the board with these
either, unless I modify (the base fits snugly around the end of th
board connector)

What craziness. Almost there... gotta go order that heatsink now

 

1.5v - as per the Slot-T instructions. Tualatin processors usually don't
overclock any better as a result of Vcore increases, so I usually test
the system to ensure stability at default voltage, then try lowering
Vcore. Most Tualatins run perfectly on 1.4v, some as low as 1.35v.

I have not done this on a P2B-S, so I don't know which resistor needs to
be changed - but it should be fairly easy to figure out by probing the
board with a DMM while referring to the application circuit in the VRM
datasheet. The formulas for calculating resistor values are also in the
datasheet:

http://www.intersil.com/data/fn/fn4567.pdf

The FSB jumpers on the Slot-T are for overriding the CPU's FSB request.
These jumpers have no effect on P2B-series boards because they ignore
the CPU and use only the FSB jumpers on the motherboard.

Example? I've never seen a Tualatin processor without a heat spreader,
unless it's been removed.

Can't help there.

I've had problems with so-called Universal Retention Mechanisms too. I
find that non-folding types usually work better with the Slot-T, and the
older one-piece types that wrap around the entire CPU slot are even better.

Getting close....

P2B

 

On a whim, I decided to pull out the box the P2B-S came in to see i
it had any nuts for the board. The box was empty, but while I wa
there, I went through some other small boxes looking for the bag o
extra screws that came with the tower. In one box, I found a complet
Asus P2B cpu adaptor kit. Holy smokes. I thought I had not receive
them with the board. I guess I missed it in my rush to assemble m
new PC in '98. I was very close to bidding on one of the two correc
adaptors on eBay today. How cool is that

Anyone need some "universal" connectors

 

I'm following your plans for adding a thermister to the heatspreader
just placed an order with digikey. However, I'm thinking abou
drilling a small hole right in the center of the heatsink, allowin
me to place the thermister in the sink over the center of the cor
(drilled from the top) for a more accurate reading of the temp of th
core heat spreader. Do you think there might be any issues with this
Also, any recommendations for a compound that would be well-suite
for sealing in the thermister, would conduct heat well, but no
electricity

 

Hi P2B!


It should work quite well with most of the PCI-Cards. 36 and a third
MHz. My most sensitive PCI Card (AHA 2940) ran at 37,5MHz (ASUS T2P4,
P54C200). Over that, it did´nt found any SCSI devices ...




Best Regards,

Daniel Mandic