Price? I doubt it would worth, even if cheap the shipping cost to my country makes it inattractive, esp. from US. The whole laptop wreck costed me 4euro on local flea market 😀

RayeR wrote on 2023-10-30, 15:55:

Price? I doubt it would worth, even if cheap the shipping cost to my country makes it inattractive, esp. from US. The whole laptop wreck costed me 4euro on local flea market 😀

Eh like 30$ last I looked. The 233 and 266 aren’t hard to find

not acceptable...

BTW I just got new x-ray images without IHS. Now there's more to see. There are 3 suspicios spots that shouldn't be there, I think that are leaked solder blobs. I must grind inside. There are no any copper traces in this area. But it would be need done very carefully to not grind CPU foil.

Following with interest.

Wow, it’s always nice to have the right tools!

I grinded off the 1st hole. And sure, blobs are there where xrays indicated...

RayeR wrote on 2023-11-13, 23:04:

I grinded off the 1st hole. And sure, blobs are there where xrays indicated...

Damn. I take it this was a irreversible invasion? Or do you think you can get some wick in there and still use it afterward?

No traces or important floods in that spot?

Thankfully there are no traces in this area. So I just crippled power and gnd planes a bit but it should not have real impact on function...
I managed to clean off the solder blobs and all 3 known shorts on address pins and VCC, VCC2 to GND are gone. I have suspection there still may be at least one small blob on next side. Maybe I should drill inside and inspect this. If everything will be OK I can fill the holes with epoxy and solder IHS back on (yes, upside down to not allow leak the solder inside again...)

RayeR wrote on 2023-11-14, 18:10:

Thankfully there are no traces in this area. So I just crippled power and gnd planes a bit but it should not have real impact on function...
I managed to clean off the solder blobs and all 3 known shorts on address pins and VCC, VCC2 to GND are gone. I have suspection there still may be at least one small blob on next side. Maybe I should drill inside and inspect this. If everything will be OK I can fill the holes with epoxy and solder IHS back on (yes, upside down to not allow leak the solder inside again...)

That's impressive. You should be a surgeon.

Hehe, sometimes I though about it, how it would be different from messing with microcircuits. This is a bit easier as the object is still/fixed and there's no time pressure, I can continue next day or a month later. And no repsonsibility for someone else life, respect to surgeons... 😀

I did through measurement and no other short was revealed so maybe I could try if it will POST. I have handy PC chips M590 that should have decent Vcore support - 2,2V. I should add some bridges for proper voltage detection...

OK, the 1st POST on M590, here we go! 😀
I set CPU settings to manual, VCORE = 2,0V but BIOS still complains 'F1 to set CPU type' every boot. I flashed before the latest BIOS I found on Retroweb but it doesn't know about Tillamook. Latet I could experiment with other S7 MBs but I don't have many that allows Vcore under 2,8V. As the central pad above CPU core gets hot quite soon I had to shutdown within a minute...

Woohoo! Confratts!

RayeR wrote on 2023-11-15, 02:23:

OK, the 1st POST on M590, here we go! 😀
I set CPU settings to manual, VCORE = 2,0V but BIOS still complains 'F1 to set CPU type' every boot. I flashed before the latest BIOS I found on Retroweb but it doesn't know about Tillamook. Latet I could experiment with other S7 MBs but I don't have many that allows Vcore under 2,8V. As the central pad above CPU core gets hot quite soon I had to shutdown within a minute...

If your cpu is getting very hot it may not be running at 2.0v or may be running very fast.

I have ran tillamooks without a heatsink before! And they barely got hot to the human touch at 2.0v 133mhz if I recall. (Running windows)

Bummer it doesn’t have tillamook support. It may still run though.

Get out your probes and verify that vcc2 is at 2.0v

After solving your thermals, see if l2 cache is working also. You may need to do some tillamook mods

Yes I will check the real voltage. But VIO will be still overvolted at 3.3V on most MBs. I also found some newer BIOS from 1999 modded by Jan S. for K6-III support so maybe it will goes better. I should also have some other SuperS7 MB from Gigabyte somewhere...
Currently I added 2 bridges at VCC2DET and VCC2H/L so BIOS should detect is as low voltage K6 but BIOS voltage should be overriden by manual setting.

Hell yeah, you did it. That's a pretty massive success in my book.

OMFG! This stupid motherboard PCshit M590 is frying my CPU with VIO at 3,86V! (so no wonder that CPU heated more than usual) Measured Vcore is OK and match the value set in SETUP but VIO cannot be changed neither in SETUP nor jumper. I belive they did it because that crappy chipset is overvolted to be able to run overclocked at 100MHz FSB. I found it;s powered out of some MOSFET with some other circuits acting as linear regulator. Maybe the voltage can be tuned by some resistor...

EDIT: the VIO regulator is simple circuit made of power N-FET with drain to 5V, source to VIO, gate is connected to collector of bipolar transistor with emitor to GND, base to GND via 10k, collector to 12V via 1k. It probably act like a zener diode with non-destructive avalanche of B-C junction where is Vdrop about 6,1V that opens the MOSFET. I'm thinking about replacing the FET by some normal LDO 3,3V like AZ1085-33 in TO252 package...

M592 is a bit better, VIO hardwired to 3,42V, Vcore_min = 2,2V, doesn't recognize Tillamook too.

Yup, had a feeling haha. Your report of getting hot was suspicious 😉

But I am surprised that vcc3 was what you found high, as it’s vcc2 that causes most of the heat I believe.

As for vcc2 (vcore)

2.2 is a little highr than I’d prefer, but Prob be fine.

Personally, I wouldn’t give these anything you wouldn’t give a k6-3+ voltage wise.

1.6-1.9 for undervolting experiments
1.9-2.1 is standard tinkering territory
2.2 is a reasonable top end,
2.3 if you are nutty, and don’t care
2.4 if you like to watch the world burn and wish to commit crimes against retro.

intel specifies 1.8V for core, 2.5V VIO for operation and abs. max ratings 2.8V for core and 3.2V for VIO so VIO is the one that exceeds abs. max. I tried it in another 2 MBs - M592 used VIO 3,4V and Gigabyte GA-5AA used 3,5V (cannot be easily changed on both). And it also gets hot soon, not much difference. There'se only little metal mass-thermal capacity...
Gigabyte also displays CPU as unknown but don't bother with stupid message every POST. I wonder what BIOSes and MBs use people whose got Tillamook detected as "Tillamook" - some custom tuned BIOS?

Also I milled second hole to check another suspicious spot on x-ray image. I found a small solder blob there but fortunately it doesn't create a short it just sits on junction of 3 traces. I hope there aro no more moving blobs. I will put it in ultrasonic cleaner to wash dust out and then finally do epoxy filling...

I know the datasheet says vio 2.5v but I really think that spec has more to do with its intended use at time of release, in laptops utilizing that voltage not an actual hardware limitation.

I don’t have any evidence other than I have never heard of a single tillamook not being perfectly happy on 3.3v. Vio

Every other s5/7 cpu is perfectly happy with it also, even late model amd, cyrix I have even encountered desktop boards that seemed to be prepared in part for tillamook spec some have bf2 intel routed, though not pulled properly sometimes. It’s no stretch that it would have been partially compatible with desktop even if it ended up only being released for mobile officially, unofficially it may have been considered/planned for both.

It just seems like the datasheet is more referencing the intended system than a actual limit of the cpu its self on that particular detail. Your 3.8v adventure seems to support this theory also.

Anyway, I’m pretty sure, intended or not, that the tillamook is 3.3v tolerant on vcc3

I belive that parameters titled as "absolute maximum' are not intended to violate. Only an intel engineer could prove it. Also it's manufactured with less um litography than std. MMX. That you dodn't see failed Tillamook i a desktop MB doesn't mean it's not degrading internal structures. Is there anybody who runs it everyday hours for 10 years or so? I doubt. Even I don't plan to use it everyday but I feel better to run it closer to specs. Other desktop CPUs was made for VIO 3,3V so they don't have problem with 3,8V but I definitely not allow to fry my Tilly this way in crappy MB with stupid regulator. I rather continue with GA-5AA and I will try to tune VIO down back to at least 3,3V if possible (by some resistor). This MB has FSB up 140MHz thats pretty high, I think I would be happy with 100MHz...

Yeah, for sure get it down to at least 3.3v 😀