Chkcpu wrote on 2024-01-20, 10:23:

I will keep my fingers crossed! 😉

Jan

It seems that helped 😉

anyway, Is this one of those Bios's that update the ESCD or something in the flash chip on each hardware change?

Just pondering what could be the cause of a broken bios aside from a failed update and a failing chip.

Retronaut wrote on 2024-01-20, 13:51:

So I plugged in my VGA monitor and turned it on, and
It BOOTS!

Im not quite out of the woods yet, its moaning about a battery not being plugged in, even though there IS one, and I tested it and it has 3v, so thats a bit of a worry. Also, I guess I should now try and installed the 2nd .exe based BIOS, as that was rev K, so I am assuming its a newer BIOS, but I think first, Ill make sure other things are working before getting that done.

Great ! the K bios is internally dated 980309, where the F bios is 961202, so over a year newer..

PC Hoarder Patrol wrote on 2024-01-20, 13:31:

You've got something now, but the original PDF is available on the FSC archive...

Ergo_e_x.pdf

Great ! I kept getting "can’t establish a connection to the server at web.archive.org." yesterday

I would assume it writes ESCD (the Award and Pheonix bios does) on a VX board at every boot AFAIK

@Retronaut: You brought the board back to life!! 😀
I admit I was sceptical about the BIOS being the issue, but I'm glad the BIOS re-load did the trick!
But I have no idea what could have caused this BIOS corruption...

weedeewee wrote on 2024-01-20, 14:09:

anyway, Is this one of those Bios's that update the ESCD or something in the flash chip on each hardware change?
Just pondering what could be the cause of a broken bios aside from a failed update and a failing chip.

The nice thing about 86Box is that, if the virtual machine uses a flashable BIOS, it makes a copy of the running BIOS in an "nvr" folder so the ESCD block can be updated in this BIOS copy and the original BIOS file in the machines folder remains intact.
When next booting this machine, 86Box will always use the updated BIOS copy with the latest ESCD data, just as on a real machine.

This feature allows us to compare the updated BIOS copy with the original one. Usually this shows us a difference because of an updated ESCD block.
On the x453 version K BIOS however, I don’t see any difference when performing this compare. This means the ESCD block is not stored in the 28F002T BIOS flashchip but somewhere else. I assume the 32KB battery backed CMOS SRAM is used for that on the x453 board.

Jan

Chkcpu wrote on 2024-01-20, 18:15:

@Retronaut: You brought the board back to life!! :) I admit I was sceptical about the BIOS being the issue, but I'm glad the BIO […]
Show full quote

@Retronaut: You brought the board back to life!! 😀
I admit I was sceptical about the BIOS being the issue, but I'm glad the BIOS re-load did the trick!
But I have no idea what could have caused this BIOS corruption...

weedeewee wrote on 2024-01-20, 14:09:

anyway, Is this one of those Bios's that update the ESCD or something in the flash chip on each hardware change?
Just pondering what could be the cause of a broken bios aside from a failed update and a failing chip.

The nice thing about 86Box is that, if the virtual machine uses a flashable BIOS, it makes a copy of the running BIOS in an "nvr" folder so the ESCD block can be updated in this BIOS copy and the original BIOS file in the machines folder remains intact.
When next booting this machine, 86Box will always use the updated BIOS copy with the latest ESCD data, just as on a real machine.

This feature allows us to compare the updated BIOS copy with the original one. Usually this shows us a difference because of an updated ESCD block.
On the x453 version K BIOS however, I don’t see any difference when performing this compare. This means the ESCD block is not stored in the 28F002T BIOS flashchip but somewhere else. I assume the 32KB battery backed CMOS SRAM is used for that on the x453 board.

Jan

I'm fairly certain that the 32K of sram was misidentified, and is in fact the tagram for the cache.

oh and the RTC & cmos ram (242 bytes) is in the AMI labeled PC97307 multi io controller. fyi

weedeewee wrote on 2024-01-20, 19:34:

I'm fairly certain that the 32K of sram was misidentified, and is in fact the tagram for the cache.

oh and the RTC & cmos ram (242 bytes) is in the AMI labeled PC97307 multi io controller. fyi

Good call weedeewee!!
That UM61256AS-12 chip must be the Tag RAM indeed. There is no other chip on this x453 motherboard that qualifies as Tag RAM for the L2 cache and the i430VX chipset sure needs an external Tag RAM.

We were mislead by this SRAM being made with (slow) CMOS technology. But it has an access time of 12ns, I never knew that CMOS memory could be made that fast! 😉

Yes, I also found the Real-Time Clock and 242 bytes of CMOS RAM for storing BIOS settings in the PC97307 Super I/O controller. Although these 242 bytes provide 128bytes of extended CMOS RAM above the usual 114bytes and can store all BIOS Setup settings, it is way to small for any PnP/ESCD data block. So the ESCD block is probably stored in the BIOS flashchip after all…

Jan

Chkcpu wrote on 2024-01-22, 11:16:

Although these 242 bytes provide 128bytes of extended CMOS RAM above the usual 114bytes and can store all BIOS Setup settings, it is way to small for any PnP/ESCD data block. So the ESCD block is probably stored in the BIOS flashchip after all…

I've often wondered, whenever this ESCD updating comes along, what the reason is for storing this information on every boot/hardware change/... ?
Shaving a few seconds off the startup time can't just be the sole reason right ?
Always/On every change overwriting the same block of the eeprom never felt like a good idea to me, aside from planned obsolescence and I've noticed on several HP computers that they often manage to mess up their bios and then only a recovery via the bios recovery procedure or an eeprom programmer will fix it. Plenty of times those computers will just get tossed without even a second look at the recovery procedure or bios reprogramming and the owners will only know their mainboard is broken, while it's just some software f'up.

J

Hey, sorry guys, I did not notice the thread had flipped to 2 pages, and the conversation was continuing.

So, the reason why I was quiet, is now that the machine is working, I had to get pen to paper as it were and make a video about it.
Right now, I don't have a setup conducive to recording me trying to fix a machine as it happens, so I have to kind of re-create that process.
Which feels pretty much as laborious as you have to go back and test everything again, whilst filming it.
Anyway enough producer hat whine from me 😀

The video is now complete
Fujitsu Ergo Pro x453 PC - CH03: The Repair https://youtu.be/jdukubiLE34

I hope it captures the overall flow of how the machine was fixed in the end. It was certainly an interesting collaboration and I learnt a lot.
I also have a friend, who is an engineer, and makes a lot of kit (mainly Amiga) and he was an advocate of removing and replacing the SRAM
Which I then dutifully ordered from China, costing £2 for the 4 chips and £26 postage 😒
And as we know, not needed in the end, so if anyone needs one of those fancy 12ns Sram chips, you know where to call 😀

Hope you all enjoy the video, knowing you had some part in it.
It was quite nerve wracking flashing the BIOS, I had this fear it would fail, and maybe break a BIOS that was in fact perfectly fine.
But it turned out to be the bullseye.

So, next step is to put it all back together, and I hope to get some new kit in there.

First I have to solder up and mount the two replacement Noctua fans. So airflow will be higher, but noise actually reduced.

I have a SCSI card, and I was going to put a BlueSCSI to replace the hard disk, but then found out they are £70 these days!
Which is funny as I bought the entire PC for £60!
So, change of plan, and now it gets a bog standard IDE to CFlash and a 16GB card.
Unless I can source a cheaper alternative, with the same perf.
The SCSI card goes in anyway, as I have a SCSI Zip drive I want to use on many of my retro machines, so its a great addition anyway.

Also, I bought a Matrox Millenium graphics card, which was actually an upgrade offered by ICL
I might try it with the on-board graphics, and maybe bench mark it.
That would make sense I suppose, before upgrading.

Oh yeah, now the machine is working, I also have two sticks of 32GB, so its not maxed out, but 64gb for a machine of this age is pretty enormous
I think it will play DOOM 😀

BTW, does anyone have an recommendations for utilities to test the machine and also benchmark it?
I want to benchmark its CPU, RAM, and its graphics card, and the hard disk.

Cheers

Chris
https://www.youtube.com/@RetronautTech

You mention using a detergent;

This might (in a long-stretch kind of way) explain the borked bios.

Detergent is an ionic surfactant, as the hydrophillic tail has a weak electrical charge, which is why it works to attract water molecules, (which have a dipole moment.)

Additionally, they can undergo ion-exchange reactions with metals and salts. If there was a electrical potential from dissimilar metals, combined with this kind of aqueous environment, there might have been sufficient voltage generated (its basically a weak saline battery at this point, with the detergent as the electrolyte) to enable the --WE-- pin, and some of the logic value/address pins, to flip some of the bits in the bios. This would corrupt the bios, and produce this scenario.

Again, it's a long-shot explanation, but maybe plausible.

Retronaut wrote on 2024-01-26, 16:27:

I have a SCSI card, and I was going to put a BlueSCSI to replace the hard disk, but then found out they are £70 these days! Whic […]
Show full quote

I have a SCSI card, and I was going to put a BlueSCSI to replace the hard disk, but then found out they are £70 these days!
Which is funny as I bought the entire PC for £60!
So, change of plan, and now it gets a bog standard IDE to CFlash and a 16GB card.
Unless I can source a cheaper alternative, with the same perf.
The SCSI card goes in anyway, as I have a SCSI Zip drive I want to use on many of my retro machines, so its a great addition anyway.

a zuluscsi is cheaper and should have better performance.

Oh yeah, now the machine is working, I also have two sticks of 32GB, so its not maxed out, but 64gb for a machine of this age is pretty enormous

Mega, not Giga

I think it will play DOOM 😀

BTW, does anyone have an recommendations for utilities to test the machine and also benchmark it?
I want to benchmark its CPU, RAM, and its graphics card, and the hard disk.

Good idea to play some Doom.
I think Phil (another youtuber) has some benchmark tool and other such stuff thrown together in a pack.

wierd_w wrote on 2024-01-26, 16:56:

You mention using a detergent;
Again, it's a long-shot explanation, but maybe plausible.

pcb assembly houses routinely run boards thru detergent bath to clean flux residue.

rasz_pl wrote on 2024-01-26, 17:29:

wierd_w wrote on 2024-01-26, 16:56:

You mention using a detergent;
Again, it's a long-shot explanation, but maybe plausible.

pcb assembly houses routinely run boards thru detergent bath to clean flux residue.

I am aware. He might have had rather hard water though, as he cleaned it in a bathtub. That, plus detergent, might make enough charge.

"Maybe."

wierd_w wrote on 2024-01-26, 17:31:

rasz_pl wrote on 2024-01-26, 17:29:

wierd_w wrote on 2024-01-26, 16:56:

You mention using a detergent;
Again, it's a long-shot explanation, but maybe plausible.

pcb assembly houses routinely run boards thru detergent bath to clean flux residue.

I am aware. He might have had rather hard water though, as he cleaned it in a bathtub. That, plus detergent, might make enough charge.

"Maybe."

Do you know of any such chemical reaction that can produce a 12V potential ?

Congratulations on the repair.
I had a Fujitsu iCL ergopro X series, with pentium 90 mhz, socket 5, with bios corupted. I gave up at the repair becouse I couldnt find any bios to flash from floppy.
help with damaged motherboard - Fujitsu ICL ergo pro X

Magnesium and Tin can get a potential of 1.3v... The corrosion table I am looking at does not list calcium.

Keep in mind, that you can get more voltage by cell-series interactions. (Similar to getting 12v out of several AA cells in series)

The principle of the razor tells us that the more compounding factors you throw at a hypothetical, the LESS likely it is to be that thing. As such, merely mentioning that something 'could maybe, possibly' result in a specific outcome, should not ever be considered to be an attestation that "Oh, it was TOTALLY that thing."

The truth is that I dont have a "Oh, I bet it was this thing!" scenario to suggest.

This leads me to grasp at straws, like this one.

A similar "you can kinda get high voltages out of it! wow!" scenario exists with your typical "Its made of pennies and dimes, with paper towels, and salt water" saline wet cell pile battery that gets used in intro to inorganic chemistry class, type setup, basically.

https://www.youtube.com/watch?v=JcRRTHkAl6A

It's a "It could maybe be that, if you have a dissimilar metal ion (calcium, magnesium) in your water" situation.

Again, the razor suggests that this is unlikely. I just don't have a better explanation, unless he was not careful with ESD when handling the board, and zapped it or something.

wierd_w wrote on 2024-01-26, 16:56:

You mention using a detergent; […]
Show full quote

You mention using a detergent;

This might (in a long-stretch kind of way) explain the borked bios.

Detergent is an ionic surfactant, as the hydrophillic tail has a weak electrical charge, which is why it works to attract water molecules, (which have a dipole moment.)

Additionally, they can undergo ion-exchange reactions with metals and salts. If there was a electrical potential from dissimilar metals, combined with this kind of aqueous environment, there might have been sufficient voltage generated (its basically a weak saline battery at this point, with the detergent as the electrolyte) to enable the --WE-- pin, and some of the logic value/address pins, to flip some of the bits in the bios. This would corrupt the bios, and produce this scenario.

Again, it's a long-shot explanation, but maybe plausible.

lmao, soapy water did not reprogram an eeprom.

kingcake wrote on 2024-01-27, 00:04:

wierd_w wrote on 2024-01-26, 16:56:

You mention using a detergent; […]
Show full quote

You mention using a detergent;

This might (in a long-stretch kind of way) explain the borked bios.

Detergent is an ionic surfactant, as the hydrophillic tail has a weak electrical charge, which is why it works to attract water molecules, (which have a dipole moment.)

Additionally, they can undergo ion-exchange reactions with metals and salts. If there was a electrical potential from dissimilar metals, combined with this kind of aqueous environment, there might have been sufficient voltage generated (its basically a weak saline battery at this point, with the detergent as the electrolyte) to enable the --WE-- pin, and some of the logic value/address pins, to flip some of the bits in the bios. This would corrupt the bios, and produce this scenario.

Again, it's a long-shot explanation, but maybe plausible.

lmao, soapy water did not reprogram an eeprom.

The issue here, is that 'SOMETHING' clearly did, and all retro did, was immerse the board in soapy liquid.
Conventional wisdom says that washing the board should not do this, as many h7ndreds of people have done this, and it has not occurred.

However,

Galvanic reactions can happen in open air with a suitably hygroscopic electrolyte present, and is a source of breakdown in electronics.

https://www.libertypackaging.com/blog/bid/104 … our-electronics

Usually, the potential created is less than half a volt. This is very far below the threshold needed to raise a logic pin, like --WE--.

That is where occams razor comes in, since it takes an additional, highly specific set of circumstancrs to get the voltages needed-- the corroding elements need to be connected in sequence, so that cell voltage is cumulative.

Science says to get more ICL motherboards, and to wash them in his bathtub, to see if the bios corrupts or not. Needless to say, that is not a valid option.

kingcake wrote on 2024-01-27, 00:04:

lmao, soapy water did not reprogram an eeprom.

No most likely not but handling a motherboard in odd ways in low humidity can, like handling it a lot in odd ways (not from the edges) could cause an issue. My guess is the eeprom is about to fail and this is the first inidicator.
Have a slot 1 that corrupted similarly, reprogrammed and worked, weeks later it hung on update escd. Luckily the eeprom was socketed, read it in my tl866 and saw two areas of non-readable, one in the escd area and another just past....
Also I have mental issue with anyone home washing a motherboard (Call it a reverse OCD thing). In the factories they are not hand washed but tank washed under specific conditions.
washing anything in a shower/tub reminds me of that episode of Seinfeld where Kramer washed the salad in the tub 😀 hahaaa....

b0by007 wrote on 2024-01-26, 17:43:

Congratulations on the repair.
I had a Fujitsu iCL ergopro X series, with pentium 90 mhz, socket 5, with bios corupted. I gave up at the repair becouse I couldnt find any bios to flash from floppy.
help with damaged motherboard - Fujitsu ICL ergo pro X

Ok, I may drop into that thread and help you there. Keep this one on topic.

Cheers

Chris

wierd_w wrote on 2024-01-26, 16:56:

You mention using a detergent;
Again, it's a long-shot explanation, but maybe plausible.

Necroware seems to wash his motherboards in water+detergent on a regular basis and I dont think he has any issues. Also, I get there might be a tiny theoretical charge from soap,
but I beleive these chips need 3.3v usually to make them do anything, and I cant see soapy water delivering that kind of volage
or we would all get a mild buzzy electric shock when washing the dishes?

As to why it went AWOL, Im not sure, its a mystery really eh?

These chips have a guaranteed minimum data retention life, and we're way past it on almost all of our retro computers. The one you have is rated for 10 years (probably at elevated temperature).