Roman555 wrote:

Also I tried the GETROM.ZIP - it works. I ran it twice and compared the images. They have difference in one address (just 1 byte). I don't know why it is, but it means that different checksums of bios could be generated even in a working system.

Faulty memory could be a cause of this.

Or incorrect memory being used (EDO, for example), or mismatched memory (latency being lower than the other). Sometimes I get the memory error with the 2x 16MB SIMM-72 RAM installed (latency is 70ns). Another example is, if you want to use 60ns RAM, disable the on-board RAM to prevent conflicts or errors. I can't even install L2 cache in my system (Windows would throw a memory fault message and the system would lock up at times when turning it on).

Hmm, I'm not sure if it's just my machine's own quirks but the memory slot furthest from the CPU has been giving me some trouble, I had to remove the RAM in that slot so that Windows could finish installing.

bjwil1991: It's worrying that you've had so many issues with the L2 cache, I've been trying to get that working too but I've only got the 128KB cache and TAG RAM chip, not the DB ram chip it apparently also requires(?) Right now, it either doesn't boot with the cache fitted or doesn't detect it.

ahtoh: Have you had a chance to try an alternative video card yet?

Thermalwrong wrote:

bjwil1991: It's worrying that you've had so many issues with the L2 cache, I've been trying to get that working too but I've only got the 128KB cache and TAG RAM chip, not the DB ram chip it apparently also requires(?) Right now, it either doesn't boot with the cache fitted or doesn't detect it.

The L2 cache requires the following:

4x 32K x 8 (or 256Kb) 20ns 28-pin or 4x 128K x 8 (or 1Mb) 20ns 32-pin
1x 32K x 8 (or 256Kb) 15ns 28-pin TAG
1x 64K x 1 15ns 22-pin DB

The cache chips must match in order for it to successfully work, and I believe it goes by the BIOS string of which chips will work with it.

I purchased a few VGA cards but they're not yet delivered.
I also sourced a working PB450 board and did some comparison between the two.
I tried to swap the bios chip and it made no difference, this makes me believe it is not the bios problem
Diag showed minimum difference but it detected monitor on the working board:

Monitor label: IBM 8514
Monitor type: Color

This makes me believe that it is VGA chip fault.
Strange thing is the low Diag performance of both motherboards, only scored 4166.7 Dhrystones (the value matches on both boards!)
But this is probably due to missing L2 cache.

Can you post some up-close pictures of the video chip area (video chip itself, and the memory for the video card)? Could need a reflow of the solder.

Also, did the other board act the same way with the lock ups or did that work as well? The dead VGA sounds like the issue there.

yes, the working board sometimes locks up as I think because it tries to detect a hard drive.
this fixes by going to bios and changing the hard drive to none

Does anyone have this bios from Micro Firmware BTW?
https://www.rigacci.org/docs/biblio/online/fi … are/mfibios.htm
pb450 version 4.05.10

bjwil1991 wrote:

Can you post some up-close pictures of the video chip area (video chip itself, and the memory for the video card)? Could need a reflow of the solder.

I inspected the chip very carefully and all the soldering is visually OK
I now plan to replace the 5428 chip with 5424 from another board (dead pb430 with leaked battery)
Not sure if I have to somehow adjust the video bios as I change the chip

ahtoh wrote:

I now plan to replace the 5428 chip with 5424 from another board

It's better to do some measuring around videochips on both mainboards and compare to be sure.
Because I've read a lot how people replaced suspect bridges, hubs, MIOs but in fact bad components were just little resistors, chokes, capacitors.
For example, as the CL-GD542X datasheet sais there are pins AVDD[1-4] that are used to supply +5 volts to the different blocks of the chip. Power goes to these pins either directly from the +5V rail or via 33 Ω resistors (or it could be chokes, IMHO).

The Micro Firmware Phoenix BIOS requires the actual Micro Firmware Phoenix BIOS to be on the chip itself. I might plan on flashing the BIOS onto a blank chip and get the video BIOS from the chip installed in my Packard Bell as well to copy it onto.

Here's a link: https://www.driverguide.com/driver/detail.php?driverid=39

According to the Readme :

This BIOS Update file will update prior revisions of Micro Firmware's P4HS20 BIOS Upgrade to P4HS20 version 4.05.10.

In order to install this update, the P4HS20 BIOS must already be resident on your PC. You can confirm this by booting your PC and watching for the Micro Firmware BIOS part number which will appear on the boot screen. DO NOT ATTEMPT TO INSTALL THIS UPDATE IF YOU DO NOT SEE THE "P4HS20" PART NUMBER ON THE BOOT SCREEN.

Update:
I received a graphics card and it works.

Will try to replace the chip

Glad you got it sorted out, man. Check the resistors next to the video chip to see if any of them are bad just in case.

I bought a videocard recently (S3 Virge DX, but it doesn't matter).
It didn't show anything, a monitor stayed in "stand-by" mode, but DOS loaded and I could get diag447 report.
Almost the ahtoh's case.
Long story short there were cold joints of videochip pins to pads. I fixed them with a soldering iron.
The videochip is OK.
So, ahtoh, if you haven't replaced videochip yet, try to inspect the soldering of the chip with a toothpick or a needle.

Replaced 5428 chip with 5424 and it did not work.
The board will not start at all now.
Though the marking say 5424/8 on the board

ahtoh wrote:

Replaced 5428 chip with 5424 and it did not work.
The board will not start at all now.
Though the marking say 5424/8 on the board

So pity. 😢 I hoped you had repaired it long ago.
The datasheet sais the both chips are pin-to-pin compatible.
I don't know if the videobios is compatible but 5428 supports more memory and more video modes.
But what about testing with the discrete graphics card while the onboard one is disabled?

PS from datasheet about videobios (page B16-4):

SOFTWARE CONSIDERATIONS A common BIOS is provided for CL-GD5422/'24/'26/'28/'29. To take advantage of the BitBLT engine, there […]
Show full quote

SOFTWARE CONSIDERATIONS
A common BIOS is provided for CL-GD5422/'24/'26/'28/'29. To take advantage of the
BitBLT engine, there are different drivers for the CL-GD5426/'28/'29. The CL-GD5420
shares the BIOS and drivers with the CL-GD5402.

I believe they sell 5428 chips as well. Make sure they match the last 4-5 characters since that also depends on latency for the video memory.

eBay has the CL-GD5428-80QC-A chipset itself.

I now noticed that I burned some electrolytic capacitors with the hot air so they leaked some black thing. will try to replace them
It could be also that I killed the 5424 chip when desoldering it from the donor board.
So I bought a new CL-GD5429-86QC-A chip off ebay. Will do an upgrade 😀

ahtoh wrote:

I now noticed that I burned some electrolytic capacitors with the hot air so they leaked some black thing.

Yes, it would be better to remove caps from the board temporarily.
And with a bottom preheater you could decrease temperature of hot air.
Although the pads look good (not overheated, IMHO)

Roman555 wrote:

Yes, it would be better to remove caps from the board temporarily. And with a bottom preheater you could decrease temperature o […]
Show full quote

ahtoh wrote:

I now noticed that I burned some electrolytic capacitors with the hot air so they leaked some black thing.

Yes, it would be better to remove caps from the board temporarily.
And with a bottom preheater you could decrease temperature of hot air.
Although the pads look good (not overheated, IMHO)

My suggestion for SMD chips and capacitors would be to use an SMD soldering/desoldering iron. I have one myself in case I need to replace something. My first motherboard repair was my IBM ThinkPad R40's heatsink fan power connector that I accidentally ripped off when I was putting new thermal compound on the processor, and it still works.