Hi Vetz, I'm sorry for completely missing the notifications about your messages and PMs.

Yes, the cache error was shown during POST. It called out the bad chip by the socket number on the cache board. I swapped two of the cache chips around and the issue followed the chip to the new socket number. I shamefully admit that I still have yet to purchase a replacement chip...

I don't have answer to the other questions at the moment. It has been over a year since I last powered on the machine, and am not at home to open it and look inside of it. I will update when I can, probably in the next week or so now that this is on my radar.

OtakuN3rd wrote on 2024-03-28, 17:02:

Hi Vetz, I'm sorry for completely missing the notifications about your messages and PMs.

Yes, the cache error was shown during POST. It called out the bad chip by the socket number on the cache board. I swapped two of the cache chips around and the issue followed the chip to the new socket number. I shamefully admit that I still have yet to purchase a replacement chip...

I don't have answer to the other questions at the moment. It has been over a year since I last powered on the machine, and am not at home to open it and look inside of it. I will update when I can, probably in the next week or so now that this is on my radar.

Thanks for replying. Good to know that it report any bad chip. Do you have 16kb or 64kb of cache? According to the service manual it could be delivered with both configs. I also have a newer revision BIOS/serial card so I'll be providing some pictures of it in this thread along with the cache board (I have the 16kb version)

I've installed a Cyrix Cx486Drx2 20/40 in my machine and the cache works fine with FLUSH and A20 activated. The performance improvement is definitely hit and miss with PCPBench (3,8fps) and Doom (5,7fps) seeing very good results, and others like 3DBench (10,1 fps) and Sysinfo (24 cpu points)/Speedsys barely seeing any improvement. I'm using an ATI Mach32 ISA 2MB card. I'll probably make a separate thread for this as I'm curious if this is normal with the Cx486Drx2 (few benchmarks are out there, and none for the 20/40 model)

Hi Vetz,

I have some pictures of the ESDI/Floppy controller attached. I also have a picture of the crystals on my main board. Changing J102 does not cause my machine to fail booting, but I don't have an OS set up to test the CPU speed.

I have not had the POST hang with all the status LEDs stay on with this machine, but I have had similar experiences (more often than not) for my Heathkit H-5100 and Rebranded Zenith version (I forget the model, but its the same motherboard as the H-5100, just a smaller riser card).

It can take 40 seconds to a minute for the POST to complete, but I suspect a lot of that is due to a dead CMOS battery inside the Dallas chip, and the fact that I have 20MB of RAM installed. Less RAM would no doubt shorten the POST time.

If I am doing my math correctly, I have 48kb worth of cache chips on my card. There's 4x YT-20C75-25PC VT20C79-25PC and 4x CY7C170-25PC chips, plus 3x SDT 6178 as tag RAM. (Its one of the tag RAM chips that is bad). There's a good chance it was a 16k board that had more chips added.

FWIW, I suspect the memory board I mentioned towards the beginning of the thread requires a DOS driver (that I don't have) to initialize.

Hopefully that helps clear up some of the mystery.

Thanks for the photos! That confirms the dual clock oscillators, one 40mhz and one 50mhz for the jumper 20/25mhz CPU speed option. I'll try and get one installed with a socket so I also can try faster ones.

Here is some pictures of my BIOS/serial card with part ID 85-3548-01. It doesn't have the botch wires your card have and looks to be a newer revision.

And some images of my cache card with part no 85-3438-02. I have 8x MCM6270P25 (4kbx4) SRAM chips for 16kb cache and 3x MCM4180P25 (4kbx4) for tag-ram. It's a very strange amount of tag-ram, 6kb. I've never seen another machine with such a odd number. Maybe someone can shed some light on this?

Your boot time seems in line with me, it's actually a bit slower, but it makes sense with your RAM size being 20 mb. When it is doing the POST, does the monitor turn on after it is completed or before/while it's running?

I don't know if you've noticed, but the machine also have little free upper memory when EMS is enabled. This is due to the Zenith custom slush logic on E000h, meaning you end up with just 91kb of available upper memory space, which again is spread across 3 blocks of 16kb, 27kb and 48kb, making it a pain to get enough memory using CD-ROM and SCSI device drivers along with smartdrv.

Thanks! Good to know about the EMS / UMB for when I start playing with mine more. I did locate and order some tag RAM for my cache board. Hopefully it shows up as I have never ordered from that site before.

The monitor comes active almost right away when I turn mine on, but it only gives an error message about a bad configuration in CMOS (again, no surprise given the dead battery). Its possible that the display is not initialized until it has a message as there is no RAM size counter. I have no way to confirm this at this time, though.

FWIW, I have a Headland HT208/A1A4978 SVGA video card in mine. There is a Zenith sticker on the ROM chip copyright 1990. The main sticker of the card has a date code of 0591, and the video memory was maxed out on it with the added RAM having 9131 for the date code (the soldered RAM either does not have a date code, or Samsung used a three digit code - 050). Given the 1990/1991 dates, I suspect it is not the original card for this machine and is an upgrade over what was originally installed.

Looking a my cache board, it looks pretty much identical, except the part number ends in -01. I'm pretty sure the number below that line is the date code, mine being 080288, or August 2, 1988.

I believe there's three different oscillators. Mine also has a 32MHz crystal, perhaps for 16MHz CPUs.

Its highly likely mine is a VERY early machine off the production line. As I mentioned earlier, I got it from someone that used to work at Heath / Zenith. (I actually know quite a few people that used to work for Heath, given I live in southwest Michigan and am active in Amateur Radio.)

OtakuN3rd wrote on 2024-04-01, 20:54:

FWIW, I suspect the memory board I mentioned towards the beginning of the thread requires a DOS driver (that I don't have) to initialize.

According to the service manual it says the following:
Add-On Ram refers to optional, plug-in memory cards (except EMSonly cards), including those manufactured by Zenith Data Systems.
Highlight the “Add-ON RAM” field and use the space bar or BACKSPACE key to make selections for the “BASE” and “EXTENDED” fields,
as necessary.
Have you enabled it in the BIOS under "Add-on RAM"? The memory board supports a maximum of 32mb in sizes of 2,4 and 8mb for each memory stick.

OtakuN3rd wrote on 2024-04-03, 23:05:

Thanks! Good to know about the EMS / UMB for when I start playing with mine more. I did locate and order some tag RAM for my cache board. Hopefully it shows up as I have never ordered from that site before.

I forgot to mention in the post that those values are with SCSI ROM loaded at DC000h (16kb) and MDA memory made available with the I=B000-B7FF command added to emm386.

OtakuN3rd wrote on 2024-04-03, 23:05:

The monitor comes active almost right away when I turn mine on, but it only gives an error message about a bad configuration in CMOS (again, no surprise given the dead battery). Its possible that the display is not initialized until it has a message as there is no RAM size counter. I have no way to confirm this at this time, though.

I see, I've seen the same behaviour once the BIOS had invalid settings. Then the monitor turns on much earlier than normal to display the error message. When you get it working I'd expect your monitor to not initialize before POST is complete.

OtakuN3rd wrote on 2024-04-03, 23:05:

FWIW, I have a Headland HT208/A1A4978 SVGA video card in mine. There is a Zenith sticker on the ROM chip copyright 1990. The main sticker of the card has a date code of 0591, and the video memory was maxed out on it with the added RAM having 9131 for the date code (the soldered RAM either does not have a date code, or Samsung used a three digit code - 050). Given the 1990/1991 dates, I suspect it is not the original card for this machine and is an upgrade over what was originally installed.

I have the original VGA card (currently replaced with an ATI Mach32 for better Windows 3.11 performance, refreshrates and resolution). Part no 152-10-C2. Western Digital 90C00-JK with 256kb RAM (8x HY53C464LS-10 64k x 4 should be 32kb each).

OtakuN3rd wrote on 2024-04-03, 23:05:

Looking a my cache board, it looks pretty much identical, except the part number ends in -01. I'm pretty sure the number below that line is the date code, mine being 080288, or August 2, 1988.

If you read my thread Re: Why would you need 3x 2kb tag-ram chips for 16kb of L2 cache in a 386? and by looking at the chipnumbers you posted earlier I believe you also have the 16kb version. Are you sure it's 48k on that board? Both the CY7C170-25PC chips and the tag-ram IDT 6178 chips are 4k x 4 (2kb). The YT-20C75-25PC chip I cannot find on Google.

OtakuN3rd wrote on 2024-04-03, 23:05:

I believe there's three different oscillators. Mine also has a 32MHz crystal, perhaps for 16MHz CPUs.

I think that oscillator is for the ISA bus, using a divider of 4. There is no jumper to switch to 16mhz FSB, nor is it mentioned anywhere in the service manual that this board supports that CPU.

OtakuN3rd wrote on 2024-04-03, 23:05:

Its highly likely mine is a VERY early machine off the production line. As I mentioned earlier, I got it from someone that used to work at Heath / Zenith. (I actually know quite a few people that used to work for Heath, given I live in southwest Michigan and am active in Amateur Radio.)

Looks like it. Very interesting backstory on your machine! I have no backstory on mine which have datecodes of early 1990/late 1989. It was obviously a very expensive machine in 1990 and it was delivered with the i387 FPU option, so it would have been interesting to know who bought it and for what purpose. When I got the machine it had a 330mb SCSI drive installed and the ESDI controller and original drive was missing. The machine had been setup for a child to play around with.

Another question, do you have any chip populated in U157?

I believe you are correct on my cache size. I thought I saw a data sheet that said the CY7170 chips were 8k x 4 bits, but after another search, they are indeed 4k x 4. The other chips are VT20C79 and also 4k x 4. I had a typo in my previous reply, sorry about that.

I do not have anything in U175 either, just the empty socket like yours.

I will take another crack at enabling the "add-on RAM" in the BIOS at some point. I think when I tried earlier, I couldn't select those fields but I wasn't 100% sure what I was doing at the time.

I'm still waiting for my replacement cache chips to come in. Looks like I'm waiting for a slow boat from China.... (I can't really blame them, I'm sure my 5 used chips are way lower priority than the bulk orders from large companies.)