Anonymous Coward wrote:

I'm assuming that when you filled both banks with 256kbit chips to make 1MB, you set the motherboard jumpers accordingly.

Initially, I pulled both the AST and the BocaRAM cards and pulled all of the 1M chips from the board. I pulled both banks from the BocaRAM (18x256K chips). I inserted them into Bank 0 of the motherboard. This makes 512KB of RAM. I used TH99 and set the two jumpers. The BIOS displayed 512KB correctly.
I pulled one row from the AST (18x256K chips) and inserted them into Bank 1 of the motherboard. This should make 1MB of RAM. I used TH99 again and set the two jumpers for 1MB. The BIOS displayed 512KB. Changing the base memory in Setup from 512 to 640 gave a memory size error.

If that is the case, and it still shows 512kb, to me it sounds like the problem is CMOS related. Are you able to take any pictures of your BIOS setup screens?

Yeah, I'm starting to think that as well. I'll take pictures when I get home around 4:30 of startup, setup, and extended setup.

I assume you have a CMOS battery plugged in too, right?

Of course. A jumper on the board changes between alkaline and lithium, mine is set for lithium and I have a lithium cell attached. Originally it had an alkaline battery that was completely dead. This one seems to keep the time (even though it can be a bit fast / slow)

Jo22 wrote:

That's understandable. Maybe the Lo-Tech 1MB card is worth a try ?
It uses modern RAM, can be configured for UMB memory, as well as acting like a normal memory upgrade.
That way, you can get hold of of more than 1MiB of usable RAM (the Lo-Tech 1MiB card is a bit slower, though).

Lo-tech is significantly more expensive, and last I checked they were out of stock anyways.

The remaining DIP RAM on the AST card then could be saved for something valuable, like an AST RAMpage EMS board.

I'd rather have XMS than EMS, it is a 286 after all. I don't have a RAMpage anyways, just the Advantage 128.

Maybe the board or BIOS doesn't support high capacity DIPs yet ?

Surely it must, the board itself has two different kinds of sockets, one for 256K and one for 1M chips. They have a different number of pins, even, and it has an extension board connector (Which I would like to populate, if I can find the board to do so). The BIOS apparently detects past that on the AST board, so I'm not sure just yet.

Anyway, it's just an idea, please don't get upset. I started with a modern 286-12 that had got a built-in CMOS utility, VGA, PAS16/CD-ROM and 4 MiB of SIMM RAM.

You're all good, mate. This is a pretty modern 286 as well, at 20MHz with built-in CMOS, VGA, and PAS16-SCSI. I don't have SIMMs, though.
Everybody starts somewhere, I started with a Pentium 100 with SVGA, ESS AudioDrive, and 64MB of SIMMs.

https://www-01.ibm.com/common/ssi/ShowDoc.wss … quest_locale=en

I'm not sure hot to interpret this. I'm not using anything from IBM here, this is onboard memory and two third party expansion cards.

Fixing the battery power could really mean something here.

The battery seems to work correctly, it keeps the clock ticking with the machine turned off, and it holds the memory size and disk drive types after over a day of being powered off and unplugged. Would that not indicate a good CMOS battery?
If so, how expensive are these batteries and where can I get them? My local Radio Shack closed down ages ago, and I haven't seen pigtail batteries since.

Use your phone and take images of all the BIOS screens, please.

Okay, done. I've been busy, my apologies.

Here's the boot screen (after the Trident TVGA BIOS), it counts RAM and shows some tests. "Extended CMOS checksum" fails.

Here's the setup screen (Control + Alt + Escape during the floppy seek):

Here's extended CMOS setup (Press left or right arrow keys on "SELECT" next to it in normal setup). Changing any of these settings in any way causes it to lock up when rebooting:

Were these screens taken without the ISA ram card installed? The the first and last picture shows it detects 2MB of on board memory. You said attempting to change the base memory to 640kb causes a problem. What about if you change both the base memory to 640kb, and the extended memory to 1024kb. Does that still throw up an error? Maybe 384kb is eaten by UMB, and when you set the extended to 1408 there isn't enough memory left to allocate 640kb base memory.

Also for a system date, you'd better set it to something in the 80s or the 90s. Setting the year as 19 means 1919, and that might cause some software to behave oddly.

The first picture was taken with the AST installed, but the second two were taken it with it removed. My bad.

The computer does not autodetect RAM size, you have to set it manually in the first screen.
The computer detects 512K and says "Memory size error - Run Setup."

The last picture shows 2MB, but when booting the machine only shows 512K. Setting the first screen to show 640K causes a memory size error, and changing anything on the second screen causes a flashing cursor until reset.

No, it's Y2K compatible. DOS shows 2019 correctly.

Well, the last screen is the most important, because something is actually sensing that there is 2MB in the system board. Notice how that screen reports 1024kb of extended memory? Try setting base to 640kb and extended memory to 1024kb in the standard CMOS setup screen and see what happens.

Without the AST installed, setting base memory to anything over 512K and extended to anything over 0 will produce a "Memory size error" message.

That second screen is not accurate - Sometimes it shows 0K, sometimes it shows 4MB. Changing anything on that screen will cause it to hang.

Other than the onboard memory issue, are there any other unresolved problems?

This board uses the C&T NEAT chipset. The datasheets are available online. You should be able to use it to identify the memory controller. Perhaps something is loose. Maybe try applying some pressure to it while going through the memory configuration to see if it helps. Test the continuity of the lines connected to the memory controller, and maybe consider reflowing some solder.

Some VGA cards (Paradise Basic, based on Oak OTI 037 or 067 I think, Acumos AVGA1) refuse to work, but the Trident TVGA8900C works fine. The AVGA1 is untested but the Paradise came out of a working machine. My guess is that the AVGA1 needs something to do with the ISA OSC that this cheap clone probably isn't doing right.

My PAS16 doesn't work and the drivers hang the machine, but I need EMU386 to install them. No PnP support, so it's probably not set up at all.

Otherwise, with some disk controller cards it hangs after POST. I tried several cards:
IBM AT Fixed Disk Floppy Diskette controller (Early) - Does not work
IBM AT Fixed Disk Floppy Diskette controller (Late) - Works
Cirrus Logic based Winchester controller - Does not work
Western Digital based Winchester controller - Does not work
Western Digital based Winchester controller with less jumpers - Works
DTC 5150XL - Works
DTC 5150L - Does not work
DTC 7287 - Works after changing jumper settings. Currently using this one.

There's no real pattern to the disk controller thing, and I feel like it's something to do with the onboard BIOS the controllers have, maybe it's picky. Nothing else memory related.

I've removed and reinserted every socketed chip on the board, where would I find datasheets for this chipset? If it's not a socketed component I probably haven't touched it.

With older SCSI and Pre-IDE HDD controllers, you couldn't normally move a drive from one controller to the other and expect it to work. It required reformatting. Each controller formatted a little bit differently, and they usually weren't cross compatible. Are you sure these controllers have their own BIOS chips? XT controllers needed BIOSes chips, but with the AT I thought everything needed was in the main system BIOS.

The non working video cards thing is interesting. May or may not be related to your memory problem.

Datasheet archive used to have the databook for the NEAT chipset, but they changed things and at present I can only find the abstract. I'll have a look around. I might have saved a copy.

Try this link:

https://www.datasheetarchive.com/pdf/download … %2520at%2520286

Hmm...seems there was a C&T chipset that predated NEAT. How many chips make up your chipset, and what are the models numbers?

I know this. I'm saying with a completely unformatted drive. It's set to boot from floppy first.

I'm absolutely certain that the Cirrus Logic, Western Digital, and DTC controllers have their own BIOS. The CL card has a label indicating a version number, the DTC 5150 cards contain a drive table and a formatting utility (G=C800:5), and the Western Digital cards have low level formatting utilities as well.
The DTC 7287 contains both a drive firmware and a BIOS, using DEBUG and G=C800:5 you get a menu to change drive types, format, and surface scan disks. The BIOS is 14KB, it's stated in the manual. The last 2K of address space the card occupies is controller RAM.
At POST, I can see "DTC RAM CHECK... 2KB" before DOS boots.

I'm not sure how many, I'd have to open it up and check. I can link the TH99 article again, but I don't know anything else off the top of my head. It's a 20MHz 286 and a 12MHz 287 if that helps at all.

DTC5150 is an 8-bit card for XT computers. It may not be possible to use those in AT computers, because there might be a resource conflict. DT 7287 should be okay. Interesting that it has a BIOS though, I didn't think mine did.

BTW, what's your ISA clock set at? I'd stick with 8MHz unless everything works.

Always set at 8MHz. The turbo switch changes between 20MHz CPU / 8MHz bus and 8MHz CPU / 8MHz bus.

The CL and WD cards are 16 bit as well, and (obviously) so are the IBM controllers.

DTC 7287 manual describes it:
http://www.minuszerodegrees.net/manuals/DTC/D … ion%20Guide.pdf

It's the Chips & Tech NEAT chipset.

This edition of PC Mag specifies it:
https://books.google.com/books?id=h6qtgYAzqDg … %20supra&f=true

Hi, thanks for the pictures! Awesome machine, btw! 😀
Have you checked the system with NSSI or CheckIt! (v3, v4) already ?
Perhaps some odd system error is causing this. Since there's a "checksum error", maybe the RTC is faulty ?

FAMICOMASTER wrote:

Some VGA cards (Paradise Basic, based on Oak OTI 037 or 067 I think, Acumos AVGA1) refuse to work,
but the Trident TVGA8900C works fine. The AVGA1 is untested but the Paradise came out of a working machine.
My guess is that the AVGA1 needs something to do with the ISA OSC that this cheap clone probably isn't doing right.

Reminds me of an blog entry of OS/2 Museum: http://www.os2museum.com/wp/the-isa-osc-mystery/

FAMICOMASTER wrote:

My PAS16 doesn't work and the drivers hang the machine, but I need EMU386 to install them.
No PnP support, so it's probably not set up at all.

The original/early DOS/Win3.1 PAS16 drivers should work just fine on a 286, at least.
Back in the 90s, I had it installed in a 286, before I startet using EMU386 (still, that piece of software is awesome)..
However, the PAS16 had trouble in some 586 machines that I had (EMM386 was loaded there).
Normally, the PAS16 should work fine, though. Provided that -5v is available and no timing issues do plague the system.

Jo22 wrote:

Hi, thanks for the pictures! Awesome machine, btw! 😀
Have you checked the system with NSSI or CheckIt! (v3, v4) already ?
Perhaps some odd system error is causing this. Since there's a "checksum error", maybe the RTC is faulty?

No, I hadn't done that just yet. Since you mentioned it, I was just about to put it on a diskette and transfer it over so I could run it, but my USB floppy drive conveniently chose this moment to pack up! Cleaned the heads, tried it with different machines XP through Windows 10 and nothing. It won't read any diskettes, even diskettes it's written and read before. The 286 machine still reads them just fine, too. The only machine I have with a dedicated diskette drive right now is a 386SX laptop with no way to transfer anything to it without major disassembly and using my USB to IDE with the hard disk. How fun!

Reminds me of an blog entry of OS/2 Museum: (link)

Yeah, I read that article after the fact and I thought that.

The original/early DOS/Win3.1 PAS16 drivers should work just fine on a 286, at least. Back in the 90s, I had it installed in a 2 […]
Show full quote

The original/early DOS/Win3.1 PAS16 drivers should work just fine on a 286, at least.
Back in the 90s, I had it installed in a 286, before I started using EMU386 (still, that piece of software is awesome)..
However, the PAS16 had trouble in some 586 machines that I had (EMM386 was loaded there).
Normally, the PAS16 should work fine, though. Provided that -5v is available and no timing issues do plague the system.

I just grabbed the ones from the Driver library. Whatever ones are listed there are the ones I tried with DOS 6.22. It showed me the base address, DMA channel, and Interrupt and then locked up.
I have an ESS AudioDrive based on the ESS688 chipset I would try, but I'm kinda in a bind with this project since I'm in the middle of a move right now. I don't have another machine with a diskette drive to use at the moment.

FAMICOMASTER wrote:

No, I hadn't done that just yet. Since you mentioned it, I was just about to put it on a diskette and transfer it over so I coul […]
Show full quote

Jo22 wrote:

Hi, thanks for the pictures! Awesome machine, btw! 😀
Have you checked the system with NSSI or CheckIt! (v3, v4) already ?
Perhaps some odd system error is causing this. Since there's a "checksum error", maybe the RTC is faulty?

No, I hadn't done that just yet. Since you mentioned it, I was just about to put it on a diskette and transfer it over so I could run it,
but my USB floppy drive conveniently chose this moment to pack up! Cleaned the heads, tried it with different machines XP through
Windows 10 and nothing. It won't read any diskettes, even diskettes it's written and read before. The 286 machine still reads them just
fine, too. The only machine I have with a dedicated diskette drive right now is a 386SX laptop with no way to transfer anything to it
without major disassembly and using my USB to IDE with the hard disk. How fun!

Ah, I see. That brings back memories. Can you format your 1.44MB disks to 720KB or 360KB (in the 286) ?
If so, this could solve some compatibility issues. I went trough the same process (with XP/7 and USB floppy drives).

If you can't get it to work, just use a serial connection or a network card for file transfer..
You can find some programs being mentioned here: Re: 386 , what can I do with it?
If you've got an USB-Serial adapter, you can make a 3-wire null-modem cable and use FileMaven in DOSBox and on the 286.

It really works, see https://www.youtube.com/watch?v=z9EoRA8ETUs 😀
If you can, use a slow connection speed or an 16550 UAR chip here.

Oh no, the 286 can read, write, and boot from high density diskettes just fine. The 286 can actually read the diskettes my Teac USB floppy drive can't.

Windows 10 was writing these diskettes with WinImage not 2 days ago - I haven't got a clue what's changed since!

Put a disk in - Inaccessible. Put the disk in the 286, it lists the directory just fine. Decide to reformat it. Put the disk back in Windows 10 - No change. Put the disk in a different machine, Windows 8.1. Nothing. Get sick of it and boot my XP laptop. Still nothing.
I tried cleaning the heads and using a cleaning diskette, nothing has helped it. I think it may just be it's time - It has a manufacture date of the 51st week of 1998, so it's overdue for problems, I suppose.

I have other machines with USB and diskette drives, just none with me right now, which is upsetting.

The Windows 10 machine has a built in serial port and I have a USB-to-serial adapter, they both work and I can sync Palm Pilots with them. They do not work with FastLynx, which is very annoying.

You know what the problem with putting a small driver on my boot floppy is? I can't read or write to diskettes with anything but that 286 and my 386 laptop.

Hi, I didn't mean to say the 286 has problems- I meant to say the PC with the USB floppy drive.
If you format the 1.44MB disk in your 286 to 720/360K, chances are better that the darn USB floppy drive can read them,
but it perhaps became missaligned. So using lower capacity formats may help for a while, so you can copy programs
to your 286 that establish file transfers (Terminal, TCP/IP, FTP software and so on).
If nothing helps you could also use XTIDE Universal BIOS to boot via serial port.
Anyway, thse are just ideas. You could try running an alternate floppy BIOS (XM BIOS TSR ?) on your 286,
maybe it causes better compatibilit< with your modern PC (unlikely, but wort a try).