You have the wrong schematic. I have the equivalent of a 256/640k board, although all of those connections do exist on that particular board schematic; it's on page 74 of the pdf. (It has three 158 chips.) But I'll take a look at ADDRSEL.

Yeah, I flicked between the two and didn't see any important differences about how XA16&17 connect to A8 (74LS158 followed by resistor). The chip numbers are all different from your board, so not much help there, but it sounds like the connections are very similar. If you see signals on both XA16 & XA17 but A8 is always low then I think it's unlikely to be a problem with ADDRSEL, since that just selects which of those two is output on to A8. It has to be one or the other. Unless something fun is happening so that ADDRSEL always selects the input that's low, and they're never high at the same time. The only way for the output to be disconnected is the enable pin 15 (not shown on the schematic). If that goes high then the outputs are held low.

Hi! I'm still learning about XTs, so I can't releally join this discussion.
However, maybe those XT related files might be useful during troubleshooting.

Best wishes and good luck,
Jo22

So if ADDRSEL is not the problem, the other suggestion one person had from the VCFed forums was to check the RAS and CAS lines.

Maybe they are not connected properly to address RAM higher than 256k?

Also, I did write the equivalents of the resistors on my downloaded copy of the schematic. They are indeed on my board.

Oops, I was reading the wrong pin. Pin 1 (A8) is indeed receiving the pulses. The 158 seems to be in working order.

We can also ignore the 24S10 PROM altogether, as this system already seems to have its addressing logic in order too.

So... I'm guessing now we move onto RAS and CAS?

Ok, if A8 is getting through then the it's likely that the 158 is ok. I suppose it could have a fault where ADDRSEL can't switch between XA16 or 17, or the input of one of those doesn't get passed through to A8, but that seems less likely. I think that then gets in to stuff I don't know, like reading the BIOS and working out how it does the auto-detection, then hooking up a logic analyzer and seeing what's actually happening. I know your board doesn't have the PAL that the IBM does, but the IBM one does change its function somehow when it's jumpered for 640k. Whatever that PAL does must have been implemented as discrete logic on your board. So if it was possible to work out what the IBM one does differently when set to 640k then it might give a clue to what changes are needed on your board. Maybe.

Ok, a preliminary check shows several pins on an LS08 (of which there are two near the memory, that control the RAS and CAS signals) not connected. This is just a preliminary, cursory check, though. I'll look at it more thoroughly later.

Just had a couple of unrelated thoughts. I know you've replaced the BIOS, but the original 5160 BIOS apparently had a bug where it tried to auto-size the amount of RAM, by writing and reading from a couple of bytes every 64k. If it failed then it assumed that was the end of the RAM, when it might have been bad RAM. Do you know if your RAM is all good?
Other thing was how did you work out which is memory bank is which? I see from someone else's photo that A8/pin 1 is routed for all the chips, when the 5160 only routed pin 1 for the 256k chips. The 5160 had bank 3 at the board edge, which from the schematic would be the 64k chips. But a different clone board I found has its 256k chips at the board edge, like yours.

I bought a RAM tester a few months ago. And that's actually what clued me into shuffling the RAM about. I used the Supersoft Diagnostics ROM to test everything. One day, after I had tested the RAM, I couldn't figure out why the system would not start up. Turns out I had used a ROM with a non-working image, which really doesn't help. 😜 I put in the other ROM that I knew I had burned, and voila, everything tested good, but the full complement of RAM still wasn't being counted, so I figured that the other portion of RAM that wasn't being counted was not being addressed/bank-swapped through the A/B pins.

A further check with the logic probe shows that Pin 10 of one of the two LS08s has no pulse or HI-LO switching. The other one is functioning. Maybe that's my problem?

Pin 1 doesn't show HI-LO switching on either of them, but the pulse light does blink. Maybe the switching is so fast as to be unable to be shown with the HI-LO LEDs.

I'll check the pinout and try to connect the pins on a schematic page, using continuity mode on a multimeter, to see if it matches with any schematics I have. (There are 4 pins connected in-parallel to a refresh gate. Again, maybe that's my problem, that one is not connected?)

T-Squared wrote on 2022-02-07, 02:09:

A further check with the logic probe shows that Pin 10 of one of the two LS08s has no pulse or HI-LO switching. The other one is functioning. Maybe that's my problem?

Skimming the IBM schematic I can see on sheet 3 an S08 U55 that drive the 4 RAS lines, used to turn them all off for refresh cycles. As you found, inputs 1,4,10,13 are all connected together, so one stuck would imply a bad solder joint. The other one I can see is U87 on sheet 2, which the gate on pin 10 is used as part of the NMI signal to the CPU. I can't immediately think (although I don't know much) how a problem with the NMI might lead to it not detecting RAM correctly, and I can easily believe that the NMI signal doesn't change much.

There may be others I haven't seen.

Just to humour me, on the S08 that appears to be all working, would you mind checking that pins 1,4,10 and 13 are all connected together, that should give a good indication if it is following the IBM sheet. Then check if pin 3 (!RAS0 output) goes to pin 14 of the first chip in each bank of RAM, through a 30 ohm resistor. I just want to double check the order of the banks, and checking that connection will confirm which is Bank 0 (if Sanyo followed the IBM design).

To complete the check, S08 pin 11 should go to Bank 1 pin 14, pin 6 to Bank 2 pin 14 and pin 8 to Bank 3 pin 14.

Pin 1 doesn't show HI-LO switching on either of them, but the pulse light does blink. Maybe the switching is so fast as to be unable to be shown with the HI-LO LEDs.

pin 1 of the RAS S08 is to do with the refresh cycle, so I'd expect it would just be occasional blips. On the other S08 it looks like something to with DMA.

I'll check the pinout and try to connect the pins on a schematic page, using continuity mode on a multimeter, to see if it matches with any schematics I have. (There are 4 pins connected in-parallel to a refresh gate. Again, maybe that's my problem, that one is not connected?)

I think if one bank was missing its RAS signal or refresh cycle that it just wouldn't work at all, since no Row Address could be signaled to the chips, so I'll be surprised if that is the problem. But then I'm often surprised.

Well, for me, The connection was missing on *almost* all the "B" pins of that one S08. (Pins 2, 5, 10, and 13)

The problem with the RAS0 output is that Sanyo didn't follow the design exactly, so the 30/33 ohm resistors (1st is IBM, 2nd is clones) are missing.

I found this last night. It's the best that I have, because no one has documented the MBC-775 very well, so the closest schematic I can use is the MBC-550/555: http://www.eriscreations.com/sanyo/diskimages … matics-only.pdf

I'll take a look at the bank order later. I swapped the order of the RAM a month or so ago, just to make sure exactly what you're suggesting; that I didn't put Bank 1 into bank 4, bank 2 into bank 3, etc.

That MBC-555 looks quite a lot different to the 5160. E.g. it looks like it's got a single RAS line that goes to all the memory. I've only spotted the one S08 (U102), which looks to be something to do with the mono video output.

I'm being a bit slow, can you explain what you mean by "The connection was missing on *almost* all the "B" pins of that one S08."? Do you mean there were cracks in the solder, or there are no tracks going to them? And which S08? Is it wired the same way as either of the S08 on the 5160 schematics? On the 5160, one S08 handles the RAS (U55) and the other handles stuff to do with Speaker, parity RAM, NMI and DMA (U87).

Ok, I got it. It took a while, but I've traced out the circuits close by the memory. I don't know how much good it will do. I also posted this to VCFed.

I think I found it! 😁

You may have been correct, snufkin! I found something odd with a 138/158 connection!

One of the enable pins has been grounded (and it's the only one that I found on my schematic that does that), potentially meaning that may be our key! I don't know the electronic logic behind it, but it seems odd that a perfectly good connection would be grounded unless that was to be disabled.

I need someone to verify this, if possible.

Okay, looking again, the 158 is supposed to be grounded. However, I don't know why that 138 is grounded too. According to two schematics I've seen (The 5160 and a Turbo XT Clone), pin 4 on that one 138 on my schematic isn't supposed to be tied to ground.

All right, it's been a while.

I have cleaned up my schematic.

Would anyone be able to decode this? I've highlighted an area of interest in the PDF schematic, because it seems all traces involved lead to these two LS139 and LS157 chips on this system.

I can attest, I had MBC-555, Not fully compatible and comes up to 256K only using 64K x 1 chips, no ISA slot or ISA signals provisions. Due to not quite compatible, I had less interest in it. Back then I was given three MBC-555 computers with no options, floppy drives only, with office software. No games to play on and that was in 1989 or so. The bios is loaded into computer via boot disk during start up. Keyboard nonstandard too.

Did people figured out to modify MBC-555 into fully compatible XT complete with 360K and executes standard Dos 3.xx and standard games without issues?

Cheers,

I think some just plopped a 5150 BIOS into it. That's what I did for my Sanyo MBC-775, except I used a Turbo XT BIOS.

pentiumspeed wrote on 2023-02-10, 02:38:

I can attest, I had MBC-555, Not fully compatible and comes up to 256K only using 64K x 1 chips, no ISA slot or ISA signals provisions. Due to not quite compatible, I had less interest in it. Back then I was given three MBC-555 computers with no options, floppy drives only, with office software. No games to play on and that was in 1989 or so. The bios is loaded into computer via boot disk during start up. Keyboard nonstandard too.

Did people figured out to modify MBC-555 into fully compatible XT complete with 360K and executes standard Dos 3.xx and standard games without issues?

Cheers,

"It's not a bug, it's a feature"

The computer line is from 1982..
Before 1985 or so, IBM was no authority in the microcomputer market.
It rather was Microsoft, well known by its BASIC products - and MS DOS.
Edit: Microsoft also tried to make a name as an Unix specialist for itself at the time.
Microsoft had Xenix as its flagship product.

Up until version 3, MS-DOS was sold to the OEM market, where it was modified for custom hardware.

In the early 80s, there were not only IBM PC-compatibles, but also MS-DOS compatibles (a forgotten breed).

PCs that were on purpose not a 1:1 copy of the IBM PC.
Not seldomly their companies tried to make better than IBM.

Better audio/video capabilities, unconventional floppy drives,
different/faster serial ports (non-8250 based),
16-Bit slots or a higher performance 80186 based design (non IBM compatible SoC).
Examples: Tandy 2000, Siemens PC-D, Sirius-1/Victor-9000, Olivetti M24 (IBM compatible, but with proprietary 16-Bit slot extension),..

https://www.homecomputermuseum.nl/en/collecti … ndy/tandy-2000/

https://www.homecomputermuseum.de/sammlung/de … /show/sirius-1/

https://www.retrospace.net/infoseiten/readm.php?id=48

http://www.duensser.com/pc_pcd
http://oldcomputer.info/pc/SiemensPCD/index.htm

Of course, to the untrained eye, the pure consumer, these PCs are inferior.
Because they don't imitate IBM's defects.

Unfortunately, the 80s and 90s were full of poorly written IT articles and magazines which didn't remember this aspect of history.

And so in our collective memory we learned and accepted that non-IBM compatible PCs
solely existed due to sheer incompetence of the other PC makers of the time.

Not considering the Taiwan clone makets of cheap IBM PC mainboards and accessories that had no trouble making duplicates.
And not considering how primitive the IBM 5150/5160 design really was at its core.
Every IT student could have come up with that, if needed.
The "ingenious" BIOS idea could have been borrowed from CP/M.
With the PC slots were borrowed from Apple II.

"To stay young requires unceasing cultivation of the ability to unlearn old falsehoods."— Robert A. Heinlein.

I mean, let's just look at the 555's bootable BIOS..
Its functionality could be updated by simply booting a different floppy diskette.
- Oh and it saves production costs (small boot ROM) and doesn't occupy precious address space permanently ("clean computer" concept). 😁

An XT by comparison had to have its ROM set replaced for an upgrade,
which wasn't easy to do for the laymen at the time.

If the Sanyo PC machine was accepted back then, a few more software releases may have been available, too..

http://www.computinghistory.org.uk/det/8932/Sanyo-MBC-555/

Edit: Just double checked this posting. I find it not to be very good.
So I feel the the need to say: Please don't get me wrong.
My response was not meant as an criticism to pentiumspeed in any way .
What I wrote were my thoughts on the topic/matter only.
MS-DOS compatibles are a piece of history in their own right, I think.
On the internet, they are too often being considered as failed IBM compatible PCs.
Which, however, was something not all manufacturers aimed for (PC compatibility).
Some MS-DOS compatibles were aimed torwards a special application.
Say, being a fast platform for running AutoCAD, dBase or Lotus 1-2-3.

Those PCs were sold with modified versions (ports) of such business software.
That's what they were primarily designed for.
Of course, having the ability to run other DOS applications was favorable, too!

That's why these BIOS simulators were being written.
To also support certain DOS utilities or internal company software.
However, these simulators were not beibg written to be running all the time.
The IBM PC personality was more of a masquerade, an imitation of a foreign accent.
Otherwise, the computer was designed to be IBM PC compatible right from the start.

Yeah, the point being that it was the BIOS that made it compatible, remapping all hardware to a common interface, so if it doesn't quite mesh with what is on the board, problems will ensue... Particularly if the BIOS was intended for a 188 or V40 chip with internal peripherals like timer etc which needed to be "broke out" to normal addresses for PC.