Page 1 of 9

MTX 512 (debugging thread)

Posted: 08 Jun 2019 11:07
by GozdniJezek
Hello, i have recently bought a MTX 512 in untested state, it will need some love before it will be in its full working order and cosmetic appearance. :)
1.) it's missing the left blank key (the q was removed by me prior to taking the picture), if i understand correctly i can replace it with modern Cherry one?
IMG_20190604_121158.jpg
IMG_20190604_121158.jpg (1.45 MiB) Viewed 26545 times
2.) My MTX also has a custom board in place of PAL A (if i read the schematic correctly)
IMG_20190604_201027.jpg
IMG_20190604_201027.jpg (2.23 MiB) Viewed 26545 times
The board itself looks like:
IMG_20190604_201708.jpg
IMG_20190604_201708.jpg (1.21 MiB) Viewed 26545 times
when in place it looks like:
IMG_20190604_115917.jpg
IMG_20190604_115917.jpg (1.91 MiB) Viewed 26545 times
3.) One thing that concerns me a bit is that the C1 is cut in half, it looks like it might be intentional (I know this should not impact the general working of the computer), this might need some attention in the future.
IMG_20190604_114134.jpg
IMG_20190604_114134.jpg (623.99 KiB) Viewed 26545 times

Re: MTX 512 (debugging thread)

Posted: 08 Jun 2019 11:11
by GozdniJezek
4.) PL50?? (middle one from the three, right of Z80)(i can't read the schematic- where can I find one in hi res?) has been replaced at some point as the soldering marks are clearly visible.
IMG_20190604_115706.jpg
IMG_20190604_115706.jpg (2.8 MiB) Viewed 26544 times
IMG_20190604_115713.jpg
IMG_20190604_115713.jpg (2 MiB) Viewed 26544 times
5.) When i turn the computer on i get the following picture on composite output (monitor output does produce composite right?), I have yet to test it with RF.
IMG_20190604_195402.jpg
IMG_20190604_195402.jpg (1.18 MiB) Viewed 26544 times
6.) A pin on Z80 was bent, so this makes me believe the machine has been tinkered with in the past, so there might be other "human induced issues" :), so my plan would be to reseat and clean the pins on everything that is socketed.
7.) When i turned it on, Z80 and TMS9929ANL are warm (not hot) to the touch. I have tried to turn it on without the ROMs the results were the same.
8.) Any suggestions/advice on what to check and where to focus?
9.) If this general diagnostics fail I'll have to start with multimeter and oscilloscope debugging (I have an electronics engineer friend that can also help)


Thank you for your help,
Rihard.

Re: MTX 512 (debugging thread)

Posted: 08 Jun 2019 12:16
by stephen_usher
As you have an engineer friend, probably the first thing to do will be to upgrade the ROM sockets to allow you to use EEPROMs:

http://www.primrosebank.net/computers/m ... eplace.htm

Then write a set of normal ROMs (you have the later three 8K ROMs) plus Martin's Diag ROM. Install the Diag ROM in the ROM A socket and switch on. As this doesn't require working memory you should at least get something and get some diagnostics working.

If it doesn't work with the Diag ROM you may need to replace the CPU, which are cheap enough and can be got from Farnell or RS Components.

From the look of the screen you may have some bad video memory chips.

As I noted on the Facebook group, someone has changed Links 6 and 7 so as to set it up for the full 64K of RAM even though the PAL is a white dot (MTX500) one. It would probably be best if you get a GAL written with the 3x8K ROM MTX512 program. This would replace that daughter board entirely.

You may also like to read my "Resurrecting and MTX500" thread: viewtopic.php?f=9&t=419

P.S. I've just noticed that you seem to have some cut traces on the underside of the PCB, near the resoldered chip!

Re: MTX 512 (debugging thread)

Posted: 08 Jun 2019 12:18
by Dave
Circuit Diagrams are in the Operator’s manual.

You can download the manual, or a smaller download of just the circuit diagrams from this page:
http://www.primrosebank.net/computers/m ... anuals.htm

When referring to specific chips, it’s better to specify the PCB location. Chips are laid out in a matrix, e.g., 2A is the 74LS244 adjacent to the keyboard connector. The circuit diagram includes the board references.

Are you referring to the chip in position 9D, the LS04 used in the clock generation circuit? You’ll need a ‘scope to check the clock signal

Re: MTX 512 (debugging thread)

Posted: 08 Jun 2019 12:24
by Dave
As I said on Facebook....
5. Regardless of how it appears on the outside, you appear to have a 32k machine (MTX500). The mod **might** have been someone trying to redo the address decode logic in the hope that the 32k chips could be made to work as 64k. (Those chips were 64k that had failed factory tests on one bank and were sold as 32k chips)
There is a reason that Memotech used those RAM chips, trying to make them work as 64k is Ill advised and we don’t know if the machine ever worked in its current configuration. You could do away with the bodge board and use a proper PAL/GAL, but I think that a better plan is to return the machine to a true 32k setup and take it from there.

Re: MTX 512 (debugging thread)

Posted: 08 Jun 2019 12:31
by Dave
If you don’t have a constant tone from the sound output, that suggests that the machine has at least tried to execute some of the ROM code, as the default startup has some sound from the chip which is silenced early on in the ROM code. That suggests that the CPU has run something and so has a good clock. As I said though, you should check the clock with a ‘scope.

If you hit <Control><G>, you should get a bell tone, even if there is no screen output, you’ll need to have the audio output connected obviously, or use the TV out.

Typical VRAM problems are indicated by rubbish on the screen, rather than a blank screen, with some working VRAMs you would likely get some, if corrupted, text. I think that a VDP fault, or a ROM fault, is more likely that a VRAM fault in this case.

The dodgy colours on the screen are pretty low down the list of things to fix first, but suggest to me a fault on the Video board as shown on this page http://www.primrosebank.net/computers/m ... _board.htm

Re: MTX 512 (debugging thread)

Posted: 08 Jun 2019 12:57
by stephen_usher
Dave wrote: 08 Jun 2019 12:24 There is a reason that Memotech used those RAM chips, trying to make them work as 64k is Ill advised and we don’t know if the machine ever worked in its current configuration. You could do away with the bodge board and use a proper PAL/GAL, but I think that a better plan is to return the machine to a true 32k setup and take it from there.
Indeed, putting the PAL back in without the extra board and changing LK7 from the '4' position to the '2' position and LK6 from 4-C to C-H (as he has the 'H' specification OKI RAM chips) would do that.

Re: MTX 512 (debugging thread)

Posted: 08 Jun 2019 13:24
by Dave
The replacement chip in 9D is a CMOS (HC) device, rather than the original high speed TTL (S) chip.

This **might** be OK, but should really have been replaced like-for-like. If you don’t get a good clock signal on a ‘scope, you could put in the proper chip

Re: MTX 512 (debugging thread)

Posted: 08 Jun 2019 14:09
by stephen_usher
I'd also be concerned about the damaged traces around the 10D/10E area.

Re: MTX 512 (debugging thread)

Posted: 08 Jun 2019 14:33
by Martin A
A couple of thoughts:

The change at 9D looks like someone's done the timing chain mod from page 47 of the service manual, R32 and R33 look to have ben replaced, you can only see one of them in the photo, but the wrong colours for a 560R. R35 and R36 have been pulled as per the mod. The CMOS chip's drive is symetrical, it doesn't need pull ups.

Page 46 also advises changing R14 to 330R and changing C6. Both C6 and R14 can be seen in the "Broken C1" shot, and both have evidence of being replaced. The resistor isn't 680R any more :) though the capacitor can't be read, there is evidence of re-work, hopefully closer inspection will show that's 33pf instead of 68pf to complete the mod.

I second the idea of trying to set the system back to factory state, by setting the links back to those for 32k high half ram and re-inserting the PAL back in it's socket. If the only issue is the dodgy memory mod, that would fix it.

If the screen output is really blue, and not some artefact of the TV indicating no signal, then the system setup must have started. The VDP doesn't do blue by default.