MTX 512 (debugging thread)

[GozdniJezek]

Hello, yesterday i was again thinkering with the machine.
1.As promised the picture of the underside of the custom PAL board.

IMG_20190611_155201.jpg (346.74 KiB) Viewed 26340 times

IMG_20190611_160810.jpg (200.53 KiB) Viewed 26340 times

2.) I have cleaned all of the socked IC pins and made some progress:

IMG_20190611_170054.jpg (163.57 KiB) Viewed 26340 times

IMG_20190611_165815.jpg (76.28 KiB) Viewed 26340 times

-at the beginning i was greeted with:

IMG_20190611_163311.jpg (141.06 KiB) Viewed 26340 times

3.) so the simptoms now are:
-after the machine warms up, it increases the number of artefacts on the screen, the picture starts to wash out (becomes increasingly grey), key strokes produce garbage, the image flickeres (sometimes it goes black for a fraction of a second), keystroke sometimes produces 2 characters, the machine freezes.
-my reasoning now is, that at least one VRAM is bad (the simptoms are very simillar to those described on: http://primrosebank.net VRAM failures)
-washing out indicates problems with video board.
-possible problems with keyboard controlers (A2&A3) ?
-possible other issues.
4.) My next action plan would be to check the VRAM (F4-F7 & G4-G7 ? ) I would check the voltages on the IC (trivial) and try to short pin 14 (with a resistor) to the ground. A "IC warmth" test would also help to identify the faulty IC. When i get a replacement i would piggyback the good chips. Any other suggestions?
5.) Of course i would socket every IC I take out.
So for the VRAM would this sockets be ok:
https://www.aliexpress.com/item/100PCS- ... e19691aa59 -is the pitch on this ICs standard? (A noob question sorry )
Since i'm ordering i would also go for:
https://www.aliexpress.com/item/20pcs-I ... 195949c748 for the keyboard controlers - 20P?

Regarding the VRAM ICs, if i read correctly they are 4116 RAM Chip, I'm thinking of ordering them from here:
https://www.retroleum.co.uk/zx-spectrum-chips
Since i would buy other stuff for C64 and so on i would combine shipping. Or is it any other (modern?) alternative you would recommend (without too much hacking)?

[stephen_usher]

Seeing as it's a problem with warming up the next course of action should be getting a can of freezer spray and trying to cool some chips down one by one to see if things change.

Doing a temperature check of chip warmth is another good test. (The video processor should be hot, that's the way it is.)

Oh, and is R10 (the big one just above the ROMs) hot? If so then the TIP transistor is dying and the regulator is trying to supply almost twice the current it's built for, so that could also be the problem. If the +5V line is dropping to a lower voltage that can cause strange effects.

[GozdniJezek]

Sorry for my ignorance, can you please elaborate what R10 is and where might i find it? Is it a resistor or something else? It would seem, it would be a good idea to check the voltages on the board at this point. Thank you.

[Dave]

Have you repositioned the wire links for LK6 and LK7 as shown in the service manual reference that I gave you?

The last photo we have shows them in the 64k RAM position, they should be in the 32k with “H” RAMs position

Keystrokes “producing garbage” could be a VRAM fault (yes, really!)

Repeating keys is likely to be a key switch “chatter” problem, best to try and fix that after everything else


You are likely to get lots of “helpful” advice here, from more than one person, and often in a pretty unstructured way - including from myself

In order for us to best help you, I suggest that you try and follow a structured approach, with a good plan before you start, and documenting/reporting back after each step. That way, we all know what’s happening.

If I may be so bold . . . .

Step 1 - repin LK6 and LK7 (take a photo)
Step 2 - check the voltages
(I think that Steve may have made a typo about “R10”, he may mean R62 - see this page for why
http://www.primrosebank.net/computers/m ... _power.htm, it has a VALUE of 10R)
Pause

[Dave]

Step 3 I agree that one or more VRAMs appear to have issues, and testing/replacement is likely to be needed

Yes, the narrow sockets in the MTX are all standard (0.3”) pitch

Those RAMs you mentioned will work, but any replacements will be pretty much the same age as the ones you have and they are not the most reliable chips in the world. You may like to consider using a more reliable 64k chip instead.

See here for how to make it work http://www.primrosebank.net/computers/m ... mavram.htm

It may be worth “piggy backing” a good 4116 on top if the current VRAM, but personally, I think this is unlikely to work and replacement will be necessary

[stephen_usher]

(I think that Steve may have made a typo about “R10”, he may mean R62 - see this page for why
http://www.primrosebank.net/computers/m ... _power.htm, it has a VALUE of 10R)
Pause

Yep... Oops. I was using my memory on the fly. The 10R came into my brain.

[GozdniJezek]

Apologies,
the current state of the machine is:
I have not changed the configuration of the memory (it runs in 64 mode with that weird doughter board).
The only change has been general cleaning of the pins.

You're not bold, i'm preaty shure you all have vastly more experiance than me, and you have my respect and you probabbly see something i don't, It's just i need to understand why I am doing something.

As i'll have to solder anyway it sounds reasonable to return the machine to 32k state. (can you just please educate, what faults we will eliminate by doing this) (might even just install a jumper, for easier modding back to 64k will that work?)
1.) If i understand correctly the 4164 is 64k so only 1 modified IC is needed? The others can be just removed. Does it function well with a socket? (a noob question sorry)
2.) as I'll only need 1 4164 and it's not that expensive i'll just order it. Does that sound reasonable?
3.) AI'll go by your suggested list, returning the machine to 32kb, and checking the voltages.
Thank you for your help.

[Dave]

Apologies,
the current state of the machine is:
I have not changed the configuration of the memory (it runs in 64 mode with that weird doughter board).
The only change has been general cleaning of the pins.

You're not bold, i'm preaty shure you all have vastly more experiance than me, and you have my respect and you probabbly see something i don't, It's just i need to understand why I am doing something.

As i'll have to solder anyway it sounds reasonable to return the machine to 32k state. (can you just please educate, what faults we will eliminate by doing this) (might even just install a jumper, for easier modding back to 64k will that work?)
1.) If i understand correctly the 4164 is 64k so only 1 modified IC is needed? The others can be just removed. Does it function well with a socket? (a noob question sorry)
2.) as I'll only need 1 4164 and it's not that expensive i'll just order it. Does that sound reasonable?
3.) AI'll go by your suggested list, returning the machine to 32kb, and checking the voltages.
Thank you for your help.

We don't know that returning the machine to a 32k state will definitely eliminate any faults. As noted previously, the issue with leaving it in its current configuration is that you are trying to use chips that the manufacturer only certified for use as 32k as 64k chips. They were manufactured as 64k, but failed factory testing on one bank of memory cells, so were sold as working "half" chips. Yes, they **might** work, but will always be suspect. They could have hard faults, intermittent faults, or no faults at all - we simply don't know. If you can install jumpers to easily swap between configurations, then that is a good way to go. As I said before, we don't know whether the machine ever worked after this ill advised mod was made.

1. No - you need 8 chips as the memory addressing is set up for 8 chips, using a 4164, you would be using a 64k chip as an 8k one, wasting a lot of space in the chip, but increasing reliability and doing away with the need for -5V (if all 8 were replaced). This is a "nice to have" change though, the simplest, though perhaps slightly less reliable, option is to just do like-for-like replacement. Personally, I just use 4116's and accept the fact that they have a slightly higher chance of failing again, but if you are fitting sockets (as you should) then that is not really an issue.

2. See point 1 - it's really a matter of personal preference, but depending on how many might have failed, you will likely need more than 1

3. Good plan

[GozdniJezek]

So using 4164 and modifying it, is just "crippling" it to work as 8k. Got it, thank you.
If I'm socketing everything i would also prefere to use 4116 as the price is the same and i don't need to modify them.

Will report back when i'll make any progress. Thank you!

[1024MAK]

Piggy backing only works sometimes, and not very often. So I think it is not a very reliable method of testing. The problem is that DRAM chips are very complex devices and there are countless ways that they can fail.

If the original fitted chip decides to output a logic low at the same time that the new piggy backed chip tries to output a logic high, which one wins the fight to change the logic level on the data bus?

For chips that only have one output pin, like 1 bit DRAM chips, you can however carefully cut the output pin of the suspect chip. Bend out the bit of cut leg still in the PCB. Bend out the leg of the output pin of the new chip. Piggy back the new chip on top of the existing chip. Temporary join the bit of cut leg still in the PCB with the leg of the new chip. Then power up...

Myself, I prefer to cut out the old chip completely (cut as close to the body of the chip as you can). Remove the legs by carefully desoldering them (so that the PCB tracks, pads and through-hole plating are not damaged). Fit a socket. Then put a new chip in the socket.

Mark