MTX 512 (debugging thread)

GozdniJezek
Posts: 28
Joined: 07 Jun 2019 20:49

Re: MTX 512 (debugging thread)

Post by GozdniJezek »

Thank you for your replays, I'm still planing on swapping out the VDP crystal, because it's easy and cheap to source. But today I watched an episode of Noel's Retro Lab, where he fixes a SVI 328, that also uses the the same graphics chip TMS9929ANL (VDP) as MTX.

https://www.youtube.com/watch?v=KSrMwfn ... e=youtu.be

His symptoms are identical to mine, missing video and no clock on VDP. So as a backup plan if swapping the crystal won't help I would also like to source a new TMS9929ANL. Chinese sell these, but from the feedback they are mostly fake/non working. Does anyone have any suggestions where i might find a good one?

My reasoning is that its more likely that the VDP is causing the clock issue than the other way around, because for a couple of seconds if the machine is cold all works fine and the picture is in sync and without artifacts. I presume if the crystal would be at fault it would not work from the begging and not fail after the machine gets "hot" (runs for a while).

Does this sound reasonable? As always any suggestions are most welcome and thank you for your help.
stephen_usher
Posts: 217
Joined: 27 Nov 2016 19:58

Re: MTX 512 (debugging thread)

Post by stephen_usher »

If it were the VDP I'd not expect it to work at all.

Seeing as it works until warming up I'd get a freeze spray and see what part of the system works when cool.
User avatar
1024MAK
Posts: 627
Joined: 24 Dec 2012 03:01
Location: Looking forward to summer, in Somerset, UK

Re: MTX 512 (debugging thread)

Post by 1024MAK »

Some designs of oscillators don’t like it when a ‘scope probe is used on any of the crystal connections.

Try probing the TMS9929 pin 37 (GROMCLK) and see what signal (if any) you get there. If the 10.7MHz clock is running, then there should be a GROMCLK signal (although I have not had a chance to confirm this on any of my machines yet).

There is a comprehensive manufacturers document on the TMS9929 (and the other versions) here http://www.bitsavers.org/components/ti/ ... _Nov82.pdf

If the clock does not seem to be running, I would certainly try replacing the crystal first.

The -5V rail is in spec. and is fine at -5.3V. So no need to worry about that.

Where in the world are you? We may be able to assist if the conclusion is the chip being faulty.

Mark
GozdniJezek
Posts: 28
Joined: 07 Jun 2019 20:49

Re: MTX 512 (debugging thread)

Post by GozdniJezek »

I've ordered a new crystal as they are cheap and easily obtainable and will try with that first. Secondly I've found a "tester" for the TMS9929: https://8bithardware.wixsite.com/websit ... anl-tester . So if the crystal swap won't work this is what I plan on doing next.

I'll measure the clock on GROMCLK before I swap the crystal and see what's happening. The freeze spray is not a bad idea, but I'll need to find one locally.

So my next plan of action in order:
-Measure pin 37 to see if out clock is giving any signal
-try the freezer spray (if I'll be able to find one)
-swap the clock crystal for VDP
-"test" the VDP
-try to find the replacement VPD (last resort)

Does this sound reasonable? I'm from Slovenia (EU) :).
Does anyone know if the 9929 is swappable with any other chips from the same family: TMS9918AITMS9928AITMS9929A ? TMS9129NL and TMS9929ANL seem pin compatible, but I'm not sure if that is the case.
From the documentation:
"The TMS9928A/9929A VDPs are functionally identical to the TMS9918A except that the NTSC color encoding circuitry has been removed and replaced with luminance and color difference signals. The TMS9918A is pin-for-pin compatible with the TMS9928A/9929A, except for three pins, the composite video output, the external video input and the CPU clock output. These pins are replaced with the Black/White luminance and composite sync (V) output and two color dif-ference pins, Blue (B-VI and Red (R-VI outputs, respectively."
If I understand this correctly, this means that only the output is different?

As always thank you for your suggestions.
User avatar
1024MAK
Posts: 627
Joined: 24 Dec 2012 03:01
Location: Looking forward to summer, in Somerset, UK

Re: MTX 512 (debugging thread)

Post by 1024MAK »

The TMS9928 and TMS9918 chips are designed for the American NTSC TV system. Only the TMS9929A can produce a 625 line, 50Hz PAL video signal.

The TMS9129 can’t be used, as it’s designed for a different type of DRAM chip. So it’s address and data lines to the DRAM is different. Well, unless you make changes to the PCB wiring.

See here

Mark
GozdniJezek
Posts: 28
Joined: 07 Jun 2019 20:49

Re: MTX 512 (debugging thread)

Post by GozdniJezek »

Hello, this weekend I had a new debugging session with my friend, this time a bit more methodical, as poking around did not bring any meaningful results. :)
I had a sense to use the oscilloscope before randomly swapping out the oscillator and other components. :D
I believe the underlying issue for no picture is indeed the missing sync in the composite signal as Mark previously suggested. If my understanding is correct, our output is missing the line sync pulse. Source of information on composite video.

Y(yellow)_composite(blue).jpg
Y(yellow)_composite(blue).jpg (30.27 KiB) Viewed 903 times
(image 1, yellow channel is Y blue channel is output on composite connector). As also Mark previously suggested, we also measured the gclk on VDP and it looked fine (it was oscillating).
resetOnVDP.jpg
resetOnVDP.jpg (185.56 KiB) Viewed 903 times
We also measured reset (pin 34) on VDP, which looked strange because it should be stable at high and not oscillating. The image appeared on the screen for a brief moment at "long" periods of high signals.
4070Input1Output.jpg
4070Input1Output.jpg (166.84 KiB) Viewed 903 times
So we took a look at the reset circuitry, so the input on pin 1 to the 4070 (XOR gate) is floating (blue channel is the output from 4070, yellow is the pin1 input to 4070). If we jumper the J1 pin 19 and 20 (reset on keyboard) the reset stays low as expected after initial boot, so i presume the 4070 works ok.
reset.JPG
reset.JPG (43.12 KiB) Viewed 903 times
We think that something on the reset line might be pulling the pin 1 on 4070 (via the resistor R64) to floating in the forbidden band and triggering reset on random/periodically . Does our findings so far look reasonable?
So the question now is, how to identify the culprit on the reset line? I was thinking on making a list of every chip on the reset line and checking/removing them one by one? If I understand correctly I can remove all of the chips that are socketed and the reset should still work? If that won't help, then one of the logic chips is the faulty one.

Does this sound reasonable or would you check something else?
As always thank you for your help.
User avatar
1024MAK
Posts: 627
Joined: 24 Dec 2012 03:01
Location: Looking forward to summer, in Somerset, UK

Re: MTX 512 (debugging thread)

Post by 1024MAK »

Reset circuit
Reset circuit
A6C6AB17-42D3-47F3-8C3D-3346B8EF306D.jpeg (51.37 KiB) Viewed 885 times
Resistor R18 (10kΩ) should pull up the input (pin 1) of the 9C logic gate (4070B) to nearly 5V. R64 (820kΩ) is only there to provide a little bit of positive feedback from the output to the input to speed up the translation when switching between logic levels.

Capacitor C11 (10uF) provides the power on reset function. Diode D13 discharges C11 when the power is switched off.

Using a digital multimeter or ‘scope, what actual voltage level do you get across J1/19 and J1/20 or across C11?
And what voltage do you get at pin 1 of IC 9C?

At this stage, I don’t see how any other chip could be causing problems with the reset line, as your photo of the ‘scope screen showing output of chip 9C shows good logic levels even if the waveform is erratic.

Can you also see what the level of the noise and ripple is on the +5V supply rail. Set the ‘scope to AC coupling and use a suitable mV range. Please post a photo of what you see.

Things to check:

With the power off, what does resistor R18 measure? Please post up your result. It may not read 10kΩ due to the effect of other components in the circuit.
Is there any contamination of the PCB in the area of any of the components in the input circuit of IC 9C pin 1? If yes, clean off the contamination using IPA.
Has C11 been renewed or is it the originally fitted component? Does it look in good condition? If it has been renewed, has it been fitted with the correct polarity?

Mark
GozdniJezek
Posts: 28
Joined: 07 Jun 2019 20:49

Re: MTX 512 (debugging thread)

Post by GozdniJezek »

Hello, thank you for your support, the reason my posts are so intermittent is that i have the machine at my friends house, that has a formidable laboratory with some very useful tools but I can't visit very often because of the distance. I've kindly asked him for the measurements and this are his findings (I must say I'm impressed with his work) :D

Values of reset circuit elements are (all measurements were made on elements in circuit without removing elements. Influences of other parallel elements could be present).
1.jpg
1.jpg (51.31 KiB) Viewed 842 times
R18 =10.22 kΩ
D13, diode measurement: 0.553V in one way, OL in other way
R16= 98.3 kΩ,
R17=32.2 Ω
R64=786 kΩ
Contamination around pin1 of IC 9C? A little contamination on bottom side of PCB near positive leg of C11 (near via and D13 leg). Cleared with IPA, no contact between via and C11 positive leg, contact between D13 and C11 (as required by schematic). Without other contamination.
C11 was replaced with new 10 uF 50V electrolytic capacitor. Correct polarity-negative leg toward pin 2 of IC 9C and positive leg toward R16/R18/R17. I could not measured capacity in circuit, but I didn't removed from PCB to check. Could be good to check? Only capacity or is also important to measure ESR?

Measurements between working:
Voltage between J1/19 and J1/20: between 2.94 and 3.0 V (measured with multimeter), voltage across C11 also around 3 V.
Measurements with oscilloscope:
Note: GND point for oscilloscope measurements was taken at video output connector. Long GND connection could influenced to measurements. For better measurements, GND point should be taken at IC 9C GND pin or negative leg of C11 (problem with space for GND clip of probe)
2.jpg
2.jpg (694.31 KiB) Viewed 842 times
Pin 3 of IC 9C (reset output)
3.jpg
3.jpg (466.58 KiB) Viewed 842 times
Pin 1 of IC 9C (input 1)
4.jpg
4.jpg (317.48 KiB) Viewed 842 times
IC 9C power supply voltage (VDD-pin 14, DC coupling). Sorry for bad photo. Informations: 1 V/dec, 500ms/dec
5.jpg
5.jpg (468.06 KiB) Viewed 842 times
IC 9C power supply voltage ripple (VDD-pin 14, AC coupling)


As always I really appreciate your help, thank you for all your efforts.
User avatar
1024MAK
Posts: 627
Joined: 24 Dec 2012 03:01
Location: Looking forward to summer, in Somerset, UK

Re: MTX 512 (debugging thread)

Post by 1024MAK »

The voltage across C11 should get very near the supply voltage (within 0.5V), certainly much higher than 3V.

There is something dragging the voltage down, or rather something preventing it from climbing up.

There are only three components that can cause this, the keyboard, C11 or the chip (9C, 4070B). But before changing anything, can you confirm that the measurements so far have been done with the keyboard disconnected.

Can you then measure the voltage across R16. Then from chip 9C (4070B) pin 1 to GND. And again across C11.

Assuming the keyboard is disconnected, if there is any significant voltage across R16 and the voltage on chip 9C (4070B) pin 1 is lower than across C11, then the chip is faulty. Otherwise the problem is C11. If the keyboard was (is) connected, then it could be a fault on the keyboard rather than being C11.

Mark
User avatar
1024MAK
Posts: 627
Joined: 24 Dec 2012 03:01
Location: Looking forward to summer, in Somerset, UK

Re: MTX 512 (debugging thread)

Post by 1024MAK »

There is one other possibility, and that is that there is an unwanted connection between another circuit and the reset circuit (either the track from J1/19 to R17, or the tracks between R17 and R18/C11/D13/R16 or the tracks between R16/R64/chip 9C {4079B} pin 1).

Mark
Post Reply