Trying to fix Mikro's wrecked TT board

[dml]

The TT will be safe to power up with just the 5V rail. There's nothing which requires a bias voltage other than audio amplifiers and they just don't work if they don't have the other voltages.

Great - thanks for the info!

[dml]

So that was extra tedious but did clear up some uncertainty.

I set up the DMM to stabilise for a few hours and same for the linear supply I'm using. I used a guard wire to help with the DMM readings.

Injected 1.022V again at a corner of the board (closest I could get to settle on that supply) and measured the decoupling caps all over the board.

There is in fact no drop anywhere on the 5V rail. There are odd readings in some places but all of the ones I checked are floating for one reason or another - most tie back to one of the other rails or to an internally generated rail (e.g. one is 8.2V).

So I think I'm done with the voltage injection measurements.

Next I'll check the board resistance again, more carefully this time.



C116 1022 CPU/FPU area
C114 1022
C117 1022
C113 1022
C111 1022

C216 1022 dip switches

C202 1022 MCU/TTVIDEO area
C205 1022
C211 1022
C212 1022
C214 1022
C219 1022

C300 1022 MIDI
C301 1022 ACIAs
C306 ~90 (+12 rail on parallel/serial schematic)
C312 1022
C315 1022 expansion area

C401 1022 RTC / FDC / ACSI area
C406 ~511 RTC crystal filter
C407 1022
C408 1022
C409 1022 DMA decoupling
C410 1022 WD1772 decoupling
C411 1022
C413 1022

C515 1022 DRAMs

C602 1022 ROMs
C603 1022
C605 1022
C607 1022

C706 1022 logic ICs, front of PCB
C702 1022
C705 1022
C709 1022

C802 1022 mainly audio systems, SNDSH, LCM1992, YM
C803 1022
C808 1022
C814 509 LF347 OPAMP
C815 1022
C816 1022 SNDSH decoupling
C821 523 DAC? area
C822 1022
C827 ~55 LMC1992 8.2v decoupling
C834 ~28 near LMC1992
C842 1022
C844 ~90
C851 ~55

C900 1022 LAN/serial area
C902 1022
C903 1022
C909 1022
C911 1022
C917 1022
C918 ~90
C920 ~80
C922 1022

CA00 1022 SCSI
CA02 1022
CA04 1022

CD03 523 video DAC / output stage
CD06 1022
CD07 1022
CD08 523
CD09 ?? ~80
CD10 1022
CD11 523
CD12 523

[dml]

And today I'm getting different resistance readings again but this time at least it is holding stable over time. After the tidy up, messing with the cut PCB edge doesn't change the readings this time.

GND ... +5V 27.5R (both directions)
GND ... -5V 70R
GND ... +12 190R
GND ... -12 70R

Still no indication why +5V would have such a low resistance after sampling less than 1mV dropout across the board. It is low enough to be suspect but high enough to be difficult to pin down.

[edit]

Maybe I should try a quick version of the 1V test again but this time with a very low current limit, so there is already a significant drop at the injection point - that should at least narrow down roughly where on the board the current is sinking (?)

[stephen_usher]

With the broken edge I would grind that at a 45 degree angle (or even finer) so as to separate the exposed edges of the layers as much as possible. I also see that there's a via linking the top and bottom ground layers. I would grind the board away so that this via is gone, in case there's a potential short along the via to the other layers.

The reverse resistance between the +5V and GND could be due to some voltage clamping zener diodes, which will conduct it the "opposite" direction, but I can't remember if there are any on the video connection on the TT. It is suspiciously low though. However, if there is a short it would likely be in a passive component rather than a chip, and if it's a chip then it would need to be changed anyway. It's at this point when you could do with a thermal camera and send, say, 1V down the +5V via a bench power supply and then look for hotspots.

[dml]

With the broken edge I would grind that at a 45 degree angle (or even finer) so as to separate the exposed edges of the layers as much as possible. I also see that there's a via linking the top and bottom ground layers. I would grind the board away so that this via is gone, in case there's a potential short along the via to the other layers.

I have it down to maybe 35' now and can see the two inner layers are separated. I will look more closely at the via though.

The reverse resistance between the +5V and GND could be due to some voltage clamping zener diodes, which will conduct it the "opposite" direction, but I can't remember if there are any on the video connection on the TT. It is suspiciously low though. However, if there is a short it would likely be in a passive component rather than a chip, and if it's a chip then it would need to be changed anyway.

Yep that's what I was thinking - there are only 5 'big' ICs left on the board - not in sockets - and maybe only one I might have difficulty sourcing/replacing. Either way I probably won't make things much worse with 5V.

I was a bit surprised the resistance didn't seem to rise with so many ICs removed.

It's at this point when you could do with a thermal camera and send, say, 1V down the +5V via a bench power supply and then look for hotspots.

I'm very close to trying that now but will probably need to wait until evening - the IR camera picks up reflections too easily during daytime.

[dml]

Since this is developing into a 'weird gear' opportunity ...I dug this old thing out of the back of the cupboard:

Screenshot 2025-04-13 at 12.51.06.png (1.84 MiB) Viewed 108 times

[edit]

Quickly went round the board while injecting 100mV with the precision source (which I figured would exclude semiconductors). There's a drop to 73mV at the injection point (power socket) because there is low impedance across the 5V rail and the precision source is for higher impedances - but everywhere else on the board is also 73mV.

To make *doubly* sure, I injected the 100mV at the ACIA and measured at the power socket - same result. 73mV at both ends. So I'm done with that angle I think

Since the current drain seems to be equal everywhere - with or without the socketed ICs present - makes me wonder if every one of those crumbly self-destructing decoupling caps is leaking something. But even then they wouldn't all leak equally.

[dml]

Ok new weird unexpected thing...

CR403 - a 1N914 diode tied to the RTC chip and very very close to the previously smashed PLCC socket for the SCSI controller - and which was also missing the SCSI IC - is reading 0.3V drop in both directions.

Most interesting of all, the RTC chip isn't even on the board. I don't think that can be explained away by R405, C405... so I might need to lift that diode and test it out of the board.

Screenshot 2025-04-13 at 14.09.53.png (91.13 KiB) Viewed 94 times

....nope, diode alone checks out ok. The 0.3V is across the CR403 pads.

Wonder if this is just a board revision issue. :/

I think this one is a herring, not lifting the diode first.... moving along....

[dml]

Daylight is fading now so I started some tests applying voltage with the thermal camera.

Already interesting - the only IC to show any temperature change at 1.5V is the YM2149.

Screenshot 2025-04-13 at 19.11.17.png (173.12 KiB) Viewed 63 times

At 2.0V, this becomes a bit more visible but a bit more unexpected is the 3x terminating resistor networks at the SCSI port are warming up.

I actually thought about removing those during the earlier resistance tests but from the schematic it didn't seem relevant unless something was plugged into the port.... hmmm.

Screenshot 2025-04-13 at 19.10.36.png (191.33 KiB) Viewed 63 times

YM2149 datasheet says it should dissipate 125mw @ 5V which is not much. Surprised to see anything visible at all @ 1.5V...

[stephen_usher]

The resistors have been installed the correct way around haven't they? Just a thought.

[PhilC]

The trouble with thermal cameras is interpreting the results. I'm presuming that the 29C is indicating the max temperature? I would have thought that to be normal for the resistors. Does your power supply give you a current draw?