[sdiy] Matrix 1000 repair, unexpected hack

[.]

Hello,

I offered myself  a cheap matrix1000 that needs some care (so good i am). 

After opening it, i stumbled across a weird thing, pin24 (Clear to Send) 
of the 68b50 ACIA (IC U809) is purposefully unsoldered and left unconnected 
(by human intervention) and the timer(UM82C54) U808 just aside it has a wire 
directly solderd on pin 14 (Gate 1). The wire is cut at 2 centimeters and 
left unconnected too. The previous owner told me he bought the unit new in 
the 80's and it was never serviced. In all logic it should be an oberheim 
staff intervention, but what was the purpose of it (it's so ugly), has anyone 
an idea of what it could be ?

Theses 2 chips, U808 and U809, are involved in MIDI handling  but so far MIDI 
IN and OUT seem to work OK, should i worry about the CTS not 
connected ? BTW pin hole 24 on the pcb is connected to GND (like pin 23), but 
the schemos are not saying that they shoud: 

http://perso.netpratique.fr/bouddha/m1kp2.jpg
http://perso.netpratique.fr/bouddha/UM82C54.pdf
http://perso.netpratique.fr/bouddha/60b50.pdf

For now i am unable to understand that, but the real pbm the M1k has is 
somehow unrelated: all patchs seem to have volume pbm depending of the pitch 
of the DCOs.

For instance a pad will play ok only on the upper side of the keyboard, then 
just below C4 for instance, the sound is quite less loud for all the lower 
keys. However, all the 6 voices can be heard, that is good news, and indeed 
on the upper region, it sounds sweet :) There is no hum and the unit seems 
pretty noiseless.

I would say the analog chips are ok, the dac and the muxes seem to do their 
job on the 6 voices, the modulations (env/lfo/etc) seem ok.

Further tests today showed me that this if i put a slow LFO on a DCO, at a 
given pitch, the volume will drop abrutply. It looks like a logic thing, the 
pitch frontier appears to be in a region that could relate to a strong bit of 
a register. 

More, if i put some noise or some resonnance, the level of theses are not 
affected and i can hear the 2 DCOs independetly with different LFO reaching 
the frontier at different time and being affected individually while the 
noise remain unaffected. So it seems its not related to the VCA and indeed 
something happens before the CEMs.

At first i suspected the CPU, but i am not sure how could happen such 
phenomens. And since all the LFO/env, etc are generated by the CPU, if it was 
bad would the synth be partly functionning as it it ?

On the other hand the CPU is quite hot, right after a few minutes, and 
something tells me anormally hot. The unit freezes completly sometimes when u 
edit parameters.

I made the basic cheking stuff, reinit the unit, calibration test (that 
don't seem to show error), mesuring the test points voltage on the circuit 
(all ok).

If anyone had a similar experience with the M6/M1k involving some logic 
(CPU/eprom/etc), i would be glad to know if some parts are know to be 
sensitive, and every other comments are very welcome. 

Thx for reading,

NR

PS: Some other quick links for the m1k schemos:
http://perso.netpratique.fr/bouddha/m1kp10.jpg    (cem)
http://perso.netpratique.fr/bouddha/m1kp3.jpg
http://perso.netpratique.fr/bouddha/m1kp4.jpg
http://perso.netpratique.fr/bouddha/m1kp5.jpg