>>> Foreword
This document is a detailed account of my progress troubleshooting the Ace Tone PS-1000. All of the debugging and repairs described below were accomplished with no documentation, not even a basic owner's manual. I had to rely on deduction, observation, information-gathering and patience...
I'm aware that this is a pretty rare synth, so much of the information contained here will not necessarily be directly applicable to many people, but throughout the text I'll try to shed light on the process and reasonings that led me to the various solutions I found. This, I think, can be useful for many people attempting to service and repair vintage synths. I believe that, when debugging faulty equipment, the most critical part of the process lies in thinking about the problem, analyzing the symptoms, proceeding by elimination, and then, attempting to test or replace a part. Knowing what to look for, and where to find it, makes a difference between five hours spent repairing, and fifty...
Before beginning this repair project, I didn’t know much about analog synth architecture and design. I’ve never designed circuits in my life, I've only done modifications and repairs… I had some knowledge of internal signal routings because of my experience playing synths, but didn’t really know what it actually looks like on a printed circuit board, what components are needed to make a VCO, VCF, etc.
After working on the PS-1000 for so many hours, I’ve learned a great amount, which will surely help me with other synth repairs in the future. There are still so many things I don’t know, but my knowledge has definitely taken a big leap forward…
If, while reading this, you feel that additions and/or corrections are necessary (I'm by no means an expert...), please feel free to leave a comment at the bottom of this page. Any feedback is appreciated.
>>> 1 - Getting inside
Opening the synth was a little complicated ; four screws at the bottom removed the bottom casing, but in order to free up the main PCB part from the front panel, I had to remove every slider cap and knob. By sliding a knife under the caps, I found I could snap them off pretty easily, and once that was done I was able to disassemble the keyboard assembly, main board and panel, bottom and top enclosures.
The main part of the synth consists of two PCBs stacked on top of another: one is for the sliders on the front panel, the other contains all the main electronics. The two panels are connected together by a series of pins, about 30 of them, which made it a bit scary to separate them, because I had to pull out the control board from the slider board, warping it while doing so. I was really worried the first time I did it, but after the twentieth time I wasn't even thinking about it anymore...
fig.1 (left) the main control board. fig.2 (right) the slider board. Notice the pins connecting the two boards
fig. 3 & 4 - Two views of the slider board
fig. 5 - Under the keyboard assembly
>>> 2 - Restoring power
I turned on the synth for the first time, and and it was totally dead, even the power indicator didn’t light up. I followed the cables from the power cable to the transformer, then to the main board where I found two fuse holders… with no fuses in them. I approximated a voltage/current rating from the output voltage of the transformer and the power consumption rating from the back panel of the unit, and popped two fuses in, hoping for a lucky break… No luck, still no power and a dead synth. But at least the fuses didn’t blow and no smoke appeared!
I spent quite a lot of time examining the power supply section: the rectifying diodes, the filter caps, the voltage regulators… The design was a bit unfamiliar to me, because it was using a center-tapped transformer and a +-15V bipolar power supply, which means two voltage rails going around the synth, one at +15V and the other at -15V, with the chassis grounded at 0V. It’s kind of like having two symmetrical power supplies, with mostly identical layout and parts.
After checking the voltages coming out of the power supply, I noticed that the negative voltage line was at -4V, while the positive was at around 15V. I thought this wasn’t normal, so began looking for the cause of this. Maybe something was shorting the negative power rail to ground, and draining its power.
With my multi meter, I traced the negative supply rail all around the synth, and checked the resistance of every component that was attached to it, especially those connecting the -15V to Ground. It took a while, but that’s how I found the culprit: a tantalum capacitor connected from the negative rail to ground was completely shorted (see figure 6). I replaced it with an electrolytic cap and that restored power to the synth: the voltage rails gave me +15V and -15V readings, and the power light lit up. Nice!
fig.5 - These are the sections I could identify while working on the synth.
I didn't have to work on the bottom left, so I'm not quite sure what each section does in that area.
>>> 3- Fixing the dead VCO
Soon after I had restored the power supply, I pressed a couple of keys on the keyboard, tweaked some sliders, but couldn’t get any sound out of the VCO (the main sound source in synths), only from the external input (I was running a signal into it for testing purposes), which gave me a signal, albeit only with the resonance all the way down, and the filter cutoff completely open (more on that later…). Good news is, the VCA, LFO and Envelope seemed to be working fine, modulating the external Input normally. So I guessed something was wrong with the VCO, and probably at its source, because the rest of the signal path seemed to be ok.
I proceeded to find out where the VCO components were located on the board. Someone with more experience would probably have found that out quickly, but with my limited knowledge, I didn’t really know what to look for. I did some research on the web and found that most analog VCOs in vintage synths come from either an IC (usually the CEM type) or a discrete circuit revolving around a pair of transistors in a common package (having them at the same temperature makes them more stable).
So I started looked for pairs of transistors, and found a few of them scattered on the board. One of them was labelled Q1 on the PCB, so I had a hunch it might be the one, since it would make sense to name the fundamental source of sound “number 1” in the circuit.
I then proceeded to do some signal tracing, starting from the VCO slider, following the traces and moving back down the signal path. I had my speaker output connected to the synth out, and was injecting a tone in various places, trying to find out where I would lose the signal. In order to do this properly, I had to spend a lot of time figuring out the map of the PCB, so I could quickly know where was the ground, the +15 and -15, and also which components were connected to which.
I was able to trace my signal all the way back to Q1 (ie. injecting my test tone around there, I could hear it at the output), so I knew I was on the right track. I unsoldered the 5-pin, common emitter dual transistor, and proceeded to check it with the diode testing mode on my DMM. I couldn’t find any problem with it, so I put it back in place.
If the transistors weren’t the problem, then perhaps some of the components feeding voltage into them were faulty. So I began checking all the components directly connected to any of the transistor pins. I checked capacitance for the caps, and compared resistor readings with the supposed values. And that’s when I found my problem: two resistors in series were supposed to read 500ohm each (according to their color code), for a total of 1k, but I got 100k on my multimeter… 100 times more is certainly out of tolerance!
I pulled the components, replaced them with some correct value resistors, then crossed my fingers, turned on the synth, pressed a key… It worked, and I was very happy!
fig. 6 - The various parts I had to replace around the VCO. That white rectangle is a piece of styrofoam
that covers the main transistor pair for the VCO. I'm not quite sure why it's there, maybe a temperature control function...
>>> 4- Repairing the noise generator
Once I could get signal from the VCO, and also from the external input, I turned my attention to the ‘Noise’ slider on the main mixer, which didn’t give me anything. I figured that the primary source of the noise, much like the VCO, was busted, so I began looking for it.
It took at while to locate the correct section of the board, but after lots of tracing, starting from the ‘noise’ slider and following the wires/traces from there, I ended up in a section of the board just above the power supply section.
From my experience repairing old analog drum machines, I knew that the usual source of noise is a transistor, of which only two legs out of three are connected, and one of them to ground. This makes visual location relatively easy, so I looked for a two-legged transistor, and found one in the area I’d ended up after tracing from the ‘noise’ slider (Q16 on the board). I removed the transistor, replaced it with a similar one (a PNP transistor I think, I don’t remember exactly), turned up the noise slider, and got some noise!
I didn’t bother to particularly ‘select’ the transistor I used, but I heard it can make a big difference in noise color (more white, more pink…), so it’s something to consider when replacing a noise generator. Personally, I was happy with the noise tone I got, so I stayed with that.
fig. 7 - The noise generator and VCF section
>>> 5- Restoring the VCF
So, now that I finally had all three sound sources working (VCO, Noise, External), I figured most of the work was done… wrong! It’s unbelievable the amount of broken stuff in that synth! I played around with the VCF (Voltage Controlled Filter), and found that both the cutoff and resonance weren’t working properly. Basically, I could move them around slightly, and get somewhat ‘normal’ results, but about one fifth of the way, they just shut down the sound completely, with a big ‘thud’ on the way out. They were basically useless, which made the synth pretty useless too, because you can’t do much if the filter’s always fully open!
This problem turned out to be one of the most complicated and time-consuming of the whole repair. I had to spend hours figuring out the signal routing on the board, using resistance measurements and signal tracing to draw a ‘map’ of the circuit. This is where schematics would have been extremely useful, but I was doing a ‘blind’ repair with no information whatsoever, so I had to spend a lot of time figuring out the circuit’s structure, until I wound up in the upper right corner of the board, right next to the noise generator area, which is where all my tracing was pointing.
fig. 8 & 9 - The two 'drawings' I used throughout my debugging. The yellow lines represent components,
the various colors correspond to both voltage rails (blue), ground (peach), and whatever signal I was tracing at the time.
This was basically the last section before the output, and I could get a strong signal from the VCO right up to that point, but lost it somewhere around there. The resonance circuitry was the last thing before the output amplifier (VCA), and seemed to be made of, among other things, and 8-pin IC and a big, black component with five-pins and an arrow printed on it (see figure 7 above). I had no idea what it was, was hoping it wasn’t the problem, which of course it turned out to be… I guess it’s the repairman’s Murphy’s law: the part you need to replace is always rare, unknown or missing from your parts box!
I did an awful lot of research, with only the number code on the unit to help me, until I could finally piece it together: what I was looking at was an optocoupler (aka photocoupler), which is basically a LED coupled with a LDR (light depending resistor) inside the same package. I learned that they’re used in modulation applications, for example tremolo or ‘Univibe’-type effects, and were obviously playing a role in the filter section of the PS 1000.
With this knowledge, I was able to test the part. One side is a LED, so I could test it for continuity as I would test a diode. It failed the test, so I was pretty sure this was the cause of the problem. Now I began hunting for a replacement… which took a couple of hours of research! Of course, I couldn’t source the exact part, the company making it (Hamamatsu) had stopped manufacturing optocouplers years ago, and couldn’t help me when I contacted them. After lots of research, I was able to find a supplier that had some stock of a more modern replacement, the Perkins VTL5C3/2, which was in the exact same 5-pin package. It cost me about 10$US to get the part, and I had to wait patiently until it arrived before I could find out if, indeed, the optocoupler was the problem.
Once I’d received it, I couldn’t put it in straight away, because I didn’t know the internal wiring of the part I was replacing, and no markings were providing info, so I didn’t know which pin corresponded to which side of the LED. And if you plug in a LED backwards, chances are it’s going to burn out, which isn’t something I want to happen to a brand new, hard-to-find 10 dollar part! After studying the circuit, and finding out which way the current was flowing into the LED, I could finally put the part in, turn on the synth, play with the Cutoff and Resonance sliders… and the filter worked. Another victory!
fig. 10 - In this picture, the new optocoupler (top center) has been fitted.
>>> 6- Fixing the Portamento
Soon after I’d fixed the filter, I discovered yet another problem with the circuit: moving the Portamento slider changed the pitch of the note, much like the ‘tuning’ pot! This was obviously not the correct behavior, so I began poking around the board once more. From the Portamento slider, I followed traces and wires until I came to a section of the board between the VCO and the octave selection button’s resistance array. I checked the parts along the way for resistance, until I came to another tantalum capacitor, which had a low resistance value while testing in circuit (see figure 6). As long as it’s in circuit, that doesn’t mean it’s a problem, but when I get a resistance value across a cap, it’s worth it to unsolder it and test it more thoroughly.
The evil Ace Tone tantalum capacitors. It looks like a chicken!
If you see one, don't trust it!
I pulled it out and tested it; it still had capacitance, but also a resistance of 600k, which is definitely not normal (ideally, caps should have infinite resistance, or at least enough to send your DMM into overload). I replaced it with an electrolytic of similar value, and the Portamento slider came back to life. This was the second tantalum cap I had to replace in the synth, so I didn't take any risks and replaced all the others on the board. Could this be the end of the repairs? Of course not!
>>> 7- Tuning the synth
After fixing the portamento, I finally had a working synth: every slider, every section of the synth was functional, but there was still a problem: the tuning was completely out of whack. And it couldn’t be solved with the tuning pot on the front panel, because the keyboard stretch was wrong (eg. you tune one C, but the other ones are not tuned – one octave doesn’t span 12 keys), and the octave switches sent the tuning all over the place.
It was easy to locate adjustment trim pots on the board, located near the VCO and connected to the incoming voltage from the keyboard & tuning pot (see figure 6). Unfortunately, these had been glued solidly in 1975, coming out of the factory, and no amount of scratching and effort could get them unstuck. Usually, I’m able to bring back glued trim pots to life, but not in this case: even after removing one of them from the board so I could work on it better, I still couldn’t get it to move without damaging it.
That meant I had to get some replacement trim pots… which of course I didn’t have! Repairman’s Murphy’s law again… So I found the parts online, ordered them, waited for them patiently, and a couple days later could get back to work, replacing three trim pots on the board, and tweaking them until I could get a tuned synth. (Don’t forget to let the synth warm up for a long time – at least 30-45 minutes – before tuning, by the way!)
Basically, one of the trim pots is for setting the zero point of the front panel tuning pot, another is for adjustement of the octave spread and the octave selection switches, and the last one for the actual keyboard voltage spread. The last two work hand-in-hand, so you have to be monitoring both of them when tuning. I started with the keyboard spread adjustment, then the octave spread, and finally one more tuning of the keyboard spread, and that worked well for me.
After all this, I closed up the synth, made myself comfortableI, and played the PS-1000 for a couple of hours, and had a blast!
>>> 8- Drifting hold voltage & the mysterious mod
During my repairs, I noticed that if I kept a key pressed down, the frequency stayed very steady, but if I used the VCA's "hold" switch (to keep notes playing without having to keep a key pressed), the note would drift considerably, perhaps by one half-tone per minute. As soon as I pressed the key again, the frequency would jump back to its correct value... I did some research on the net and found that this was caused by the hold circuitry.
So what's a hold circuit, exactly? Here's what I could gather from my internet readings.
When you press a key on a voltage controlled analog synth, it sends a specific voltage to the synth circuit, which defines the frequency of the note to be played. If you let go of the key, the synth has to keep this voltage value stored until you press another key. This is necessary to enable, for example, the EG or VCA sustain to decay normally. Also, in the case of the PS-1000, the VCA has a 'hold' function, which basically keeps the note sounding forever, even with no key depressed.
In order to store this control voltage, a 'hold' circuit (somewhat related to 'sample and hold') is designed to catch the voltage when a key is pressed, and keep it. At the heart of this circuit are two
components: an op amp and a capacitor. Together, they can sample the incoming voltage and keep it. The op amp is setup as a 'voltage follower', while the capacitor is used to store the electric charge.
In theory, this simple combination works very well, but in reality many things can go wrong, causing the voltage to drift, and thus the frequency to change over time. If the capacitor is leaking current, or if the traces/wires pick up voltage from nearby components that induce current, the held voltage will change.
Normally, this isn't really a problem, because the drift isn't that severe, but for the PS-1000 I quickly realized that there are more serious problems with the hold circuitry; on the longest envelope sustain setting, I can actually hear the note drift noticeably before dying out. This only happens when I let go of the key and the EG sustain or VCA hold are active; the note doesn't drift as long as I keep a key held down, because then the voltage isn't 'floating' in the hold circuitry, it comes directly from the keyboard.
Interestingly, If I press the bottom low C (0 Volt), the note drifts up, and if I press the high C (9 volts), it drifts down. It seems to want to stabilize at around 2.5-3V. I also find it a bit surprising that the voltage range of the keyboard is 0-9V, which is three times more than the typical 1V/Octave; I don't know if this is related to the hold problem or not...
I did lots of testing around the circuit, but always I come back to the main voltage follower IC and the holding capacitor. I was able to disconnect pretty much every thing around that part of the circuit, including the front panel sliders, but still could measure a drift in the voltage. I replaced the IC and the capacitor (with a low leakage polypropylene cap), to no avail.
Finally, I noticed that, if I pressed a note on the keyboard and then quickly unplugged the whole keyboard assembly from the board, the drift was reduced considerably, by a factor of five at least, which makes the PS-1000 more than stable enough. With that amount of drift, I can't hear a pitch change even with the longest sustain setting.
This seems to imply a few things: first, the cause of the drift isn't capacitor leakage, it's induced current from all the 'floating' wires (including the keyboard bus bar) still connected to the hold circuitry. Disconnecting the keyboard assembly basically solves the problem. Which means that a solution would be to isolate all this conductive metal from the hold circuitry. And that's where the "mystery mod" comes in...
You see, someone played around in that synth before I got my hands on it, replacing electrolytic filter capacitors in the power supply (good idea), and removing/bypassing a cluster of components directly connected to the voltage follower/capacitor hold circuitry (bad idea).
The missing parts: one diode (D20), one op amp (Q46), two resistors (R43, R44) and a balance trim pot (bottom left).
IC3 is the voltage follower, the hold capacitor is on its right.
I don't know why this was done, my guess is that perhaps the IC that was removed was busted and impossible to replace, or some other strange reason I'll never know... One thing's for sure, without schematics or another PS-1000 to inspect, and with my limited knowledge, it's very hard for me to figure out which components were there. I tried looking around for a 7-pin can-type IC, but couldn't find a single one on the net...
I'm not sure, but my guess is that this part of the circuitry was designed to isolate the keyboard from the hold circuitry, thus stabilizing the voltage hold much in the same way as unplugging the keyboard did. I haven't attempted to reconstruct this part of the circuit, I'm not sure I would know how... Looking at the traces on the PCB, this is what the circuit looked like before being modified:
Click to enlarge
I would very much like to restore the circuit to its original state, as it was designed. Any help in solving this little puzzle would be greatly appreciated!
>>> 9- The External Modulation Input jack
Both the VCO and VCF can be modulated by an external signal. This is pretty much the only connectivity that the PS-1000 offers, as it has no CV/Gate or Trigger ins/outs.
Because I lack documentation for the synth, I'm not sure what kind of signal I'm supposed to feed it. I'm guessing some sort of LFO or variable voltage source, but it's hard for me to guess what specs I should use. Using a 9 volt battery, a resistor and pot, I fed the circuit 0 to 2 volts, and could get a modulated signal when turning the pot, albeit not very strong. A higher voltage didn't do anything.
So I'm guessing maybe a +/-5V is what it wants, but I don't have modular synths, so I can't try feeding it a LFO voltage output...
>>> 10- Slightly unstable octave spread tuning
I expect old analog synths to need tuning every time they are used, but in this particular case, it isn't the main tuning of the VCO that changes from time to time, it's the keyboard and octave voltage stretches. This is more inconvenient than the general VCO tuning, because the adjustment pots for the voltage spread are located on the PCB, which means opening up the synth every time I want to make an adjustment.
The problem isn't totally unbearable, because the synth does settle near its intended values, but still I wonder why the voltage stretch settings are detuning, and not the main oscillator. It's still a mystery to me...
.jpg)
.gif){kind=small}
