2025-04-21
I reflect on a synthesizer DIY project that did not go as intended. My goal was to replace a humming power supply in an old synthesizer but I accidentally damaged the machine, leading to an unintended repair side quest.
Original goal: do something about the humming power supply of my MKS-80
The Roland MKS-80 is a rackmount synthesizer that was produced from 1984-1987. Mine came with a 117V step-down transformer because it was not originally sold in the western European market.
Annoyingly, both the internal transformer of the MKS-80 and the step-down transformer had started to hum. The humming does not get recorded into my music but it's annoying to have to listen to it in the studio all the time.
All this was just annoying enough for me to decide to modify the MKS-80 to get rid of the big transformers. I am actually sceptical of "improving" vintage equipment that is not broken because I feel that people underestimate how sturdy the old stuff is. But in this case I felt like doing it anyway because the humming was so annoying, having a step-down transformer is clunky, and because it would be a new challenge for me. I have only done one power supply replacement before, for my Nord Rack 2, in that case out of strict necessity. The MKS-80 power supply is a bit more challenging but it looked doable. I like knowing that I know how to do this and this seemed like a good opportunity to learn.
The simplest fix would have been a new transformer but I learned from the Nord Rack 2 that this sort of hybrid analog/digital device often uses a custom made transformer. This makes it hard to find a suitable replacement. It's easier to find switch mode power supply modules that replace the whole power supply rather than just the transformer.
I soon found out that there is a commercially made power supply replacement for the MKS-80 but I was put off by its price tag: over €500 including VAT. Something like a MEAN WELL RT-50C costs €20 and gets you most of the way there. In the end I chose to use a TDK CUT 35 which cost me €60. It is more compact than the RT-50C which made it easier for me to fit it into the MKS-80. This way the total bill for parts would stay under €70, still quite a bit less than €500.
The original power supply
The original power supply in my MKS-80. On the left you can just see the mains transformer. The brown L-shaped board converts the low voltage AC power coming out of the transformer into low voltage DC used by the synth. On the right you see the heat sink of the power transistors.
In this picture you get a better view of the big humming transformer, in the lower left.
There are quite a few internal power cables in this machine: 8 power cable connectors going to 5 different circuit boards. The cables are all soldered to the L-shaped power board so it seemed like a good idea to me to re-use that board.
The new power supply
The new power supply. The mains transformer is gone in the lower right hand side because it is no longer needed. At the top of the picture you see a silver box with honeycomb holes, this is a TDK CUT 35 power supply module. DC from the CUT 35 is injected into the power board at the bottom where it passes through the three original fuses. Then it bypasses most of the original power supply with the dark blue wires emanating from where the bridge rectifiers used to be.
And this all worked! The converted power supply board generated all the required voltages: +5V for the digital circuits, +/-15V for the analog circuits, an additional +10V for the voltage controlled oscillators, and +8V for the optional MPG-80 programmer.
Details
I performed the modification by removing the power transistors (Q1, Q6, Q8), several resistors (R21, R25) and a lot of wire links. By doing this I disconnected the original linear regulation circuit from the power board outputs. I removed as few parts as possible in case I ever want to reverse the modification.
The +10V is generated by the original power supply board circuit. It uses an opamp (IC1) that was originally powered by the unregulated +18V line but I verified before I started that the opamp would still do its job when powered at +15V. So all I had to do to get the +10V working in the new situation was to route +15V to pin 8 of IC1.
The original +8V came from the unregulated +8V line that the +5V was derived from. Because the unregulated power is gone on the new board I had to improvise something here. I used an LM7809 regulator to tap power from the +15V line and send that into the original unregulated +8V part of the circuit. Because I removed the +5V power transistor Q1, this does not end up accidentally powering the original +5V regulation circuit.
In the original situation, the black and white live and neutral mains wires are routed through a filter and fuse in the upper right corner of the power board, then through a 2-pole switch on the front panel, to the mains transformer. Taking inspiration from the commercial solution I mentioned before, I left this routing intact. I added two new brown/blue live/neutral wires where the transformer used to be and routed them back to the back of the machine where I installed the CUT 35.
The original machine had a 2-pin mains socket, meaning there was no ground connection. I replaced this socket with a regular 3-pin C14 socket because the CUT 35 manual stipulates it must be grounded. You can see in the upper right where I bonded the ground wires to the case.
An unintended adventure
For my initial test of the new power supply I mounted the various components into the case, soldered back the mains wires to the power board, and powered it up without the rest of the synth connected to the power board. Everything looked good. Then I re-connected the power board to the rest of the synth to do further testing.
This is where I made a mistake that would cost me a lot of time.
Remember that I said there are 8 internal DC power connectors that connect to 5 different circuit boards? I forgot to plug in one of them. Usually when you do something like this, you find out that the synth isn't working, you discover the unplugged connector and plug it in, and you're done. But this time the synth mostly worked, delaying my discovery of the overlooked connector, and there was immediate damage.
The first sign of trouble was that although the synth worked, there was a hissing noise mixed in with the normal sounds it was producing. At first I thought "maybe that was there all along and I missed it because of the transformer hum" but I kind of knew that wasn't the case. I then thought, maybe I need to recalibrate the machine because the power supply voltages, while in spec, are not the same as before. That also did not fix it. Then after a while the sound started to distort and I was no longer in doubt that something was wrong.
The red arrows highlight 5 power connectors going to 4 boards in this picture. The one marked with "!!!" is the one I overlooked, causing damage to the synth.
It really took me a while to spot the disconnected connector because it is hidden underneath a bunch of other wires as the picture above shows. It was a big relief to find it because before then, I was utterly confused why the synth wasn't working correctly. Did I do something wrong in putting together the new power supply? Should I have bought the commercial product after all? Reassuringly, the answer was no. The power supply was fine and even if I had bought the commercial one I could have still forgotten to plug in that connector and damaged the synth just the same. To avoid this mistake I should have had the power supply conversion carried out by a professional, which flies against my DIY ethos.
So what whas the damage? Before I go on I need to give some more context about the internals of the synth. The high level structure is that there is a single digital "brain" which controls two identical "module boards" which are really 4-voice synthesizers. The sound coming out of the lower and upper module board goes to an "output board" where it is mixed and sent to the output sockets.
The output board has 2 power connectors. This is because it has an analog audio section and a digital MIDI section and you don't want digital logic noise leaking into the analog audio. The analog section of the output board was unpowered. Somehow it still worked but it was now drawing current along the audio signal lines coming from the module boards. After a lot of investigation and trial and error I discovered that on both module boards, a voltage regulation circuit physically close to the audio signal connector that the current was unintentionally going through got damaged. Many hours later, and after replacing less than €2 of parts, the hiss was gone.
The green board on the left is the upper module board. The brown board on the right is the output board. The red arrows show the suspected unintended flow of current. I suspect it went along the thick red, brown, orange and yellow signal cables from the module board down to the output board. The damaged parts were all near the left red arrow on the green boards.
Details of the repair
The faults were clustered around IC50 on the module boards, the dual regulator of the +/-7.3V rails. On both boards the -7.3V power transistor failed. To be safe I replaced the power transistors for both +7.3V and -7.3V on both module boards (Q5 and Q4). The original Japanese transistors are no longer available new so I used BC640 and BC639 transistors which work fine here.
Before replacing Q4 I could not even get -7.3V on one of the module boards. After replacing that transistor on both boards, one module board was fine but the other still hissed. I replaced C130, because the datasheet of the regulator says C130's job is to reduce noise on the voltage outputs, but that did not do anything. The hissing finally went away once I replaced the filter capacitors that follow the power transistors (C125 and C126). Although only one module board was hissing I replaced the C125 and C126 on both module boards.
So the parts I spent on this are five electrolytic capacitors, two BC640 transistors and two BC639 transistors.
Conclusion
I have mixed feelings about this project. I am pleased with how the power supply conversion turned out. The machine is quiet now (apart from the whine of the LCD backlight inverter), the bulky step-down transformer is gone and I saved quite a bit of money by doing it all myself. I am also happy with how the repair went. I patiently used trial and error and I solved the problem with targeted replacement of parts. It's easy to panic and blindly start replacing parts thinking maybe it will help but I did not fall into that trap.
But at the same time I am frustrated by the humbling reminder that I am not an experienced repair professional. I said at the beginning I wanted to learn; I got a bit more than I asked for.
Tags: diy