A detailed guide of how I resolved a couple of issues with my Roland SH-101 synthesizer. Biggest problem was – no sound except for filter resonance. Some of the faults of my machine are described in other places on the web but a good bunch of them lacks resolution. So read on to find out what I did.
If you have the modulation grip installed, you may want to remove it. Remove the wiring either way. If you don't own a grip you can defintely leave the four screws for Grip and strap holder in place (left and right side). Go on removing all screws from the bottom plate. Note, in the middle row of screws there are two with metric thread connecting to two brass spacers underneath. All other screws are of the same type, more or less.
Before I acquired my SH-101 it must have had a hard time since there was rust all over the place, signs of battery leakage in both the battery case and on the boards, a crack near the pitch bender, several fader caps loose and others bent to one side or the other.
So first of all there was some mechanical work to do:
- derusting housing and keyboard frame (metal brush, sand paper, zinc spray)
- thorough cleaning of all keys (household cleaner) and keyboard rubber contacts (benzine/isopropyl alcohol)
- straightening of bent sliders with two pliers, one holding the slider right above its housing to prevent damage to slider board and housing, the other doing the bending
- adding small portions of ABS plastic to the guide slots of worn out fader caps by dissolving remains of an old keyboard key or broken LEGO in Acetone and applying the solution with a tiny brush – note, a very thin layer inside the slot helps already a lot to tighten loose fader caps
- fixing the crack with Acetone (use sparingly and keep an eye on prints on the housing - they could get damaged, so do fixing from bottom side), reinforcement with ABS remains from bottom side, cutting small pieces of original grey ABS from inside and dissolving it to make up the crack on top, filing, polishing and finalising the structure of the make up after drying
SH-101 PSU check
First thing to check electrically is always voltages. Pick up Ground from the jumper closest to the edge of the board between IC2 and C11.
An SH-101 is powered by a 9 - 12V wall wart (DC, center negative). One interesting thing is, these 9 or 12V are used to drive some of the ICs directly, on the mainboard e.g. IC5 (9022 Op Amp) or the most precious IC14 (IR3109 Filter). This means on the other hand, don't exceed this voltage too much or damage may happen. I recommend using a good (low noise!) regulated power adapter.
The other interesting thing is that the Roland engineers built an oscillator (TR1 SD571 and TR2 SC1815) to provide AC that is stepped up to above 20V with a tiny transformer and then rectified back to DC (D1/D2). Next is a regulation circuit to provide the other voltages needed in this 2½ octaves beast: ±5V, 14V, 15V. The PSU got rather complex to allow for running the instrument on batteries.
In this particular SH-101 all voltages were present, although I had slight irregularities. There is a tinkered image available on the internet showing multimeter voltage screenshots in different places of the 101 PSU. That image is not completely wrong but it shows measurements from places not well chosen. If you are not sure what the correct and working voltages should be please have a look at my commented schematic below. I also hope this helps to clarify the various voltage symbols in the schematics.
Please note, in place of C7 (1µ/50V) the original schematic says "C2 .01c". Maybe the available schematic is a bit outdated, there are a few other hints. I know there are at least two revisions of the mainboard. Later boards contain coil L1 in the PSU replacing a jumper, and resistors 199 and 210 with a wire connecting these to CPU pin 26. There might be further improvements I did not notice at a quick glance. My 101 is an older model.
A strange thing is, IC1 (5218 Op-Amp) is driven by the 20V rail on pin 8 although this component is rated for 18V max. I doublechecked this with another One-oh-One (20.3V, thanks, Christian!) and this seems to be correct. Still, it is making one a little nervous.
No sound problem
Actually, this was resolved easily. IC13 is the VCO, the famous CEM3340 found also in devices like Prophets, Oberheims or Jupiters. On pin 16 check for 15V. Voltage was ok but output was completely missing on pins 4 and 8. Shorting the corresponding sliders didn't help. Since the keyboard was working (I could play with Freq set to 5 and Resonance and Keyboard sliders all up) and pin 15 received flipping voltages while playing I had to conclude, this Curtis 3340 was all dead. Luckily these ICs are available again through a variety of sources. I installed a socket and a new VCO – sound was back.
Funny thing was, the Pitch Bender and Tune knob worked in reverse – bending upwards or raising the pitch with the tuning knob resulted in a lower frequency and vice versa. I did not care for the moment.
Sad, by the way, that the third available 3340 waveform (triangle, pin 10) is not used at all by design.
Ok sound, but where is the noise?
Next on the list was the noise circuit. Noise is doing two things: one can add white noise to the sound and use noise as a modulation source. Both was not working so it wasn't likely that just the noise slider was faulty.
The noise circuit is fed by the 14V rail, voltage was ok at R83/R84 and IC12 pin 8. So I took the oscilloscope and probed IC12(a), realizing that noise was only present on the input side of the Op Amp (non inversed pin 3), an old 9022. To doublecheck I shorted the noise to the filter side of R109-R112 (not the ribbon side), and voilà, noise was there. I installed another socket and replaced the IC with a TL022, which is a modern days equivalent. Check.
Side effect: The pitch bend and tuning knob issue was gone and this makes sense, since both controllers are connected to the b-side of IC12. So if you have a pitch bender functioning the other way round, this could be the solution.
Note, TR23 (SC945) is a selected component and got a white dot for that reason. If you have to replace that install a 3 pin socket and check several transistors for the best noise result.
VCF Bender Modulation
While the pitch bender worked again I realized that filter modulation did not work. This time I really suspected the slider and shorted it to check and yes, that was it. And no, oh no. There is no source on the internet for 20mm 100K linear sliders. All those 30mm sliders would be available but not the short ones.*
So I desoldered the pot and opened its housing to do a cleaning first. Eventually, the problem turned out to be a broken path. Luckily, it was not the resistor path but the ground side. So I fixed this with some layers of liquid conducting silver painted on the tiny and clean slider board. I don't expect this to last forever but hopefully until I found a source for replacement sliders.
Btw, the crack near the bender was the result of a bad or unselected part, Roland is to blame. The bender frame is kinked on the edges and fixed with two screws. These kinks come a bit late, maybe by a quarter of a millimeter. So there has been some pressure on the housing right from the start. I solved this by attaching thick paper as a washer.
Bouncing notes keyboard
Now basically everything was working and I thought of calibrating the synthesizer. I hoped this would help resolve another funny issue. While playing the keys from low to high it felt like octaves were bouncing. Ok, not really octaves, the pitch would drop or raise extraordinarily every 8 notes. So I had a few strange intervals on the keyboard between keys next to each other. Another symptom was the impossibility to adjust the three D/A voltages according to the procedure described in the service manual (see SH-101 D/A Converter Adjustment in box on this page).
Since basic D/A circuit calibration did not help and since the keyboard symptom followed an 8-scheme I assumed a digital problem and one of IC7 (4050 hex inverting buffer) or IC10 (4052 de-/multiplexer) to be the culprit. I had both ICs in stock, installed sockets once more and gave that a try without further measurements. The outcome was, only IC7 was defective. So back to calibration.
Calibration: Problem with VCO Width
Now I got the voltages right quite easy by adjusting VR1/VR2/VR3. But I could not get the VCO pitch spread right. Trouble was, one octave was always slightly more than an octave while VR6 was already turned fully counter clockwise which means 0 Ω. Keyboard and CV input led to the same result. First I thought some other VR might need some further adjustment but after some fiddling it was clear, there must be another fault, or the new CEM3340 Rev. G (IC13) had slightly different specs. Remember, Roland used selected ICs marked with a white dot, presumably due to better linearity (Aha!) and low frequency performance. See Roland Service Information 100248. This sheet also advises to lower R108 from 470Ω to 330Ω to increase the tuning range, which became standard from serial number 553.000 and up. However, tuning range was not my problem.
Sadly, I could not nail the reason for this deviation in intonation. So I looked at the schematic and decided to lower the resistance of R106 which is in line between VR6 and IC13. Since I had installed the board already in the synth I decided to put another resistor in parallel following Ohm's law. The schematic says 1.69K, my Multimeter said 1.67K and I thought let's bring it down to around 1.5K. To achieve this I piggibacked a 15K resistor on R106, see image above. This helped absolutely, the synth is now in tune and wonderfully playable. Well, in tune for most of it's range. There is still some deviation in the highest octave but since I use the synth for basslines mostly I can live with that.
A Roland SH-101 is such a great and flexible mono synth. And I love the ability to program sequences very quickly and to trigger them by e.g. a DR-55 DBS Out or a TR-707 Rim track. I enjoy it a lot.
I hope this writeup could help you. If it did or if you have questions, I'd be glad if you leave a comment.