SEM Power Section

July 22, 2011 Leave a comment
SEM Power Section

Always, always take the time to replace the electrolytic capacitors in your vintage synths if you find yourself opening them up for a repair anyhow.  It is relatively cheap and your synth will generally sound and behave better after you have given it a fresh set.  Electrolytics have a limited lifespan and therefore *all* your vintage babies probably are due.  Go ahead and splurge on some nice high temp and high quality caps to ensure you don’t have to do this for another twenty years.  The SEM is particularly easy as there are only two electrolytics on the whole synth.  To open an SEM, there are three srews on the PCB to remove and then you have to pull the two boards apart.  I hate this part.  They are connected with a series of pins around the edges and it takes a good bit of force to get them seperated.  This is another bad design decision as there is no way to seperate the boards without flexing them at least a bit.  I have had solder connection go bad from this process so try and minimize the number of times you remove the PCB.  If you are following along at home, try and open it, make *all* the modifications I’m documenting and then close it up rather than pulling it apart for each one.  There are not many mods to do (at least here) and chasing down intermittent solder joints is not fun.

There is not much to the power section on the SEM since the main work was handled off the SEM and in my case in the external supply as you can see in the schematic.  I’ve pointed out the interesting bits in the pic.  I’ve had two occasions over the years where a malfunctioning SEM was due to one part in this section.  Note the white Molex power connector – this is one I had to repair as it had broken off.  If your SEM is acting at all flakey (or if it is completely dead), start by checking the voltages coming out the power section first.  If they are not correct, your problem is likely to be here.  Once when I was probing around with a 5v control voltage trying to find where to insert the pulse-width modulation mod into the VCO1 circuit, the probe slipped and shorted across something on the board.  My head was turned at the time looking at the scope so I didn’t see it.  The result was a non-functioning SEM.  The gate LED would flash constantly and there were some random gates going on but none of the voice sections were outputing anything correctly.  The second time was more horrific.  It was late and I was trying to troubleshoot a problem that had derailed my day and I accidentally reversed the power supply voltages.  This of course released the magic smoke in the circuit and the SEM was completely dead.  In both cases, the voltages coming out of the power section, particularly the negative side, were way off.  As all the parts are readily available from Mouser and cost a total of something like $5 I decided to just replace the whole power section’s ICs and caps.  In both cases it turned out to be the 2N3638A transistor and therefore an easy fix.  I am not sure if I just got lucky or if the design of the SEM is such that the transistor sacrifices itself in the case of shorts or other power problems. I would not replace anything besides the electrolytics as a matter of course, but if you are having issues, check the voltages in the supply section.  You might get lucky and have an easy fix!

Categories: SDIY, Synths Tags:

SEM Molex Connectors Repair

July 16, 2011 Leave a comment

The bummer of most restoration or modification type projects is that often the piece you are working on has some other defects and they generally tend to not be minor either – at least that has been my experience. It typically takes some fairly major repair to even get me thinking “while I am in there anyhow fixing X, I might as well go ahead and do that whole Y thing I have been thinking about forever.” The most common problems in SEMs are broken Molex connectors on the back or a whole host of diffuse symptoms related to the power section. In this post, we will address the Molex connectors.  The picture shows you every Molex connector and pin type you will need to make any connections or repairs to your SEM.

First off, as much as I adore the SEM, the Molex connectors are complete crap. For as well as everything else seems to be designed, the use of the Molex connections on the rear of the unit and the interboard pin headers always kind of baffled me. True, the SEM is an older piece from the late 1970s but even then more standard 0.156” and 0.100” headers were available and common. If you have any broken connections on the rear of your SEM, you might be thinking what I was thinking: “I’ll just replace these awful Molex connectors with 0.156” headers.” It is a good idea except for the fact it will not work. The pins are spaced close enough that the 0.156” headers would fit but the pins themselves are too large to fit through the PCB. You would need to find a header with 0.156” pin spacing but with the smaller pins of a 0.100” header. I searched connectors high and low and never found a solution.  So, we are pretty much stuck with the Molex connectors as they are still made and available.  Until recently, they still made regular appearances inside computer cases. 

 I say “pretty much stuck” because you could opt for two alternate routes. First, you could just hardwire everything to the board and remove the Molex connectors entirely. This works great except that it makes removing the SEM from whatever housing you are using a major pain as you would have to unsolder all those connections.   It is however a common response and I’ve seen plenty SEMs go this route.  Alternately, you could use some sort of mid-cable interconnect but I did not find any that I particularly liked and they made the wiring messy to boot. Consequently, as much as I hate them, I had to order some of the old Molex bits.

Making Housings for Cables

A quick search of the web will turn up that what you need for wire housing are the 3 and 4 position Molex housings – Molex part number 03-06-232 and 03-06-2042 respectively. Mouser and Digikey both carry these.  These are just the ones you need to plug into the back of the SEM. You need to order the pins separately and these are male crimp pins, Molex part number 02-06-2103. You really should use a crimper for these and not solder them. Just go and buy a cheap crimper online. There are tons of them around $40 that work great and beat the hell out of soldering all those damn pins. Now that I’ve railed against soldering, I’ll let you know that if you are using 24AWG or 26AWG wire, you probably will need to solder and crimp these. The pins themselves are rated for 18 – 24 gauge wire but I’ve pulled the 24 out fairly easily so I end up crimping and soldering them. Here is the easiest way I’ve found to make the connections.  First, strip just enough off your wire so that the wire sticks slightly past the crimp section of the pin. Tin the wire with solder. Do it. Really. Always. I know you hate it but just do it – your life will be easier. Next slightly tin the inside of the pin. Just a little in the crimp section and slightly below. You are not trying to fill up the pin. If you put too much on, you will not be able to crimp it well. Next crimp the tinned pin onto your tinned wire. Now apply heat to the crimped assembly until you see the solder melt. This creates a great connection and is easier than soldering and them crimping.

Replacing Broken Terminals on the SEM

Now, the more likely problem and the one that for some reason never comes up on the Google searches, is what to do about those broken off terminals on the SEM itself. The above instructions will let you make plugs to interface the patch points but that only works if there is a terminal there to begin with. Pretty much every SEM I have seen either has missing terminals that have broken off (because they are crap) or some that are so weak that they are going to break off any second. My old tech (who did original warrantee work on these) once told me they pretty much expected to replace 2 of the housings any time a SEM same in because invariable one would break off just trying to get the SEM out of the housing. For this, you need the 3 and 4 position housings – Molex part number 03-06-1032 and 03-06-1042 respectively. Here again, these are sold without pins. Once you order the replacement pins, you will see why these are so fragile. The part is a female PCB pin with solder tail – Molex 02-06-7103. First, clean out any remaining pins from the broken one, use wick or a pump to clean the PCB holes. Next, insert the pins as far as they will go into the housing and then use a pliers to pull them back down. You do not want there to be any play in the connector when you solder it down. If the connector can rise up from the PCB, it is going to break off. Next, insert the connector back into the PCB observing the correct orientation of pin1 and resolder.

Categories: SDIY, Synths Tags:

SEM Resurrection – The Patch Points

June 19, 2011 2 comments

If you are following along at home, here are the patch points on the back of the SEM available for your SEM patching pleasure.  These are from the SEM Owner’s Manual.

A1: External Modulation Input – VCO 1
A2: Sawtooth Output – VCO 1 Orange
A3: Pulse Output – VCO 1

B1: Control Voltage Input #1 – VCO 1
B2: Ground Red
B3: Control Voltage Input #2 – VCO 1

C1: Sync Output – VCO 1
C2: Ground Yellow
C3: Sync Input – VCO 1

D1: Control Voltage Input #1 – VCO 2
D2: Ground Red
D3: Control Voltage Input #2 – VCO 2

E1: External Modulation Input – VCO 2
E2: Sawtooth Output – VCO 2 Orange
E3: Pulse Output – VCO 2

F1: Sync Output – VCO 2
F2: VCA Control In
F3: Sync Input – VCO 21

G1: LFO Trigger Input
G2: Ground Brown
G3: LFO Output

H1: Control Voltage Input #1 – VCF
H2: Control Voltage Input #2 – VCF Grey
H3: External Modulation Input – VCF

I1: VCA Output
I2: Ground White
I3: Output Amplifier Input

J1: Hi Pass Output – VCF
J2: Ground White
J3: Bandpass Output – VCF

K1: Selected VCF Response Output
K2: Ground White
K3: Low Pass Output – VCF

L1: External Audio Input #1
L2: Ground Blue
L3: External Audio Input #1

M1: External Audio Input #2
M2: Ground Blue
M3: External Audio Input #2

X1: Output – ENV 1
X2: Gate Input – ENV 1 Green
X3: Trigger Input – ENV 1

Y1: Output – ENV 2
Y2: Gate Input – ENV 2 Green
Y3: Trigger Input – ENV 2

Z1: +18.5 Regulated Voltage Input
Z2: Ground
Z3: Ground Black
Z4: -18.5 Regulated Voltage Input

Categories: SDIY, Synths Tags:

SEM Resurrection – The New Plan and Patch Panel

June 19, 2011 Leave a comment

General Approach and Guidelines 

So, with the reemergence of my SEMs there is the need for a new plan.  For one thing, I only have a vague memory of what my original plan was.  Although my temporary panel  gives me some idea of what I was thinking, looking over it there are several things I no longer think are worth doing as I had envisioned them.  Mainly, my setup and interaction with my gear has shifted since I first pulled these so I have a different set of needs.  In particular, my modular rig is substantially larger and has a full complement of modules.  This means that I need the SEMs more as dedicated voice units than as a sum of their individual VCOs, VCF, etc.  My new guidelines then are that the modifications and patching needs to respect the integrity of the SEM as an independent unit.  No mod can break any existing functionality of the unit.  Thus, modifications may be added but nothing can be broken.  This means no traces are to be cut, opamps swapped out for “better” components, resonance in the filter monkeyed with etc.  Also, any mods must be 100% reversible.  I don’t foresee these units ever being used in any situation other than how I’m going to rack them but I still ideally want the process reversible.  This being said, I have several modifications that I think are essential and meet the above requirements.

   Here is my current panel design which shows the direction I’ve decided to take my SEMs in.  This will be for two of the SEMs as I am thinking of integrating a MIDI->CV module into the panel for the other two and this will require getting rid of some of the features on this panel to make room.  I envision these first two SEMs sitting next to my modular rig and being largely driven by it as individual voices.  Alternately, my Cirklon when it eventually arrives will pair with these as monosynths.  Therefore, I wanted the option to use the SEMs VCOs, filters, etc as standalone modules but mainly always together rather than as disparate modules. 

VCO mods

   The standard SEM patch points provide outputs for the VCO waveforms of saw and pulse and inputs for three v/oct control voltages (cv1, cv2, ext) and sync.  I’ve decided to add a waveshaper pcb board driven by the saw output to derive the sine and triangle waves.  Also, there is no CV input for pulse width modulation so I am going to have to insert into the circuitry.  A look over the schematics and the SEM PCB shows that this shouldn’t be a problem.  Likewise, I wanted a total of 3 CV ins for frequency modulation.  I’ll add another resistor and input off the junction of the CV2 input prior to the opamp to accommodate this.  This technically gives me four inputs as I could use the EXT mod input.  Unfortunately, the EXT is attached to a switch on SEM the front panel which means you must decide between using EXT, ENV1, or the LFO.  There are times when it is desirable to be able to flip a switch to suddenly turn on a CV so the EXT feature is useful.  I do not have adequete panel space to accomidate the EXT with a dedicated input.  I generally would prefer a third cv input not tied to the EXT so I am going to use a pot with a pull switch to select whether the input on the patch panel is routed to the 3rd CV input I’ve added to the SEM or to the standard EXT patch point.  I’ll use the CV1 in as my straight v/oct input on both VCOs.  Since I use banana jacks throughout my system, there is no need to provide any switching to route a single CV to both OSC1 and OSC2 as this is trivial with bananas.  If you are considering patching yours out in 1/8” or ¼” jacks I would recommend some sort of routing switch. 

The other feature to notice is that I have added extensive inverting attenuators throughout the patch panel.  I could have gone the easy route and just use positive attenuators but I always get annoyed having to go to an inverter rather than having them available on a module.  Rather than cobble this together on protoboard, I’ve laid out a PCB that contains the circuitry for 8 inverting attenuators and 2 inverters that will mount in the back of the rack alongside the waveshaper pcb.

VCF mods

While I like the idea of adding triangle and sine waves to the VCOs, they obviously cannot be integrated onto the front panel of the SEM.  On the one hand, I could use the external input on the filter to patch them in but the SEM only gives me a single input.  Rather than add another external channel and patch, my plan is to just connect the outputs from the waveshaper through a volume pot on the patch panel and tap in with a resistor into the audio input connections for the filter.  This way it behaves more like an integral feature of the module rather than a patched in hack.  This means that the VCOs will behave independently with all waveform outputs available but the SEM will also behave predictably when being used as a standalone voice as well.  I think it is a elegant solution overall.  To save panel space, the volume controls for the sine and triangle waves will have pull switches to select the waveform.  Additionally, the VCF only has two CV inputs but plan to have three – two with inverting attenuators and one straight input so an additional CV will have to be added.  Like the VCOs, I will use a pot with a push/pull switch to route the input either to my added CV in or to the EXT patchpoint so that it will appear on the selection switch for ENV2, EXT, LFO.

LFO mods

I’m not 100% certain but it seems to me that the SEM uses a sine waveshape internally but ouputs a triangle wave on the patchpoint (or the other way around but they are different and I don’t have my oscilloscope handy at the moment).  There is a square wave generated as part of the LFO circuitry but it swings between -18.5v and 18.5v, excessively hot to use.  So besides bringing both the sine and triangle waveshapes out to the panel, I’ll add a voltage divider to bring the square down to standard modular voltages.  This will make for 3 LFO waveshape outputs instead of the standard 1.  Considering you also get a reset patch for the LFO, this should be a cool little unit once properly adapted.

Envelope mods

I had original thought about bringing the trigger and gate out to the front but I really seldom retrigger envelopes and I have other envelopes elsewhere if I really need that functionality.  I was running out of panel space and had to choose between separate triggers or inverted envelope outputs.  I had space on the CV Utility PCB so I built on a couple of inverters and that is what is on the panel.  Additionally, i discovered while testing the raw SEM that since I have the two envelopes unjumpered (you can set the SEM so that ENV2 fires from ENV1’s gate) that it would be nice to know if ENV2 is firing.  I debated between adding a bicolor LED to the SEM panel to show the gate status of both envelopes but my Sequentix P3 had a similar indicator scheme and I hated it.  Instead, I’ve added a LED to the panel for ENV2.  I hav not yet decided if I want it to show the gate status like the one on the SEM or have its brightness change with the envelopes intensity.  I’ll probably have it follow the env output.  Perhaps I’ll mod the SEM LED to behave likewise.  

Categories: SDIY, Synths Tags:

Oberheim SEM Resurrection – The Power Supply

June 18, 2011 Leave a comment


  The power supply is the one piece of the project I did right away when I pulled the SEMs out of the 4 voice.  At the time, I was in no way able to make a power supply on my own without potentially electrocuting myself.  I had recent acquired a Serge modular system and they use a remote supply in a case that connects to the panels via a standard XLR connector.  Remoting the supply means that the cases are lighter, there is much less heat buildup in the cases, and you remove a potential source of noise away from the panels.  I figured I certainly did not know what I was doing but Rex Probe (the maker of Serge modulars, owner of Sound Transform Systems) certainly did.  If it was good enough for Serge, it was a good solution for me.  I opened the Serge supply and discovered cloning it was going to be a piece of cake – it was an off-the-shelf commercial linear supply. 


   The SEM requires +/-18.5v which rules out any standard voltage power supply.  There are only two models available currently by International power.  Both are adjustable from 19-24v and output 9A or 1.8A depending.  Mouser part numbers are 597-BB24-1.2 for the 0.9A version and 597-CC24-2.4 for the 1.8A version.  There is a slight problem in selecting between these two.  The original SEM power supply schematics show that it is designed to deliver 0.25A per SEM.  I do not know how much overhead there is in the supply specs but I suspect there is a good bit.  This means that the 0.9A supply is technical underpowered for 4 SEMs and the 1.8A is overkill.  It wouldn’t be so bad except that the smaller supply costs $63 while the larger runs $99.  In the end I decided 0.9A was close enough.


   I bought the supply, a standard power cord connector, 4 Switchcraft XLR panel mount connectors, a fuse, a banana jack, a lighted power switch, and a project box.  Came together perfectly.  I needed the banana jack as a common ground source so I could patch the SEMs with my Serge system which uses banana jacks instead of grounded ¼” or 1/8” jacks like other modular manufacturers.  Using an off the shelf solution had many advantages.  First, it greatly simplified making the supply and took the guesswork out of me building a regulated supply.  Secondly, it is a great supply.  In the intervening years I have used many of these supplies in my modular rig and *none* of them have ever failed.  Ever.   In the intervening years since this project went on the shelf, I have spent much time learning more about electronics in general and synthesizers in general.  I now certainly could build a regulated supply safely.  However, there is really no point.  These supplies from International Power, Condor, and Power One are so good you quite frankly are a fool to make your own.  Making your own will not save you any money either.  Take a look at the cost of good transformers and you’ll see what I mean.  So, in the end, just buy a linear supply and save yourself the hassle.


   While I had originally planned to run all of my SEMs from this one supply, I’ve decided since restarting this project that it would be nice to have 2 of the 4 more stand-alone independent units.  I plan to use an individual supply put into the racks for two of them.  Part of this is also because while I think 0.9A is close enough to drive 4 SEMs, my current plans involve adding at least three additional PCBs into the case to enable the functionality I want from my SEMs.  This includes another power board to derive a regulated +/-15v supply to power a CV ultilty board that provides 8 inverting attenuators for the patch panel and a waveshaper board to generate sine and triangle waves from the VCOs.  In the stand-alone SEMs I am contemplating adding an integrated MIDI->CV converter.  These additions may well tax the supply if I ran all 4 SEMs from it.  Consequently, even though the supply has connectors for 4 SEMs, I only plan on driving two fully modified units with it.  I’m using a smaller International Power supply (Mouser 597-AA24-0.6) that delivers up to 0.4A for a single SEM.  They are currently priced around $50 and are small enough to fit inside the rack enclosure easily.

Categories: SDIY, Synths Tags:

Oberheim SEM Resurrection – The Backstory

June 18, 2011 Leave a comment

Oberheim 4 Voice

   So, once upon a time I had an Oberheim 4 Voice synth. A Beautiful monster but still a monster. If you haven’t seen one before, imagine a white smart car and you are about in the ball park of the girth of this instrument. I bought it from a local guy who in his desire to make it patchable had drilled a series of holes in the synth case behind each SEM and installed a series of 1/8” jacks for the various i/o functions of the SEM. On the one hand this seems like a good and reasonable idea. However, there are more i/o points on the SEM than there is room for behind the SEM and so the result was a series of two rows of extremely cramped and unlabeled jacks running the entire length of the synth. I got a good deal on the 4 Voice as the programmer wasn’t working properly. This was not a particularly big deal for me as even working it is fairly useless and secondly I tend to make my patches as I go and so storage and recall just aren’t how I traditionally work. The programmer was not essential as recalling a patch would set the SEMs into an approximate state for the patch you were looking for but you still needed to spend a substantial amount of time retuning and setting each SEM to sound identical if you were to use them as a poly synth. It turned out that while the 4 voice sounds absolutely amazing (and it really does), 4 note polyphony while revolutionary at the time of manufacture, isn’t really anything to brag about when it comes to playing pretty much anything. As a result, I usually ran the 4 Voice as four monophonic voice blocks. In the end, someone offered to trade me a working Korg Wavestation A/D for my dead programmer. I had done a preliminary exploration of the fixing the programmer myself and determined that it was going to be no simple fix – several major blocks appeared to have malfunctions and most of the programmer appeared to run on electronic black arts involving experimental, esoteric, and unobtanum componentry. Around this time, James Husted, a fellow member on Analogue Heaven made a run of metalwork for racking SEMs.  His cases seemed to provide an obvious solution so I promptly bought four of them and they arrived as advertised. I thought this was going to be a quick in and out project. That was roughly 12 years ago.

   The problem with the SEM is that it’s patchability is incredibly seductive.  Major synth guru Kevin Lightner had a good brief overview of the problem in 1998 around the time I was struggling with this same issue.  There are all kinds of patchpoints brought out to the back of the unit on Molex connectors that make the whole process seem simple. In a sense it is. You stick some signals in, hook some outputs up to a panel and you should be good to go. The problem is that the patch points are somewhere in between a dedicated monosynth and a fully modular voice. If you want the monosynth with limited patchability, everything is fine. However, if you want it to function more as a modular voice, there is a good bit of work and planning to be done – but I’m getting ahead of myself. First, you need to power the things. Not really a problem except that the SEM uses a +/-18.5v supply. A quick search of the Mouser/Digikey/Allied/Newark catalogs will show you that there really are not many options out there. I opted for a remote supply ala Serge. I bought a small project box, bought a Power One adjustable supply and wired it up. That went well enough but as soon as I moved on to design the patch panel everything came to a screeching halt. My DIY and electrical circuit skills at that time were paltry at best. I certainly was skilled enough to wire the patchpoints up but extending the functionality by adding or modding the SEM in any significant way was out of the question.

   While I contemplated my design and my next move, I sent James’ metal work off to a local powder coater to transform them from bare metal to a nice industrial finish black. I know what you are saying, the SEM is white. Well, it is a white – an off-white in fact. Further, it is a little bit of an unusual off-white at that. In any event, it wasn’t a particular shade of white that my powder coater had. If you ever wanted to feel really emasculated, try having a discussion with a powder coater about matching particular shades of off-white. I settled on black. The powder coater was doing this as a side job for me and let me know right off the bat he was backed up and so my little project might be a while. The price was right so I decided to wait it out. Two months later, I had wrestled with the panel design problem and come to the conclusion that what I found minimally acceptable for a patch configuration would require a substantial amount of modifications on my part that while not risky, certainly would require a significant amount of time and effort on my part to accomplish. By the time my racks came back in their manly black, I decided to put the project aside while I thought about it some more. Thus, the SEMs sat in my closet for the past decade plus until a month ago.

   A good friend of mine, Brooks Rongstad came over and like he has done many times in the past inquired about the SEMs. Now that Oberheim is making them again he was curious what all the fuss was about and wondering if they were worth considering. He pestered me enough to get me pull one out of the closet over my protesting that I was not sure if they worked any longer, where the power supply I built ended up etc. Turns out that after putting off dealing with these things for 12 years, it took me all of 15 minutes to get one up and running. The power supply still worked perfectly, and I remembered I had actually cobbled together a temporary panel maybe 7 years ago that had most of the rough functions patched out and was still attached to the SEM. My wife was out of town that evening, my son was spending the night at his grandparents, and I didn’t sleep a wink that evening. The sun came up and found me still sitting in the basement studio with the SEM and a sequencer in my modular rig grinning like a complete lunatic – which I am. Now, this project *has* to be completed.

Categories: SDIY, Synths Tags: