SS30-M - A Yamaha SS30 in a rack-mount, with MIDI

Friday, September 12, 2014

Five years - and then an oscilloscope

Five years is long time between posts. I'm not going to try and explain that gap. Life has it's own priorities and this project wasn't one of them, I guess. I mean, I have done other things with my time beyond work, family, home-life and friends. In fact I have done a bit of work on the project in that period, but I didn't post about it. I can't really explain why but I do know that I get interested in something, spend a lot of time on it and then lose momentum suddenly. Or more accurately, I get distracted by something more important or interested in something else.

However, this project was always intended to be long-term. It took me years to go from a pile of bits to getting a case and thinking seriously about how to progress it. So, a bit of a gap is no great concern to me, although every year I don't have the finished article is another year I could have been using it.

On the other hand I'll be forty next year and if I aim to finish this project by then it would give me a date to work towards. As I type I have around 11 months to go, so it's attainable. Finishing would also free me up to do something else. I can't think what that would be though. 

Anyway, enough of this introspection.

One thing that has held me back from continuing with the project is having a decent scope to work with. In my early twenties I spent around three years of my work-life with a scope or soldering iron in my hand. When I'm working on electronics it's a scope that I reach for to see what's happening. It's the right tool for the job. I say a decent scope because I do have a scope, a GBDSO - Elektor Gameboy Digital Sampling Oscilloscope. This project was fun to do and produces tolerably good results but it's also a far cry from the professional kit I'm used to and, whilst a bad workman might blame his tools, a bad workman usually has bad tools. The issue is that I need this to be a pleasurable experience and using the GBDSO can be frustrating. I also want to be able to see audio traces cleanly and even some fairly pricey and professional digital scopes do a poor job of that.
So I bought a second hand Philips PM 3050 60Mhz analogue scope instead.

As you can see it's dual-trace and there's a nifty LCD display to show you the current settings. The traces on screen in that photo are the square/pulse and sawtooth output of the CEM3340 voltage controlled oscillator chip on my Roland MC-202.

The 202 was modified by me with some CV inputs years ago and has always been a bit flaky. As part of a general sort out in my studio I resolved to do something about that. I also realised that an SH-101 that I have on loan from a friend isn't working any more. No, I didn't break it. Well, I don't think I did. It's hadn't been used for years so I'm not sure what happened to it. So, I'd like to repair that too. It was these repairs that set me thinking about a scope again and how annoying the GBDSO was to use.

I've fixed the 202 now. The main issues were actually to do with removing some of the battery circuitry and disconnecting the internal sequencer. The sequencer was zapped when I did the mods originally so I decided to live without it but made a mess of the way the battery state is monitored and disconnected when the mains power is applied. I also fixed the filter audio input which never worked because I hadn't realised that the 1/4" headphone jack socket re-purposed to be the input was shorting the input to ground!

The 101 is next on the repair list and then - back to the SS30-M.

Wednesday, April 01, 2009

How are you to switch negative voltages?

I've been wrestling with the problem of key switching again. Because the key driver circuits switch a negative voltage to ground it creates a bit of a problem.

When I first looked at the j-Omega MPT8 I thought it coudl switch negative foltages but after a thinking again and e-mailing then it seems not.

What's all the fuss about th though? I can use a solid-state relay, optocoupler or CMOS switch package right? You don't even have to think to hard to get it working. The issue here that I have 49 keys and very limited space. I'd really just like a transistor and maybe one or two resistors per switch. CMOS switches only coem in quad packages at most so I'd need 12 and all the trracking back and across each other to get everything wired. If I must have a PCB at least I'd like it to be simple.

Generally switching negative voltages to ground is not something you get a lot of talk about when looking up these things. Everything is geared to positive voltages and how to bias you trnasistor that way. It's not impossible just less usual and if you wan t o use a simple +nv/0V logic level your options are limited.

The reason for this is not that you can't do it (just switch from n-channel to p-channel FET) but that transistors that switch negative voltages themselves need a negative voltage to switch. which takes you back to square one.

Well, If a CMOS switch can switch a negative oltage with just a +nV power rail how does it do it? I've been wondering.

Google books have a Modern CMOS Circuits Manual online and chapter four has the answers

Wednesday, March 18, 2009

External Power Supply Unit

I'm thinking life would be much easier if the power came from an external unit, rather than having the PSU inside the chassis. Space is really at a premium and that's before squeezing in the Midi converter and switch circuits.

There are six lines (rails) coming from the PSU so if I had an external uni I'd need a six-way connector.

These ones from Rapid look good...

I powered up the PSU last night it looks good but I am worried about the extra cutrrrent draw when I add the new modules. It's all on a 0.5A fuse which seems quite low. Might have to do a bit of measuring...

Saturday, March 14, 2009


Is it practical to get a custom made PCB built up?

These UK company's do a custom service:

They both accept Gerber and Excellon format files.

pcb train recommend using Easy PCB for layouts but that runs to hundreds of pounds.

Instead, I found KiCad which does the job and is GPL.

Thursday, March 12, 2009

Switch circuits

I will need 6x SPDT, 2x DPDT and 2x SPST switches.

The 4066 contains four SPSTs which can be connected as required to make SPDTs and DPSTs.

2 x SPSTs = 1 SPDT
4 x SPSTs = 1 DPDT

So, I will need (6 x 2) + (2 x4) + 2 = 22 SPSTs in total. So, I'll need 6 4066s in total giving me 24 SPSTs to use.

It would be simplest to make a single circuit board up with all 6 switches on but you also need to accomodate the logic invertors. The 40106 invertors have 8 not-gates per package and you need two per switch (irrespective of what type you're making) which makes 20 gates or three hex invertors.

In total thats 9 x 14 pin ICS so in a single strip there'd be at least 63 rows on the board or in other words it'll be about 9 x 20mm = 180 mm. Which is okay, as Maplin do an 81 strip board 213 mm long .

Switches cont...

After looking around at switches I can see that there's alot of choice but it get's quite limited when you start narrowing down to DPDT types. This is another reason for looking at the switch IC idea.

I want to choose switches that will look nice and the biggest choice is in SPST.

Like this one from Rapid Electronics which has built it LED. Not blue though :-).


I'm going to need ten switches plus the power switch.


Speed - SPDT
Cello - DPDT
Violin - DPDT


Attack - SPST
Cello 1 - SPDT
Cello 2 - SPDT


Attack - SPST
Viola - SPDT
Violin 1 - SPDT
Violin 2 - SPDT

So, that's 6x SPDT, 2x DPDT and 2x SPST.

When I breifly worked in pro-audio I noticed that they almost never had audio signal going through the actual mechnical switch. The switch was usually controlling a DG201 or DG211 audio swicth IC. This was high-end gear designed for the live use so I suppose there was a concern that meachical switches would degrade over time geting damaged or dirty. You don't really want audio going through anything that might colour the sound or produce noisey clunks on switching.

The original switches are all mechanical though, so why worry? Partly because I won't be using the same kind of switch and partly because there may be a way to improve on the original.

This guy calling himself The Tone God has a design called Wicked Switches that uses 4066 analogue switch ICs. He also explains how to add debounce and LED indicators.

Wednesday, March 11, 2009

Spec reset.

Over the years, I've been thinking I should really make this project easier to finish.

The two main issues have been the awkward meachnics of fitting the original switches and knobs and the challenge of conjouring up a 49 output, velocity sensitive MIDI decoder.

The meachnics iossue could be resolved at stroke by moving to a set of new off the shelf compoenents.

Similarly there are several polyphonic MIDI decoders available from one company or another.

Therefore, I will continue this project on a simplest is best principle in order to finish it sooner rather than never.

j-Omega Electronics make a reasonabley priced MIDI decoder that will meer my needs.

j-Omega Electronics MPT8

The MPT8 has 64 outputs in an open-collector 'pull-down' arrangement. I'll be pulling up, from -7V, but to ground which is okay.

Tuesday, March 10, 2009

Matrix synth comments.

I added some comments to Matrix Synth:

"some sources describe this synth as a duphonic/ multiphonic ,others describe it as a 4 voice polyphonic. I have a feeling it is a 4 voice polyphonic but the 4 voices are split between cello (two tones) and violin (2 tones) . It then uses divide down chips enabling all keys to be played at once)"

Nearly. It has two master oscillators which are both dived down to give you the full set of notes, twice.

You need two waveforms to get a stringy effect so the same note outputs of the dividers are mixed.

When you press a key you're switching in a mix of two divided down oscillators.

The detune affects both oscillators but not equally hence you can detune one against the other and get to the good old thick sound but more importantly set the PWM type character you need for strings.

The outputs of these mixed waveforms are sent to either the Violin, Viola, or Cello mixers. In some cases the outputs can be sent to either one or the other mixer as set by the split selector switch.

The cello, viola or violin character is set by filters after each mixer. Violin 1 & 2 just are different (1st order) filters fed from the mixer, so when you switch them in you're just mixing and matching different filtered outputs of the same thing. the same goes for Cello 1 & 2.

So far it's just a paraphonic organ but with two tones per key. However the SS30 rules over other string synths because it has an AR envelope generator per key. When you set a long attack each time you press a key it starts a new attack just for that key. same with Release.

There's no VCF per key but that's what we have Polymoogs and, err, GX1's for.