Here are some wonky photos of the Roland System 700 clone VCA boards that arrived yesterday. The layout was ordered through iTeadStudio on Sunday 4th January and verified as ready for building the same day. I got a message to say they’d been dispatched on Friday 9th, and received them on Monday 19th, so it took just over two weeks. It’s a bit of a drag but I could’ve paid for the expensive postage if I’d wanted them quicker.

Check the comically thick traces round the back:

Compare with the SH-2 VCO board:

…I probably went a bit over-the-top with the trace thickness, but hopefully it’ll be ok.

I’ve stuffed most of one of the boards already, including the rare-ish 2SA493 and 2SC1000 transistors, I just need to order a couple of things I’m missing. Thankfully all the holes seem big enough, and despite the outrageous size of some of the pads it hasn’t made it particularly difficult to solder it. I’m pretty impressed with iTeadStudio so far.

With the parts in place, it’s easy to see it’s definitely quite a, um, relaxed layout – certainly compared to that 808 bass drum I did a while back. The boards are wayyyy bigger than typical Eurorack things – this is one of the smaller boards and it’s just under 15cm x 10cm (so no Euro version, soz). With a different layout this could be made an awful lot smaller, but I was keen just to stick to the original.

While I wait for the PCBs to make their way from China, here’s a preview of how the Roland System 704 VCA finished up, previewed using Kicad. Not sure about C6 now I look at it – although I think the red circle is the solder mask cutout, so it might not be so bad.

The service manual has the board layouts, but I wasn’t keen on just copying those straight out; it seemed more reliable to draw out the schematic and then route that with the parts placement and track layout as a guide.

Wanting to avoid learning potentially hugely expensive software (thinking of large boards in the future) I had got used to drawing schematics in Kicad, but the one time I tried to route a board using it before was tortuous.

I really wanted to use Osmond PCB – something to do with it being designed for OS X, and also the possibility of old-school curved tracks seemed attractive – but there didn’t seem to be a way to convert the netlist output from Kicad to Osmond. It seems that Designworks Pro might well output Osmond format, but it’s £150 less a penny. I don’t mind paying for really good software, but the comments on Muffs were hardly glowing.

I went through the convoluted process of installing the gEDA suite to try and use their converter program, but didn’t get very far with it.

Finally I opened up the output netlist in a text editor and realised it wasn’t the incomprehensible voodoo I thought it was, so I knocked up a quick and dirty program in Java to convert from the Orcad netlist to Osmond format. The workflow I’ve ended up with is probably not as smooth as using something like Eagle, but it seems to work ok.

Here’s another preview using Mayhew Labs rather lovely 3D Gerber Viewer:

In a fit of excitement because it almost seems like fun (although it takes me ages) I’ve since routed the 702 VCO and 705 dual ADSR, but I’m holding off ’til I get these boards to see if I’ve made any horrible beginners mistakes. Hopefully they’ll arrive this week, I’m pathetically excited.

Continuing my suddenly obsessive System 100 stripboarding with a boring bit, the voltage controlled amplifier.

I wasn’t expecting this to be amazing sounding, being not much more than a CA3080, but I thought it was worth a go as I’m trying to replicate the original as far as I can.

Here’s the original VCA, with the blue marking on the 3080 presumably to show that it has been selected for low CV feed-through. I’ve seen circuits for matching OTAs, but I’m not clear how best to select for low CV bleed. Maybe it’s just a case of trying a bunch of 3080s to see which is the least clicky.

I wanted to keep the mix input from the 102, so I built the thing around Q337. Also not having a TA7136 SIP opamp to hand,  I just subbed in the usual (non-inverting) TL072-based CGS DC mixer on the end, which was a mistake.

As I understand it now, the System 100 VCF inverts the signal from the VCO, and the VCA inverts it again, so it’s all back the right way up. As it stands, my layout inverts it twice, so input to the VCF eventually ends up inverted.

Incidentally I can’t imagine the TA7136 will make much difference, but now I’ve just had some arrive in the post, I’ll build a version with it.

To give an idea of my general confusion, I spent an evening wondering why the output from the VCF was crazily offset by a number of volts, before realising that this would be sorted out by the capacitor C324, which is on the VCA board. Just because it seems all nice and modularised doesn’t mean that it is.

Here’s a demo with the envelope on the System 100 101 keyboard modulating the VCA cutoff.

It gets clicky like my SH-5 with the envelope on a short release and the filter cutoff low, it might be that I need to find a better 3080.

One mod that might be interesting is to boost the input to the VCA from the VCF to see how it overdrives. Maybe messing with R378 on the input to the CA3080 might do the trick.

Here’s a stripboard layout for the inverting version using a TL071, any corrections or improvements joyfully received in the comments below, taaaa.

Update 9th Feb 2023: Corrected capacitance values on decoupling capacitors on IC305 (C325 and C326) from 100nF to 10nF – thanks Mike in the comments below.