Delta Build Blog #1 (The truth hurts)

These days I build synths for a hobby, yes I sell them, so it’s a “self funding hobby” and I’ve been very lucky to be in that position.
So I thought I’d share some of the joy, and pain, of designing and building a synth in today’s world. This will be an ongoing blog, so if you’re thinking about building synths professionally this might help you in some way.

As you may, or may not know, I like synths and I really enjoy making new and interesting synths, I like modular synths because they allow you to connect things in ways that conventional mono and poly synths don’t always permit. On the plus side for mono and poly synths, you can usually store and load patches.
So there are pros and cons for both, and both modular synths and fixed architecture synths have their place in my life.

One of my passions is making something new. 002 did that in many ways with it’s internet updating (not new to us, but new to synths), HTML editable patches and control morphable ladder filters in hardware and a bunch of other things.

The WTF (Window Transform Function) Oscillator also came out of my wondering what it would be like to switch at audio rate between two waveforms, it didn’t imagine how I thought it might, it sounded totally different, but it’s a very cool thing.

The waveplane oscillator was a way to allow people to sort of create their own tables, at least a way for people to morph between waves in any order they wanted, something that’s difficult to when you create wavetables (believe I’ve created lots of wavetables in my life and it’s a lot of work involving hex editors, audio editors, spreadsheets and code editing to make them come to life).

The natural step, for me at least, was to move to a poly synth. So whilst I’m finishing the MuSeq I’ve begun working on what I’ve tentatively called the “Delta” polysynth.

What are the specs? Well I’m keeping what I’m working towards close to my chest, but it will have both the WTF and Waveplane oscillators in, along with some new features (remember I like making new things, move forward, not backward).

First step on the development path was to prove the oscillators, Having done that by selling both as modules for Eurorack and MU systems I’m quite pleased they sound good.

Next step was prototyping the VCFs, this time I did this on breadboard.

Then we moved on to prototyping the voices for a poly synth, I found that I can get 4 voices of oscillators and 12 CVs from a single 600Mhz processor.

So we have all the parts!
The next challenge is getting down the cost, and here is were things get tricky, thanks to some budget brand companies, people now expect the moon on a stick for $300.
Ok, I get it, I’m on a really tight budget (still paying of debt thanks to a certain someone from a few years back) and I certainly can’t afford £3000 for a synth.

So how do you get the cost down?
Well there are a number of ways, firstly you go all digital/DSP or you could compromise on the quality of the parts used, You could use lots of menus and reduce the panel complexity (complexity = cost), you could reduce the feature set and so on.
So you have to find what works for you, I have a target retail price in mind, now one thing consumers of synths probably don’t know is how the synth market works.
Let’s say a synth costs $1000 (nice easy number to work with), that is split 4 ways… $250 to the dealer, $250 to the distributor, $250 for the raw parts which leaves $250 for the manufacturer.
So next time someone moans about the cost and has a go at a company for the “excessive pricing” please, keep in mind that the company in question has only made 25% of the retail, if that.

How I’ve chosen to keep costs down?
I’m trying to keep things small and light, yup big boxes and weight cost a fortune to ship to distributors, and to dealers (this all has to come from somewhere!). So my Voice PCB is 100mm by 110mm for 4 voices currently, I hope to make it smaller.

So, we’ve chosen a design approach, we’ve proto-typed our oscillators, our filter and worked on the cost down. Now we roll a PCB.

Delta 4 Voice Prototype PCB

Now, let’s be clear, this stuff isn’t cheap, I have 4 of these quad voice PCBs, that cost £1100, in volume they’re much cheaper. Remember that rule of thumb? as a one off this synth is already at a £4400 retail price.
Which brings us to the next bit of your “mega synth project”, how many are you going to sell? if you can sell a thousand of them, then yup you can get the cost down, if you can only sell 20 then the cost is going to increase.
Let’s say the board in 1000 off costs you £50 each, that’s £50,000.
That’s the world we live in. And don’t forget, YOU have to pay this upfront!

Yes, my aim is to make this commercially available, how possible that is depends on potential sales, pre-orders, investment and what the final cost will be, at this stage I have a target cost but nothing firm.

So your PCBs are here, you power them up and…… nothing quite works…
Welcome to Research and development. The first PCB you roll in any project will not be the one that goes into production, there will always, always, be a change. The WTF had 3 iterations of PCB, the Waveplane had 2.

In my case, I have two mode wires (I got two chip selects the wrong way around for my sample and hold circuit). and I will be changing the value of some resistors (the range on the cut off doesn’t quite cover what I’d like, it’s close, but not right).

Mod Wires (not intentionally blurry, it’s just hard to photograph something that small)

So, having figure out why half my CVs were not working or in the wrong place, they’re now fine.

I’m now tweaking the resistors, and one of the tiny little ones gets lost whilst I’m modifying the board (they’re 0402 which is 1mm by 0.5mm), I think nothing of it…. carry on, power up and the CPU is dead… ugh, guess what? that resistor? shorted two pins killing the processor.

Thankfully I had the foresight to NOT put the CPU on this first generation of board, so another one is on order.