Stellaris Galon

Dragged an old friend out of the junk pile and press ganged it into Galon operation. I will put a 2 x 8 0.1" connector in the corner and try the UI as well. It will never really be a easy DIY build as any PCB will have to have the 0.5mm pitch processor... It is available in 64pins so not too many connections to get right and  maybe I will do the whole analog front end in surface mount as well...

Rise of the Dotz...

Well the Dotz Display PCB's were back in no time at all! So quick they have beaten the parts by about a week! Oh well...

I also decided to pension off the old DIY Braids that was a bit of hacky mess of Nucleo board and this rats nest of wires. It never really worked very well and was particularly noisy... Still it got me on this adventure.

Folding the Wave

Slowly cranking through the remaining Oscillators that I hadn't got going before. I am now into the Analog Oscillators as appose to the Digital Oscillators.  These are used to make more traditional Modular VCO sounds. I am always amazed that when I think "Oh I can add..." to the oscillator I find that Oliver has already got it in there! This is a case to point, I was thinking lets make a wavefolder like the Buchula 259 as a voice... While rummaging through the Folded Sinewave Oscillator FOLD I find that it is already here! The wavefolding transfer function that Oliver uses is a clever mixture of a modulated sinewave draped over a arctan transfer to achieve:

Which bears a close resemblance to:

From Jürgen Haible's Wavefolder A, described here. Jürgen typically went on to make Wavefolder B which is a fairly insane beast!

Where does that Highway go to?

Have some more boards in the works... Currently doing some interesting comparisons of Cortex M0+ Vs M4F performance... It would be nice to work in the two on the left in as well.

New processor...

I have a couple of new processors in the pipe-line... The one that I think will be the most successful is the STM32 M4F. It has a lot of similarities to the NXP one that I have been using as far as the peripheral set is concerned. I am thinking that I will crank this one out to 16+ bits in the DAC department though. I was a bit disappointed with the DAC on the KV10... The QFP32 in the STM32 family dosen't have I2S but you can fake it up with some counters and the SPI. So I am thinking of using the CS4334 that is easily available here. The other advantage is that STM32 M4F that I want to use is available in a Nano footprint eval board. So some testing is under way...

 There is also the PT8211 -- very cheap but maybe a bit wobbly in the quality department... Is available in DIP8 though... I also found some MAX5556 in the old sweet tin...

Display #4 (Virtual)

Whilst waiting for deliveries and other stuff I finished off this:

You can see it here as well.

Display No.3

 

Got the 3rd display to work -- I managed to swap the nibbles to the display somewhere along the line... Quick fix in the code but there back of the display is a bit messier than I would have liked. Wrong line to reset and completely forgot about the contrast pot! Anyway it works...

The XR Sawtooth is an attempt to implement Thomas Henry XR Oscillator in software. The original is a pretty cool oscillator and one of the first that I made. I was really after something that I could compare the against as I now have both... Still as little bit of a way to go. You can read about the XR VCO here.

Output Quality...

Well the advantages of using a processor with on board DAC and ADC is it is well integrated, cheaper and simpler... Down side is that the DAC is only 12bits and using an ADC on the same substrate as a lot of digital logic is always going to be a compromise. To see what we are getting out of this baby I have made a digital oscillator in the code that is A440 and it is independent of any analog input control so we are just testing the DAC. It looks fine on the Oscilliscope:

However having a look on the Spectrum Analyzer shows that there is a moderate amount of noise there:

So a after having a fight with the torturous UI of the spectrum anaylser the signal-to-noise is about 45dB rms. This isn't too bad as the theoretical best for a 12 bit converter is about 72dB (6dB a bit) and reading the datasheet we see that we have an INL of +-8 LSB so probably what we would expect... The software currently just prunes off the bottom 4 bits, we could do a nearest neighbor and also accumulate the error to perhaps push it a bit closer... The distortion comes out at around 0.5% which compares quite favorably with the sinewave accuracy of most Analog VCO's output from transistor curve sine approximations...

How does it stack up against the original? Well the original Braids has a external, rather nice, high quality TI DAC. I don't have one available to measure but your going to have to be 20 to 25dB better off... But it's not all that bad when you consider the Oscillator is designed to be +-5V out in most modes and the filtering and dynamic processing is down stream in the VCF/VCA chain. It is more the modes which do physical modelling that it will be more noticeable. The Orgone Accumulator actually runs on the same family (albeit the M4 version) and uses the same internal DAC.