Testing frequency changes related to pulse width

Update: Solved - it is not pulse width that is the issue, it's the pulse amp CV, just as in the waveshaper

 
The AS3340 suffers from a problem with the pulse width circuit where the oscillator frequency changes when the pulse with changes. 

I have made a separate circuit to prevent this, and tonight I did some measurements on the VCO to see if the fix worked.

Surprisingly, it did NOT:

 

The frequency is clearly directly related to pulse width. I don't know why, but it definitely means I have to look into this. Again. At least, this time I have a good tool for measuring stuff.

 

For reference, without modulating the pulse width, the frequency is stable:

Update - found the reason

For simplicity, I had connected the pulse width CV and the pulse amp CV together, even if the pulse amp is not in use.

When grounding the pulse amp, the frequency became stable:

 And then, with a stable pulse width (CV at 2.5V) and a changing pulse amp CV, frequency changes

 

Testing with updated VCA

With the new component values, the frequency is stable:

 

Pulse wave amplitude is changed a lot without showing up on frequency

Linear FM

And while we're at it, I checked linear FM (and routing of exponential FM), it works as expected:

Yellow: Output of DCO B, blue: output of VCO. DCO modulates VCO pitch

 

 

Red is VCO frequency, it clearly follows the DCO above. Note also that it is capped at 0, a through-zero FM VCO would work better here. That's for the next version of the VCO.

All in all, I'm pretty confident the new VCO version works as it should.
 

Improved CV inputs for VCO

As mentioned earlier, I'm not satisfied with how the VCO CVs work. Noise on the CV probably has a higher effect on the pitch than necessary. Here are my current thoughts on how to improve this:

First of all, move all summing resistors to the VCO board. This requires an additional pin for separate pitch and exponential FM input.

Second, reduce the Pitch CV resistor from 39k to 100k

Third, add a second pitch CV, also with 100k input, but make sure the CV is inverted (and 5V added to it) after it is generated. In this case, any noise present on Pitch CV 1, will be negated by the same noise on Pitch CV 2. Again, this requires an additional pin on the VCO board.

I'm pretty sure adding two extra pins here are possible, especially when removing the two resistors to the left of the top square (moving them to the VCO board).

 

The two 15p CV headers on the mainboard and the CV generation board can be replaced by 32 pin ones, like this female one:

https://jlcpcb.com/partdetail/XKBConnection-X6511FV_32C85D32/C2883749 

NB:

I am not sure how this is affected by noise on the ground plane, and how the VCO is affected by ground plane noise - because, if the noise is present at the ground plane (as seen by the VCO) but not on the CV generation board, the CV value will fluctuate, and vice versa. If it is present both places, I'd assume it would cancel out, but I'm really not sure about this.

Also, when adding 5V to Pitch CV 2, I need to think about how noise would affect this as well.  

Cem3340 four trimmer tuning

I can’t remember if I’ve written about this before, but there is apparently a way of tuning the CEM3340 (and AS3340) that is done in a single go, without the need to go back and forth between different potmeters. The downside is that it requires four trimmers instead of just two. Also, I don’t know exactly how it reacts to temperature changes.

The procedure is described in detail (with explanation of how it works) here: 

https://cabintechglobal.com/tune3340