CV & Gate Expander circuit voltage issue

[sixbyseven]

Hi There,

Has anyone built Ray's MFOS CV and Gate Expander circuit?

I am using an Arturia Keystep. and the voltage is fairly decent up and down 7 octaves, but the circuit seems to have an inconsistent drift

I am getting a voltage drift on the CV voltage outs.

Example (steps in octaves)

Voltage IN _> Voltage out from circuit
0.746 _> 0.746
1.746 _> 1.747
2.745 _> 2.735
3.741 _>3.607
4.740 _> 4.139
5.735_> 4.436


The drift after 5 octaves is almost 1.3 volts!

The circuit is here http://musicfromouterspace.com/analogsynth_new/CVNGTEXPANDER2015/pdf/cvngtexpand2015_schem.pdf

I am thinking it has to be either with the zener diodes (double checked the are both 1N5231B, oriented correctly and the resistors are all 10K 5%). The voltage on all 4 outputs (out 1 thru 4) is exactly the same. They all drift the exact same way. Ray mentions that there could be some offset voltage, but it should always be the same offset as the input voltage changes.

Should I try a different manufacturer for the 1N5231Bs?

[sixbyseven]

The 1N5231B Zener diodes have a reverse current leakage of 5 micro amps at 2 Volts. Ray mentioned that he did not use 5Volt zeners, but that he used 10 volt ones. The 1N5240B is the 10 volt version and does not starts the reverse leaking until 8 volts. This is why I got an exponential looking shift to the output voltage

I removed Z1 and Z2 from the board and retested and the voltage outputs are perfect, exact copies of the input. Anyones else building this take note:

Choose a Zenier where the reverse current leakage starts above the range you need, so if you want 7 octaves, use the 1N5240B instead of the 1N5231B.

[LukeLabs]

I made two of these recently and can confirm replacing the Zeners with higher voltage versions drastically improves the performance.

I'm using 1N5240B's too and now have 10 octaves of usable CV output.