I made some tests:
Note that I changed the value for the timing capacitor (that defines the sampling time) from 470p to 1.5n as stated in the description at Rene’s homepage to increase precision. The sound of the videos is only recorded with my phone, so maybe you need to turn up your volume a bit to really hear it, sorry for that!
This one is the default:
C_hold = 1nF → a lot of droop!!
This one is with
C_hold = 47nF → a lot better!!
I also tried C_hold = 100nF, but It was pretty similar to C_hold = 47nF, but I only checked by ear, so maybe there is a difference. Also: the other caps I tested were regular film caps, not polypropylene.
I also varied the resistor to change the charging time (I tried 10k, 560k and 1M), but I could not see any difference there, but again, maybe when really measuring it makes a difference and I only tested with a maximum trigger rate of 16th notes at 300BPM.
In all cases (also with the default) I noticed a wobble when triggering at higher clock speeds:
Conclusion:
For my purpose (I want to use it to sample CV to get more complex sequences from a simple 4 step sequence or from my Turing Machine) I will go with 47nF or 100nF. That should be sufficient to keep the note more or less for about a minute, fine with me, especially if I use a quantizer after that 