Jay first suggested the resistor cure last year: viewtopic.php?f=25&t=472&p=3048
Another option is a separate actuator battery, splitting the positive at the DRV8838 as described in the DigiSpark recoder thread: viewtopic.php?f=25&t=599
I've only just realised that the very original recoder thread was lost with singlechannelersreunited - its a bit late in the day now but time permitting I'll see if I can find anything. It also had the 'separate battery' option, and even if the thread is lost I do at least still have the old diagrams...
Basically its a case of splitting the pin 2 to pin 9 link on the DRV8838 and supplying a separate pos to pin 9.
16 every time Mike
I find 16mhz to be very stable and repeatable, its just an opinion but I think it might be because the PLL, whilst locked to the input frequency, has a lock time-constant so it can can free-run over any minor input hiccups or noise, just like any textbook PLL. Its just an opinion