Hi all,
The good news is - I've found what's causing my servos to chatter continuously while stationary. Unfortunately the bad news is that I haven't found how to cure it.
Here's a video showing my testing and results:
http://www.youtube.com/watch?v=nTKaGRf7g5kHere's the analysis:
I started by examining the PPM signals at the input pin to the Tx module and at the servo outputs on the Rx, looking for any instability in the servo control pulses. Both proved to be rock solid, with <1us jitter on the encoder output and none at all I could measure with a 1Gsample/s scope on the FrSky Rx outputs. A single click on the trim would cause a very obvious 4-5us shift in the pulse length. This rules out any encoder problems (such as noise coupling through the power supply from the Tx module that some others have noted in this thread) or receiver instability as the cause of the servo chatter.
In the process of fiddling around however, I stumbled on the cause - RFI (Radio Frequency Intererence) from the Rx telemetry transmissions directly affecting the digital servo amp. I found that as I handled the Rx the chattering would come and go, and I worked out that it would stop completely if I held the active antenna.
In the video I show the chattering reliably stopping when I hold the antenna, or cup the RF end of the Rx in my hand. I tested with both digital and analogue servos - JR DS811 digital servos, an old JR NES507 analogue servo and a cheapo no-name mini-servo out of a park flyer. The analogue servos are completely unaffected, only the digital servo shows the problem (I have 4 DS811's - they all do the same thing).
I then did some testing to try and work out if the RFI was coming in on the servo lead or being picked up directly from the RF field. I added ferrites on the servo lead (the video doesn't show it, but I tried ferrites at both ends of the lead), and then tested with extension leads to space the servo away from the Rx. The results suggest there is a bit of both going on, but that direct RF coupling is the dominant problem.
So I reckon the RFI is either:
a) Affecting the operation of the pot a/d, causing the pot position to be misread.
b) Pushing the quartz oscillator frequency around, so the PPM counter miscounts.
c) Doing both.
I opened up a servo and had a bit of a probe around to see if I could find a fix, so far without success. Some 10nF/100nF bypass caps in the right places or better groundplaning on the servo amp might help, but for now the only solution I have is not to use the DS811 servos with the telemetry system. That's not really a great solution though, so I'll keep looking.
I'll post a follow-up video soon with some additional testing I did - and showing the antenna diversity and RSSI working. A cool and totally unexpected by-product!