vtolnut
Heliman
Location: USA
My Posts This: Topic Forum | JKos
I think you are on the right track on two of these your ideas. I believe the problem is likely on both ends. The tx is reading the pots in a sequential manner (likely and A/D converter and a multiplexer to save costs). There is a finite amount of time required to calculate the pulse widths required for each servo. Even if the algorithm updates the servo positions in the TX data stream simultaneously, as you have stated and I have measured independantly (actually 5 years ago) the generation of the servo command pulses are interleaved on the RX and not sent simultaneously ( Disclaimer I have not scoped a 5019 DPS 14 rx yet so I cannot say that it conforms to this).
This means even if you had "Ideal" servos that were match to .000001 percent variance in specs they would still not begin moving at the same time with current technology except with one cheat. If you calculate the latency on the RX side and the algorithm execution speed is fast enough you could update the "slow" channel first and then after a couple of iterations update the "fast" ones. by fast(early update) and slow I mean based of firing order of the RX. Since it is and order like 1,5 .. 2,6 .. 3,7 .. 4,8 ..(varies by mfgr.) and is fixed this is possible provided that the Microprocessor in the TX can run the code fast enough that multiple iterations can occur in the time between sending the encoded frame to the RX.
If you scope the bus voltage of your reciever a 10 microsec or less sampling you will likely see a significant difference between the behavior of the voltage when all servos are moving when you compare V. reg vs. battery power. Also you will be surprised at the marked difference in the variance in battery voltage during servo sweeps on a full charge battery and one that is 50% charged.
If you want a real treat do this test with JR Futaba and airtronics equipment (different tx update rates, different interleave of servo pulse). from our research ATX had clearly the best CCPM (especially 120 degree). The 14MZ has closed the gap and may be slightly better (still testing). I know that the 14MZ is touted to be the only one that you can mix out all of the interaction, but on a machine with a good CCPM implementation we were always able to get purer swashplate motion on the ATX equipement before the 14 came out.
One RX that I have seen actually sequentially sends the command pulses out for each servo (not a great thing for CCPM) in a 1,2,3,4,5,6,7,8 manner. Not one of the big 3 though. |
