Replies to This Discussion

Permalink Reply by Curt Olson on August 15, 2012 at 12:57pm

Just a quick follow up to my own question (I still want to know the answer because I suspect I'm not seeing something I should be.)  I can find references in the ArduPlane code (APM_Config.h.reference) saying MANUAL mode is full manual control via the hardware multiplexer.  But elsewhere in the code (Attitude.pde) I see if control_mode == MANUAL, the code appears to copy the receiver input values to the servo output values in software.  In my own experimentation (i.e. running a modified APM_radio demo), I toggle the selection switch (CH_8) but I never get straight pass through from my reciever ... the servos are always driven by the software outputs no matter which way the selection switch is set (and I verify the sensed value is changing from one extreme of the PWM range to the other.)

So hopefully someone can clarify -- is there something I need to do to enable the straight multiplexer pass through?  Was this feature removed from the hardware?  Are the comments in some places in the code lagging?  Am I correctly understanding the code, or am I missing something?

Thanks!

Permalink Reply by Paul Feely on August 15, 2012 at 1:01pm

Curt 

Manual/Auto is software based, there is not separate hardware mix like original Ardupilot or the APM 1, check out similar query on subject.

If it is important to you have a look a a separate mux

 

Paul

Permalink Reply by Curt Olson on August 15, 2012 at 1:18pm

Hi Paul,

Thanks for clarifying.

Hmmm, bummer, no one got permission from me to make this change!  Echoing the comments in the other thread...

• This could affect the ability of any APM2 based UAS to ever pass safety checks unless the description of the manual failsafe can be sufficiently convoluted that no one catches this.
• I assume we add a bit of noise and also a bit of lag in the controls by doing manual pass through in software -- the system has to read the RC receiver, update the PWM output registers, and then regenerate the signal.  I assume the hardware signal generation is clean and good, but timing the RC pulses is a place where some noise can be introduced (even though the APM does this pretty well) -- and then there will be a frame or so of extra lag in the response.
• I understand this simplifies the board, but this is going to bite someone somewhere sooner rather than later I think -- who writes perfect code all the time?  (other than myself) :-)
• Thanks for pointing out the external mux -- that my be a workable option -- although it adds significantly to the rats nest of wires in an install, increases the number of connectors, solder joints, etc. -- more external points of failure.

So I guess for now I'll quit complaining and work with the system, but I hope Jordi and Chris and others have their antennas up for horror stories, run away props, and otherwise unexplainable crashes and will consider adding the hardware mux back into the APM2.6 (or 3.0 or whatever comes next.)

Permalink Reply by Fred Forbes on August 17, 2012 at 2:32pm

Hi Curt

Bummer indeed!  I learned awhile back that APM 2 had software controlled manual.  I was unhappy with that for the same reasons as you.  What I didn't know was what I learned from the reference given by Paul, APM2 is primarily used by Arducopter fliers and Arduplane fliers tend to use APM1.  This was the first time I became aware of that.  I would not have spent my hard earned money on APM2 had I known of these two facts.

Oh well, back to my old APM1.  However I would support a future version of APM with hardware manual, perhaps as an option or maybe user-configurable

3D RoboticsPermalink Reply by Chris Anderson on August 17, 2012 at 2:54pm

Curt, as others have mentioned, hardware MUXs don't make any sense for multicopters (which cannot be flown manually), which represent the majority of users. Unless we release a different board just for planes, we can't have a hardware MUX built in.

We do however sell a standalone MUX (just $18) that you can add that will do just what you want.