HobbyKing.com New Products Flash Sale
Reply
Thread Tools
Old Jan 01, 2014, 08:30 PM
Registered User
Mesa Falcon Fld, Arizona, United States
Joined Sep 2004
426 Posts
Question
Does Anyone Have a Recording/Log of a Servo PWM Input with Glitches?

I was wondering if anyone had a log of a glitchy servo signal. What I am looking for is the length of time for the on cycle of the pwm. I would try and do a setup myself, but I just moved, and I do not have all my tools with me yet and will not have them here for a while. Some data with actual commanded moves in the servo would be great.

If no one has some actual data, I guess that I could simulate this by adding some random noise to a representative signal that was designed to have some commanded movements in it. I would have to make some estimations of how fast a person could move from one extreme position to the opposite extreme.

The end goal is to look at some filtering routines that might help for reducing glitches being sent out to the servo. It is nothing too high tech, and most likely is implemented in reputable products already, but I am making a device from scratch and hence I will have to do it myself.

Thanks in advance,
Adam
Aeroengineer1 is offline Find More Posts by Aeroengineer1
Reply With Quote
Sign up now
to remove ads between posts
Old Jan 01, 2014, 11:20 PM
We want... Information!
Bruce Abbott's Avatar
Hastings, New Zealand
Joined Jan 2001
5,169 Posts
If the receiver does not have filtering built in, ie. it's decoding ppm with a standard shift register circuit, then the servo pulse could be any width - from less than a microsecond to permanently stuck high. Noise pulses can also bump channels down the line, so for example the aileron channel (even if glitch free itself) might sometimes get fed to the elevator or rudder servo.

The fastest I can move a channel from one extreme to the other by hand is about 10 times per second. That's for a switched channel (eg. retracts, flaps). I can move a stick full deflection in both directions at about half that rate, but I would never want to. So you could limit the maximum deflection per frame to 20% of full travel and it wouldn't bother me.

That's not to say that it wouldn't be noticeable. A standard servo takes about 200ms to deflect 60║. With an additional 75ms delay due to rate filtering it might slow down to 275ms, almost 40% worse. The delay percentage would be even greater for faster servos. This could make a highly aerobatic model feel a bit sluggish on the controls, but does it matter? If you want the best performance you shouldn't be using outdated ppm technology anyway...
Bruce Abbott is offline Find More Posts by Bruce Abbott
Reply With Quote
Old Jan 01, 2014, 11:32 PM
Registered User
Mesa Falcon Fld, Arizona, United States
Joined Sep 2004
426 Posts
Bruce, thanks for the response. Thankfully I am not needing to work this for high performance models. Your numbers on the switching rate are a good start for what I am trying to do.

On a side note, it is unfortunate that I cannot update to a more modern methodology, but I need to stick within 75MHz and maintain compatibility with older radios.

Adam
Aeroengineer1 is offline Find More Posts by Aeroengineer1
Reply With Quote
Old Jan 02, 2014, 06:25 AM
"MAYONNAISE"
Acetronics's Avatar
Le Treport, France
Joined Jun 2004
1,397 Posts
Hi, Aero

as Bruce said ... a signal between 800 and 2200 Ás which changing rate " not too fast " can be considered as a "classic" valid servo signal ...
depending on servo brands 250 to 2500 Ás signals give a possible servo response ...

I made, years ago, a PPM dedicated failsafe module based on that filtering type ... Elektor magazine also published an 8 channel Failsafe some time ago.

some trails have also being made by sending a kind of CRC before the PPM sequence ( see Francis Thobois's site ) and allowing new output only if CRC Ok ...
not perfect but catches " heavy problems "

no miracle the PPM way ... but it may help !!!

Alain
Acetronics is offline Find More Posts by Acetronics
Reply With Quote
Old Jan 02, 2014, 11:03 AM
AndyKunz's Avatar
Illinois
Joined Sep 2001
22,326 Posts
Peter Berg and I independently developed similar technology for this back in the mid-90's. Here's a link which will give you some info: http://www.castlecreations.com/produ...7-channel.html

There are several filtering methods you need to apply.

1) Throw away any signal which is too long or too short. Alain's point is good, but in general you can narrow it down to the 750-2250us range and cover 99.44% of the radios out there.

2) Determine the frame period. This is done usually on startup, but may be adjusted during operation. The frame period will usually be very consistent, especially with newer radio designs. Any signal which starts too far before or after the expected (scheduled) start of the next pulse is presumed to be invalid. You do need to allow a window (typically < 100us in my experience) as there are valid reasons for a signal to have some start jitter.

3) For a single channel, expect that adjacent pulses be similar widths. If you have a long string of pulses at 900us +/-5us, and then you get a 2100us pulse - it might be valid. But if the next pulse is 900us again, you can ignore it. Doing this introduces a frame-period delay in the signal (5-25ms) - probably acceptable. If not, send the pulse through anyway - the servo will have a hard time making the transit. Yes, you'll see a glitch.

It was a fun project. On the systems I was working on (this was for FMA in the mid-90's) we tested two transmitters on the same frequency at the same time, and the rx was able to properly lock onto the correct transmitter even if its signal level was only 1/4 of the interfering transmitter's. You would see jitter as the two transmitters phased through one another (like two combs sliding) - you'd have control of your model, and you'd know somebody was on your frequency, but you probably wouldn't crash.

Andy
AndyKunz is online now Find More Posts by AndyKunz
Site Sponsor
Reply With Quote
Old Jan 02, 2014, 05:24 PM
Registered User
Mesa Falcon Fld, Arizona, United States
Joined Sep 2004
426 Posts
Thanks for the info guys. I will take a look at that link a little later. I am looking to design myself a small 75MHz RX for my subs. It is getting harder to find good ones anymore. The last good vendor stopped selling them because his RFIC was no longer produced. I am looking at a couple of different concepts, ranging from the simple low pass recursive filter methods to something complex like an extended Kalman. I am going to run some quick tests of each method in Octave and see the difference in phase shift that would be observed in each filter method for a real world scenario. After that, I would just need to determine the processor cost, though with the uC that I am planning on using, I think that it would be able to handle an EKF for what I am looking to do. We shall see. It is all a concept right now, just doing some feasibility studies.
Aeroengineer1 is offline Find More Posts by Aeroengineer1
Reply With Quote
Reply


Thread Tools

Similar Threads
Category Thread Thread Starter Forum Replies Last Post
Help! Does anyone have the fuselage build plans for Dare Design DC6 ? redtail RC Airliners (Commercial, Cargo, Troop Carrier) 5 Mar 27, 2014 08:02 AM
Discussion Duratrax 835E, Does anyone have one? rwestbrookjr Off-road Cars 9 Jan 11, 2014 10:13 AM
Question Does anyone have any experience with E-flite's universal light kit controller? SkylineFlyer Night Flying 0 Dec 28, 2013 09:29 PM
Discussion Does anyone have this plane? waywardson Electric Plane Talk 11 Dec 28, 2013 11:46 AM
Discussion Does anyone have any in-flight video from mass warbird launch? Zagabond NEAT Fair 2013 4 Oct 01, 2013 06:05 PM