Odd one: retract sequencer on FreeWingRC Mig-21
09-24-2017, 09:17 PM
#1
Thread Starter
Odd one: retract sequencer on FreeWingRC Mig-21
Hi all,
I have a somewhat unusual dilemma. I'm working on a friend's FreeWing Mig-21. This is not to be a flying model, he's going to use it as a display piece for a museum exhibit (he has the actual cockpit section of a Polish Mig-21 two-seater/trainer, which will be the main part of the exhibit). We're trying to make this a "hands on" bit of the exhibit by connecting a real Mig-21 gear selector control to the FreeWing model, to actuate the landing gear.
Obviously, we won't be needing a full transmitter/receiver installation. Rather, I'm trying to run the retracts with an Arduino. I understand these are not conventional retracts, but the "on/off" kind. Having said that, since the retracts receive normal servo signals from the RX under a flying installation, I see no reason why the Arduino's native servo library can't run the retracts.
Here's my problem: the retracts seem to be nuts. I've tried connecting my Arduino directly to a retract servo and sending various timings (servo position commands), but the retract's response is inconsistent: sometimes a certain angle command gives me a retract or extension sequence, sometimes not. I tried going into the onboard gear/flight control mixer that comes with the model, with similarly frustrating results. At one point I had the gear cycling OK except the nosegear would lead the mains on retraction (not really a problem, just odd). Then after a while, the gear sequenced perfectly. But at this exact moment.....I'm getting one or both mains working at various times, and the nosewheel is not working at all.
Is there some magic beyond just hooking the mixer to the landing gear channel of your RX? Some more background: system voltage is about 5.15vDC, provided by a regulated ATX power supply. I can hook this exact rig straight to a standard servo and it works *perfectly*. I would suspect a bad mixer, but even going straight to one of the retracts is unreliable (bypassing the mixer and just connecting the retracts directly to three signal pins on the Arduino would be fine, if it worked). The retracts, the PSU, and the Arduino all share ground, of course.
I'm kind of out of the RC hobby now, or else I'd have an RX unit handy to plug into this and make sure it's not just some weird conflict between the Arduino and these retracts. At this point, I'm seriously considering gutting these retracts and just rigging up conventional servos directly to the gear....plus that would allow me to slow gear cycle time to something more realistic than 2 seconds (should be 6 to 8 seconds for the Mig-21).
Matt
I have a somewhat unusual dilemma. I'm working on a friend's FreeWing Mig-21. This is not to be a flying model, he's going to use it as a display piece for a museum exhibit (he has the actual cockpit section of a Polish Mig-21 two-seater/trainer, which will be the main part of the exhibit). We're trying to make this a "hands on" bit of the exhibit by connecting a real Mig-21 gear selector control to the FreeWing model, to actuate the landing gear.
Obviously, we won't be needing a full transmitter/receiver installation. Rather, I'm trying to run the retracts with an Arduino. I understand these are not conventional retracts, but the "on/off" kind. Having said that, since the retracts receive normal servo signals from the RX under a flying installation, I see no reason why the Arduino's native servo library can't run the retracts.
Here's my problem: the retracts seem to be nuts. I've tried connecting my Arduino directly to a retract servo and sending various timings (servo position commands), but the retract's response is inconsistent: sometimes a certain angle command gives me a retract or extension sequence, sometimes not. I tried going into the onboard gear/flight control mixer that comes with the model, with similarly frustrating results. At one point I had the gear cycling OK except the nosegear would lead the mains on retraction (not really a problem, just odd). Then after a while, the gear sequenced perfectly. But at this exact moment.....I'm getting one or both mains working at various times, and the nosewheel is not working at all.
Is there some magic beyond just hooking the mixer to the landing gear channel of your RX? Some more background: system voltage is about 5.15vDC, provided by a regulated ATX power supply. I can hook this exact rig straight to a standard servo and it works *perfectly*. I would suspect a bad mixer, but even going straight to one of the retracts is unreliable (bypassing the mixer and just connecting the retracts directly to three signal pins on the Arduino would be fine, if it worked). The retracts, the PSU, and the Arduino all share ground, of course.
I'm kind of out of the RC hobby now, or else I'd have an RX unit handy to plug into this and make sure it's not just some weird conflict between the Arduino and these retracts. At this point, I'm seriously considering gutting these retracts and just rigging up conventional servos directly to the gear....plus that would allow me to slow gear cycle time to something more realistic than 2 seconds (should be 6 to 8 seconds for the Mig-21).
Matt
10-05-2017, 07:48 PM
#2
Thread Starter
Just a quick follow-up.....I fixed this by ditching the Arduino's native servo library and just rolling my own signal timings to the servo. It ended up being very simple.
There is obviously something "special" about the Mig retracts that makes them less forgiving of odd input signals, and there's obviously something odd about the Arduino library's servo timings, but I don't have the time or energy to figure out what. I'll just carry on with my own method. My solution is simply based on the standard RC hobby servo signal as described in many online sources (except the one that said to use 40ms cycle times...).
In case anyone else has to deal with this, here's what I did: I'm not using the Atmel's onboard PWM timers, I'm using software PWM. The input signal to the servo is held HIGH for 1ms to 2ms, depending on desired servo position (or retract condition, in this case). This is followed by 18ms LOW. We end up with something very close to a 19ms to 20ms servo signal, which is fine, because servos are somewhat tolerant of not-quite-perfect timings (I don't know the exact tolerance, but anything reasonably close to 50hz seems to work).
My exact timings are 1.9ms and 1.1ms for gear down or up, respectively. The retracts work perfectly with this.
Matt
There is obviously something "special" about the Mig retracts that makes them less forgiving of odd input signals, and there's obviously something odd about the Arduino library's servo timings, but I don't have the time or energy to figure out what. I'll just carry on with my own method. My solution is simply based on the standard RC hobby servo signal as described in many online sources (except the one that said to use 40ms cycle times...).
In case anyone else has to deal with this, here's what I did: I'm not using the Atmel's onboard PWM timers, I'm using software PWM. The input signal to the servo is held HIGH for 1ms to 2ms, depending on desired servo position (or retract condition, in this case). This is followed by 18ms LOW. We end up with something very close to a 19ms to 20ms servo signal, which is fine, because servos are somewhat tolerant of not-quite-perfect timings (I don't know the exact tolerance, but anything reasonably close to 50hz seems to work).
My exact timings are 1.9ms and 1.1ms for gear down or up, respectively. The retracts work perfectly with this.
Matt
