|
Post by tronicgr on Nov 19, 2016 15:17:45 GMT
The issue with the backlash is that it messes up the PID, increasing speed too much then it has to brake the motor to avoid overshoot.
One solution is to reduce the max motor speed, and P values. This will make it smoother with the backlash.
The best solution though is to go around the backlash by attaching the sensor on the back end of the motor (where its fan is mounted) with a small no-backlash gearing that matches the reduction of the main gearbox.
Thanks Thanos
|
|
|
Post by riton39 on Nov 19, 2016 18:17:08 GMT
Yes, it's a good idea, I thought about that, but with our potentiometer of about 270 °, it's not easy. Moreover the engine turns to 3000 tr.
It is necessary to be sure that it is really a backlash problem ...? It may be just a noise problem, what does the oscilloscope say?
|
|
|
Post by Trip Rodriguez on Dec 2, 2016 6:13:20 GMT
I'm afraid I don't know how to use the oscilloscope. With step by step instructions "for dummies" I could try find out! I'm about to record a flight for you guys in a few minutes here and post it. I know Stevant was very happy with his after you tronicgr helped him tune it so I figure I have to be able to get decent results. I am using the shorter (130mm) torque arm setting as I thought that would be helpful but maybe it's a hindrance?
|
|
|
Post by Trip Rodriguez on Dec 2, 2016 6:59:13 GMT
Here we go:
A013 is 0 (default) b083 is .5 (default) ACL and DCL both set to .50 as per Stevant, I had them at .15 previously and this feels a bit better. A054 Braking is set to 80%, tried lower settings and didn't notice any improvement.
PID= in this video is 18-0-0 But lower P seems better, I've gone down to less than 10. Max motor speed is 50%, feels better than if I set it faster. I've been running it at 40 even, less than that seems to deaden it but maybe if I turn motor speed down and then turn cue settings up and reduce smoothing?
My seat platform is going to get a lot more rigid and a bit heavier in the coming days, we will see how that improves it.
Thanks for any help folks!
|
|
|
Post by Trip Rodriguez on Dec 2, 2016 7:26:52 GMT
Sorry the game sound is so loud guys, I turned it way down in X-split but apparently it didn't take. =(
Ok so watching the video myself it really looks pretty good. I think the shaking from my platform not being rigid enough might be a significant part of my complaint and that is about to improve a lot over the next week or two.
Aside from that I think it's the direction reversals that are too abrupt. I'm thinking about turning the external braking off and try that. What else can I do to make it take longer (make it smoother) to reverse the direction of the arms?
|
|
|
Post by paulg100 on Dec 2, 2016 18:04:13 GMT
"What else can I do to make it take longer (make it smoother) to reverse the direction of the arms? "
More washout or smoothing in bff?
OK looking at your settings, you still have b083 set at 0.5.
You can test yourself, run the platform and put BFF in manual mode. Use the slider on one of the dofs and see how quick the platform responds. Now lower B083 to eg 0.1, then move the slider again, Can you see how much smoother the movement is and how it responds to the slider, more like 1:1 rather than a big deadspot which you get with 0.5 at 0.5 my platform is like its playing catch-up with the motion cues which creates more jerking.
What do you think?
it seems logical to me that the more you slow the platform ie slow acl and dlc and higher start frequency (b083) the less 1:1 the platform will be with the motion cues. Surely its better to get these parameters as low as possible then use the AMC max motor speed to cap the speed?
Sorry I havnt had time to contribute more I'm tied up with other DIY and work until xmas vacation when I have 2 weeks to get back into things on the motion sim.
|
|
|
Post by Trip Rodriguez on Dec 3, 2016 9:51:01 GMT
No worries PaulG, as I said you've been a great help.
I will play with the B083 as you said. I did not yet test it the way you say but I did fly with it for a couple weeks at 0.1 and then set it to 0.5 to compare and back again and didn't notice any difference from a purely subjective standpoint. Since I didn't feel a difference and my wife complains more with it at 0.1 it is currently back to default. I'm covering the motors which should greatly reduce the noise soon.
Today I took apart the platform to weld some square tube framing to the main upper hex frame in order to give me a nice rigid platform and get rid of the shakes from flexing MDF. Tomorrow or Sunday hopefully I'll cut the tube and weld it up.
I also took the opportunity to press the downsize 5/8" to 1/2" spacers into the 160mm and 190mm settings on my torque arms (they were previously at 130mm). I'm considering trying 190 next to go from one extreme to the other, but with those inserts installed now I can switch arm lengths in minutes so experimenting should be easy.
Using the longer arm settings should be akin to having a heavier platform, but as usual there are checks and balances and in this case it's that with the longer arms the gearbox backlash is bigger at the platform.
Another reason for trying out my max length torque arm setting is to get more heave travel. I'd really like more heave and think I can definitely deal with a little less pitch and roll which I'll have to do to prevent the platform from hitting the motor/gearbox housings.
Trip
|
|
|
Post by Trip Rodriguez on Dec 10, 2016 3:36:08 GMT
Figured I'd post a little update here. I've finally got my steel frame upgrade welded together and will be installing it tonight on the livestream. Lots more to do, but this is a big step in the right direction. Hopefully eliminating the shake from flexing MDF will go a long way toward making the sim feel smoother. I'm also hoping the added weight might help.
|
|
|
Post by Trip Rodriguez on Dec 21, 2016 14:34:43 GMT
Problems, problems, problems! So I just did my first tests after the frame upgrade and I'm having serious issues.
First let me tell you about how I foolishly changed too many things at once! The main thing I did was expand the steel frame as seen above (from just the hex) to give rigid supports for my pedals and side consoles. This also significantly increased the weight. While I had it apart I decided to also try going from 130 mm torque arm setting to 190 mm. I also moved my seat a few inches forward to make it easier to climb in and out.
So the first thing I did to test (while it was still just the frame bolted on with no seat, floor, etc. attached yet) was move the platform around using BFF to move individual actuators and just check range of motion. I did find that in certain combinations of actuator positions I was able to go beyond the limits of my heim joints so I'd have to use less than 100% range of motion in BFF settings.
The next round of testing is where things went bad. I tested using the BFF World version of BFF this time with everything fully assembled and me on the platform. I set my max heave to 90% and everything else to about 50%. I did my testing by enabling the "by axis" manual mode. The first thing I noticed was I tripped some of the VFD's a couple times so I slowed down the motors by reducing the P and online motor speeds to 8 and 35 respectively.
The movement seemed all wrong by axis, with lots of instances of one actuator being delayed or not moving at all etc. I started suspecting that this was maybe due to the DCS version of BFF being a sort of Beta so I switched to P3D for further testing and had similar results. In particular actuator 3 and 5 seemed to be acting strangely. At this point I smelled that dreaded hot electrical smell so I shut things down and found that several of the motors were very hot, especially 3 and 5 but a couple others as well.
I gave it a few minutes to cool off then tried to use the buttons on the AMC1280USB to move the platform to the up position so I could swap my torque arms back to the 130 mm setting but actuators 3 and 5 wouldn't even make it all the way up.
After letting things cool for an hour or so, at which point motors 3 and 5 were still warm to the touch, I powered up the sim and was able to achieve full up position and change the torque arm settings back. After doing this using the AMC1280USB buttons the sim once again moved nicely like it should.
I then attempted testing again with DCS and P3D using the manual mode "by axis" and again got odd behavior with the #3 motor especially not doing much moving. I then tried "by actuator and got strange behavior that way as well, with actuator 3 moving little or not at all. I made this test session short, then I checked and some of the motors had gotten pretty warm again, especially #3.
I checked with the AMC1280USB buttons again immediately and all actuators are still working perfectly that way.
So what do I do?
Important Note: I've never used the "by axis" manual test mode before, only the "by actuator" mode so it's possible there are issues there I don't know about.
|
|
|
Post by Trip Rodriguez on Dec 21, 2016 14:56:26 GMT
Flying in P3D seems to work fine (or rather same as it always has) though, odd. Problems might be limited to manual mode by DOF? Still this makes me suspect that maybe the motors aren't exactly doing what they are asked which could explain some of my dissatisfaction with the motion. One more thing that might be relevant, I just remembered I'd disabled A051 braking just before I took it apart to see if that would smooth things out so that was OFF for all these tests. Just now turned it back on. Another thing, can anyone tell me what the "feedback limits" setting is exactly on the AMC1280USB? tronicgr
|
|
|
Post by Trip Rodriguez on Dec 21, 2016 15:29:26 GMT
Just did some flying in P3D and some motors got hot enough that I can smell them again. #3 is hot enough I have to take my hand off it after a few seconds. I only added 40 pounds or so to the platform so shouldn't be that big a deal. The one thing I must say though is I moved the CoG forward and the rear two motors (1 and 2 for me) are cool to the touch but the other four (all attached at the front) are warm/hot.
Thing is that even with my CoG moved forward the COG is still behind the center of the top hex but maybe further back is better.
|
|
|
Post by Trip Rodriguez on Dec 21, 2016 15:54:03 GMT
Alright, seems I made a classic mistake there. Motors were working too hard because I slowed them down too much. I did this first by lowering max online speed, then after by reducing the P to 5. I actually really liked the motion at a P of 5 (nice and smooth) but the platform actually pretty much wouldn't pitch up! When I climbed off and lifted up on the front it pitched up way higher!
|
|
|
Post by vicpopo on Dec 21, 2016 17:38:05 GMT
Hi Trip ,
First , many other members building a 6 dof didn't use so lower P values .Personaly I have P= 32 and I know other member using P also around this value. Eventhough you use your sim for flying , the setup to get smooth movements is not 100% a question of the P value.I agree that P has a big influence but for exemple if you once tried Fabi's software to setup the PID , I would be very surprised if the setup choose a P value like 5.
Second correct me if I'm wrong , you use the feature in your inverter DC braking ? If yes the overheating comes frome this feature because to slow down the motor the inverter send current to the stator .The result is that the motors become quick hot. I work for a AC motor manufacturer , I'm not an expert in electonical question regarding motor controling .But some other people in the factory yes. When I talked to an expert in this area ( motors driving ) regarding my application ( I bought all stuff for my sim in my factory ) and knowing that my inverters have got the feature dc braking he told me to forget this feature . He said that this feature usaully used to slow down a payload but for exemple in lifting applications , and not applications where you can brake several times per second. He said that I will loose completly the control of the regulation because the inverter become the master. In our application the master is always the amc and the inverter is only slaved to what the amc send .The PID regulation is made in the amc and not with the inverter. For your information my inverters are setup in vfc mode ( Vector flow Control ).
So I would recommend not using dc braking feature in your inverters because to slow , motors overheating and you loose the regulation !
|
|
|
Post by Trip Rodriguez on Dec 21, 2016 18:24:54 GMT
vicpopo Thanks for the reply! I have been experimenting with many things to try to get smoother motion, very low PID was just the most recent and not something I stuck with. On the DC braking interesting thing here is when I had the bad overheat problem initially DC braking was actually turned off. I now am pretty happy with the motion after some more fiddling with the settings in Ians BFF but motors 3,4,5,6 still getting hot. I think I need to move weight more to the rear which will be a pain but I guess it's got to be done. I'm not sure how much heat the motors can tolerate, they get hot enough I can only keep my hand on the housing for maybe 10 seconds and I have to let go. Currently the COG is actually slightly to the rear of the platform still, but less so than before and I wasn't smelling hot motors before. About Fabi's tool. I could not get my motors to oscillate using his tool so it did not work for me. Others with similar setups to mine have had the same issue. As a result my PID I and D settings are zero. Right now I'm running 19 P and 100% motor speed and toned it down using the BFF settings and pretty happy now. For anyone following the thread, making the platform much more rigid was definitely a huge help.
|
|
|
Post by vicpopo on Dec 21, 2016 19:34:18 GMT
For the motors it depends on their temperature class factor . F is a comon class Factor , Class H is a level higher ( the max temperature is 180 °C I believe). But sure If you can put your hand on the motor housing it's quiet " normal " for a ac motor ( 70°C is a normal temperature for an ac motor , if you have a tester you can check this temperature) Normally in most of case with a inverter configuration there is a temperature sensor into the stator winding .The sensor ( named kty or pt100 or pt 1000) is connected to inverter .When the motor become very hot the inverter stops the motor ( security mode ). On my simulator I'm always surprised because the motors stay quiet cold ( max 30°C) .
edit : corrected class F and H , I made a mistake . Class F is until 155°C and H 180°C
|
|