My 3DOF project is born!

Hi dudes, the 33ms are used to avoid my PID exceed to accumulate errors in a duty cycle...
haha yeah it's just a my personal idea about the mysterious matter of PID..
In fact my old (over 1 month ago) PID was 10ms and this actual 33ms is just one of the thousand of little modifications I made to avoid the connection lost problem.
Sincerely I don't feel the delay of the motion if I am using 33ms.. but to cut the problem I will back to 10ms!.

bsft make me thinking about this:
X-SIM working with 33ms of delay. So If I am using 10ms from jrk config utility so.. what is the ACTUAL and working delay? the one from jrk config utility or the one from X-SIM?



Today I am ready to new tests where I will to try these new things:

1) I will to test these PIDs :
1/0 - 0/3 - 0/0 I think the most brutal configuration
7/2 - 0/3 - 5/1 I think the better safety config (by eaorobbie)
(actual PID is 4/2 - 0/3 - 0/0)

2) I will try to remove or definitely reduce the mean value filter because yes, this filter make the vertical movement more clear and linear but using it I lost a lot of little bumps!

3) I will put a camera really near the spring to watch exactly how serious the problem is.
About the spring I notice that seems is not a problem about rotation of the spring but seems more a deformation problem. I hope not but I need to know precisely the situation!

4) I back to use more sensibility when I make little movement of my wheel. From last vid you can notice that I am using a too high lateral effect minimum value.
I make little wheel correction while I am driving but the platform don't calculate these little wheel correction since I don't touch the too high lateral minimum value.

See later for update!

I USE 16ms AT JRK CON UTILITY AND XSIM PROFILE WITH NO PROB MY PID SET IS 2/2 0/3 1/1 AFTER ADVISE FROM AEROBIE AND BSFT BUT YOUR PROJECT IS TO HEAVY MY MAX CURRENT AMP SET AT 35 I TAKE POWER FROM APC PSU 1000W AND NEVER PROBLEM ONLY AFTER HOURS PLAYNG WIPER MOTORS BECOME HOT . WITH JRK AND EVERY PROJECT HERE NEED A START POINT AND IF U HAVE SOME TIME U WILL FIND WHAT U WANT EVERY PROJECT ISN THE SAME I SPEND MORE TIME FOR ALL THE TESTS AND PROFILE AND NO MORE TIME TO PLAY MY WIFE TELL ME I NEVER SEE U TO PLAY RACE GAMES ONLY YOUR SEAT UP AND DOWN HAHAHAHA

Ok, I made the tests and I feel a mix between joy about performances and sadness about a bad mechanical problem I got..



When I finished the test showed in this video so I made another one test and I get my left motor crazy and it went really over the minimum position, the ABS MIN and the FEEDBACK DISCONNECT security system did not work as if there was no security system... So the motor stirrup going down it's finished to push a winch bolt trying to continue the crazy motion.... I got my pot gear with 2 or 3 broken theet..
I notice too that X-SIM made an automatic STOP of the simulator but the motor continue to want to go... then I stop the problem only when I cut the current..

So tests stopped and I need another winch ( (In fact I need just another iron gear but the only way I know to obtain a gear as this is to buy another winch...

I tried to use aggressive profile (very reduced mean value filter so a lot of peak movements and little bumps too) and aggressive PID (I think the PID is the cause of the bad problem):
P = 2/0
I = 0/3
D = 0/0

From what I understand since here the P down parameter is the speed of the motor and I think that 0 is unlimited.

I used 0/0 (unlimited/unlimited) on the D zone probably making the STOP motor and OVERSHOOT variables going to get some damn bad error ... then the overshot movement and the gear damage.

I also notice that the problem is happen when I had my car drifting... Never in my driving section I get drifting my car, it's not my style, but today I done 2 corners both drifting REALLY MUCH and (just a case?) in the moment in where I was drifting to the max so I obtained my left motor CRAZY since the mechanical damage..

Here the video of the last NO PROBLEMS test section of this morning, the following test (lucky no video..) is the test in where I got the gear broken

http://www.youtube.com/watch?v=0VrW5kCGTws&feature=youtube_gdata

Which gear do you need? You can have one of mine, cause I don't use them all since I make my own when I modify.

I dont know if i can help becouse i amnot expert like others here but when i have befofe problems with feedback disconnect and after the motor goes around like crazy i put some stops up and down to the motor becouse i found that when the motor go up or down with much force broke my pots and the motor crazy

Yes... I think I will to use this system next time... because at now I don't know WHAT THE HELL is happen..

Hey dude you would save me!!! I need just the gear attached to the motor shaft, not all the winch!
Please tahustvedt, make me know (with a private message)how can you send me the gear and how many money the shipping need!

You should can really save me 'cause at the moment I have really overtake the budget and I can't buy another one winch.... ... .. .

Can you post a picture of the gear you need?

Here the pic showing the gear. In my system this is the gear where the pot is mounted!

I have an unused gear like that here. We can swap. You send me your broken gear and I send you my new one. I need it to make a hub for my plastic gear later out of the broken gear.

Send me your address in a PM.

Done mate!

I dont know much about the workings of the JRK but it will be much the same as any other feedback driven loop. A tighter response loop will actually create less wear on your gearing, providing you are using a less aggressive PID. Your error accumulation will become smaller between setpoints hence your proprtional gain will not need to be as aggressive as it responds to the change in a shorter period of time much faster.
Basically you want your refresh loop as fast and tight as you can possibly get it, this creates a faster response to the change in condition( the error). With a faster response you will not overshoot, but you will with an aggresively set P which you will need more of with a slower response.
Think about this a smooth driver is a faster driver, a driver that is jerky hard on gas hard on brakes aggressive with steering change creates more stress and wear on a car and will actually be a slower driver.
I don't know the JRK to tell you what to change the PID numbers are also weird looking to me on the JRK they dont look constant with the math I think they must factor other things.
So...
1). get a very tight refresh loop
2). get yourself a very good PID tuning guide, I believe there is also a graph in the JRK program which should make tuning visually easier. The less time it takes you to hit your mark the better at tuning your getting.

Hi BartS, from what I understand about the PID you are right.
But today something strange has happened.

For the first time I notice the follow:

1) The feedback disconnected secure system when the pot reached the Abs Max value was as non-existent... as it did not exist..
When the motor exceeded the abs max value so it continue his run since it touch a bolt of the winch...
I obtained the pot damaged too because the motor run turning the pot on his limit turn breaking it.
Then first time I see my feedback disconnect emergency stop not working..

2) X-SIM realize that there was a problem and automatically has STOPPED the comunication (as if I had pressed the x-sim stop button) but for the very first time this was not enough to exit the emergency... in fact the motor continue his run to the destruction...

3) I get the problem WHEN I done 2 consecutive drifting corners... In the very last moment of the drift I notice my platform behaviour strange.. and some moment later I listened a bad crack crack crack sound.. (the motor stirrup trying to move throught a winch bolt...)

All this is very bad because make me really confused about a solution...
as jimakosgr2002 told I think I will to use 2 turn/off buttons for each motor..
If the motor stirrup goes crazy and without control so it will finish to touch a MAX or MIN stop current button...
at least I will to avoid to get again a problem bad as the one of today..

Ps:
I want to publicly give a BIG thanks to tahustvedt because he will send to me another winch gear (the same type I damaging today! without this gear I must to stop all tests and no more money for a new winch...) :thbup:

In the event of safety you should always have max / min limit switches that are triggered a little further past your absolute max / min pot position in software presets. This is a backup safeguard system that you can use incase your software should fail which in your case has and unfortunatly cost you damage and money. These switches can trigger a circuit breaker to your power but if your motors spin and carry momentum it may not be enough to stop intime so maybe better to trigger emergency stop command on JRK. On a sabertooth motor controller this is called a dead man switch where you are no longer incontrol and a full stop command is initiated this can be triggered by limit switches. You have to safeguards there in the event of something going wrong, if the Absolute Max /Min position is overun and fails, your limit switches will save you as backup.

This does not solve your problem tho, it only prevents damage in the event of failure.

You need to find what has caused this problem and it can be a million things.

Start with...
Is there anything on your computer that could stop the smooth operation of X-Sim, X-Sim needs to be priority of your simulator and computer. Virus scanners, system routine tasks, windows updates, processor, memory and graphics card all working at limit etc can all affect X-Sim. Windows performance monitor can show you events that occur at the time of your double drift. Are your JRK's recieving enough Volts to operate effectively, are they powered by USB or by the battery?

If your system resources are fine....
Sounds to me like you have Absolute Max /Min limits set in both x-sim and JRK's could this be conflicting with each other its probably best to set the limits only on JRK, or set the limits a few positions before on the JRK than in X-Sim. You have 2 softwares doing the same thing, can have a negative effect on your closed loop.
X-Sims signal disconnection! is the JRK responding quick enough with an emergency stop command after no signal.
If your period is 33ms that means your simulator could already moved a few more positions in the time by the next 33ms.
Hence I recommend you send data faster to JRK say about 10ms, if no data recieved after 20ms you should emergency stop.
You could probably detect and respond 3 times faster to a connection problem at 10ms than 33ms.
Are your JRKs programmed to emergency stop on signal loss?
Could the problem be the pot broke first before anything else went wrong, or do the pots suffer from some kind of magnetic interference?
Did your JRKs overheat ( too many amps) and cause a software failure and reset during your double drift, because you are at the near end points of motion and pulling hardest on the spring at that point can also cause greater spikes in Amps and cause JRK voltage requlator trouble and force reset connection loss.

I have provide some key areas where you should focus your investigation further, hope it helps good luck.

Ok Aldoz, I have done some PID fiddling .
p lower number controls speed of motor, upper p number softens it s bit.
i lower and upper numbers control overshoot and braking effect, The less the number the harsher the braking, the bigger the number, the softer the braking and more overshoot. It seems to work well keeping each number the same.
Just an update from what I have found, I put on 5 turn pots and had to re-do PID.
Now, too harsh a motion and thats where JRKS get upset and crash I have found, especially on big spikes. The JRKS are having to control a much bigger motor than they were designed for so they are likely to spit the dummy.
And before Barts pipes up, I have a 32 amp 15 volt PSU running at 13.8 volts less than 1/2 a volt drop. And a car battery in parallel, so power may not be an issue. I had to raise the ms rate in JRK utility from 5 to 25ms as the JRKS had to have time to think , this alone stopped a lot of errors in the beginning of my big motors being used.
Eaorobbie and I have spend a LOT of hours sorting PID so we have a fairly good idea of what works. No need for mathematical calculations, just trial and error.
As for JRKS overheating, unless you do not have a heatsink and fan on them and are running them in an oven, this is unlikely, I have had mine for 1-2 hours and way overdriven current, like 32-38amps plus some extra. Yes the heat sink get HOT, but never shut down.

Guys, in first I am really happy about your interest and help.
I very carefully read the last BartS and bsft messages and I taked a lot of points of interest.
But unfortunately I find the problem.. and yes, it's my fault.

Just all these months of work and my attention slowly fall and I forget to set ON in the input disconnected and feedback disconnected error page of the jrk config utility precisely for the motor that went mad.
(I modified to OFF these options, one week ago for some tests and never forget to re enable...)

Each other 2 motors have these options enabled but just now I noticed that the left motor jrk have these 2 options disabled.

I think if these options are disabled then is normal that If the motor exceeds the pot abs max level nothing will brake it..
I do not know how I could forget this fundamental security step.. I know just this has happened.

if I had enabled these securities options from jrk config utility so I have gotten just the motor position over the abs max then the motor stirrup stopped before to make damage.
But both the security options was not enabled.. and bye bye.

Probably I need a bit of time of relax and shut-off from my project, or just to use more relax and operative way to continue to work.
I can not really afford these expense damages..

From a side of this sad discover, this is a good thing too because now, probably, we know WHY this is happen and no more fear about strange jrk behaviour.
Just a damn my error.

HEY! Aldoz its better a broken gear from a bumper if its real race car that cost too much hahaha i destroy about 10 pots to find the right way and about 4 wiper motors

Glad you found the issue Aldoz, its always a relief sorting a problem out.
I believe my troubleshooting tips help you think to find your problem faster, I know nothing of the pololu software so I couldnt really suggest anything what to look for in the pololu options, but all closed control loops work in the same way and I would troubleshoot the problems in the order I suggest.
And Aldoz you are way ahead of me mate so dont feel bad, take the weekend off and do some sim racing without the simulator. I am feeling really drained all the work I have put into the programming it been a very long 7 months problem after problem, I havn't even seen any motion yet for my hard work or started building it.

Get those limit switch backup measures built in and you will have no more expensive problems.

@BSFT: to be honest Aldoz simulator require more force to operate over the same period of time, 1 of his JRKs does more work than 1 of your JRKs whether you are driving more amps through your JRK or not. It isn't wise to guess recommending he use the same amps + volts as your sim and it be ok. You should only use what you feel you need and dont get greedy. He has plenty of movement and speed already.

BTW whether you have good cooling or not more volts / amps than the JRK is recommended to use is not good, maybe 7-10% over should be the cap, but consistantly over this and degredation will soon plague your microchip and circuits.

Hello Aldoz, I forgot about the settings in the error tab of JRK utility, stupid me should have thought of that. I'd better check mine.

Barts, I know Aldoz sim works differently to mine. I NEVER recommended that Aldoz use the same power/current as me. HE is fine with what he has.
I am only noting that from what I have used, there has not been a problem as YOU stated there may be a problem with power supply, of which, by the way, I have had such a problem in the past.
This forum is for those of us to share experience, not just opinions.
I deliberately have been overdriving my jrks and motors to see how far they can go. Yes I am aware of long term damage.
Finally Barts, where is your simulator. You talk of such a great build, yet we have not seen as so much as a bolt posted about it?
All you have shown us is that the Arduino will drive a few small servos. Thats not hard to do.
Where is this be all and end all code that you are supposed to be sharing? Still waiting to see that as well.
And before you raffle on about how perfect and reliable you are designing it, the people whom post problems on here actually have BUILT a project or two and they would actually have hands on experience about sim building, unlike you. You have had the nerve to comment on peoples builds on this forum and comment that they could look more professional, well lets see yours then.
Shut up or put up.

aldoz quiet, what matters is that you very well, come on! :