PINNED - Roboteq Controller - developing for powerchairs

[Burgerman]

OK will do.

[Burgerman]

Did that, nothing changed. Instant disconnect, script doesent run, and refinds the controller.

[LROBBINS]

OK, here's the script with no motor compensation. Let's take out subroutines one by one till we find whether one of them has the problem, or whether it's in MainLoop or in the preliminary stuff.

[Burgerman]

Thanks! But you forgot the script!

[LROBBINS]

Oh well, it's been a long day. Here are two scripts. One has no motor compensation. The other one has a main loop that has nothing in it except the Wait statement, so it will test whether the disconnect happens in the "preliminaries".

While doing this I did have one other thought. Early in the program it reads the Roboteq settings for motor maximum amps (a.k.a. amp limit).

Code: Select all

vMaxAmps1 = GetValue (_ALIM, 1)
vMaxAmps2 = GetValue (_ALIM, 2)

I don't know what happens if you haven't set a value for these in the Roboteq configuration.

Ciao,
Lenny

[Burgerman]

Will find out.

[Burgerman]

Strangely, they both do exactly the same as before.

The origina old one still works fine.

Are you sure they are different?

[LROBBINS]

They sure are different, but both have the same "preliminaries" section. Even the one that never calls any subroutines does, however, still have the subroutines present (though they should never run). I'd like you to try two more "do nothing" programs. The first, called "prelims only" is just like "no main loop", but I've physically deleted all the subroutines and all comments. This really shouldn't make any difference, so this one should still fail.

The other one, called "no _ALIM" has one other little change that I won't try to explain now except to say that looking in the Roboteq software file "constants.h" and watching the simulator output I may have spotted a bug in the Roboteq software that makes _ALIM unusable. If I am right, this script will run, even though it actually won't do anything at all useful and will sit there seeming to do nothing at all. If it does run, I will bitch to Roboteq about the bug (actually, should be easy to fix by them) and come up with a workaround so that _ALIM isn't needed. If this one also doesn't run, I'll have to think again.

[Burgerman]

NO ALIM doesent misbehave, but doesent do anything useful.

The other one disconnects and does the same as before...

[LROBBINS]

OK. Those test scripts were not designed to do anything, just to help figure out where the problem is. IT IS DEFINITELY a bug in the handling of the _ALIM variable by Roboteq. It's not actually a bug in the software or firmware, but a bug in the manual. In several places the manual uses GetValue (_ALIM, 1) as an example, BUT THIS IS NOT AN ORDINARY NAMED VARIABLE. It is a configuration variable, and it has to be called with GetConfig and not GetValue! (I had a much more complicated explanation of this as a real bug, until I realized that there's a, repeated several times, mistake in the manual.)

So, here's a new version of a working script that has the two offending lines fixed. Then I'm going to bed. Tomorrow (actually later today), I'll try to explain why we may not want to use _ALIM at all, depending on just what Roboteq meant by one sentence in the manual. But no worry, once you have this hooked to some motors, I'll tell you how to simply test whether we want to use _ALIM or not.

I can't guarantee it, but I think that running this version will bring a smile.

Ciao,
Lenny

[Burgerman]

OK. Just in from pub. Wet, windy, and cold... Thanks!

Will test tommorow afternoon. Then we are bringing all my tools, powerchair stuff etc inc BM3 bits back her for rebuild and testing! Next few weeks will be fun.

[LROBBINS]

Hey John, I'm anxiously waiting to hear whether I did good or not. Ciao, Lenny

[Burgerman]

Just got back in. Will test in a minute!

[Burgerman]

Looking good!

Its smooth (but not quite as smooth on the graph updates as before?) and turn and reverse is limited.

100 percent at TX and calibrated to full stick movements. But reduced as it should be on graph. Now, will the compensation work?

We can but test. But will not be for a week or so until I get that far!

Thanks much for your efforts!

[LROBBINS]

Good, at least it's functioning, the rest can be tweaked as needs be. If I don't get an answer from Roboteq about how to interpret the phrase "the command value is the desired Motor Amps relative to the Amps Limit configuration parameters", Roboteq manual p. 168, I will tell you, once you have motors hooked to the controller, how to find out experimentally. My feeling now is that scaling _A in accord with _ALIM should actually NOT be done, but I'd like to be sure.

In what way to turn and reverse seem limited? To slow? If so, just up the values in

Code: Select all

vReverseScaling = 33 'REVERSE MAX SPEED = 33% of 90% SpeedPot = 30%
vForwardTurnScaling = 33 'FORWARD MAX TURN RATE = 33% of 90% SpeedPot = 30%
vReverseTurnScaling = 33 'REVERSE MAX TURN RATE = 33% of 90% SpeedPot = 30%


or does it seem to not respond to the full range of reverse and turn stick movements? I don't see why that would be so, but if that's the problem I could start thinking about what's needed to fix that too.

Ciao,
Lenny

[Burgerman]

They are not too slow or too fast. Need to sit in it to test! Working fine so far.

[Burgerman]

Pod now looks like this after a coat of paint (was old and grey) and after throwing away all its PG Drives electronics and it joystick.

Now has APEM joystick, and so far just 3 buttons for speed reduce, lights, off/on. Unlike all the latest HUGE pods with built in wide screen TV this one is small and does the job! Stolen from the Pilot Plus system obviously! Even though its just the casing.

[teebee]

Hi guys. New to this forum.
Am I right in thinking that the Roboteq could (theoretically) replace ANY controller with the right programming ?

[Burgerman]

Basically yes. And some safety concerns and a bit of creative thinking/wiring.

[dr_fisher]

Did John implement the safety switches as shown as mandatory in the Roboteq documentation? I have read through vrious suggestions but have I missed the final soltion. eg. relay, cord swtch, crowbar etc?

[Burgerman]

As yet undecided... I have it on my bench at the moment. Testing with wheels in the air tipped on end.

[dr_fisher]

Aha, thought you may be further along, I saw script seemns to be working as of November. Devices that will swith 2-300A dc are pretty pricey.

[Burgerman]

I am not in a rush, have 3 built, shiny chairs... This one is one to play with. Doing water cooling blocks at the moment for the hot end of the motors.

[dr_fisher]

What's your feeling so far about the equipment chosen. How does it drive?

[Burgerman]

Fast! Acceleration from stopped same as any other chair, but it continues to 3x the speed. So you can go along at say 3 or 4 mph and then wheelie away like a mad thing! Of course double the volts gives 4x the power. Hence water cooled motors...

But that was with RC transmitter as joystick, unpainted (temp build) 45 volts, and no motor compensation, and control trhows set on the transmitter. And only for 10 or 12 mins in the street. So it wont really help you decide.

Would be better with longer wheelbase for normal users normal use. And with suspension as well as fat tyres if it didnt make it wider.

[rcbook]

This is my first time so let’s start
My name is Robert I am the equipment manager / coach for the San Jose steam rollers
A wheelchair power soccer team, I have been using Roboteq 2550 and 2450 controllers for 5 or 6 years now in my wheelchairs with 2.4gh RC radio the problem I have found is when to many chairs
Get too close together or when the chair hits another chair the rc system loses signals a lot or the program will change in the transmitter I have got the best rc system Dx-8 now still the same problem.
One time when we went out for dinner the chair took off on its own Have you had this problem before?
I am now switching to an Arduino microcontroller in place of RC inters face still working on dead band code still.

[Burgerman]

When a user is in the chair you do not want to be relying on a RC link. The 2.4ghz band is far too crowded and weak, with masses of other things sharing 11 channels of data. You are surrounded by phones (bluetooth is 2.4) wireless networks in buildings use the same 13 or 11 channels and laptops and phones trying to talk to it. And the failsafe if too slow to act at times of signal loss! Works great for hobby stuff but less than good indoors due to reflections (multipath distortion) and because of all the wireless networks etc.

You need to be using a well sheilded DC swing analog input joystick for input, (or a pulsewidth/frequency one) with some roboteq scripting to control both motor compensation as well as turn rates and reverse speed. This is what I am doing. Albeit only in testing, and it seems to be as good if not better than any normal powerchair controller at least. Especially if still using 24v and lead batteries or control will be a little vague to put it mildly.

So no I have never had that problem. I use 2.4 ghz radio link on all my chairs, but only for moving them about while not in them. Its really not a safe option to use with a human on board.

Theres no need for dead band that I can find. In fact mine is set to have prctically none. Its at 2 percent at the moment and thats not really required. Adjustability Is already built into the inputs in the roboteq software. Why would you need the added software / code when the roboteq can also run this if required?

[woodygb]

You really want to get a Spektrum reciever like the AR500 that failsafes to null output on loss of signal.

[Burgerman]

It still takes 1 second before the failsafe kicks in. I have several fancy 12 chanel receivers here. Came with the JR12X but they work the same as the better 9 channel recievers from spektrum. That 1 second delay could be under a train...

[rcbook]

Ok I think I know what you are doing.
You are connecting the joy stick directly to the controller and then using the scripting to control
Or tame down the jumping of the motors at low speed with the compensation Right?
I could really use your help. Can you get me the scripting?
Some player’s need the dead band open up more than others because of lack
Of muscle like my son we have dead band set at 15% with only 30% total through movement of
Joystick would make full speed on the chair. If the dead band is to close and he gets hit or hits a bump in the road as you know any slight movement of the joy stick causes the chair to react right now.
Like you I have all the p@g factory level programming. We find the r-net can be set up to do what
It’s not supposed to do like react fast turn in a dime set power levels up so on
But it still does not compare to the roboteq.
And yes I have been in rc airplanes for 38 years it was easy using rc gear and you are right it’s not safe
But I am not good at all with scripting or open source codding. Back in the day when I first
Started using 2550 I connected the joystick directly to the controller and it was way to jumpy
And when we lost one leg /joy stick wire broke OMG hold on for a wicked ride. Yes it hurt lol
When I finely stopped ran to something that did not move or break down. Of course after that
A kill switch was installed for safety.
I hope I am not asking for too much I have spent the last 15 years working with the soccer teams
Across the US In my little world I am the go to man.
Thanks for the quick response
Robert