R-net powerchair continues to 'creeps' when it should STOP!

[Scooterman]

It's an intermittent problem with old r-net system I assume. In that sometimes when I decelerate and come almost to a stop (joystick is upright off/neutral position) and electric brake should clunk on. It doesn't and the chair continues to 'creep' forward under residual power. Even if I let go off the joystick chair will sometimes creep on, and it's not gravity. As I say it's as if some residual power is still going to motors. I tap the joystick back and it stops it, but it's still a little disconcerting if an immoveable object is in front of you? The speed of the creep is a fraction of an mph I would guess?

Any ideas?

[sergey0]

Could it be that the joystick itself doesn’t return to the neutral (zero) position when released?

[Scooterman]

I was wondering that too. Iwas thinking of removing it and squirting some contact cleaner through one of the threaded holes in the base.

[LROBBINS]

The joystick has Hall effect sensors - no sliding contacts. It may, however, be out of calibration and that can be fixed with the Rnet programmer.

[Burgerman]

I am not sure it can. It has a calibration setup built into the CJSM2 or CJSM screen. But think it only calibrates the end points and not the centre.

Do you have anything magneting in close proximity? That will make it drive.
It an also make it go out of calibration. Presumably you can make it go into calibration by messing around with a strong magnet too... Dont be indoors when you try that!

[Scooterman]

Perhaps it’s my magnetic personality affecting it!
Nah, not in the least!!! Only joking I’ll live with it for now but if it does get any worse I’ll try messing around with a magnetic. Outside is good a point, safety first as they say.

[Scooterman]

I’ve been thinking…

I’ve not got a magnet near the joystick but i have a neodymium magnet fixed to the inside rim of one of the rear wheels. Every time wheel rotates it activates a reed switch which is connected to cycle odometer. Cos I only use chair outdoors I use cycle odometer milage as a ballpark recharging guide. Magnet is passing motor and motor cable every rear wheel rotation

[shirley_hkg]

Your magnet is too far to act on joystick.

Check if the boot is compressed (lowered), that obstruct joystick returns to neutral.

[Scooterman]

Ooh good idea Shirley.

Maybe rubber boot has stiffened with age.

[LROBBINS]

John,

Are you sure that RNet joystick calibration doesn't silently calibrate the neutral point as a reference when it calibrates throttle and steering swings? I don't have any RNet systems (at this point just Roboteq, even got rid of all Dynamic stuff) so can't check.

[Burgerman]

I didnt think so. But it is possible.

On hobby stuff that centre position is calibrated seperately (centre all trims, sticks, pots, 3 position switches and press x button) as not all pots or josticks are 100% linear or mechanically perfect But they also have zero deadband and 4096 resolution.

[emilevirus]

I've had that once. The spring under the boot had moved from its space somehow. But you'd notice, it was harder to push.

[LROBBINS]

Does RNet let you set the deadband? If so, widening a bit might sole the finding neutral problem. BTW, the CANbus implementation for the Roboteq does have adjustable deadband as well as joystick zero and ranges (independent of the Roboteq calibration.

[Burgerman]

Yes you can. Default is 10% which is already 9% too wide if you ask me though. And it cannot be reduced. Makes it very "sloppy and non linear" feeling as it is. Compared to RC systems.

[LROBBINS]

Deadband on Rachi's (CANbus) chair is set at 7%, but there's also another parameter MinimumStick set to = Deaband that moves throttle and steering joystick output to 7% as soon as the deadband is exceeded. Without that, it was indeed hard to know when the chair would start to move when the stick was moved slowly, with that it became very predictable. I never noticed the problem when I was driving, but watching my granddaughter driving the chair I could see her struggling right near zero stick.

[Burgerman]

You should try a fast RC jet at a few hundred mph on a low pass (or inverted low pass)... Or a RC 3D heli. Yes not a Wheelchair. But it really focuses your attention on both latency and things like deadbands very quickly! Those things begin to be critical for control.

ELRS RC transmission for e.g gives me a choice in the menu of latency (frame rate) versus range.

So I get to choose between 50 frames or updates of servo position per second up to 1000 per second. Most competition flyers want the 500 or 1000 fps. And this reduces Rx effective sensitivity. And so range.

So we have a chart. This shows frame rates V Receiver sensitivity. We have to choose.
You dont think that an update at 50 times per second in a servos position would matter. It really does! Not for slow things, but for super high performance planes and race quads or 3d helis.
This is of course with very high resolution, and zero deadband already because that would make many things intolerable.