PINNED - Roboteq Controller - developing for powerchairs

[Burgerman]

I imagine the current suppliers exec's getting angry that you are so far ahead of them in performance - it is silly - and disgusting.
Especially with the current 'news' of the cheap batteries they have decided to supply in their chairs.
You must give them bad dreams -

They don't care. They are about making $$$ £££ as easily as possible. As such they are only interested in the easy targets. The mass of users that don't go far, don't do much, don't understand, don't really care, or get the cheapest chair they can find, or that are paid by insurance. Why bother doing any real development or actual improvement, when the average user is more impressed by cup holders while they watch TV. Or sit in a retirement home looking out of a window.

People like myself, will, and many others that are on this forum are a tiny minority of users that don't "fit" the pattern. My local government wheelchair services say that I am ruining the chairs, that I am unusual. Lead batteries for e.g. I get less than a year. They were horrified. They say that almost all their users get 4 to 7 years or more. Some are older. They allow less than £100 per year for maintaining across the whole area average based on years of real world costs. I use much more in tyres alone!

Motors, 18 months? They say that they have almost never had a motor fault or replaced one. I was calling them every 18 months.

This is about 7 years worth... And there's still a few more in the garage. Brakes worn till they no longer hold, gearboxes with tons of backlash, noisy motors, burned brushes / commutators, cush drive rubbers failed, etc. If you USE a chair hard, they fail...

[Vitolds]

[Burgerman]

Nice. Caused by a short circuit? Or overloaded? Is that a 60A unit?

[Vitolds]

this 2360
I do not know what happened. cruel experiments

[Burgerman]

http://www.roboteq.com/index.php/robote ... 360-detail

Small electric vehicles, electric bikes sugested use.

60 peak, 40A continuous. And 2.5mOhm all sounds a bit too weak for tracks... Maybe it just didnt like it.

And no script? Wont steer properly. Use this:

http://www.wheelchairdriver.com/BM3-con ... Script.txt

[Vitolds]

http://www.roboteq.com/index.php/robote ... 360-detail

controller HBL2360
I had this script with my changes.
he It burned in the engines 2х2000 watts

[Burgerman]

The motor watts do not tell us anything useful.

We need:
What is the motor impedance?
free running Amps?
Stall Amps?
(at your chosen battery voltage) to determine motor power.

But the controller is very likely far too small.

Max Amps per Channel 75?
Continuous Amps per Channel 50? Large motors/skid steer? Not enough Amps.

If motor impedance is low, and it probably is, on larger or more powerful (2000w???) motors, you will hit the MAX amps those MOSFETS can handle (50 to 60 Amps) at just a few % pulse width. A very small stick movement. The roboteq will rapidly correct this and try and reduce power output but I think that it will/may be the cause of your problem. Roboteq warns of this:

Important Warning
Do not connect a motor that is rated at a higher current than the controller.

https://www.roboteq.com/index.php/docma ... anual/file page 29...

[LROBBINS]

John and Will,

I'm attaching scripts for both of you to try that include a "Damping" user setting. It does a moving average of the joystick readings as I described earlier. If set to 1, the stick is read only once, so it's like your current scripts. I have set it at 10, which is what I'm using on the attendant joystick on Rachi's chair, but you may want a lower or higher number. Maximum is 20 (that could be increased, but I doubt if it will be needed). As usual, change the filename extension from zip to mbs.

There have been other changes made in my version of the analog script and there are some differences in the Will and John versions. See the comment for 2015_06_28 at the end of the files.

There is one thing that both of you will have to edit before using them:

Instead of using a timer routine, the script now just reads the brake pin flag bits to avoid MotorCompensation starting the chair moving if it's jiggled while stopped. BUT, I don't know what pins you are using for the brakes, so I just put in absurd numbers so I could compile the script. You must change those to the numbers of the pins you are using.

The biggest change, other than damping, is the way speed pot user settings are specified. Instead of percentage scaling, and percentages of pot effect, we now set min and max speed and turn values for forward and reverse. There are four settings for forward -- SpeedPotForwardMax, SpeedPotForwardMin, TurnPotForwardMax, TurnPotForwardMin -- and four settings for reverse (0 to 100 for each). Otherwise, the changes just shorten the code a bit and make it more readable.

Now, a warning. Given that my Roboteq and CANbus nodes are installed in Rachi's chair I have no way to test these scripts. So try them with some caution, and if there's a problem revert to the ones you are using now. I don't intend to buy another Roboteq, so at most I'll have only the motor simulation subroutine available if debugging is needed.

Ciao,
Lenny

[Burgerman]

OK thanks!

Will need to be out of this chair and into a different one and on blocks to test. Will take a look now.

Only using 1 brake pin here? Problem? Should I set both the same?

I am using: Dout 1, to operate both brakes via the relay.
So, do I add: 1, in both fields? Or the pin number on the connector? This pin thing always confuses me! Should be input/output number instead I think.

EDIT. I could add another brake output on the next free Dout if needed and just not connect anything.

[LROBBINS]

Here are version b of these scripts that can be set to use 1 or 2 brake pins. You need to set Brake1Pin to whichever DOUT is used (same "meaning" as for the other pin definitions - the DOUT numbers, not the DB physical pins). If using only one brake pin, set Brake2Pin to 0.

Ciao,
Lenny

[Burgerman]

Thanks!

Presumably "print" at the to should be commented out?

[LROBBINS]

No, you needn't comment out that print line. It's there so that if you ever have doubt about which script is installed in the controller you can just start it from the Roborun console tab and it will tell you (as long as you remember to update that line every time the script is changed). Because it runs only once when the controller starts up, it doesn't affect loop time so doesn't affect responsiveness. Ciao, Lenny

P.S. Microbasic doesn't have it, but in C there's a macro that actually reads the program name and download date from the processor itself so that this info automatically gets updated.

[Burgerman]

OK.

[Williamclark77]

Thanks. I'll give it a try once my new motors arrive. It will probably be a few weeks. They said mid July. I hope sooner. I need back in it.

I have no brakes. Will that make any difference? I haven't had a chance to actually look over the script yet to see.

[Burgerman]

You can just set up a digital out in roboteq configuration as if to work a brake, to release brakes on both motors. You dont have to actually connect any wires though. If need be.

[Williamclark77]

Thanks. Simple enough if needed.

[LROBBINS]

John's right - you can just set some pin as a brake pin and set the profile to make it active after BrakeDelay time when motors are not moving. Alternatively, I could put back code that uses a timer instead of reading the brake pin, but I'd rather not have to maintain multiple different scripts.

[RichAC]

While looking at the PowerLab8 I ran across this. It is a dual PowerLab. It appears that you can get an expansion channel which is about $100 cheaper than buying two units. Also if you use one 240VAC power supply at 40VDC/3400W you don't have to worry about floating grounds. You could also use two power supplies. Let me know if I am reading this right. I hope this is the right thread to post this.

Configuration D
Summary: two Dual PowerLab channels charging separate packs, one power source, Expansion Channel Mode.
Components:
 1 Dual PowerLab 8x2
 1 or 2 power source(s) (one shown)
 2 standard adapters or Safe Parallel Adapters (1 adapter per pack) or MPAs
 2 or more packs (Safe Parallel Adapters and MPA require same cell count for all packs)
 1 optional FUIM3 for computer support
 1 optional “Y” servo cable (required for Expansion Channel operation with computer support)
 Male-male servo cable (required for Expansion Channel operation with or without computer support) 
Dual PowerLab 8x2 mode: Expansion Channel.
Dual PowerLab 8x2 maximum charge power: up to 1344W to each pack (8s) using 40VDC/3400W power source. Standard house circuit would likely not provide enough power; would require 240VAC power supply for full power. Optionally, power each channel using 40VDC/1700W power sources on 2 x 20A breakers.


http://www.google.com/url?sa=t&rct=j&q= ... ipDoZLYJkQ

[Burgerman]

Its basically 2 PL8 chargers in one box.

You still need to keep 2 floating power supplies to charge more than 8 cells, in the same way as before, and it lost the ability to charge and balance at up to 3A via the balance tabs alone. And if one charger fails, you are left with no option but to replace the complete thing. So on balance its better to use 2x PL8 chargers.

[Burgerman]

Lenny, do you think that the script we are using will run on the small controller here:

http://www.roboteq.com/index.php/robote ... 130-detail

Getting a robotic mower but they dont do edges.
Thinking of building a small grass cutting bot for the edges... DIY

[LROBBINS]

I think it could be made to work, but some changes would be needed because it has fewer I/O pins. There are only 4 analog pins and you are using 5 - you'll have to leave off the speed pot, or use estimated motor current instead of sensors (maybe an OK option for a lawnmower). There are only 2 digital outputs, you are using more and some would need to be eliminated. Your latching contactor needs 2 all by itself, so you'd have to get rid of your indicator light(s). But I think it can be done without too much effort; the basic algorithms would stay the same.
Ciao,
Lenny

[LROBBINS]

Just noticed that the data sheet defines acceleration on this controller rather differently - as 100 to 65000 milliseconds to full power, as though larger numbers mean less acceleration. No, just looked at the HDC2450 datasheet and it says the same thing, but that's obviously not correct. The user manual defines it as 100 to 65000 power unit change per second with 1000 power units = 100% PWM (or for brushless, 100 to 65000 RPM change per second). Milliseconds to full power would be 10000msec (10 seconds) at accel = 100, to 15 msec at accel = 65000. My guess is that the datasheet writer had a minor mental pause.
Ciao,
Lenny

[Burgerman]

RC Overide only, otherwise I am just going to let it drive slowly on its own and run into walls/trees/fences and things. Then current rises, which triggers reverse in roboteqs setup. Repeat over and over. May add an analog input for .5 sec occasionally with a timer chip to make it change direction randomly...

No current sensors, no contactor, no anything much!

Ok will order one. This will only be 300mm x 300mm and low... With a spinning sharp blade to do the edges behind a strong thin metal sheild.

Built onto this: http://www.ebay.co.uk/itm/281470497176
with one extra vertical brushless hobby motor, with http://i.ebayimg.com/00/s/ODAwWDgwMA==/ ... T/$_35.JPG
and larger wheels...

[LROBBINS]

Don't need to use an analog input to have a timer - timer functions are built into MicroBasic. Doesn't sound like you'd want much of the present script at all. No contactor - do you intend to just let it burn up if something goes wrong? You could use a 40A automotive relay as contactor, an ordinary toggle to turn it on, and one dig output to at least shut it off with MOSFET fail. No motor compensation? Then let it do "normal" Roboteq tank steer and it will pseudo randomly change direction every time it hits a bump. Pretty much no script at all: a variable to set speed if you wish, a timer to reverse direction, no inputs except RC shut off, MOSFET fail if you add a relay (and regen diode and pre-charge resistor).
Ciao,
Lenny

[Burgerman]

I need the script because the roboteq mixing makes it unusable by RC!

With it, it steers controllably. Without it it is anyones guess what its going to hit. You really should try it!

Whats worse, almost all these robots that people build are the same and they think its normal. Not only roboteq.

[Irving]

Lenny, do you think that the script we are using will run on the small controller here:

http://www.roboteq.com/index.php/robote ... 130-detail

Getting a robotic mower but they dont do edges.
Thinking of building a small grass cutting bot for the edges... DIY

Was thinking of getting one. My daughter has one, but doesn't mow right to the edges as you say cos the wheels fall over the edge unless the flower bed is raised, and of course you cant get up to a wall, needed a strimmer for that. Maybe a design based on an auto-feed strimmer head slung out the front of a three wheeled, rear-steered base?

[Burgerman]

I am doing this.

4 wheeled base.
A lightwight strimmer blade (not string) out ahead of my RC /autonomous robot driven by a brushless hobby king lowish rpm motor. But with a metal sheild stopping it hitting a fence. Like a protected propellor... All small. Like a shoe box. Stacked with cheap lipo batteries and RC overide so I can do it myself if needed.

[Burgerman]

http://www.ardumower.de/index.php/en/

The shop sells everything. But I cant build one because I have no idea what/how all that arduino stuff is/works/programs etc...



DIY here http://www.mowbot.nl/Mowing/New_UDO_5/E ... udo_5.html and this does edge to edge.

[Burgerman]

Code: Select all

    ILeft = GetValue (_AIC, LeftCurrentSensorPin) 'discard first read to avoid cross talk between analog pins
    ILeft = GetValue (_AIC, LeftCurrentSensorPin) - ILeftAdj
    IRight = GetValue (_AIC, RightCurrentSensorPin) 'discard first read to avoid cross talk between analog pins
    IRight = GetValue (_AIC, RightCurrentSensorPin)-IRightAdj

Do the space differences matter here?

[Irving]

http://www.ardumower.de/index.php/en/

The shop sells everything. But I cant build one because I have no idea what/how all that arduino stuff is/works/programs etc...



DIY here http://www.mowbot.nl/Mowing/New_UDO_5/E ... udo_5.html and this does edge to edge.

Ardumower looks straightforward enough, most plastic parts seem 3D printed and it's pretty much plug n play if you don't want to build the electronics yourself. Software shouldn't be a stumbling block, its readily downloadable and no more complex to upload to the mower than what you've already done with Roboteq scripts etc., just needs the right uploading tool. I'm sure Lenny or I could help you with that. Since when has a little lack of knowledge phased you