PINNED - Roboteq Controller - developing for powerchairs

[Burgerman]

This applies to the sensor script in the bot that is working good too?

Because I think thats the one I may end up having to use on the BM3 as well. Because it seems to calculate things very oddly...

[cdb0ewm]

I was looking at configuration gif file you have in the index of BMS construction. When I compared the AIN 7 & 8 values I notice the Input Use value is 0 for one and 16 for the other. Since my understanding these 2 drive left and right direction, can you explain why theY are different and the 16 value is which option.

Thanks

<Node Text="AIn7" Value="Sens-R"><Nodes><Node Text="Conversion Type" Value="2"><Nodes/></Node><Node Text="Input Use" Value="0">
<Node Text="AIn8" Value="Sens-L"><Nodes><Node Text="Conversion Type" Value="2"><Nodes/></Node><Node Text="Input Use" Value="16">

[Burgerman]

I dont know what 0 or whatever refers to. What does that mean on the roboteq settings options themselves?

Just took a look. Next to the box "input use" for what its used for, which is "no action" one motor has a tick - but its disabled... But its unused so not relevant. Its sensor is now only read by the script. Take the tick out... Set to 0 if you want!

[cdb0ewm]

My bad.. its the Roboteq xml file, not gif file. See this link http://www.wheelchairdriver.com/BM3-construction/Roboteq%20BM3%20Profile.xml.
Regardless I just trying to make sure I have the right info for the two left and right motor analog inputs - ANA7 and ANA8

[Burgerman]

The two above were for the sensors.

The motor ones are different. But that settings file is for my joystick and sensors calibrated to my equipment. Dont just copy it. You must calibrate yours. And depending on how you wire it it may be opposite sides/direction too for both sensors and motors and joystick.

[cdb0ewm]

Now I'm really confused. I thought these two inputs were for the x y inputs of the joystick.

[Burgerman]

<Node Text="AIn7" Value="Sens-R"><Nodes><Node Text="Conversion Type" Value="2"><Nodes/></Node><Node Text="Input Use" Value="0">
<Node Text="AIn8" Value="Sens-L"><Nodes><Node Text="Conversion Type" Value="2"><Nodes/></Node><Node Text="Input Use" Value="16">

Sens means sensor

[cdb0ewm]

What sensors are these

[Burgerman]

100A Allegro, bi directional, that measure the motor current. They work to tell the script how much motor load compensation to add.
They work great. Lenny is trying to get the roboteq to do this internally based on battery current. But it behaves very oddly on my bot, and on the BM3 when I tried it the measured current was vastly different side to side. So not sure how well that wil ever work.

http://www.wheelchairdriver.com/BM3-construction/ look at current sensors...

[cdb0ewm]

Yesterday you asked me to remind you re the latest script
Thanks

[Burgerman]

The reason I did was because I was stuck on y bed, and the script is on my PC. And still am. But there may be a backup copy on my laptop. So give me 5 mins!

[cdb0ewm]

ok...sorry bout the bed

[Burgerman]

OK things to note.

This has been tested on my BOT not my chair,but they behave and respond the exact same way. Other than with the waythey use different sensors, voltages, and the bot responds much faster!

But every user setting is set for the BOT. That means voltages, sensors, accelerations, inc all input channels, etc and the turn and speed-pot settings. (The bot has no speed-pot.)
So every one will need adjusting to give you a starting point. Or it will likely go into orbit.

AND

it has had 2 wait states added that I have not had a chance to test yet. To remove a glitch that Lenny found, and that I never saw. Presumably that will not change anything as far as we are concerned.

This is an actual zip, not a rename.

Go through line by line and I will see if I can suggest a sensible setting!




I can give you a few starting point settings, later. You will need RC and your wheels in mid air to set this up. And a lot of time to waste.

[Burgerman]

What voltage?
What motors?

And I can suggest a starting point to specific questions.

[cdb0ewm]

Its for a BM3 clone

[cdb0ewm]

without radio control

[Burgerman]

You need it for testing, and safety in case...

[cdb0ewm]

looks like I have some homework

[Burgerman]

What would happen if you sit in it, and accelerate like this:

http://www.wheelchairdriver.com/BM3-con ... g-test.mp4 Download to desktop first!

At 26 seconds. What happens if when I accelerate theres a bug in the script, a incorrect setting, a fault in a current sensor setting, or something else. And it wont stop! Or turns hard right and into my van? I have had several runaways while testing, configuring etc. I can test without broken legs, by RC. When or if it runs away, or flips on its back, I can turn off via RC by flicking a switch and drive the contactor to cut power instantly. Once confident and thoroughly tested and know everything is correct I get in it!

[Burgerman]

Same as mobot script, but settings changed to give a STARTING POINT for the BM3. Make sure you choose the setting that represents your inputs, sensors etc. And start with wheels off deck. Then test with great care outdoors till motor compensation correct. Start low.

[Burgerman]

I have to mention that Lenny is convinced that we can do this without current sensors in the future. And is trying to get his head around a script bug in a sensor free version at the moment. I am less convinced... We will see!

[robert97]

another power hiccup

The glitch I experience which is intermittent and unpredictable
is a power drop out for a fraction of a second. I'll call it
a hiccup. The hiccup occurs during steady state driving.
Often I'll be going at about one-third speed and then I'll
feel a lurch where Motors stop producing power. At high speed ,
12 miles per hour the hiccup causes the wheels
to lock up and the tires to screech.

I began implementing John's suggestions of July 21st. Foolishly
I applied too many changes at once and ended up with an
undriveable chair. Subsequently I have more methodically applied
changes.

While up on blocks, I captured the following Graphs. As
you can see in both images there is an instance where even
though motor control says apply power, the motors stopped
producing power for a fraction of a second.

i am using Mowbot_2016_Dec_21.mbs with Accel = 700, Decel = 900

[robert97]

another issue.
with script print ("John_2016_11_24 ,
changing TurnBoost = 4, to TurnBoost = 2
helps drivability, but motors stop producing braking when joystick released.
any ideas?

soon ill try it on Mowbot_2016_Dec_21.mbs and try to get a graph.
robert

[Burgerman]

You have a wiring fault. Or a connection fault. Or a configuration fault. And of both of those screenshots there is no script running. So it has nothing to do with script. So I know that the above is correct. And you have serial and analog lit, analog testing, no script, should be with mute ticked. That will cause spikes that are not related to your in use issue.

Send me the settings file again, with the inputs marked as to what you are using them for. EG so I can see what they are like THROT, STEER, SENSOR-L or R, S-POT, etc.

And also are you SURE that you have the current sensors configured in the correct sense? And on the correct wheel? Does motor compensation actually work? When tested on blocks? In use?

That motor on/off will wreck the motors. Even airborn. I had that problem when there was an issue with th script. But the script should be fine. I have tested that version extensively on the bot. In any case those graphs above are with no script running.

[Burgerman]

READ ABOVE CAREFULLY!

And here. Examine CAREFULLY!!!
Note that EVERY INPUT and EVERY CORROSPONDING SHARED pin/output is named so I can see the lights go on/off and what that does. For eg if an input is a shared pin - say a digital in or out, you can see why it illuminates as you add speed with an analog input. You can then go to its settings and asign it some suitable values. Mostly OFF! If its set to read that and so do something (like stop!) then you may get your issues.

So label EVERYTHING you are using. Note the run screen. Everything is either labled or blank white (add a single empty space in its name).

[Burgerman]

And are your sensor, and analog signal cables screened properly? Does the joystick/current sensors have a seperate 5v and 0v supply that nothing else uses? They should.

If tested by RC then it rules out the analog noise or edge bands protection issue. Or if it stops the problem it shows bits IS the issue. And testing by RC is safer. What if instead of stopping motors for an instant it ran away instead? Hope you only test in a very open space!

[Burgerman]

soon ill try it on Mowbot_2016_Dec_21.mbs and try to get a graph.
robert

You should ONLY be using that one!!! The earlier ones had issues... I have a dozen between the one you are using and the one above! We found bugs and fixed them later on. The Mowbot_2016_Dec_21.mbs is the only fully working bug free one.


We thought it was all done. Then lenny started trying to make a version that works without current sensors. After having one fail and a runaway... So we are testing again. But for now use Mowbot_2016_Dec_21.mbs ONLY. Start withmotor compensation set low.

[robert97]

john,
thanks for helping. i tend to agree that the problem is
not script related. i'll add one more reasonable candidate:
something wrong with the roboteq.

maybe it is a wiring or connection fault but the hiccup
frequency doesn't seem greater when i travel on rough ground. ie, vigorous shaking.

i think motor compensation is working because driveability is reasonable
while one wheel is going over a 2" stick.

here is a video on smooth street surface, and one on rough yard surface.

boulderlabs.com/bob/TRACK/speedyC350street.mov
boulderlabs.com/bob/TRACK/speedyC350yard.mov

as soon as i get help, ill annotate the config file and send.
robert

[Burgerman]

I doubt anything is wrong with the roboteq apart from some setup or configuration. Or actual input error/bad connection/noise. You didnt asnwer many of my questions! Details. One detail. May be it. Send your profile file so I can look.

The problems you show on the graphs are seperate, caused by not ticking mute. Its reading both the 0 signal for stop serial input as well as the analog signal for go depending on port/firmware timing, The problem you show in use is different. It may even be the old script you shouldnt be using.

Dont worry it will be something simple. Are you on 45 or 48v? If you dont post back I will presume you are dead now.

[robert97]

i am waiting for a helper to connect to roboteq.
but to answer some questions...
i currently have four lead/acid batteries for
about 50-52 volts. recall i have the 60volt roboteq 2460.
hopefully within 2 weeks i'll be using two 7s5p packs for 46-47 volts.

my cables are not shielded (screened?). a pod/joystick rebuild
with shielded cable has been started. i take 5v from roboteq
for joystick, current sensors, and 10k speedpot. nothing else.
maybe some despiking capacitors are in order.

as of yesterday i use only Mowbot_2016_Dec_21.mbs .
robert