Tutorial Diagnostic and hardware help

I spent some time working on my set up tonight, after migrating to hall effect sensors and I'm still having some issues.

My motors are Jazzy style wheel chair motors. The brand seems to be Quantum Rehab. They have a part number of 64-006-002. There is also a Pride Part No. of DRVASMB7110010. They have a continuous power of 350 watts (24v x 14.583A, right?) and a peak power of 1,250 watts (24V x 52.083A). They actually list a voltage of 22.5V

My control box uses as many as 4 cheap 24v 15A power supplies. I've spent most of my testing before tonight using only 2 power supplies, all of which are bussed together in parallel. Tonight at the end of my testing session I bumped it up to 3 supplies and seemed to have better results. I was having other electrical issues in the house while also working on these so I'm monitoring that as well.

I'm also using the IBT-2/BTS7960 motor drivers, an Arduino Uno, and the latest SMC3 firmware. I havent done much in the way of tuning my PIDs but my motors will just sometimes shut down, unsure if this is due to too many amps being pulled through my drivers or due to momentum extending the motor arms beyond the digital limits as set by SMC3.

I decided to grab another Arduino and use it more or less as a datalogger, and I wired two of its analog read pins to the current sensing outputs from one of my two motor drivers (initially set it up to read two analog signals, one for each driver, then realized that each driver has two current sense pins). Based on probing with a multimeter and reading the BTS7960 datasheet, it seems that the pin should output about 0.117V per amp, or roughly 5 volts for 42.5 amps, right near its 43 amp limit. While running sine wave or motion, it wasnt uncommon for the current to spike and hit max current, but only for an instant. I know when I sold RC cars for a living, our speed controls were rated in continuous current, with a peak current typically higher than that, same with our lipo batteries. For the price, I gotta believe that the IBT-2 actually has an undisclosed continuous current and a max peak of 43A. I plan on integrating this current sensing or maybe even shunt resistor current sensing into the finished control box, outputting to an LCD or a 7 segment LCD set.

I've got a bit of a deadline I'd like to meet and some folks had mentioned other motor drivers. I feel like lots of folks have great luck with the BTS7960s but I'm guessing they're maybe pushing less aggressive motors? And so, I'm left wondering whether I should keep trying these, try to find a reliable source on some good ones or try and find a different, more reliable driver altogether.

Excellent, thanks for quick response, lemme see what I can find real quick.

This is the power supply I have. I spent some time when I first put everything in the box trying to get them all as close to 24V as possible. I have a power strip in my control box, and each PSU has its own power cable and so I can plug in and out whichever units I want or as many as I want for testing.

AVAWO® DC 24V15A 360W Switching Power Supply Transformer Regulated for LED Strip Light, CCTV, Radio, Computer Project etc.



I will swap hardware side to side tomorrow if I can, as well as extend functionality of my makeshift amp meter to show for both sides. I have I think one or two spare units as well, as I planned on this being a 2DOF + Traction Loss rig.

At least once tonight during my testing, the output of the one motor I was pulling current info on pegged at 5V, which I dont think is actually a constant 43A but instead a diagnostic indicator. I'll see if I can find the datasheet I was reading, and see if it has a truth table. I know only one of the two sides of the h-bridge was pegged, and it was the side that is tasked with pushing my seat "up", ie, never has the assistance of gravity pulling anything down.

At least once tonight as well, I had an instance wherein that same motor was extended beyond my preferred range, surely beyond my software limits.

With respect to SMC3, its a bit late at night right now for me to power things up and not wake roommates but I can show you my PIDs and all other settings. Ignore the buzziness of the green line, my pots arent actually hooked up right now and I assume thats just the ambient ADC noise without any pull ups or pull downs.

I can say with decent certainty that its not a USB disconnect, I would hear it and I dont hear any notifications. Usually, power cycling the power strip, which only supplies power to the supplies, is enough to get things working again, but not always (like when I think the momentum of my motors swing arm moving upward pushed it beyond the soft limits before the PIDs could reign it back in). It almost always stops working while I'm playing in-game and while sitting on the chair. I can run it in test mode with SMC3 Tools all day long, no issue.

I will run it in sine, square, and motion tomorrow as soon as I can and post the results.

Ok, just got off work, starting to produce images requested.

(Long post incoming, I dont want to get flagged at all for spamming)

Some photos of my rig, its an unashamed near-clone of the original DOF Reality H2/P2 design.







First test of the evening, hooked everything up, ran sine wave test on motor 1, within the second full waveform the thing flatlined.



Note that the COM13 graph below is not scaled horizontally to match the graph in SMC3, and its still a prototype sketch on that arduino, and therefore its accuracy is still questionable. Whats important to note is that the reddish color is the amps drawn while pushing that side of the simulator upwards, and that it pegged the same time the motor stopped responding. Note that SMC3 still has the motor as "on" so it hasnt exceeded any limits that would warrant a shutoff.

SineWave Test with Motor 2.

https://youtube.com/shorts/9V737wDXwzk

Note no real issues. Orange line represents the upwards direction on this bottom graph, green represents the downward direction.

It also seems like I almost always have issues with motor #1's control, I feel like the one I replaced a few weeks ago was also motor #1, and so now I'll go ahead and swap them to see if the problem follows the motor (potential internal short) or if it follows the driver.

Motor 1 test on Motor 2 driver


This wasnt wholly scientific. I swapped motor and pot wires between the two sides and ran the test on Manual mode first just to make sure polarity was fine. Found that when centered, my Motor 1 was a bit high, and so when I started pushing it even higher, it would near this point in the arm rotation where I believe it was reaching an imposed mechanical limit. I recentered the pot and ran the test again, as seen above, with no issue. I will say however that the motor arm will hit the steel structure on the bottom of the travel before hit hits the red Max Limit/Clip Input lines above, which I believe means that the game may try to push the hardware beyond its physical limits. I'm thinking of maybe shortening my connecting rod a little bit and moving my "center" position on the motor a little higher (currently set at 9'oclock, maybe move it to 9:30 or 15 degrees above horizontal). I could also maybe alter the potentiometer code a little bit so as to "expand" the range of it to get it to effectively hit those limits.

But that leaves the question of motor driver 1 on motor driver 2:


It would seem rather foolishly that my issue was related to my centering, as I've been able to run Motor 1 on Motor 2's driver and Motor 2 on Motor 1's driver no issue. I'm gonna revert these conditions, get motors 1 and 2 on their respective drivers.

Square Wave Tests with motors back on correct drivers, motor 1:


Motor 2:



Sim Tools Settings:




However, the problem has not entirely changed IN game. I booted up Dirt Rally and within a couple turns, motor 1 yet again stalled in the upward position, but not as far as it had before. Its motor driver is outputting full 5V on the upward side.

Working on the Sim again today. Great information and I'm taking it all into account.

I started by setting my Kp to 400 and set my PWMRev same as PWM Max.

I went into my garage and grabbed the digital angle finder I use for my table saw blade. According to my current sensor output, "centered" is about 4.3 degrees down. I then zero'd the angle finder to test how far I swing the arm in either direction and yeah, its pretty close to 40 degrees in either direction. It hits roughly 35 degrees on the bottom end before it hits the frame.

Updated Settings, Motor 1:


Initial testing shows that my current seems to be coming down. The red line here is my downward direction, presumably spiking because of the extra PWMRev I dialed in. The Blue line represents pushing the rig up against the force of gravity.

Already things are performing better but I need to do more reading on tuning SimTools still. Unsure when I'll be at a point in SMC3 tools to move on though. Ie, when are things "sorted out"?

Thanks again

Excellent. I used to tune PIDs on drones way back when, but my PID tuning knowledge is a bit sketchy, so I'll start there, I feel like thats where I can find the most room for improvement in matching those sine waves.

Should I be running or begin running these tests with me in the seat? All screenshots from above have more or less been with an empty chair

Also: I have fuel injection line mating my pots to my output shafts, with hose clamps tightening them down. Is this suitable?

So I've spent some time tuning and starting to add my hardware. I am really afraid I'm spinning my wheels and wasting my money on things that arent working.

Center of Gravity
I run this simulator at an event one of my rally team sponsors puts on in celebration of the local rally events. I have always brought a home made steel static rig that has always kinda been alright. I discovered this hobby at the beginning of the pandemic and bought the motors and the power supplies and the motor drivers and got to work, and then the events were cancelled. The event is now coming around, and if I can't make this motion sim work, I have a way to make it static. The event is March 4th, and I'd like to have time before that to tear it all the way apart and paint the chassis. I will have to tear it somewhat apart just to get it there, and so I need a day or two before that event to accommodate that.

However, that brings us around to my two points about CoG. When I first designed it, I was pretty confident I had my CoG figured out, but that was BEFORE I upgraded to a Moza R5 wheel base from my G27. That was BEFORE I lost like 40 pounds off of my small 5'7", 190lb body. And none of that accounts for the fact that I plan on this machine being the centerpiece of an event where people of varying sizes will try and use it.

I'm spending some time learning about where weight is distributed in the body and trying to figure out how much a difference in a persons weight and height will affect the ideal CoG of the machine. I have the seat on sliders as well, I guess if I have a range of like 150-300lbs, I would wanan make sure that the CoG still sorta moves around that area. I thought about making the pedals and steering wheel arms moveable, I still can, but that means for each person making an adjustment and then also, the further those are from the U-joint, the more torque/leverage they induce on that point of movement.

Motor Drivers
I spent some time last night working with the sim in-game, fully loaded, with and without me in the seat. I spent time putting limits on the axes of movement to prevent hitting my chassis with my motor arm, and in that process I realized that even if I dont hit it at max pitch forward, if I also induce roll, that will extend the motor beyond the pitch limit, which totally makes sense. This draws too much current, and I think once current has been drawn to max beyond a small, even clipped spike, the motor driver shuts off.

The motors draw upwards of 1250W, 50-something amps, at burst/peak. I was looking at the Cytron dual Smart Drive 60A, and I am afraid to just throw money at the problem but it would seem that it would handle whatever current my motors wanna draw, no matter what. I just dont know how to begin wiring those up through SMC3, or if I can get it here fast enough to get installed and wired.

Game Choice
I'm a rally driver, I got into sim racing to supplement that, even though I know that Dirt Rally isnt entirely representative of the real rally experience, its still a great tool for practicing hearing pace notes and visualizing roads. Naturally, rally is a rough sport, where you can flip a car pretty easily and its usually flipping a car where the game over extends the motors. Should I be testing and tuning this thing with the most extreme scenario, being a rough rally stage where I'm on and off the road, flipping the car end over end and rolling side to side, or should I maybe try the sim out on Project Cars or Assetto Corsa, something road course based, and then slowly start ramping up to Dirt Rally?

Feedback Pots
I'm running hall effect sensors now and have found they're much quieter than the 50k or 100k pots I was using, but I feel like I still see some jitter. Should I run these with shielded cables? Right now I'm using servo extensions because they had a real easy connector on them, and they are very much not shielded. The run from pots to control box is like 3 or 4 feet right now. I'd have to tear apart an electrical taped wiring harness for each motor to swap them out, but I guess I can do that. If I do that, I would wire a ground to the shield wire in that harness but how would I shield that sensor at the sensor end? Or should I even concern myself with that?

I have all day to spend on this thing today and would like to make what progress I can with it, all things considered.

I did some quick math to figure out if I could bypass my problem with hitting my chassis rail by cutting my tie rods shorter and making my motor arms centered a bit higher, ie, offset the center of the motor arm by x degrees.

I quickly modeled all the important bits of this calculation in a profile view in Sketchup. My motor arm point to point is about 6.25"



I then screencapped this and put it into Paint.Net because I dont know how if at all this simulation can be done in sketchup easily, with moving parts and pivot points, etc. Note, the white square box in the upper left is my U joint

I then moved the motor arm to -15 and -35 degrees, -35 being about where it would hit the chassis arm.


I found that my simulator has about a 5:1 ratio of motor arm movement to seat rail movement. I cant recall what trig I would need to use to express this as a complete formula, but that doesnt matter at the moment.

I then decided to see what would happen if I offset the center position of my motor arm by +10 degrees:

The resulting linkage would be shorter by about 3/4" of an inch. Next, I simulated it again for -35 degrees from center position:

and I found that my seat rail moves just a fraction of a degree more because of this. I have some spare threaded rod and could cut shorter linkages but reducing by that little I would have a lot of thread left in my rod end and would just need to jam nut it, so I think I will do that today as well.

I'll look into doing what I can with sketchup. I havent got my whole thing modeled entirely, and I have whatever free version is available and so I dont know if I have all the features.

When it comes to over engineering, that was sorta my intention with getting these motors. They seem beefier than the DOF Reality H3 motors and beefier than the windshield wiper motors most folks use. I tried to build it with slightly heftier stock than the DOF Reality system as well, however, I am looking at areas I can cut weight when compared to my original design. My wheel and pedal mounting plates are currently wooden and my steering upright arms are aluminum, and I'm sure this has contributed something.

I might try and see if I can find the new CoG of the system with it fully loaded, including me. I might see if I can use a scale and put a 2x4 under the main seat beam, and I would imagine that at such a point the CoG is perfectly centered about the point of movement in my U-Joint, the scale should read 0 pounds, right? No force forward or backwards?

Yeah I cant exactly do that though, my top frame bends down like the DOF Reality H3

Also I was looking into that post about the visual meter and it seems like if my CoG is perfect I should be drawing almost no amps from either motor, which makes sense because theres no force working against the motor, and if thats the case, I already have a visual meter for current being drawn through my second Arduino reading the current sensing output from my motor drivers, but I've never looked at them with the intention to get them reading as low as possible in aims for perfect CG