Roboclaw 2X15 getting too hot in PID mode

Questions about using encoders with the Roboclaw product line
George
Posts: 12
Joined: Fri Jul 01, 2016 6:05 pm
Roboclaw 2X15 getting too hot in PID mode

Post by George »

Hi,

I am using the RoboClaw model 2X15 to control speed of DC motor rated at 24V and 1/3 hp (250Watts or about 10A full load current). The motor is rotating a metal ring that has significant angular momentum but still at full speed of around 7000rpm motor is using below 10A (it is actually just around 8A). For this reason I limit M1 Max Current to 15A to protect the controller and be safe.
I am also using the Teensy micro-controller with the Arduino library in packet serial mode to communicate with the Roboclaw to issue commands. When I just change the PWM settings using drive forward m1 command and ramp pwm from about 10% to 90 % ( 1% increase every 2 seconds, so it is fairly slow) and decrease from 90% to 10% at the same speed and do that over and over again controller temperature is at nice and safe 33-35 degC.
On the other hand when I used the micro controller PID algorithm (command 38) to control motor speed the controller quickly starts to get hot (over 60 deg C). It seems to me that controller is actively braking while it is trying to control the speed of the motor. How can this be avoided if I am correct?

I also noticed when using the IonStudio in "General Settings" and setting the pwm at 100 % that motor will run nice and fairly cool (35 deg C) but if I press stop all to stop the motor temperature very quickly will jump over 70 deg C. Is this safe on the controller? Do I risk damaging it?

I am mostly concerned about the controller reliability and I event tried bridging the channels to increase current carrying capacity since we are dealing with only one motor but it seems that the eeprom settings that configure it in that mode can get erased/compromised to easily and go back to default mode and then risk of damaging the controller by having bridged channels is even higher.

Thanks,
George
User avatar
Basicmicro Support
Posts: 1594
Joined: Thu Feb 26, 2015 9:45 pm
Re: Roboclaw 2X15 getting too hot in PID mode

Post by Basicmicro Support »

Your PID tunign is probably too aggressive based on your description. Though it could be your motor is too large for the 15a. When the PID is tuned too hot/aggressive it ends up fighting itself. Reduce your PID setting by equal ratios(eg reduce each setting by 10%) and retest. Do it again until you are no longer overheating. If your control becomes too sloppy at the point it isnt overheating you probably need a more powerfull controller.

Note that a 250watt motor rating(at 24v) means the motor can pull 11amps continuously without overheating the motor. It does NOT mean the motor will not pull more than 11amps. In fact the motor will proabbly pull between 5 to 10 times that amperage when stalled. If the PID is too aggressive it's nearly as bad as if you stalled the motor at full power. But in general PID will run a motor harder than PWM since it is constanly compensating for errors in the feedback(eg accelerating and decelerating the motor) which means it will pull more current then simply running the motor with a constant power(eg like your PWM only tests).
George
Posts: 12
Joined: Fri Jul 01, 2016 6:05 pm
Re: Roboclaw 2X15 getting too hot in PID mode

Post by George »

Thanks for your feedback and I tried the following:

1. I ordered and received your 2X30A controller from Pololu robotics
2. Configured it the same way as 2X15A contoller (default settings, except for single channel mode); saved settings
3. Tested basic functionality in Ion Studio and it worked well.
4. Performed auto tuning (default medium level) . Got the P= 5057, I= 150.8, QPPS = 850,000
5. Since you suggested less aggressive tuning I used P = 100, I = 1.0 in the Arduino control program
6. I used the same motor and decoder as last time and powered everything from 12V 7AHr batteries in series to give me about 25.5V. I also maked sure to use 1K to slowly charge caps on the controller when connecting it intially.
7. On the Arduino controller I set up a program using the roboclaw.SpeedAccelM1 command to step from 50,000 PPs to 750,000 PPS in increments of 50,000 PPS (with 30 seconds at each step) and then back down to 50,000, and stop, 5 minute break and then again ramp up and down.
8. It took about 5-7 minutes to finish the first cycle and was monitoring roboclaw temperate and RPM in ION Studio (arduino was issuing commands via serial port and roboclaw was also connected to the computer USM port. I was keeping a close eye on in these first 7 minutes and PPS were right on target, roboclaw temperature went from around 29 to 34 deg C. Motor was audibly running very smoothly and it look really good. Then the motor came to a full stop and waited for 5 minutes
9. At this point I decided to let it run on its own few more cycles to just make sure it is all good.
10. I left my testing area for a few minutes and then came back and motor was running and I focused my attention to another task on a different computer.
11. The sometimes after I heard a popping sound (not very loud at all) I directed my attention to the controller and saw that it was starting to smoke and catch on fire. I was very lucky to be there when it happened to put the fire out. It was quite bad and the controller completely melted in 2-3 seconds.
Please let me know if you have any idea of what might have happened here. I am not sure what to do at this point. I invested quite a bit of time and money into your controllers and I really like the features but I am not sure for how to proceed ahead. I would really be glad to hear if I made any obvious mistakes in my setup that you can point to. Please let me know if you have any detailed questions before you cam make some suggestions.

Thanks
User avatar
Basicmicro Support
Posts: 1594
Joined: Thu Feb 26, 2015 9:45 pm
Re: Roboclaw 2X15 getting too hot in PID mode

Post by Basicmicro Support »

I recommend you send a motor and the two controllers in to us to test/repair. We will repair the first one and replace the second one and then test the boards with the motor to determine the cause of the problems. My suspicion is low inductance motors.

Based on the high speed of the motor it probably has a very low inductance. This causes excessive current ripple in the electrolytic caps on the driver. This can cause one or more of those caps to fail. They make a popingf sound if they fail. They also tend to cause a short when they fail hence the fire. An electrolytic cap failing matches your description pretty well. Of course I cant be sure until I can go over the board.

IMPORTANT: You should always have a properly sized fuse for standalone applications. You are dealing with potentially alot of power(dead short on a 24v battery system) so a fuse is absolutely required in that type of setup. Someone running a remote controlled robot can get away with out a fuse because they will always be near the robot but not in a walk away application. ALso you need to make sure the fuse will actual blow if the power supply shorts. You need to look at the fuse timing table which tells how long it takes for the fuse to open at a given current. You want a fuse that will blow in .1 to 1 second at around 90% of the maximum battery current(or lower) but that wont blow at the maximum application current.

Ship it all to our contact address and include your return address and contact information in a note inside the package.
George
Posts: 12
Joined: Fri Jul 01, 2016 6:05 pm
Re: Roboclaw 2X15 getting too hot in PID mode

Post by George »

Hi,

I did have a 20A fuse in the circuit which blew at some point but I should have sized it more conservatively. I will ship you all the equipment you requested today. Thanks a lot for being willing to take a look at this.
User avatar
Basicmicro Support
Posts: 1594
Joined: Thu Feb 26, 2015 9:45 pm
Re: Roboclaw 2X15 getting too hot in PID mode

Post by Basicmicro Support »

The external amp rating of the fuse is usually much lower than the point it would blow at(eg a 20a fuse is garunteed not to blow at 20a). For example a 20a fuse may not blow until 50amps and may takes 10s of seconds to go even then.

There should be a timing table available for any fuse you buy. You do want a fast fuse.

Standard automotive fuses may be slower than you want.
George
Posts: 12
Joined: Fri Jul 01, 2016 6:05 pm
Re: Roboclaw 2X15 getting too hot in PID mode

Post by George »

Hi,

Yes I was using automotive type fuse but I will do some research on faster acting fuse with appropriate rating.
Do you at all recommend any type of breakers or are they inherently too slow.

I also shipped all the requested supplies and you will hopefully receive the box on Friday. I also included the prepaid FedEx shipping label to easily ship things back to me.

Tahnks
User avatar
Basicmicro Support
Posts: 1594
Joined: Thu Feb 26, 2015 9:45 pm
Re: Roboclaw 2X15 getting too hot in PID mode

Post by Basicmicro Support »

The fuse/breaker is just to protect the external world from a damaged motor controller/motor(eg fire). It isnt going to prevent the motor controller from being damaged. As long as the breaker is fast enough to prevent extended periods of shorted current(eg fire) it should be fine. But you do need fuses/breakers that will work under the given load/power source(eg no open when they shouldnt and open wihen they should).

The safest option is to use a fuse/breaker that will open as just above the maximum current you will draw from the power supply but that can be hard to determine. The current the motors pull are usually alot higher than the power supply will see so I usually recommend basing the rating on the battery/power supply maximum current - some derating factor(10% or more).
User avatar
Basicmicro Support
Posts: 1594
Joined: Thu Feb 26, 2015 9:45 pm
Re: Roboclaw 2X15 getting too hot in PID mode

Post by Basicmicro Support »

I got your boards and motor. I cleaned the burned board. The damage was not in the power circuit or the DC Link caps at all. Those look nearly pristine and the heat sink has no burn marks where the mosfets are located. The damage is in the power plane directly under the screw terminals upto where the DC Link caps leads come through the board(which could indicate a short through that section of the power plane happened but didnt have any direct connection to the DC link caps. The ground plane doesnt look bad though so doesnt look like the same power went through it(hense my suspiscion it was a short through only the power plane). One mosfet driver chip was burned but none of the associated mosfets where damaged. Driver chip failures happen either because a mosfet fails(not indicated in this case) or an over voltage happens(eg above 50v spikes) so still no definitive cause for this boards failure yet though it doesnt appear to have anything to do directly with the power circuit/h-bridge though one mosfet driver chip may have been involved(the burn damage was such the driver chip is missing on one channel).

One the two other boards you sent back, one was never opened(a 2x30A board) and passed all the basic tests. I will use it to test your motor to see if I can reproduce the problem with your motor.

The second unit(a 2x15A board) one mosfet is damaged(can see the burn mark on the heatsink). That could have been caused by a damaged driver chip though I wont know until I replace the damage mosfet. If the driver chip is also dead it could have been the driver that killed the mosfet or vis-versa.

I dont see anything in your motor itself that would cause this failure. Its resistance measurement was 1ohm using a milliohm meter and its inductance was 275uH which is pretty low but not so low I'd consider it a problem(we have a minimum 150uH requirement on this board).

What was the length and gauge of your power wires?. My guess at this point is either something shorted at the screw terminal across the power plane. The amount of power required to delaminate several square inches of PCB copper is more than a damaged mosfet driver could source through its package and the mosfets themselves appear to be undamaged.
User avatar
Basicmicro Support
Posts: 1594
Joined: Thu Feb 26, 2015 9:45 pm
Re: Roboclaw 2X15 getting too hot in PID mode

Post by Basicmicro Support »

I've run some tests with the motor using velocity control now. Can you describe what the target application is? Based on what I am seeing for normal speed control applications(accelerating to speed crisply and decelerating crisply) a 30A controller is not sufficient. Possibly a 2x30 in bridged mode but since you are worried about the bridge mode getting reset(which can happen if you reset the settings from the buttons on startup, reset them using a packet serial command or update the firmware) then I would recommend moving up to a 60A controller(probably the Solo 60 since you only need one motor channel).

When I was testing any speed changes would drive the control well past the 30a continuous current limit. For example changing speed from 10% to 30% would take a second or more and peg my power supply at maximum current during that period. I'm limited to around 40amps to the motor with my current test setup so to do anything beyond that I'll have to setup my battery setup but it definitely shows the motor wants significantly more power than a 2x30a would comfortably run at if you will be accelerating and decelerating a lot.

If you application has very slow accel AND decel then it shouldnt be a problem but as you found if you slam on the brakes(eg stop the motors fast) all that power has to go somewhere hence the large heat spike.

Post Reply