0
\$\begingroup\$

I am trying to run a BLDC motor made for model airplanes by locking it to a 3-phase rotational field. Setup:

  • motor spec'd as 10V, 1700 V/RPM, max current 9.5A, without a significant load attached, screwed down to a piece of hard wood

  • driver IC PCB widely available on ebay, using the ST L298N variant, i.e. 2 H-bridges supporting permanent 2A per bridge - since you can control the halves of the bridges, I actually (ab)use the PCB as 3 half-bridges, keeping the 4th part unused. What is peculiar is that it has 2 electrolytics, but not the 100nF decoupling on the Vs and Vss pins as prescribed by the datasheet.

  • microcontroller PCB with 3 PWM outputs of 1 timer peripheral. Currently using a 20 kHz PWM - to avoid the hellish noise of the apparently more standard 3 kHz used among RC plane enthusiasts. It may have more losses, but as I do not intend to use this for model vehicles, but rather stationary, reducing acoustical noise is more important.

    • the modulation of the pulse width is done using s sine table, with PWM channel0 having 0, ch1 120, and ch2 240 degree phase offset.
    • the frequency of table sweeps is adjustable (roughly) from 0.1 Hz to 200 Hz, by using a potentiometer attached to the MCU PCB for experimenting
    • the maximum PWM value is determined by a sedond potentiometer, to hand-adjust / determine the power needed at a certain rotation frequency

The funny thing now is: This runs as expacted at e.g. 1 Hz rotation frequency, 10V, adjusted PWm to drawing 700..900mA for that frequency, for maybe a minute or two, then the previously smoothly running motor begins to "stutter" every couple seconds to every other second. If I switch everything off and try again 20 minutes later, the same thing happens.

I wondered whether heat could be a problem, either directly, so I blew air on the heatsink that's mounted to the IC, or indirectly - raising the logical level that I maybe was operating at the edge with my 3V MCU, so I built 3 discrete NPN inverters to bridge my 3V MCU and the 5V logic, but that didn't help.

Resetting just the MCU does not help, and I don't see any difference at the beginning vs. when the malfunctioning starts in the PWM signals on the oscilloscope.

When the maltunction starts, the lab PSU does show current spikes which appear to be in sync with the metallic sounding stuttering noises of the motor, I detect no mechanical problem per se in that motor, though. The heatsink does get warmer than I'd like to touch for longer than the blink of an eye.

I don't have thermal imaging equipment, but get my lips down to ~ 5mm distance to the motor coils, and as far as I can tell, they don't get particularly warm.

Any ideas what may be the cause?

\$\endgroup\$
  • 1
    \$\begingroup\$ Guess: When your NCO phase wraps around, you set it to zero. \$\endgroup\$ – peter Dec 18 '16 at 20:13
  • 1
    \$\begingroup\$ Nope, it's a uint32 which overflows on purpose. And if it was such a simple software bug, resetting the MCU should help immediately - it does not. But pressing reset for 1 minute and then releasing lets it run smoothly again for a few seconds. Switching it all off for more minutes lets it run for longer afterwards. It all screams temperature, I'm not sure how exactly yet, though. I have some suspicions, but that's only crazy ideas some rookie electronics hobbyist could come up with ;-) I will try within the next 1..2 weeks or so and post updates if I have anything interesting to tell... \$\endgroup\$ – user1847129 Dec 19 '16 at 20:33
  • \$\begingroup\$ Somebody told me he thinks the heat sink on the IC is far too little for passive cooling, I wouldn't know how to determine what's enough, but it does get rather warm. The seller of the PCB quoted the IC specs with regards to supported current, probably not the one for the whole product... Chinese electronic product with grossly undersized parts is not unheard of exactly. \$\endgroup\$ – user1847129 Dec 19 '16 at 23:18
1
\$\begingroup\$

The IC used on the PCB I have is often quoted as providing 2A ber bridge. According to experiments, in actuality, it may be more like 2x 0.8A, before the thing goes into thermal shutdown, as I already suspected.

Details here: http://www.maelabs.ucsd.edu/mae156alib/electronics/The-Motor-Driver-Myth.pdf

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.