3
\$\begingroup\$

enter image description hereenter image description hereenter image description hereThat was a pdf when originally uploaded, but am now going to try to zoom in on the sub-circuits. Please let me know if there's a better way to do this. enter image description here enter image description hereI'm trying to make a PIC-based BLDC motor driver but am drawing ~0.5A even with no motor plugged in. I can move a motor, variable speeds, clockwise, counter clockwise, have validated my drive tables, but am not doing it efficiently...like at all.

I suspect that I'm permitting some current to shoot through a transistor pair. MOSFETs heat up crazy fast when I try to PWM at frequencies greater than 1kHz (I'd ideally want this to drive at about 30kHz). A lot of time on the o-scope has shown me that the gate voltages cross over (high side turns off and low side turns on) for about 1us and then for about 2us when high-side turns on and low side turns off.

Because all the MOSFETs are N-Channel, I have to use a gate driver (LM5109BMA) to boost voltage and be able to turn on hi-side MOSFETs. Unfortunately for this to work I need to keep on charging a boot-strap cap which means I have to do that business where the two MOSFETs of a pair toggle opposite one another.

I guess my question is, is that 1us to 2us of crossover really that bad? Is there any way to practically mitigate that?

see also: AVR443 and AN957 application notes from Atmel and Microchip

\$\endgroup\$
  • 1
    \$\begingroup\$ Which PIC are you using? Some of them have "dead time control" built into the PWM hardware to address precisely this issue. But if you're using six separate output compare modules, you should be able to achieve the same thing by tweaking the pulse widths in your firmware. \$\endgroup\$ – Dave Tweed Dec 22 '12 at 2:26
  • \$\begingroup\$ As ever, a circuit diagram is about essential. If the circuit is 101% standard then show a 101% std cct diagram (not just say so). If non std, show diagram. As ever, a LINK to the ICs used helpeth heaps. Does 2uS crossover matter?. Who can say? What voltage. what rise times, what ... ? WHY see also xxx app note? There are enough different "wholly std" ones that we cannot tell with certitude which wholly std problems you have. Richman's advice is probably good - but does it apply well to you? Circuit please... \$\endgroup\$ – Russell McMahon Dec 22 '12 at 20:19
  • \$\begingroup\$ I'm using a PIC24FJ256GB110-family PIC, and haven't found anything in terms of dead time control. I am now only using three output compare modules and just dynamically assigning which three control pins those output on. \$\endgroup\$ – tarabyte Jan 3 '13 at 17:21
  • \$\begingroup\$ In response to the the second comment, I'm pretty sure Richman's comment was produced by a random sentence generator (and is deleted now). Attaching schematic now... \$\endgroup\$ – tarabyte Jan 3 '13 at 17:44
  • \$\begingroup\$ schematic is unreadable. Please fix it so we can help. \$\endgroup\$ – akohlsmith Jan 4 '13 at 1:09
0
\$\begingroup\$

Yes, it turned out that that crossover was significant. I have now manually tuned things so there is no overlap, but now too much dead time is preventing me from going to higher frequencies.

Please feel free to comment on the circuit diagram if there are things that you see I might be able to improve.

\$\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.