0

I want to build robot with 2 DC motors to move like a tank. So I need 2 H-bridges (one per motor) to be able reverse the rotation.

I used https://www.pololu.com/product/1117 motors rated for stall 0.8A at 6V, and rated to 3-12V. I also used some H-bridges rated 2A+ and 7.6V LiOn battery, but eventually I burned 3 H-bridges :(

On the last I was able to make some measurment and it showed -30..30V spikes (on DSO138 osciloscope) and sometime about 1.6A (at common mutlimeter - slow).

So I decided to build more robust H-bridge (to not burn at all) and make measurement on that. something like BJH H-bridge going in reverse but not forward, LL transistor overheating with transistors rated 10A, 100V ( TIP142, TIP147 ).

But I found in datasheet, that for Resistive load Turn-on time is 0.9us while Turn-off time is 4us - this lead me to conclusion, that when swithing eg. left side top/down it would shortcut the rails for 4 - 0.9 = 3.1us

Is that right conclusion? Is that a real problem? And if it can be corrected/used, what is good PWM speed to drive it effectively?

Thank for your help

Community
  • 1
gilhad
  • 657
  • 1
  • 5
  • 15
  • Make sure you have free-wheeling diodes in place, and consider introducing a dead time period in your control signals. – JimmyB Feb 27 '17 at 12:29
  • I have 4 free-wheling diodes there (both sides of motor one to +, other to - in closed direction to drain over voltage) and small capacitors in paralel (to hit before the diodes start to open). I am not sure, what is the best way to introduce dead time period - I considered to manipulate all 4 transistors with AVR (over optocouplers) directly and put it in the program somehow, but it can be error prone, when interrupts hits (like from communication, measurements and so) – gilhad Feb 27 '17 at 12:32
  • 1
    You must have a dead time between the two states! – winny Feb 27 '17 at 12:35

2 Answers2

3

Is that right conclusion? Is that a real problem? And if it can be corrected/used, what is good PWM speed to drive it effectively?

Yes this can be a real problem and one solution is this: -

enter image description here

The RC keeps OUT1 and OUT2 from being on at the same time. Make RC equal to a few microseconds to avoid the "clash". See also this stack exchange Q and A.

Community
  • 1
Andy aka
  • 434,556
  • 28
  • 351
  • 777
  • please note (for OP) : this is a common hardware deadtime generation scheme BUT you must ensure your modulation depth is such that a minimum pulsewidth is still created. Failure todo so can & does result in failed freewheel diodes due to recovery time clashes –  Feb 27 '17 at 13:36
  • @JonRB Thanks, I will take it to calculations too (Also it is part of "what is good PWM speed to drive it effectively" part of my question) – gilhad Feb 27 '17 at 13:55
  • @andy-aka: Thanks, It is probabely what I need. I think about adding another AND gate on each OUTx to be able to close both transistors (for free-wheeling) - or would be better use 5. Darlngton on Power Line (as Enable)? (But this would take another voltage from the motor) – gilhad Feb 27 '17 at 14:02
0

Is that a real problem?

You may want to consider dead time management. Or shoot through is possible.

dannyf
  • 4,272
  • 1
  • 7
  • 9
  • You've been downvoted. This answer doesn't provide any useful information, it just introduces some technical terms that the OP probably isn't familiar with. I've noticed that, recently, you posted a lot of answers to many questions; most of them have very few votes or none. That should tell you that you need to put more effort in your answers, otherwise it's just noise.. and people will just continue to downvote your answers or at best ignore them. The quality of your answers is, IMHO, way more important than the quantity.. – m.Alin Feb 27 '17 at 14:56