brushed pwm controller motor "surging"

Dec 31, 2008 8 Replies

i set up a basic lm324 15kz uni-directional pwm motor controller. this goes to a tc4422 mosfet driver and then to 2 irfp4668 mosfets that are in parallel. there is one 10 Ohm gate resistor that both fets share .. selected at random since i have no idea why one would use a powerful driver and then reduce its power.(NOT the question at hand) and a gate diode bypassing the gate resistor to allow fast turn off.



i'm using a second voltage source for the motor/mosfets so i can use more voltage. when both the lm324 and secondary motor/fet loop get the same voltage(+-12v) it works great.



when i give the mosfet/motor loop like 45v and the lm324 12v, as i turn the pwm pot it seems to jump from 0 to 10% instantly (not so at lower voltage) but most importantly, the motor surges between the pwm corresponding to the current pot setting and 100% pwm at a very specific pot setting around



50%. this is not the case when i use 24v for the motor loop and 12v for the pwm generating part.

i dont know why, but im guessing i need a bigger gate resistor. i will try that because i think it solved the problem for me in the past with a tiny motor, but i dont know if its the right solution. heck, i may just need a



100nf cap somewhere(there are none on the power side of it besides the cap required by the driver.. v+ to gnd to help with instantaneous power to gate)

thanks alot!



end goal is 150vdc, 55mph bicycle. everything but the controller is finished.



I played with PWM several years ago and made this page:

formatting link

My knowledge of analog stuff is not great, but I had a lot of fun with Tina making up circuits and getting graphs of the result. I'm not sure what your circuit is like, but it sounds like something Tina could do easily. The fun part is changing the circuit and then instantly seeing a graph of the results.

I think they (designsoftware.com) have a free demo version which might be enough for you. Their website is a mess, though.

As I said, this was years ago, I wonder if there are better and possibly free things out there now.

-Kevin

.

er

de

ore

d
m

Sounds like you've got a spurious oscillation in your control loop.

Schematic? (most of us don't get the alt groups any more, so try to find a website where you can post it.)

Good Luck! Rich

have you looked at the gate signal via a scope?

I know that you say at 24 volts it works how ever, the extra voltage could be showing you an error that you may have due to the extra voltage thus allowing the motor to accelerate better.

It's my guess that the gate's signal is truncated a bit and the LM324 may be current limiting ..

If you don't have a scope to check this, lower the carrier freq by a lot!.

formatting link
"

i put a 220 OHM gate resistor on the mosfets and it works great. gave the lm324 its little 12v source. ran 60v 20amps through the mosfets and the motor responded very well (linearly and smoothly) to the speed adjustment.

thing is, i have no idea why! hahaha

it seems that when mosfets and pwm generation use separate sources with common ground, as the voltage through the mosfets is increased, more resistance at the gate is needed.

LM324 doesn't like capacitors (ex power mosfet gates) connected to its outputs.

yeah, or use a mosfet driver.

sorry, im using a tc4422 driver, so i got 9amps peak to drive fet. im having alot of trouble calculating an estimate for the gate resistor, that 220ohm was a random guess. i found a paper stating Rg=2*sqrrt(Ls/Ciss) - (Rdrv+Rg,i)

for my irfp4668 Rg,i = 1ohm (internal gate R) Ciss = 10,720 pF (input capac) Rdrv= im assuming this is driver resistence.. which is ~3ohms Ls= SHOULD be on the mosfet datasheet, but its not! so im stuck.

is there an easier way to estimate gate resistor?

also, as far as a anti parallel diode across Rg to enhance turn off, i realize the diode is only working when Ig>[(V,diode fwd)/Rg] so i should look for a diode with the lowest "forward voltage" on its datasheet ? or is it "forward recovery voltage" ?

oh.. and im the original poster.. too many google "things"

ok. i just used [5*(mosfet rise+fall times)]/mosfet input capacitence. i will finalize exact value with experimentation.

i hear about "distance"=bad from mosfet driver to mosfet... i realize what this is dependant on everything, and why its bad.. but do you guys think 4" of 12 gauge solid copper wire from tc4422 9amp driver to the mosfet is too much distance?

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required