Victor logistics.

[ubermeister]

I am also looking to design a speed controller, and have thought for a while about this problem. The biggest issue is switching polarity based on the length of the pulse. I assume the Victor achieves this with a microcontroller, which also lets it have a deadband and the ability to be calibrated. If you don't want to get into that, you can do the following much simpler design. I haven't tried this, but plan to, and see no reason why it wouldn't work.

Have the PWM signal control the speed, but not polarity, of the output. This just requires some big FETs. Have two separate digital outputs control the polarity via an H-bridge. More info can be found here: http://roko.ca/articles/hbridge.php?page=3

[DonRotolo]

Note that you don't *need* a microcontroller for this - a servo amp, a 555 circult, and other ways exist that are entirely in hardware.

But, let's say you wanted to do it in a uC just for the learning experience.

First, you need a comparator to clean up the incoming PWM signal and slice it at a good decision level, say 2.5 volts (for a 5v PWM signal). Feed the output (TTL level) into an input pin on the uC. Set a timer to start when the pin changes state (low to high, for example) and read the timer when it changes state again. Store that value, it'll translate to a value between 0.9 mS and 2.1 mS (or so).

Meanwhile, reset the timer and set it to trigger at the next pulse.

Then, using one of several methods (I suggest a look-up table), set the on-off ratio of an output pin to a value between maybe 0/5% and 95/100%. Pick a pulse rate between about 500 Hz and 4 kHz, but that can realistically be between 60 Hz and 20 kHz or more.

This output pulse controls an H-Bridge - homebrew with discrete FETs or bought as an H-Bridge module.

When the on-off ratio (duty cycle) is 50%, the motor won't move, since forward voltage and reverse voltage cancel - the motor won't move forward or backwards, the pulses are too short and evenly balanced. At 5% duty, it'll move backwards, and at 95% it'll move forwards.

If that's not enough to get you started, write back.

Don

[yongkimleng]

One of the larger problems would be to actually choose the right MOSFETs and drive them correctly. I did a bit of intensive reading and found two things

1. high side mosfets (near to +ve) and low side mosfets (near to GND). A pair of high side and a pair of low side MOSFETs make up a HBridge.

2. MOSFET drivers - faster u saturate the mosfet, lesser heat generated, lesser power losses

3. proper timing to prevent "strikethroughs" where +ve is shorted to ground during switching.

4. balancing resistors to provide proper distribution when running MOSFETs in parallel

5. dropping in a hall effect sensor for current sensing is good too.

Being unfamiliar with discrete MOSFETs, I did run them off a TLE5205 (5A constant) Hbridge which takes TTL inputs. Next step would be to run some 5205s in parallel - or try discrete MOSFETs.....

Unfortunately TLE5205s are rather hard to find in my area recently. Not sure for you guys.