Hardware Peripheral Input/Output

Switching Losses in Chopper type Stepper Motor Drivers

Chopper drivers control the coil power by rapidly turning a drive transistor on and off. This allows for much higher power drives, but comes with it's own set of costs: In addition to the standard chopper squeal, whine or hiss which is not only annoying, but can actually cause midrange resonance issues and missed steps at specific speeds, it also introduces switching losses into the motor coil which causes the motor (instead of the driver) to heat up. As heat and magnatism don't mix well, this can lead to unexpected loss of power when the system is under load or has been operating long enough for heat to build up.

Comparison of Linear vs Chopper drives

posix says: [ed: The Linistepper is a linear stepper motor driver, which does not chop power to the motor coil]

[The Linisteppers] only deficiency, it seems, is the transistor heat. I have a chopper here and a linistepper and they both run at 24v. Linistepper "sounds" sweeter and it seems it runs faster as well before stalling. This is strange as I always thought a chopper would go faster and stall higher up in the rev range.

...

I don't know what I like better, burning my finger on the driver board or burning my finger on the motor. Which do you prefer? P.S. it seems that once I switched to a chopper driver all that heat that used to be on linisteppers just transferred to motors themselves!

...

[ed, at this point, we find out that there was NO heatsink on the linisteppers power transisters]
Since I burned my fingers on a motor that was driven by a chopper (and that motor was actually HOTTER than linisteppers driving other motors) I have decided to stick to linisteppers. It's easier to worry about heat in one place than 3 separate locations. Went down to my local PC shop and they'll have some 2nd hand pentium II coolers with fan in by monday.

Measuring Switching Losses

Mariss says:

Switching losses shouldn't be ignored. [For example, ] We have good in-house lab documentation on IRF540N MOSFETs because, well, they are the ones we use: The major switching loss source is intrinsic diode reverse recovery current. We use an average current of 20A for 100nS; the current waveform (measured with a Tektronix TCP202 current probe) is roughly triangular with a peak current of 50A. 20A times 80V is 1.6 kW of MOSFET dissipation for 100nS.

At a 20kHz switching frequency (50uS period), the loss amounts to 3.2W per H-bridge.

If we used only the I squared R value, the loss per H-bridge would only 3.1W typical at 7A and 25C. Add switching losses and the total becomes 6.3W per bridge at 7A and 80V.

There is a way of measuring total losses (switching + conductance) in drive without using expensive lab equipment. All you need is a good multimeter, a power supply and a pair of 1mH ferrite-core chokes.

1) Substitute the chokes for a motor; the chokes must be ferrite-core and rated for the current you intend to use. Mine look like something a mad bomber would use; 7 150uH 10A ferrite-core chokes wired in series and taped together with electrical tape.

2) Set the multimeter to DC Amps and put in series from the power supply to the drive.

3) Set the drive to zero Amps per phase and run it at the equivalent of about 1 rev per second. Write down the current the meter indicates. That will be the drive's quiescent current.

4) Set the drive to maximum Amps per phase. Measure the current again and subtract the quiescent current reading from the new measurement. Multiply the result by your power supply voltage to get your drive's losses (switching and conductance) in Watts. I'll guarantee it will be significantly higher than the calculated I squared R loss.

Reasoning: The zero Amp quiescent current is what it takes to light-up the drive. The max Amp reading times voltage (Watts) is entirely dissipated in the drive. The chokes won't get the least bit warm so you know every Watt goes up as heat in the drive. Subtracting the quiescent current cancels what gets lost in the control circuitry. You then left with an accurate measurement of the true H-bridge losses.

Mariss