Old discussion on SED, tracking original post

I found that google groups can in fact filter on author, but did not find the original post anyway.

Might be hard, since for some reason on google groups, all the posts in that thread is from 0900 at the same day.

There were no 'Google Groups' in '98.

Don't know what news handler employed '34BB81EA...' type message identification.

The mirrors that lifted threads from usenet are not always consistent with date and time stamps.

For loss in silicon steel, check with steel vendors.

Am 25.06.23 um 18:24 schrieb legg:

J.Woodgate is often on the LTspice list.

Cheers Gerhard

Google bought the Deja News archive.

I saw an article by him recently, in EE Times or some such.

The post you're referring to had no info on hysteritic core losses in stepper motors. Mr Woodgate's interests have been in the audio field.

Comments about pre-biasing speakers reflect physical practicality. Large subwoofers that are excursion-limited, when pointed at the ground or at the sky can be prebiased in order to compensate for the static displacement of the cone/ diaphram due to gravity.

Hysterisis loss in that application are a small fraction of copper loss - and they don't tend to use the same type of steel or magnetic structure as in steppers.

RL

I would expect them to be small compared to copper loss, and energy that creates the mechanical output.

Are you full stepping, half, or microstepping? DC or PWM?

Microstepping, 256 steps. Just measured the hysteresis loop today. Quite bad, but most motor laminates are lossy. I have contacted the manufactor to get the data for the laminate and the turns quantity, then it's simple to check with a calculation.

It's also possible to measure the loss with an impedance analyzer, but that process is hard.

You can apply sin/cos weighted high-frequency PWM drives to the stepper windings, and let them fiter it. Or use some external inductors so the motor lams don't see the PWM and have corresponding losses.

When the motor is not moving, one can reduce the drives and save some power, with maybe a tiny angular error at the transition.

Yes, I know it was for another application. But the general idea, how to measure the properties is the same.

Yes, I have seen that done on GaN motor drives, but in that case it was done to reduce the dV/dt that would otherwise damage the insulation of the enameled wire

In this application, there is a linear thread attached, so during standstill the motor is powerless

The frequency's got to be pretty low, though; isn't it just the step rate? Otherwise it's DC.

RL

The frequency is 25kHz. The stepper driver drives the current into the coil as a hysteretic buck converter, where the current demand signal to the buck converter is set by the sine lookup table incremented by the Step input.

A plot is seen here:

If the pwm is continuous, then hysterisis loss will be hard to calculate, as it varies the farther you are out from 0 and is non-loop (ie a spiral) so long as the average is changing.

Under steady state, around zero, loss is related to the area of the loop. Published Steinmetz coefficients only apply in that region.

RL

Yes, it will vary some depending on how far it is driving up on the BH curve.

Measurement of the curve is seen here, compared to a ferrite drum core inductor.

If you say so.

At best, this comparison shows the limitations in your present measurement and data processing technique.

There are two orders of magnitude difference in the units employed for 'comparison'.

RL

Yeah, crappy 12bit picoscope. In fact, I have pretty expensive Siglent and Tek scopes, but the Picoscope SW has good math functions, so I find myself using that for more complex work. If I had time, I should build a real integrator.

Yes. Two different components. 6mH winding vs 100uH drum.

The real comparison is looking at the hysteresis area, much lower for the ferrite, which is should be, since ferrite is way better than laminated iron.

Even so, ferrite will exhibit a loop, not something that crosses over itself a couple of times in the sweep.

At 20Vpk, reversing at 20KHz, I don't see a 6mH choke exhibiting much loss, no matter what the material. With steppers, it's usually the copper that dominates, because it costs more by weight/volume.

Sinusoidal exitation is only intended to reduce mechanical noise - HF pwm to reduce losses in generation of the sinusoidal form or moderate the static torque / power loss.

RL

Yes, I think you are right. I will try to get the data, to calculate the loss.

Also intended to provide high step reolution..