100 kRPM Brushless Motor Design 4

[sdk_imported]

Hello,

I would like to see some discussion about high-speed brushless motor & controller design(60-120kRPM, 1-5kW). Our biggest challenge is preventing rotor heating and maintaining stable speed control. We currently are using two-pole segmented magnets, segmented magnet sleeves, high resistivity materials, high speed switching, and external inductors ... but, we are still having issues. Will any of the following help me?

Slotless configuration
Non-PWM based speed control
Laminated rotor yoke
Something else I'm not thinking of..

If anyone has successfully designed and tested a motor in this speed and power class I would very interested in hearing from you. Finally, do you know any companies that make motors like this in large quantities?

Thanks.
Scott

 

[Skogsgurra]

cbarn,

We are talking about sine filters, not du/dt filters. And I would not use electrolythic capacitors in a sine filter. Would you?

 

[cbarn24050]

I thought we were talking about an output filter for a pwm inverter.

 

[Skogsgurra]

Yes cbarn,

We are talking about the filter that sits between inverter output and motor. There are several classes of filters;

1 Motor reactors that are used mostly to avoid overvoltage due to reflected waves and also limit the charge dump that the transistors have to carry when switching on and long cables.

2 du/dt filters that serve the same pupose, only a little bit more efficiently. They are usually composed of a reactor and some capacitors. Some manufacturers depend on the cable capacitance and connect low ohm resistors parallel to the reactors to avoid ringing, which would otherwise heat the reactor cores. Other manufacturers have very elaborate du/dt filters with reactors, capacitors and feed-back diodes connected to the DC link.

3 Sine filters. They produce a pure sine wave for the motor and removes - as you say - the carrier frequency and also the EMI (mostly diverting it to PE, which then gets polluted).

4 There are also common mode filters that are ferrite or amourphous magnetic material or compacted iron powder cores (toroids) that are used to reduce the common mode voltage at the motor. The common mode voltage is the cause of bearing currents, so the CM filter is mostly used to reduce that kind of problem in VFDs.

5 Some manufacturers also use small ferrite cores to try and avoid EMI, to no avail (in my opinion).

All of these filters are output filters for PWM inverters. So we do not talk about mains filters or PFC. And since the motor voltage is always AC (except for standstill) you cannot use electrolyts. And even if you use them (putting diodes parallel to them and connecting two of these combinations in series) they would not work very well because of rather high ESR and internal inductance.

 

[cbarn24050]

It seems we are talking about the same filter then. Electrolitics work fine (I'm not sure why you think we need diodes) as long as carrier frequencies don't go much above 15KHz ESR is not too much of a problem. Ripple current is much more troublesome but you can get around that by paralling smaller caps instead of using big ones. Of course you still may need to add small film caps for EMI suppresion if that is a problem.

 

[Skogsgurra]

Cbarn,

That is very interesting. Do you connect the capacitors from DC link to output (after the chokes)? Can you really (really) have a pure sinewave with small components? Do you have any references where I can see for myself?

 

[cbarn24050]

Yes, they go between the choke output and the dc link, use 2 caps 1 to each side of the link for balance. These filters were used years ago before mosfets and igbts were around on inverters with thyristor output stages ( switching at around 1-2KHz). The motor current wave is quite good as the motor itself is a filter as well, you could allways cascade another stage if you wanted to be picky but it's unlikely to be needed.

 

[Skogsgurra]

OK cbarn,

I remember them. But I do also remember that the switching frequency with thyristors seldom was as high as 1 - 2 kHz. Those frequencies are IGBT territory. Thyristors were not able to switch more than once (or a few times) per output cycle and the resulting current was not very similar to a sinewave - and the voltage was, of course, even worse.

The filter I mentioned is available from one of the dominant European inverter manufacturers and it is built into a housing that adds to the weight, but I doubt that removing the housing would reduce weight or space requirements very much.

If we go back to the original question, it is obvious that much research is going on in the field of high-speed drives and it is also obvious that filters have been tried in many configurations and that they do not deliver an acceptable set of performance parameters. The problem is not only cost, weight and space, but also waveform, losses and EMI as well as getting good dynamics out of the drive.

 

[cbarn24050]

I'm afraid your very wrong there, 1000amp thyristors with 20uS commutation times have been around for a least 20 years (they would run at about 5KHz). Even today the thyristor is untouchable for the big inverters.

 

[Skogsgurra]

I appreciate this discussion. It brings out many aspects that I havent even touched for years and certainly not in conjunction with high-speed drives in the 1 - 10 kW range.

Yes, there are so-called "inverter grade" thyristors with "short" commutating times. But to my knowledge, which may be limited, they are nowadays only used in CSI inverters for low speed high power drives. My first inverter (around 1970) used thyristors with commutating circuits with an LC combination and an extra thyristor. This technology was then replaced by GTOs.

I would be very grateful if you could point me to any URL where thyristors are used for high-speed drives (30 000 RPM and above) with good sine voltage output since that would be a great contribution to the high-speed technology. I have searched the web for this, but have not found anything. Have you?

 

[cbarn24050]

I don't recall saying thyristors were suitable for this design.

 

[Skogsgurra]

Well, but the subject for this thread makes one temted to think that you were talking about high speed drives. Thanks for the views, anyhow.

To round off: 1 -10 kW 30 000 up RPM drives do need sine wave drives and PM synchronous motors. Especially if efficiency, space, cost and weight are important.

 

[avigodner]

If I got it correctly soft switching was mentioned in the context of clean sinusoidal output voltage. I think soft switching itself rather reduces semiconductor switch losses and has nothing to do with voltage harmonic contents.

Soft switching does allow for higher switching frequency (cleaner high frequency voltage sine waves). Is there any additional motivation for switching frequency increase like PWM loss reduction?

 

[Highspeed]

100kRPM=1.66kHz,
A clean sine wave needs at least 32 steps driving to PWM minimum time = 1/53kHz = 18.75µs this will also allow a reduced size filter to avoid bearings currents
based on a 8 bits resolution, smallest PWM element = 73ns
it will be tough to get that with "soft" switching specially if using a full bridge MOSFET configuration where some dead time is required to avoid cross conduction.

Another tip that we used in our design is to have a "dynamic" balancing :
1st step : Each rotor is balanced (mass & excentricity) on a reference stator.
2nd step : when rotor in the final device, PWM generation is tuned during final test to compensate windings unbalance
HS

 

[sdk_imported]

avigodner,

I've done some back-to-back testing at 19kHz and 39kHz soft-switching. Higher switching frequency will reduce rotor temperature. The testing I performed was at 30 kRPM, 1kW, running at partial duty cycle condition, 4-pole motor, with a trap drive. The magnet temperature reduction was less than 10degC by going to the higher switching frequency. I'm sure that this finding is highly dependent on motor and operation variables. Adding external inductors did far more to reduce the rotor temperature than did increasing the switching speed.

sdk

 

[jbartos]

Suggestion to skogsgurra (Electrical) Dec 10, 2003 ///\\
We are talking about the filter that sits between inverter output and motor. There are several classes of filters;

1 Motor reactors that are used mostly to avoid overvoltage due to reflected waves and also limit the charge dump that the transistors have to carry when switching on and long cables.

2 du/dt filters that serve the same pupose, only a little bit more efficiently. They are usually composed of a reactor and some capacitors. Some manufacturers depend on the cable capacitance and connect low ohm resistors parallel to the reactors to avoid ringing, which would otherwise heat the reactor cores. Other manufacturers have very elaborate du/dt filters with reactors, capacitors and feed-back diodes connected to the DC link.

3 Sine filters. They produce a pure sine wave for the motor and removes - as you say - the carrier frequency and also the EMI (mostly diverting it to PE, which then gets polluted).

4 There are also common mode filters that are ferrite or amourphous magnetic material or compacted iron powder cores (toroids) that are used to reduce the common mode voltage at the motor. The common mode voltage is the cause of bearing currents, so the CM filter is mostly used to reduce that kind of problem in VFDs.

5 Some manufacturers also use small ferrite cores to try and avoid EMI, to no avail (in my opinion).
///It appears that the above filter classes tend to overlap in the operating frequency range. For example, if a filter is designed to attenuate frequency above certain cutoff, then the higher frequencies do not have to be filtered by another class of filter for the higher frequency since those frequencies are supposed to be attenuated by that low pass filter. Ideally, a notch filter would pass 50Hz or 60Hz only, leading to a pure sinusoid. It would suffice for all unwanted frequencies attenuation.\\

 

[Skogsgurra]

Yes jb,

Many things "appear" to you. And a lot of your "suggestions" are difficult to understand for us technical people that live in the real world and have first-hand experiences.

Has it never "appeared" to you that real world components have parasitic properties? Like an iron core having hysteris and eddy currents that can be represented by a parallel resistor and that windings have parasitic capacitance that also are in parallel to the winding.

These parasitic properties act as parallel roads for the high frequency signal components and need to be reduced by dedicated HF filters. So even if the different filters overlap in theory, they do not in practice.

 

[Highspeed]

jb,
any reference or drawing of a notch filter that is efficient for EMI for the kind of load we are talking about?
I'd like to test that
HS

 

[jbartos]

Eng-tip to the previous posting: The notch filter is meant to be for the 50Hz or 60Hz fundamental frequency pass not for EMI only. I have never posted such a statement.
Please, exercise care what you are writing about.

 

[jbartos]

Eng-tip to Highspeed (Electrical) Dec 13, 2003 marked ///\\jb,
any reference or drawing of a notch filter that is efficient for EMI for the kind of load we are talking about?
I'd like to test that
HS
///Attached is a low pass filter with a cut-off 10xfundamental frequency. If there are not subharmonics, then the low pass filter is adequate. This should filter out EMI on the output too. However, the technial details may be proprietary.
Visit

http://transcoil.com/documents/I_O_M_Manual.pdf

for:
Figure 3: Wiring diagram

Visit

http://www.transcoil.com/#klc

for:
Sine Wave Filter
KMG High Performance Output Filter

The KMG filter is designed to attenuate the carrier components present in the output waveform of typical PWM output power supplies. The general filter topology is an L-C-R low pass circuit. The circuit input is a three phase reactor of sufficient impedance to control the capacitor charging below the inverter peak current fault point. The filter cutoff frequency is set approximately ten times the max allowed fundamental frequency of the inverter to minimize the fundamental KVAR absorbed by the filter while attenuating the carrier components at the rate of 40db per decade. This allows carriers greater than 2KHz to effectively be eliminated from the output of the filter. The purpose of the damping resistor is to control the over voltage excursion at the cutoff frequency to a reasonable level and keep the peak capacitor currents within design limits. An added benefit of the low pass configuration is the capacitive reactance at the load will provide motor power factor improvement thereby improving the overall filter insertion loss to that of the resistor and inductor thermal levels, typically about 2-3% of the inverter full load level.\\\

 

[jbartos]

Eng-tips to skogsgurra (Electrical) Dec 13, 2003 marked ///\\Yes jb,
Many things "appear" to you.
///Yes, especially, when presented popularly without references or more rigorous way. What is wrong with that?\\ And a lot of your "suggestions" are difficult to understand for us technical people that live in the real world and have first-hand experiences.
///I do post many references, links and analytics. I do not see that in your postings.\\Has it never "appeared" to you that real world components have parasitic properties? Like an iron core having hysteris and eddy currents that can be represented by a parallel resistor and that windings have parasitic capacitance that also are in parallel to the winding.
///Yes. However, they tend to be neglected in many engineering and conceptual phases. Often, a detail design phase consider them. Are you by any chance a designer?\\These parasitic properties act as parallel roads for the high frequency signal components and need to be reduced by dedicated HF filters. So even if the different filters overlap in theory, they do not in practice.
///Not quite true. Visit

http://www.transcoil.com/#klc

where KMG filter has the same feature as KLC among other features, i.e. KMG is somewhat overlapping the KLC functionality.\\