brushless dc motor drive from the mains / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

brushless dc motor drive from the mains

previous 1 2 3 next

Move Thread

LAN_403

radiotech

Mon Sept 26 2011, 10:17AM

Registered Member #2463 Joined: Wed Nov 11 2009, 03:49AM
Location:
Posts: 1546

With three phase circuits involving motors there are several issues. The concept of impedance matching
takes on new meaning with a new lingo. Power factor is considered referenced to the entire network
of things connected.KVARs are moved from motor to the network and change the voltage at places where
it may be too low by reducing current in the conductors while maintaining KWs.

The other is the effect of unbalance in phase currents causing circulating currents at multiples of
line frequency. These harmonics represent themselves as torques of opposite direction on the motor
shaft. This result is I^R losses in rotor bars and windings. Added to harmonics produced sub-cycle
modification of current by switching regulators tied to speed control loops, the conventional design
of motors just wont work well. The big energy users, pumps for example, whose optimal speed follows
a flow loop, benefit by continuous modification of power factor.

1) The switch will apply 170 volts to the motor and it will have to shut off once the measured current reaches 15 amps. This huge current fluctuation will back feed all kinds of noise into the mains without proper filtering, and provide me with only a small portion of the motor's rated power.

The problem with this is a motor turns at speed that determines power output delivering specific torque.
Motors have speed torque curves when connected to constant voltage as they ramp up. Loads have
inertial characteristics that can be handled by tuning PID loop programs found in cheap off-the shelf
controllers. This will work, in a manner, using motors not originally intended for variable speed duty.

One of the most on-the-point sources of information on motors used in industry is the
Electrical Apparatus magazine, a Banks publication.

Where the electrical codes are involved (public utilities are not bound by these codes in most jurisdictions)
since they are under national rather than local laws), with power factor correction of motors, the placement
of capacitors , i.e. at the motor itself, determines current in feeder conductors which dictates what size may
be used.

Dr. Slack

Mon Sept 26 2011, 12:56PM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

3l3ctrici7y wrote ...

If we assume that the RMS current from the mains is <= 15 amps, can we use an inductor rated for 15 amps and expect it to stay out of saturation? That is; we have a very tall peak, but the total integrated area under the curve is equivalent to a 15 amp resistive load? (not to suggest that I expect to look resistive, but rather conceptually for purposes of sizing the inductor).

No. Peak current is what saturates an inductor. The peak that matters is the peak *after* you have added the inductor to the circuit, not before.

What I suggest is that you simulate with a suitable circuit simulator (see the HVwiki for a few suggestions if you don't have one already) the current flow into a rectifier/capacitor that's having your target DC current drawn from it. Increase the value of the series inductor and watch the peak current fall. Choose a suitable value of inductance and peak current, and then build or buy an inductor accordingly. But how to choose a suitable value?

You might think you want to go for minimum peak current, but that is no good as a criterion, as it reduces assymptotically as the inductor becomes infinite. You might choose peak to be 2x your mean current, or perhaps input RMS current to be 50% above your load current. Basically there is no optimum, the inductor needs to store more and more energy as it becomes larger, for smaller and smaller improvements in the peak current. What most people do here is one of two things - a) look for a "knee" in the curve of desired outcome versus cost, and choose to operate just below there or b) find the largest core they've got in their junk box and wind some wire on that. If winding your own remember, an inductor needs an airgap.

3l3ctrici7y

Tue Sept 27 2011, 03:26AM

Registered Member #1806 Joined: Sun Nov 09 2008, 04:58AM
Location: USA
Posts: 136

Ah. Now that you mention it, that makes sense. It is the current thru the inductor that leads to saturation, and is therefore whatever the current thru the inductor ends up being.

Thank you.

3l3ctrici7y

Tue Sept 27 2011, 05:45AM

Registered Member #1806 Joined: Sun Nov 09 2008, 04:58AM
Location: USA
Posts: 136

The motor is a permanent magnet unit, intended for battery powered operation over a wide rpm range. I apologize for not mentioning that earlier.

I was thinking hard about this situation, and...
It occurred to me that if I use a series inductor in each of the three phases, that it may function as the impedance matching I need to fix the power factor problem. I would then omit the input capacitor and simply run HF PWM into the inductors. It would still have the 120Hz torque ripple, and it would still have a PF of < 1, but I don't see this producing those large input current spikes that the input capacitor would. Also, the high switching frequency would allow for the usage of small ferrite inductors, and would help to produce something actually resembling a sinewave going into the motor.

The switching frequency would remain very high but I could control the speed of change of pulse width to achieve the output frequency I need for the desired motor speed.

What I'm picturing is that the added inductance will give the overall system an inductance suitable for operation at 170 volts at whatever the design frequency happens to be, and would also prevent the motor from seeing 170 volts which its insulation is certainly not rated for. The design frequency would be based on the usual trade offs between; cost, component size, power losses, and so forth. Most obvious is that the higher the frequency the smaller the inductors, but also more losses in the fets due to greater per-cent of time in the ohmic region.

Is this good stuff, or am I way off?
Thank you.

Dr. Slack

Tue Sept 27 2011, 06:11AM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

I think you're way off.

If you are happy with 120Hz torque ripple on the motor shaft, then you can get theoretically perfect power factor, with no inductors (or at least PFC -specific inductors). Run three half-bridges, with only input ripple filtering rather than a "big C" input supply resevoir.

Input PFC, output speed and torque, with instantaneous input and output powers equal (as there is no DC link capacitor to act as a resevoir) is too many constraints for the degrees of freedom you have available with PWMs. So the one you can't control is the instantaneous output torque - hence the output torque ripple.

What you could do is use the output speed error to give you a figure for the output power required, then draw this input power in a sinuosoidal fashion, in sync with the input voltage, to get unity power factor.

This sort of thing is what happens in "Matrix drives", where you have a three phase supply, a 9 switch crosspoint, and three phases out to the motor. However, you would be doing it with only one input phase.

As you are talking about turning switches on until the current reaches whatever limit, I fear you are thinking at too low a level, even to understand how a simpler DClink driven motor PWM works. Add to that the input PFC, and you'd have to think very hard about how to do it.

Don't add extra inductance to your motor phases. If you can control them with a PWM, then it will be easier with no added external components.

3l3ctrici7y

Tue Sept 27 2011, 07:35AM

Registered Member #1806 Joined: Sun Nov 09 2008, 04:58AM
Location: USA
Posts: 136

OK, I think I get it.
Adjust the output pwm as necessary to achieve the desired motor output, while watching the input current and adjust the output as necessary to shape the input current waveform as a sine wave in phase with the voltage. Adjusting bridge output waveform to keep input current in phase with the voltage takes precedence over adjusting bridge output to control motor output parameters. ??

Thanks.

Experimentonomen

Tue Sept 27 2011, 06:22PM

Registered Member #941 Joined: Sun Aug 05 2007, 10:09AM
Location: in a swedish junk pile
Posts: 497

This sounds like a brushless dc motor, google on brushless dc motor drives.

I'm developing a 3 phase motor drive that will have thge capability of running pretty much any 3 phase motor hooked to it, beeing a small ipod harddrive motor or a 1kW industrial induction motor ort anything in between.

It will likely be microprocessor controlled when im done.

Bored Chemist

Tue Sept 27 2011, 06:29PM

Registered Member #193 Joined: Fri Feb 17 2006, 07:04AM
Location: sheffield
Posts: 1022

Could someone please correct the title of this thread. If you impedance matched something to the mains at the supply frequency you would blow the fuse.

3l3ctrici7y

Wed Sept 28 2011, 03:46AM

Registered Member #1806 Joined: Sun Nov 09 2008, 04:58AM
Location: USA
Posts: 136

I found a high level write up where a group of people did almost exactly what I was picturing doing. Well, they did it correctly and did not abuse the mains supply.

They also link to an interesting pdf entitled; "A robust digital PFC control method suitable for low-cost microcontroller". Unfortunately the file has been removed. I found a copy in the Internet archive, which I have attached to this post for the convenience of anyone interested. ]01_03.pdf[/file] (602KB)

This is clearly going to require more R/D resources than I have available to me at the moment. Oh well. When I get some more time, I will probably try the technique that Dr. Slack described in post 125007.

Thank you so very much to everyone that considered and responded to my post :)

Dr. Slack

Wed Sept 28 2011, 07:03AM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

Oh well. When I get some more time, I will probably try the technique that Dr. Slack described in post 125007.

Whoa, slow down a minute, do remember that all the suggestions I make are contingent, rather than absolute. In other words, *if* you want to do such and such a thing *then* this will acheive it. That last post of mine didn't make any value judgements on what you were trying to do, merely answered the theoretical question of is it possible.

As it happens, I think that intentional torque ripple on a motor is a bad idea, the idea of unity PFC is not necessarily to be promoted to number one consideration (at least for projects of this small size), and attempting to do it in one stage of PWM will lead you into a very difficult control loop design.

The idea of having 2 stages, a DC power supply and a motor inverter stage, has a lot going for it, the only drawbacks being total efficiency, weight, size, stored energy, elegance and cost (but those are the only ones). Everybody else does it that way, so it must have something going for it.

For you, the main advantage is simplicity/debugging, and allowing incremental development. You can bodge one up with recitifiers and a capacitor to get you going, then improve the bad PF to reasonable with an inductor, then improve it to unity if you want with a controlled boost converter - there are plenty of commercial controllers out there as it is becoming a more and more used function. Note that once you are using a boost converter, then controlling your input current waveform to be more of a square wave will allow more power through a thermal breaker than would unity PF, but the utility company would frown on the third harmonic so introduced. But for a few kW, would they even notice?(so thats *if* you want to maximise power drawn *and* don't mind pi$$ing the power utility *then* draw squarewave current into your DC power supply).

For everybody else, it's the absence of output torque ripple.

I'll wager that a PWM to drive the motor is somewhat easier to design if fed with reasonably constant DC rather than half-wave rectified mains.

previous 1 2 3 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob