Universal motor control, best solution?

[NikWing]

Hello all

I came here because of the Rigol DSOs but just looked around the forum some more and found this branch.

I've read so many stuff about speed-controlling a universal motor that I currently don't know what to do.

This is what I want to do: control the rpm of a universal motor, using Arduino.
Parts that I have to use and prerequisites:
- Arduino, hall sensor, potentiometer, universal motor (230VAC, wattage usually around 250W, but might peak around 400W, max rpm: 16000)
- circuit should run on 115V@60Hz and 230V@50Hz
- the motor will be connected to a gear box (1:, then after that the rpm will be measured by a hall sensor board and at least 1 magnet (don't know if 1 or 2 yet).
- the potentiometer will be used as some kind of switch (for example, let's say it's a 360° POT, every 60° will result in a different fix rpm value. so 60° will be 200 rpm, 120° will be 500 rpm etc)
- these values might change in the future
- the motor will be secured by a over-temperature switch which can only be reset by disconnecting the mains (I'll have to check this also using a port of the arduino, but this part is kinda easy to solve)

I found 2 major solutions:

solution A)
setting speed using a triac.
some people say this works, some people say it won't because of the current-voltage phase shift
some say it's enough to measure the 0-crossing of the AC mains using an optocoupler, some say I have to measure the voltage over the triac, because of the phase shift etc ...
for this solution I have to get the mains frequency into the equation and set the timings accordingly, I think

solution B) is what I prefer.
Making a PFC that works from (let's say) 85V to 265V input, so we'll have an intermediate circuit voltage of around 380-400V DC.
The universal motor can run with DC voltage without problems it seems (I hope that's really the case)
So I can control it using a PWM and power MOSFETs or IGBTs or something and an optocoupler to decouple the Arduino from the mains.
(a small PSU using a VIPer16 or something similar could be realized on board, too, to generate the VCC for Arduino)

With this solution I would avoid some problems: circuit can run from 85-265V AC mains, no need to measure the mains frequency, DC voltage with PWM to control the motor's drive.

But this is all I currently have on my hands, along with too many questions.
So let me start this slowly asking this:

- Is solution B indeed the way to go?
- Will the motor have any problems with square-wave DC voltage?
- Do I need to ponder something like a snubber, high initial current and RFI?

Thanks a lot for any reply that helps me starting this project!
BTW, I'm from Germany, so if you don't understand something I wrote at first, please ask me about it

With best regards,
Nik


PS: Additional stuff:
I'm an Arduino beginner, but I guess the sketch would be quite "simple" for solution B.
From my current knowledge I have to read the potentiometer using an ADC port, define a rpm value for the read-out ADC level.
Then I have to measure the rpm value, I guess this is done via interrupt (or I count the hi level on a digital port for 1 second, using millis() and a previously stored millis() value, then calculate the rpm)
after that I risen or lower the PWM value that's fed to the MOSFETs/IGBTs.
But again, I don't know the best way to do that without running into timing problems or cause slow reactions to speed changes ...

[MaxPower]

what kind of motor you're planning to control??? single phase or 3 phase???

[SeanB]

You do not really need PFC with rectifying the mains input, and then just using the DC to control the motor. Depends what regulation figure you want, but generally most of the speed controllers just use a triac triggered after mains zero, and this is used to vary the voltage to the motor. Best speed sensing is done direct on the motor, not on the gearbox output, using either a magnet and a coil to detect each rotation or using a hall sensor giving one pulse per revolution of the motor. Regulation with using straight AC mains is about 5% of full speed, and is usable from about 20% of full speed to 100% of motor speed at rated voltage.

You would probably be better off using one of the dedicated speed controller IC's that are around for universal motors, and use the micro to deliver a desired speed voltage to it and get a sensor reading for speed. Using a DC intermediate rail will also help with improving speed regulation, though any universal motor is not exactly stellar in this area. You will need snubbing on any power switches, and with chopped DC you will need a very high voltage device as well to handle the induced voltages, or you will need to have a LC low pass filter to smooth the chopped waveform into something closer to DC, and this will reduce the otherwise horrible RFI issue you will have.

There are many implementations using triacs and speed feedback, you might want to look at only using 220V as a starting point.

[NikWing]

the motor is single phase.

I'm no engineer but someone asked to find a solution for this problem (that's why it has to work on both, 115V and 230V mains and the hall sensor board is after the gear box, the motor is used to rotate a stirrer)
I saw some circuits and controllers, for example the TDA1085, but I'm not sure how to control it using a µC.
There's just a potentiometer connected and also the tach signal is connected to the TDA, too.
(see attachment for example)

So using the zero-crossing of the mains is enough? I'm puzzled because some people on the www say this doesn't work for inductive loads.

[oldway]

Triac control with the new high reliability 3 quadrants triacs best suited for inductive loads.
http://www.nxp.com/products/thyristors/3_quadrant_triacs_high_commutation/

For control ic's: see U208B, U209B, U210B, U211B, U2008B, U2010B, ...

[temmi_hoo]

Apparently my recent reply got somehow lost into the winds of internet tubes.

Anyway, universal motors are basically DC motors with wound stator magnets instead of permanent magnets for stator.

This means they work nicely from AC or DC and pretty much don't care what you feed them. Controlling the torque and speed are very similar to a normal DC motor.

You can use a triac to control this but the EMC implications will be horrible.

I think you're better off rectifying the input supply to intermediate dc of some hundreds of volts and then just chopping that to a controlled dc current with a low side N channel mosfet or igbt.

Oh btw I repair all kinds of machinery using electric motors for living.

In both cases the feedback current sense and zero crossing detector and gate drive need to be isolated.

[SeanB]

They work fine with mains control, as the inertia of the motor is so high that 20ms pulses are so short that the motor inertia will integrate it, at the expense of a little more audible noise, and needing a pretty good mains filter network on the input to reduce noise they emit onto the supply.

Speed control is simply a voltage fed to C1 from a source, but you just have to note that the reference for this is the one leg of the mains, so you need an isolated solution or float the entire control at mains level. Look at the datasheets for internal block diagrams and how they work.  If you want fixed speeds you can simply use a few optoisolators to switch in resistors to vary the voltage at the set input to implement speed changes. You will have on the live side a 12-15v supply capable of giving about 5mA to drive any circuitry needed.

[Zero999]

How about full wave rectifying the mains and PWMing the motor with a MOSFET or IBGT?

[temmi_hoo]

How about full wave rectifying the mains and PWMing the motor with a MOSFET or IBGT?

This is exactly what I meant but you said it better. I'd definitely do it this way.

[NiHaoMike]

Anyway, universal motors are basically DC motors with wound stator magnets instead of permanent magnets for stator.

This means they work nicely from AC or DC and pretty much don't care what you feed them. Controlling the torque and speed are very similar to a normal DC motor.

You can use a triac to control this but the EMC implications will be horrible.

The brushes are going to be far worse than the triac as far as EMI is concerned. It's also possible to do zero cross switching to avoid EMI, depending on how sensitive the application is to vibration.

[temmi_hoo]

The brushes are going to be far worse than the triac as far as EMI is concerned. It's also possible to do zero cross switching to avoid EMI, depending on how sensitive the application is to vibration.

In the rectifier pwm method you can design a filter for the rectifying part (pfc) and a separate filter for the motor emissions. Depending on the motor pole count the commutation might make a lot or a whole lot of electrical noise. The DC bus must have buffer caps and most likely an inductor as well. I'd put buffer caps near the motor terminals as well and perhaps a hefty schottky diode too.

[MrsR]

Have a look at pcbheaven.com you might find a schematic drawing to do what you want. Last time I was there I saw a PWM controlled fan. Easier to have a base  and change to suit your needs than starting from a fresh sheet.

ALL THE BEST
Rachael

[NikWing]

hey folks,

thanks for replying!

yes, that kinda is what I wrote in solution 2.
at work we manufacture electronic ballasts and I have some schematics for PFC/intermediate circuit etc.
my problem is, I'm no engineer, there's also none at work I could ask
I googled a calculator for power factor pre-regulation(?), on a German website:
http://schmidt-walter.eit.h-da.de/smps/pfc_smps.html

I entered a few values: Ue_min = 90V, Ue_max = 265V, Ua = 380V, Ia = 1.5A (if the motor uses 400 W, it's around 1 A, so I added 0.5A extra)
I guess the load cap should/could be around 450µF, I'm not sure about the switching frequency, I tried either 50 kHz and 100 kHz.
So L would be  236µH or 472µH ...

Alright. Now I guess I need an optocoupler + MOSFET driver connected to the µC port and a fitting MOSFET (or IGBT?) that drives the motor. Once should be sufficient, right? Because it's only around 1A max.
The free wheeling diode over the motor poles, ok - and the caps? directly on the poles?

I'll look at pcbheaven and then I'll probably make a basic schematic which can be expanded later

(I'm writing "I guess" because I'm not 100% sure if it's correct)

[Zero999]

I doubt you need power factor correction.

A simple rectifier will do the job. You may need some EMI filtering, if so just put that on the input of the rectifier.

[Paul Price]

Most of the advice given here is pure nonsense, suggesting complicated circuitry, power factor correction, powering the motor with DC etc. to solve a simple motor control problem that washing machine mfg's have cheaply and simply solved for tens of  years.


Many, many mfg's of Washing machines still use simple, but powerful universal motors to turn the tub and just use simple Triac phase control to control the motor, albeit with feedback from a tach-generator mounted on the rear of the motor to adjust speed according to motor load. It is possible to construct a reliable and practical motor control using abut a dozen cheap and easily found components, and this is including resistors, capacitors, diodes, and a single LM339 comparator and a LM358 dual op-amp(Optional for speed control tach signal conditioning.

Myself, I have hi-jacked the control board out of a discarded washing machine to obtain a Triac already mounted on a suitable heatsink, relays for direction/power control, bias circuitry for Triac triggering, and the powerful motor already included in the package. All I needed to do is to create a simple circuit to opto-isolate the triggering of the Triac and use a few other optoisolators to control the direction relays with the relay control transistors already neatly provided on the control board..

Getting a zero-crossing reference for the triac is easy, just use a small 12.6V AC transformer fed to a diode bridge to give you a unrectified DC output that is fed to one LM339 comparator stage - inputs through a 1K resistor, (connect a LED anode to -input, cath. to ground to limit the positive voltage swing into the - comparator input.).  Bias the + input of the same comparator with a simple voltage divider to set this pin at approx +40mV.  This arrangement will give you a <1mSec positive spike to trigger an interrupt on your Adruino at the power mains zero crossing.

If you need power for your project, add an additional rectifier diode to the output of the 12.6V transformer bridge and connect the cathode of the diode to a electrolytic filter cap so to isolate the zero-cross voltage excursion circutiry, all the while providing a DC power source. In this way you can use the same power transformer/bridge to power the logic of your MCU and optoisolators for control just by placing a 78L05 regulator to this point.

I connect the output of the zero-cross comparator to a 1K pullup resistor to +5V and feed this zero-cross spike to the IRQ input. I also connect a 1-meg resistor from this output back to the +input pin to give some schimdt-trigger noise rejection.

To control the motor, you apply the least amount of power to the motor by phase-triggering close to the end of each of the 50/60Hz power half-cycles and then lower the triggering delay time ( form a little less than 10mSec.. that is  for 50Hz) to increase power to the motor. You can create a software 100uSec/count timer tics to create the Triac triggering delay. This trigger counter is reset by your software each 1/2 cycle.

To isolate the control circuitry from mains power I connect a output pin of the MCU I use to a 817-type 4-pin optocoupler with NPN transistor output.  On the mains side Triac circuit I use an isolated -12V power supply(again, 12.6V 100mA or better transformer, LV bridge rectifier, 470uF/25 electrolytic filter cap) to feed the emitter of the optocoupler and a 330ohm series resistor connected to the gate of the Triac and the +12V supply ground.  NOTE: Triac's triggers best with a negative voltage gate bias for triggering and should supply at least 25mA of trigger current. The trigger pulse need not be longer than a mSec.

Since I used in my project Universal Motors I have salvaged from discarded washing machines, they always have a tach motor mounted on the rear of the motor and this tach motor is electrically isolated from the mains. You can feed the output of this tach to 50 or 60Hz notch filter and feed the ouput to an op-amp with x50 gain stage to get an amplified tach signal to feed to one other of the LM339 comparators and use the output of that comparator to deliver a tach IRQ for speed control. The tac generator gives 8-12 sinewave output signals per revolution(typically.)

Be careful not to attempt to run the motor too slowly at high load as the heating of the armature, stator and motor brushes can cause the motor to overheat and fail.

[oldway]

- circuit should run on 115V@60Hz and 230V@50Hz

Triac solution will need to use an auto-transformer 115/230V.

[Paul Price]

Oldway:  An autotransformer is not required.

A 120V AC motor can be used without a problem at 240VAC with the same phase control circuit. The Triac phase angle of the trigger pulse must be limited to the half-cycle waveform decay area of the 240VAC sinewave so that the voltage applied to the motor does  exceed the 120 VAC rating of the motor. This is easily accomplished with MCU control.

[Kremmen]

Most of the advice given here is pure nonsense, suggesting complicated circuitry, power factor correction, powering the motor with DC etc. to solve a simple motor control problem that washing machine mfg's have cheaply and simply solved for tens of  years.

While washing machines do use an easily understood concept of speed control, it does not make the alternatives nonsense in any way. The triac based control is deceptively simple in that you still need to implement zero crossing detection and firing phase timing and control. Bear in mind that the firing angle and motor speed are not linearly correlated.

Many, many mfg's of Washing machines still use simple, but powerful universal motors to turn the tub and just use simple Triac phase control to control the motor, albeit with feedback from a tach-generator mounted on the rear of the motor to adjust speed according to motor load. It is possible to construct a reliable and practical motor control using abut a dozen cheap and easily found components, and this is including resistors, capacitors, diodes, and a single LM339 comparator and a LM358 dual op-amp(Optional for speed control tach signal conditioning.

This is where the simplicity is starting to fade. While you are basically correct you now have introduced extra signal conditioning circuitry and feedback control with all its potential pitfalls. Depending on the intended usage this may be a necessity or just a nuisance. In any case you are now along the road that you criticize above.

I put forward the argument that of these two alternatives:
1: create a zero crossing circuit; a phase delay timer; a speed control loop with tach feedback; opto isolated triac main circuit or,
2: create a DC bus by bridge rectifier + smothing cap; a 555 based PWM circuit; opto isolated MOSFET switch
we can argue that there is no big difference or if there is, it is alternative 2 that is the simpler to implement. Washing machine mfgs have spent years to whittle away the last cent of expense so there is considerable design effort in the simple looking circuit. Replicating it offhand on blank paper, while of course possible may not be trivially easy. Alt 2 will mostly work out of the box.

I deleted the rest of your description which rather underlines the opposing viewpoint to the one you are advocating...

[Zero999]

Most of the advice given here is pure nonsense, suggesting complicated circuitry, power factor correction, powering the motor with DC etc. to solve a simple motor control problem that washing machine mfg's have cheaply and simply solved for tens of  years.

I agree with you about there being no need for power factor correction but powering the motor from unfiltered DC is a perfectly valid solution and has the advantage of not requiring zero crossing detection.

Oldway:  An autotransformer is not required.

A 120V AC motor can be used without a problem at 240VAC with the same phase control circuit. The Triac phase angle of the trigger pulse must be limited to the half-cycle waveform decay area of the 240VAC sinewave so that the voltage applied to the motor does  exceed the 120 VAC rating of the motor. This is easily accomplished with MCU control.

If you opt for that, just make sure the motor is protected against overheating, in case the MCU output freezes high or the switch jams on.

[Paul Price]

Hero999:  If you choose to add expensive, bulky, hard to construct and insulate not needed circuitry to overly complicate Universal Motor control by using D.C. power  then the higher voltages will not permit PWM control to be used with with a 120VAC motor on 240VAC circuits, the voltages are just too high.

As for MCU lockup with an active high trigger, that is why MCU watchdog timers are used to failsafe program code.  I have never seen my MCU circuit lock up with proper shielding, using the built-in MCU watchdog timer,  and using a well-filtered power 5V power supply, so I have not ever experienced MCU lockup problems, but the overly paranoid can always add a single 555 one-shot to buffer the Triac trigger MCU output control pin to limit pulse width to 1mSec.

Kremmen: My control scheme is not a on-paper only phase fantasy. I use this technique and it works reliably as I have described.

And while phase-angle and speed are not linearly related, a MCU does not need to know that, phase angle is adjusted by software to develop the motor speed required by software control. In most cases, and especially with a gearbox, this is not that difficult to accomplish with simple software.

[Zero999]

Hero999:  If you choose to add expensive, bulky, hard to construct and insulate not needed circuitry to overly complicate Universal Motor control by using D.C. power  then the higher voltages will not permit PWM control to be used with with a 120VAC motor on 240VAC circuits, the voltages are just too high.

How are the voltages too high?

PWM is used in inverters to control AC motors so should work fine with a DC motor.

[Kremmen]

Hero999:  If you choose to add expensive, bulky, hard to construct and insulate not needed circuitry to overly complicate Universal Motor control by using D.C. power  then the higher voltages will not permit PWM control to be used with with a 120VAC motor on 240VAC circuits, the voltages are just too high.

I am not quite clear to what expensive and bulky circuitry you are referring to. Also, PWM control with bus voltages significantly higher that the max allowed motor voltage are entirely feasible. You just need to limit the maximum duty cycle according to the armature inductance. Granted, it is a small complication to the trivial circuit but entirely doable and in fact done all the time in stepper drivers.
If the supply exceeds the maximum allowed for the motor then you need to limit the voltage regardless of what control strategy you use. That doesn't differentiate between PWM and phase angle control (it is the same peak voltage in both cases).

As for MCU lockup with an active high trigger, that is why MCU watchdog timers are used to failsafe program code. 

Kremmen: My control scheme is not a on-paper only phase fantasy. I use this technique and it works reliably as I have described.
And while phase-angle and speed are not linearly related, a MCU does not need to know that, phase angle is adjusted to develop the motor speed required by software control.

Sure, for your information i have experience of phase angle control starting from the halcyon days of late 70's (and they were not washing machine motors). So the functionality of the scheme was never in doubt. But the more you write the more you argue against your own case - not only have you now introduced a MCU, now we are also discussing watchdogs to mitigate the risks that option brought along. How much more do you plan to simplify this problem? 
In a "serious" case where proper control and safeguards are needed, a MCU is a natural candidate to handle many of the control and monitoring tasks. That however is something else than the main power circuit of the motor. A MCU is equally applicable (or otherwise) to both of the competing alternatives here so in that sense it is irrelevant to the discussion regarding the complexity of the control scheme.

[Paul Price]

Hero999:  120VAC has a rectified peak rectified value of 1.414 * 120V and 240 VAC develops 341 VDC rectified and filtered, twice that of 120VAC. Do the math. PWM does not change the peak amplitude applied to the motor, only the duty cycle. Brushes will arc and fires will start.

[Paul Price]

Kremmen:  The original poster is starting out using a MCU to make his idea work.


"This is what I want to do: control the rpm of a universal motor, using Arduino.
Parts that I have to use and prerequisites:
- Arduino, hall sensor, potentiometer, universal motor (230VAC, wattage usually around 250W, but might peak around 400W, max rpm: 16000)"

[Kremmen]

Kremmen:  The original poster is starting out using a MCU to make his idea work.


"This is what I want to do: control the rpm of a universal motor, using Arduino.
Parts that I have to use and prerequisites:
- Arduino, hall sensor, potentiometer, universal motor (230VAC, wattage usually around 250W, but might peak around 400W, max rpm: 16000)"

Ah fair enough. I didn't check back to the original post any more.
But can we agree that generating a simple (possibly modultion limited) PWM signal using Arduino standard library is anyway simpler than detecting mains phase an synchronizing the triac firing to that? Both are doable of course, and with Arduino if the OP wants that.