PWM AC Induction Motor Control

Thread Starter

nekojita

Joined Nov 19, 2010
75
Hi,

I posted in an older forum did not receive a response so I thought I should start a new thread. Where to start and keep it brief. I replaced an old 16" table fan recently that has 3 speed switches. Unfortunately, at the slowest speed switch selection, it is still too fast and loud, especially at night.

Online, I purchased several fan motor speed controls and all caused the fan to hum at very low speeds which was annoying at night. I did however, purchase a Staco 201 Variac for fan speed control and it works perfectly!

I am not convinced that there is not a solidstate solution that will provide quiet operation in a more compact form factor. After much searching this lead me to a PWM AC controller.

Adding a triangle wave generator and massaging the control circuit values, I was able to get a beautiful PWM AC controller. I am attaching a photo of the controller output and the LTSPICE circuit schematic.

All is great, right? Well, no. I measured my fan motor resistance at 263 Ohms. With that resistive I obtain the response. shown in the photo. Here's the rub: I also measured the motor winding inductance. It is 80mH and change.

When I add the inductance in series with the motor winding resistance, the simulation load voltage goes crazy - not at all well behaved and certainly not what I was wanting.

Any help from this Forum on circuit mods that would work well with the series motor inductance would be greatly appreciated.

Thanks,

Neko

1596071266167.png
 

Attachments

Alec_t

Joined Sep 17, 2013
11,549
When I add the inductance in series with the motor winding resistance, the simulation load voltage goes crazy - not at all well behaved and certainly not what I was wanting.
It's the load current which is important. Plot that and you will see something more like what you're expecting (according to my crystal ball) :) ).
 

schmitt trigger

Joined Jul 12, 2010
312
The problem with induction motors is that its impedance, both the real and imaginary terms, changes with the slip, thus with the motor speed.

Have you tried a much slower PWM carrier frequency?
 

Thread Starter

nekojita

Joined Nov 19, 2010
75
It's the load current which is important. Plot that and you will see something more like what you're expecting (according to my crystal ball) :) ).
Hi Alex,
Thanks for your response and comments. So, where can I get a crystal ball? You are about the load current. Any thoughts as to the load voltage looks so crazy? In the plots below, I added the 80mH series inductor:
1596125847451.png
Since the load current appears as expected, I'll build up the circuit to see how it performs.
Thanks again,
Neko
 

Thread Starter

nekojita

Joined Nov 19, 2010
75
The problem with induction motors is that its impedance, both the real and imaginary terms, changes with the slip, thus with the motor speed.

Have you tried a much slower PWM carrier frequency?
Thanks for your response and comments. The original research paper talked about using a 32kHz PWM frequency to control an induction motor. As you point out, motor impedance is complex and varies with many factors like motor speed, load, etc. My plan was to optimize PWM frequency for minimum hum, if that's possible with this circuit. All the common Diac-Triac controllers I tried hummed or buzzed at low fan velocity. It's easy enough to change the PWM frequency in the sim to see if it cleans up the load voltage. Alex had a good point about the load current which looked fine in the sim. Any other thoughts? Thanks, Neko
 

crutschow

Joined Mar 14, 2008
25,273
If you don't need an adjustable speed (other than the 3-speeds on the fan), I have reduced the speed of a similar table fan to a low, quite level for sleeping by inserting a non-polarized (film) capacitor of a few μF in series with the motor.

The value depends upon the motor size and how slow you want the fan to go, but a value of 2 to 6 μF should work.
You can buy several 1μF units and connect them in parallel until you find the right total.
 

Thread Starter

nekojita

Joined Nov 19, 2010
75
If you don't need an adjustable speed (other than the 3-speeds on the fan), I have reduced the speed of a similar table fan to a low, quite level for sleeping by inserting a non-polarized (film) capacitor of a few μF in series with the motor.

The value depends upon the motor size and how slow you want the fan to go, but a value of 2 to 6 μF should work.
You can buy several 1μF units and connect them in parallel until you find the right total.
Hi Crutschow,
Thanks for your response and idea. :) In my searching for speed controls, I saw that idea several times but I think I really do want more speed range control. Also, the circuit that I cobbled together is at least worth a try.
Thanks,
Neko
 

Thread Starter

nekojita

Joined Nov 19, 2010
75
Hi Crutschow,
Thanks for your response and idea. :) In my searching for speed controls, I saw that idea several times but I think I really do want more speed range control. Also, the circuit that I cobbled together is at least worth a try.
Thanks,
Neko
Well, I'm back. Seems that I fat fingered in 10mH instead of the 80mH that was intended. Looking at I(R5) with the correct value, it doesn't look too good. See below:
1596141924255.png
Zooming in, the switching is really slow, likely the time constant of Lmot/Rmot. Also needed to change the FET to get any reasonable switching. See below:
1596142145766.png
There's no differentiation in PWM width. Any thoughts? Thanks, Neko
 

crutschow

Joined Mar 14, 2008
25,273
The inductance will tend to keep the current flowing when the PWM signal is zero so the current waveform may not turn off as fast as the voltage, but that should not be a problem in the real circuit.
 

Thread Starter

nekojita

Joined Nov 19, 2010
75
The inductance will tend to keep the current flowing when the PWM signal is zero so the current waveform may not turn off as fast as the voltage, but that should not be a problem in the real circuit.
Hi Crutshow,
We all Eli the Ice man, so your points has merit. :) Generally, I believe SPICE results- been using in some form since Berkeley SPICE in 1982. Good models give good results. GIGO, of course.
So, the question here is whether to get the components, build it up and test it out. I usually only build a circuit that's SPICE proven- easier to tweak in SPICE than on the bench.
Thanks,
Neko
 

crutschow

Joined Mar 14, 2008
25,273
I also have used SPICE since running the early Berkeley versions, batch on a main-frame, and wouldn't build a circuit without simulating it first, if possible.

The problem with electric motors is that they provide a significantly varying impedance with load so, for accurate simulation, you need to know what that is at the desired load.
If you have a power meter, such as the Kill A Watt, you can measure the fan power (or VA) and power-factor at the desired speed with the Variac to calculate the impedance for the sim.
 

Thread Starter

nekojita

Joined Nov 19, 2010
75
I also have used SPICE since running the early Berkeley versions, batch on a main-frame, and wouldn't build a circuit without simulating it first, if possible.

The problem with electric motors is that they provide a significantly varying impedance with load so, for accurate simulation, you need to know what that is at the desired load.
If you have a power meter, such as the Kill A Watt, you can measure the fan power (or VA) and power-factor at the desired speed with the Variac to calculate the impedance for the sim.
Hi Crutschow.
Thanks for your feedback. Yeah, don't have that equipment; just a DVM and basic Impedance Analyzer. Well, it's not a complicated circuit; could build in a couple hours and test after I collect all the components.
Maybe it's just me, but I just want to know if this topology can work in this application. As you suggested, the sim may not be telling the full story. :)
Will try it out on the bench and report back.
Thanks,
Neko
 

MrAl

Joined Jun 17, 2014
7,764
Hi,

I posted in an older forum did not receive a response so I thought I should start a new thread. Where to start and keep it brief. I replaced an old 16" table fan recently that has 3 speed switches. Unfortunately, at the slowest speed switch selection, it is still too fast and loud, especially at night.

Online, I purchased several fan motor speed controls and all caused the fan to hum at very low speeds which was annoying at night. I did however, purchase a Staco 201 Variac for fan speed control and it works perfectly!

I am not convinced that there is not a solidstate solution that will provide quiet operation in a more compact form factor. After much searching this lead me to a PWM AC controller.

Adding a triangle wave generator and massaging the control circuit values, I was able to get a beautiful PWM AC controller. I am attaching a photo of the controller output and the LTSPICE circuit schematic.

All is great, right? Well, no. I measured my fan motor resistance at 263 Ohms. With that resistive I obtain the response. shown in the photo. Here's the rub: I also measured the motor winding inductance. It is 80mH and change.

When I add the inductance in series with the motor winding resistance, the simulation load voltage goes crazy - not at all well behaved and certainly not what I was wanting.

Any help from this Forum on circuit mods that would work well with the series motor inductance would be greatly appreciated.

Thanks,

Neko

View attachment 213517
Hello,

One of the problems i see is that the switch stage is just a MOSFET and a rectifier bridge. I am hoping that you used high speed diodes for the bridge, but that's probably not the main problem. The main problem is that there is no snubber and no low Z 'off' state and there is no DC buss so there is no way to recover spike energy, and so spike energy gets wasted.
The way this is usually done is to use an H transistor bridge which then has a regular DC buss, and there is also a low impedance 'off' state of the bridge.
If you want to do this right, check that out. Look up an H transistor bridge either MOSFET or bipolar.

With an H bridge first the input line is rectified and filtered, then the H bridge does all the PWM by chopping up the DC into a synthesized AC waveform.
Since you dont need isolation you dont need a transformer except maybe for the low voltage supply but even that you can sometimes get around.

A cheap trick is to see if you can find a light bulb that when put in series with the fan causes a drop in power to the fan. You can try different wattage light bulbs, which will depend on the wattage of the fan and how slow you want it to go.
 

Thread Starter

nekojita

Joined Nov 19, 2010
75
Hi Danko,
Thanks for your response and circuit contribution. :) I must say that I've never seen a motor drive configuration like this before; very interesting! Considering the 1.24H motor inductance, your circuit is a very capable driver. It appears that the back drive turns on the snubber at an appropriate time, correct. Presumably when the current stops flowing during one half cycle?
The PWM and Back drives have no ground reference. How would you implement these drive signals in a working circuit?
Thanks,
Neko
 

Danko

Joined Nov 22, 2017
972
Hi Neko,
The PWM and Back drives have no ground reference. How would you implement these drive signals in a working circuit?
For example - using some of these drivers:
Isolated gate driving solutions
It appears that the back drive turns on the snubber at an appropriate time, correct.
Not exactly. M3 and M4 are controlled "free-wheeling diode".
That "diode" is connected in parallel to motor, as should be with PWM, and changes its polarity every half cycle.
Presumably when the current stops flowing during one half cycle?
In such emergency cases independent snubber C2, R3 works.
 

Thread Starter

nekojita

Joined Nov 19, 2010
75
Hello,

One of the problems i see is that the switch stage is just a MOSFET and a rectifier bridge. I am hoping that you used high speed diodes for the bridge, but that's probably not the main problem. The main problem is that there is no snubber and no low Z 'off' state and there is no DC buss so there is no way to recover spike energy, and so spike energy gets wasted.
The way this is usually done is to use an H transistor bridge which then has a regular DC buss, and there is also a low impedance 'off' state of the bridge.
If you want to do this right, check that out. Look up an H transistor bridge either MOSFET or bipolar.

With an H bridge first the input line is rectified and filtered, then the H bridge does all the PWM by chopping up the DC into a synthesized AC waveform.
Since you dont need isolation you dont need a transformer except maybe for the low voltage supply but even that you can sometimes get around.

A cheap trick is to see if you can find a light bulb that when put in series with the fan causes a drop in power to the fan. You can try different wattage light bulbs, which will depend on the wattage of the fan and how slow you want it to go.
Hi,

Thanks for your comments. So, this journey started out with trying several diac-triac controllers- they caused the fan to hum at low speeds. Next I bought a small Variac on Ebay. (Staco 201) I boxed it up as shown below:
1596227670756.png
It works fine even when I can see the blade rotation- no hum. With some time on my hands with the Covid thing, I thought I'd try to reduce size and make it solid-state.

The basic circuit that you commented on came from: https://www.electronicshub.org/pwm-based-ac-power-control-using-mosfet-igbt/
It looked simple to simulate and I relaced the Arduino with my own PWM circuit. But, as you can see, it doesn't work as expected with the measured motor inductance.

Next choice may be Danko's suggestion if I can get the drive implemented. The question that I won't know until I try it is whether the fan will hum at low PWM duty cycles.

Thanks,
Neko
 

Attachments

MaxHeadRoom

Joined Jul 18, 2013
21,274
Personally with all that effort, for a silent solution, I would have sought out a low voltage DC solution such as a Radiator fan motor from an automotive wrecker.
The controller is easy to make or CHEAP on ebay, a small mains transformer to 12v and a bridge would all that is needed for supply.
Max.
 

Thread Starter

nekojita

Joined Nov 19, 2010
75
Personally with all that effort, for a silent solution, I would have sought out a low voltage DC solution such as a Radiator fan motor from an automotive wrecker.
The controller is easy to make or CHEAP on ebay, a small mains transformer to 12v and a bridge would all that is needed for supply.
Max.
Hi Max,
Thanks for your reply and comments. What intrigued me about going the AC PWM route is that there is an abyss of circuit design available for it and cheap commercially PWM units sold on Ebay or Amazon. I asked myself why?- no good answers. Was able to find some research papers on the topic, but the circuit descriptions were vague. Looked at patents- again, a lot a smoke but no fire. What gives? So for me, it's a quest for the unknown. :) Neko
 

MaxHeadRoom

Joined Jul 18, 2013
21,274
For one, AC 1phase induction motors in general are notoriously poor at being speed controlled.
For the most part it is a compromise.
Apart from AC 3ph induction versions, which, VFD's work well, the general choice is with the PM motors of some kind as the prefered format for RPM control.
Max.
 
Top