Hard Switching AC?

I was looking into speed controlling an AC fan motor and was surprised that it used a multi-tap transformer and an array of relays to select the drive voltage. As I looked more in depth, controlling AC fans are still usually either are the 3 / 4 speed multi-tap motors, or thyristor phase angle.

Old school!

Are there really no good modern options for hard switching an AC waveform? SSRs have limited power handling capability, physical relays are slow, IGBTs/FETs need to be doubed up (is this the way?) and SCRs can’t hard switch.

I saw some exotic topologies like MCT, but couldn’t purchase any!

Any reasonable way to make an AC/AC “buck” regulator ?

It seems to be either SCR or frequency inverter. I don’t recall seeing anything in between like IGBT.

Scale is important here. What voltage and current are we talking about? Kinda sounds like pretty small.

Is this a one-off or a production design.

If this is a small product application you might consider (multiple) DC fans.

Why do you want to “hard switch?” Is 20 ms uncertainty that big a deal in your application? Fan mass would seem to be a bigger issue that switching uncertainty.

Zero crossing turn off is generally undesirable in AC switching, not that turn on phase angle ‘dimmer’ style control is all that desirable. Big (6 HP) fans I worked with used a variable frequency drive. I suspect that was only because the tapped transformer used in the earlier generation was expensive both monetarily and weight.

Look at variable frequency drives (VFDs.) Not simple, but can offer a lot more degrees of control freedom like constant speed, torque, anticogging, etc.

Beware of the simple VFD descriptions. The descriptions of concept vary a lot from the reality of implementation. Most concept presentations describe a AC->DC switcher, and a DC->AC PWM. The reality is a lot more clever. VFDs are really common. Most washers, dryers, and some high end AC systems rely heavily on VFDs. Pretty much any design that used a mechanical transmission with an AC motor up to several horsepower 20 years ago uses a VFD instead now.


1 Like

For motor drives that has a tangible output, meaning torque ripple matters, hard high frequency swithing is the way to go

For a fan where ripple doesn’t matter SCR or other techniques are good enough and cheap

There is good info here:


" When a DC fan or motor is available and DC power is available, nothing beats a DC control for energy savings, cost, size, safety or controllability. When circumstances do not permit the use of a DC fan or motor, CRI offers 3 AC motor speed control technologies to consider; TRIAC (voltage or phase control), VFD (Variable Frequency Drive), or ECM (Electronically Commutated Motor)."

Note that any electronic speed control that involves switching, like phase angle or VFD, has the potential to generate vast amounts of EMI.


I generally agree, but size matters. Small electronic cooling applications, sure. Industrial blower, not so much.

The fact that the original solution is tapped transformer implies either mid 20th century, or larger than typical small cooling applications…


SCR / Triac is not done with tapped transformer. They are used directly on the line

I agree, but I think we’re talking past each other and may be close to violent agreement.

The original post said that the original config used a tapped transformer. My point was that Thyristor options have a practical upper power limit even for fans. This application may exceed the practical thyristor power limit as the original used a tapped transforner which might imply a higher power class application or just a very old one.


I have done megawatt thyristor control, so there is no real limit. Granted, this was for a electronic brake for a wind turbine, so no limits on the harmonic content

Other than that I guess we are talking about the same thing :blush:

I did once a back to back mosfet to control the fan speed with a PWM signal. One thing I found out is audible noise. So there are reasons they might use a tapped transformer. One other interesting concept I came across is these three speed controllers were they used no transformer, but with every tap of the rotary switch they switched in series a different capacitor value. So basically you got a capacitor in series with your fan. This also was 100% silent what is quite important when a fan is mounted in a household.

I was absolutely flabbergasted that multi-tap would be a solution in a modern high end appliance; For work I do a lot of DC/DC, and a bit of AC/DC DC/AC… and I was blindsided that AC/AC is so different. I’m in the EE generation where transistors are fast and amplifiers start at class D!

Scale-wise, this started as a one-off project adapting existing equipment for a research project - I needed to remotely activate the fans in an existing product as small part of a larger system. No need to make it productizable, purely research at this phase. I thought I’d be able to buy an off the shelf home automation solution (don’t reinvent the wheel when the wheel store has 2 day shipping), but I’ve been surprised how disappointing that search has been.

The difference in audio is overwhelming - driving it with the multitap transformer + relay array that it came with is effectively silent below 70% power, but using an SCR dimmer makes an annoying loud buzz at any power level. Or to put it in market pressure terms, I would have gladly paid 80$ more for a solution that didn’t add this noise… and I’m much less spendy than a typical home automation nerd.

I wound up just hacking in on the low voltage digital side instead. It isn’t as portable a solution and required a few hours of unexpected firmware work / reverse engineering, but it works good enough for the project requirements.

If I had more time on my hands and a good high-frequency hard switching solution I’d go ahead and productize it myself for those spendy home automation nerds.

I agree that ECM is the way to go when possible - the brushless fans in other parts of this project are beautiful. Not possible for this sort of retrofit.

@Bikkel - Interesting approach! It sounds really similar to a 3speed / 4 speed fan, which is essentially just turning the motor winding into a multi tap transformer… ish.

Apart from any audible noise, any digital switching scheme is going to generate RFI. The complete lack of RFI is a huge benefit of tapped windings.


Sure, but rfi mitigation is well understood and straightforward to handle. It shouldn’t be product-definingly scary!

As my first boss once told me, RF is the easiest thing an electrical engineer will ever do, as long as you do it right the first time and you don’t f it up. If you do f up, break out the robes and the sacrificial chickens.

Lesson two was don’t put bandaids on before there is blood. Footprints are allowed, but premature mitigation snatches defeat from the jaws of victory.

Tell that to all the clients who come to me for emissions mitigation after failing testing time after time.


+1 to what @JuliaTruchsess said. It might be well understood in this forum circle, but experience suggests that it’s not that well understood if you go that far outside of that circle.

Heck, my track record suggests I’m pretty damn good at RFI mitigation and I still get stumped by it from time to time.