DIY'ing a VFD with 3x transformers and audio amplifiers?

[LooseJunkHater]

This idea has been in my mind for quite a long time and I'm wondering if it's possible.

If I obtained 3x identical transformers (say each are 500w capable), could I drive each one using an audio amplifier and essentially create 3 phase AC? What would be the limitations of this (aside from a LOT of wasted heat)?

[barshatriplee]

It is not practical or advisable to create a three-phase AC system using audio amplifiers and consumer-grade transformers. If you require a three-phase AC system, it is recommended to use dedicated three-phase power sources and equipment designed for power distribution.

[Jwillis]

I just guessing that you want to try 3 single phase transformers to create 3 phase? Or are these 3 Phase transformers?
3 single phase transformers wont give you 3 phase because the phases of each won't be 120 degrees apart. And I'm not sure an audio amp can achieve 3 phase without a heap of aggravation. 
You might look either into a VFD 3 phase convertor or a RPC (Rotary Phase Convertor) . There are pros and cons to either type. All depends on your needs and application. 

[tooki]

I think what the OP means is that they’d have three signal generators creating the sine waves 120 degrees apart, into three audio amps, and is wondering whether you could run the output of each through a transformer to raise it up to mains voltage.

Why wouldn’t that work?

[langwadt]

I think what the OP means is that they’d have three signal generators creating the sine waves 120 degrees apart, into three audio amps, and is wondering whether you could run the output of each through a transformer to raise it up to mains voltage.

Why wouldn’t that work?

within the limitations of the amplifiers and transformers it would work

I think with the right transformers you could by with two generators and two amplifiers, once you have two phase 120 degrees apart the third phase is just those two phases added and inverted

[bdunham7]

Why do you need transformers?

[strawberry]

use 3 phase motor to generate 3 phase

[langwadt]

Why do you need transformers?

you don't, but you'd need some pretty beefy audio amplifiers to do mains voltages without them

[bdunham7]

you don't, but you'd need some pretty beefy audio amplifiers to do mains voltages without them

OK, I didn't see that he specified the voltage.  So as I see it, the main problem with such a system is that the source impedance of the resulting 3-phase supply would be pretty high.  We'd need to know what was going to be powered by this supply to really comment further.  I've heard of people doing stuff like this to run 400Hz avionics at very low power levels. 

[LooseJunkHater]

Alright so based on the responses it seems "possible" albiet stupid.

My next problem would be trying to keep them all 120° apart and ensure that when I vary the frequency, it maintains the same phase angle. I just have no idea how I'd do this. Is it possible to somehow use an NE555 (or some other basic chip) to create a basic variable frequency sign wave (from 0hz-75hz) but then split the signal from the NE555 to somehow create the 3 phases? Or essentially create 2x "delayed" phases? I couldn't find any information on this but I'm probably not using the right keywords.

I guess I have an additional question; would this only be possible with a class A/AB amplifier? Or could I do this same project with a class D amplifier (with an appropriate filter???)? I'm thinking I can't use a Class D amplifier because the class D probably wouldn't react appropriate to the varying 120v output load?

[KaneTW]

Regular VFDs are nothing more than a class D amplifier running on a particular waveform (selective harmonic elimination PWM).

[LooseJunkHater]

I guess, but they use sPWM from mosfets which require a complex microcontroller to vary the frequency and/or duty cycle and more complex driving circuitry... Which I don't believe I'm smart enough to do.

[Circlotron]

I think with the right transformers you could by with two generators and two amplifiers, once you have two phase 120 degrees apart the third phase is just those two phases added and inverted

I have done exactly what the OP is talking about.
Wired the primaries of three separate transformers in delta and grounded one corner. Then drove the other two corners with a pair of class D amplifiers each fed with a sine wave 60 degrees apart. Yes, you read right. 60 degrees. With one corner of the delta grounded the other two corners must be 60 deg apart. 120 deg simply will not work. Maybe someone else can explain why.

Connect the transformer secondaries in star for extra voltage.

A safety concern - do not run a motor at stupid high speeds like I did. There is a danger of things coming apart and doing damage to both nearby objects and yourself.

That said, here is a 6 pole nominally 1000 rpm motor running at 16,800 rpm.

[KaneTW]

I guess, but they use sPWM from mosfets which require a complex microcontroller to vary the frequency and/or duty cycle and more complex driving circuitry... Which I don't believe I'm smart enough to do.

I've done it for ultrasonics before. It's not very hard. Precalculate the coefficients and feed it to a HRPWM engine. I've published my Github repo for doing it on a 2837xD C2000 MCU: https://github.com/KaneTW/ultrasonic_code (no guarantees the comments are accurate, this was never meant to be published).

Rough overview (this was 4 years ago, so might be forgetting something):
* Precalculated angles in she_pwm_coefficients.c (for 5 angles and 100 amplitudes, eliminating 2*5+1 harmonics). I have the Mathematica file somewhere, but there's plenty of ways to go about it.
* Compute the HRPWM configuration values in compareConfig/periodConfig (see the C2000 docs for it) using the angles and the desired frequency.
* Set up the DMA engine to feed the updated PRD/CMP values to HRPWM
* Set up HRPWM/EPWM engine to use the values/trigger properly/etc

[Infraviolet]

LooseJunkHater, "NE555 to somehow create the 3 phases? Or essentially create 2x "delayed" phases? I couldn't find any information on this but I'm probably not using the right keywords. "

I'd suggest you might do well with one 555 generating a constant waveform of the right frequency and duty cycle (50% here) using astable mode, then use two more 555's (one for each delayed phase) with preceding edge trigger RC circuits* to respond to the falling edge of the first 555's waveform. Each delayed output from that pair of 555's then triggers another 555 (so 5 555's in total) and these last pair of 555's each generate a single pulse of the same length as the positive part of the waveform from the astable. You might need to insert some logical inverters in this logic chain too, either using a logic inverter IC or using the single transistor logic inverter method.  If necessary the monstable generated pulses could perhaps be a cycle or two "behind" (so long as you didn't vary the frequency so quickly that having the other two phases generated on the next waveform to the initial one would mess up timings severely), that is to say one following the initial by 360+120 degrees phase shift and the other following by 360+240, or you could use an inverter between the initial 555's output and the first pairs inputs to make the falling edge in to a rising. This 5 x 555 solution is almost definitely not the most chip-count-efficient way to do this, and if you want to change the frequency you'll need to vary multiple resistors at once (can be done with a group of MCP42100 digital potentiometers), and it will be limited to the max frequency you can achieve, and it will give square wave outputs not sine waves (though a low pass filter following each phade may make it more sine like, again the resistors on these would need to vary for changing the freq), but I'm pretty sure the principle would work. I've done similar things in the past to create pulses delayed behind constant waveforms.

*necessary whenever you want a 555 in monostable mode to ouput a pulse which lasts less time than it takes the waveform who's falling edge triggered it to return to high

[Circlotron]

This looks pretty simple.

[Doctorandus_P]

Using a class-D audio amplifier, and then bumping up the voltage with a low frequency transformer certainly is a valid option, but it does have it's own limitations. One thing to keep track of is transformer saturation. It's much better to use a transformer at is rated for 400Vac or so, so you have a decent margin to prevent transformer saturation.

Another limitation is of course the extra cost and weight of such an transformer. but such an transformer also delivers inherent isolation, and this can be a real bonus for a lab instrument. For example, when you are designing mains connected circuits, you can use it to measure how your circuit reacts to both mains voltage and frequency variations.

[james_s]

The proposed idea WILL work, indeed California Instruments, Elgar and a few others have made 3 phase AC power sources that are more or less this internally. They age huge heavy rackmount things that produce a ton of heat but they do work.

Most consumer type audio amplifiers are not robust enough to take this kind of abuse for very long though. The commercially made AC power sources have banks of power transistors on enormous heatsinks with forced air cooling.