class-D driver chip

500 watt-capable driver, 1 amp gate drivers, over-current protection, programmable dead-time, audio input to PWM modulator is on-chip, all the things. Available at Mouser:

300 watts is still two-mosfet territory these days if you can boost the supply to 100.

That's half-bridge. Maybe I can use two of them somehow. I have a single +48 volt supply.

I was thinking of making an open-loop full-bridge amp.

We might not even need the DC balance feedback.

If it's beefy enough, we don't need to be in any hurry to shut anything down. It takes a while for transformers and inductors to melt.

You can also buy class D bricks off-the-shelf that can be customized, these are about the size of a credit card:

If it's a big-money job $80 won't break the bank

Another bipolar supply half-bridge.

Half-bridges are popular for audio because speakers are mostly ohmic.

--

John Larkin         Highland Technology, Inc 

Science teaches us to doubt. 

  Claude Bernard

Not sure I'm able to tell the difference between full bridge and half. Is BTL (bridge tied load) full bridge? I found this on DK

George H.

I think that H-bridge, full bridge, and BTL are the same. 4 fets.

Half-bridge only needs two fets, and one side of the load is grounded.

--

John Larkin         Highland Technology, Inc 

Science teaches us to doubt. 

  Claude Bernard

Is this for full range audio frequencies or is it just for low frequency ? LF is eaasier to do of course.

120 Hz to maybe 4K. This is a 3-phase power supply.

This works,

but I have a customer who needs more power. We're using a TI TPA3255 per channel, and it limits at 17 amps. This is rude and crude, so I figured that an open-loop switcher would be good enough.

--

John Larkin         Highland Technology, Inc 

Science teaches us to doubt. 

  Claude Bernard

WoW ! Cool product !

I am surprised though that it doesn't go down to 60 or 50 Hz

I do inverters but they don't do the high frequencies that your boxes do.

More and more inverters and audio amps are HF class D like switchers these days. I'm sure that GaN will play a big part in newer class D amps and especially ones like yours.

The transformers would have been huge, and the systems that we are simulating are nominally 400 Hz. Aircraft "wild power" is roughly 300 to 800 Hz. Wild is what you get from an APU or a ram air turbine or, in our case, an engine-driven alternator.

Probably so. GaN switchers can run at many MHz. I don't need that now.

We do use the EPC GaN fets in fast pulse generators. It takes some getting used to a part with essentially zero Cdg and zero lead inductance. Circuits that look really stupid just work.

--

John Larkin         Highland Technology, Inc 

Science teaches us to doubt. 

  Claude Bernard

Oh, OK. I thought you were using HF only so no LF transformer.

You could do that too I suppose. Maybe another product down the road.

Yes, these new GaN and SiC parts are awesome and gettng more affordable all the time.

Later on, you said that the load is 3 phase. Is the load a motor ? To use a full bridge, you would need access to both ends of each winding.

If the motor is fixed wye or delta connected or you are feeding a wye/delta network, I don't see how you could use full bridge, but half bridge would be OK.

What is the spectral purity requirements for the sine wave ?

Each class-D amp drives a transformer to make one AC phase. The outputs usually simulate a 3-phase permanent-magnet alternator. This is for testing FADECs. A FADEC starts up from aircraft 28 volts DC but, once the engine spins up, runs off a dedicated PM alternator.

Practically none! It's just an alternator, which should make an open-loop class-D amp OK. The waveform from an alternator depends on the shapes of bits of metal and tends a bit towards the triangular.

The alternator manufacturer claims all the specs are proprietary, which sure didn't help. We had to get one and spin it and measure stuff.

--

John Larkin         Highland Technology, Inc 

Science teaches us to doubt. 

  Claude Bernard

That may be by design. The inductance varies with angular position too. The varying inductance, high at the voltage peak, probably smooths the triangle towards sinusoidal current. Clever.

We'll make a constant-impedance sine.

--

John Larkin         Highland Technology, Inc 

Science teaches us to doubt. 

  Claude Bernard

=1

so a couple of irxxxx type half bridge drivers some beefy fets and a few pins doing PWM on a MCU would do?

not that different from a BLDC driver except you don't need at the stuff to do sensor-less commutation

IRS2108 maybe. Analog/Linear have some parts, but they are expensive and weird. The upside is the LT Spice models, and the downside is that the models only cover some of the parts and seem to be pretty bad when they do. I suppose something this simple doesn't really need to be simulated.

The existing board accepts a differential analog input, so I'll have to do the PWM myself. Probably a classic triangle generator and a comparator. Maybe a slow DC servo to make sure there isn't much offset between the half bridges. Maybe even a trimpot for offset.

I'm driving a big toroidal transformer that really doesn't want to see any DC on its primary. So we have 60 mF of DC block caps, 6V electrolytics, that wouldn't like much DC either, especially in the reverse direction.

Nope, dumb open-loop PWM is probably good enough. The easiest control loop to stabilize is no loop.

I know little of 3 phase, but watching machinists flip the spindle on a mill back and forth it's gotta both give and take power... maybe your gizmo doesn't need to do that?

In the distant past I helped run a helium liquifier, whose brake on the expansion engine was to send power back into the AC line... 3-phase motor/ generator.

George H.

mandag den 2. november 2020 kl. 19.18.57 UTC+1 skrev John Larkin:

dsniptechnology.com:

t l e .

w=1

f

and some FETs ...