class D amp

The caps will force the volt-seconds to average to zero, but not necessarily within a cycle. There may be some DC on the caps because once the average current in the core gets off zero, it'll walk in the direction of increasing magnetization. It does have to stop on account of the capacitor, but maybe there could be a bit of distortion as a result.

Seems like you might get some subharmonics too.

Cheers

Phil Hobbs

How about using a VFD? Like this :

!20 VAC in, variable volts and frequency out and three phase also.

Cute idea; an idling motor (or better, motor with a flywheel) would be a really fine low-Z filter, too. The old days (1930-ish) before 'electronics' got well established, saw motor/disk/generator audio generators as standard off-the-shelf items. You'd tune 'em by moving the driven wheel from rim to center of the drive disk.

I wouldn't normally expect enough voltage drop across the caps to turn the diodes on, unless something goes wrong. I could use a shorted bridge rectifier and get two junction drops in each direction.

That TI part has a max output DC offset of 60 mV, which is pretty good, but that's still about 1 amp into my toroid. It also has a shutdown if it sees DC load current, 1.5 amps, which is pretty bad.

The data sheet is medium terrible.

lt.

the filter inductors must also add some DCR

how about using a stepper motor driver?

Keep it simple. For that sort of voltage output, any audio amplifier with suitable output swing should get the job done, and distortion will be insignificant at that sort of frequency. Either that, or find one of the variable output lab ac supplies from Ebay or similar.

Do you want to design it, or use it :-)...

Chris

The transformer shown isn't centre-tapped so it doesn't lend itself to the usual Baxandall oscillator scheme, or my low distortion variant of it, but the voltage-switching variant - Figure 8 in the original paper - might work

That sort of scheme isn't susceptible to progressive saturation. Because th e core permeability does vary with flux, you can get significant harmonic c ontent from that alone, which would be less of a problem if you were using the transformer purely to step up a lower voltage output.

I did put together a current-switched low-distortion quasi-Baxandall driver for LVDT in the - extremely precise weighing head of the Metals Research G aAs crystal puller back in 1987-88. It worked fine.

Driving the toroid from a 50 ohm fungen, at 400 Hz, it only takes 10s of mA of DC to make serious distortion. It will be better driven from a low-Z source, but I'm thinking that DC is bad.

The TI part has a lot of nice features and protections.

Use it and maybe sell it if it works nice. It would complement our synchro/LVDT boards, which don't output much power.

There aren't many low-power 400 Hz sources around. 1 KVA is small in this world.

I don't know the TPA3251... Says 13.2 V_max?, maybe you mean the tpa3221?

Is it a one of? (I would do class A, a heat sink and fan... ('cause that's what I know.)) Is the class D partly for fun and future profit?

George H.

---------------------------

The obvious problem is core saturation, if the input frequency goes too low PLUS inrush surge if the drive signal starts suddenly or is interrupted. T he variable sine wave oscillator was on the same PCB and thus permanently c onnected.

Powering up produced misbehaviour of the power IC which was solved by addin g a fade-up feature in the signal path consisting of a red LED and CdS cell . The cell was in series with the signal and the LED came on delayed and gr adually but would go out fast if the supply was interrupted. A few diodes, a zener and a 1000uF cap did the job.

..... Phil

-------------------------

The class D power stage spec is 36V !!

Only a $10 part but rather complex to use.

..... Phil

Ouch, sorry Phil, no need to yell. (I only looked at the first stage of the spec sheet and misunderstood.)

Sorry, my own hammer/nail problem. I've used someone else's AB and complained about the cross-over. (I'm assuming you by class AB you also mean a push-pull output.) (Well i've done push-pulls for one of stuff, just never sold one.)

Of course, I mean I don't know a lot, but I'm not a complete idiot.

Complex from the micro-p programming side... (ignore that I'll go read the spec sheet.)

George H.

Why even filter prior to the xfrmr at all? Since you're going to be generating the signal using DDS on a uP anyway, why not use a part like:

that takes a PWM input to begin with? Then you get significantly lower offset voltage specs and should be able to ditch all the caps n such on the primary side and not worry about that "devil's staircase" stuff.

Intuitively it seems like you'd get the best performance by keeping everything digital/PWM until the last moment via an LC on the secondary side.

----------------------

..... Phil

No, No, it's 50 volts max. Or maybe 38.

Sometimes the signal could come from our VME module, so I need the option to accept an analog input.

The LC on the primary keeps the HF stuff out of the toroid. I'd expect it to be very lossy at 600 KHz.

------------------------------

The data sheet make this very clear, Mr Pedant.

..... Phil

I would use the transformer to generate a 90V DC voltage on the secondary side. Then add the class D circuit on the output instead

No staircasing, and very simple filter

You can do an unregulated DC/DC converter which simplifies the converter. It would add more active components since you need a half bridge and 2 diodes on the converter

You would need communication across the barrier, but you need that anyway to monitor the output voltage

Cheers

Klaus