Question on Class H amplifiers

[jonjara]

Hi,

I am new here so please bear with me. I am have essentially no experience with audio amplifiers and equipment, but I need to construct a high power amplifier for research purposes and this seems to be the place with the best resources. I do have quite a bit of experience in low power DC and high power pulsed-based electronics work, but not high power DC stuff, although I've spent the past week reading up on amplifier design. The requirements my amplifier needs to meet are:

a) 0-200V, capable of sourcing ~40A at all voltages
b) DC - ~30kHz, the higher the bandwidth the better
c) reasonably small (eventually we will need 34 of them so size optimization is best)

It is to be powered from a 200V DC bus so I do not need to work about a rectifier.

Based on these requirements, I think that a class H style amplifier would be best, and I've read all of Workhorse's thread on class H amplifiers, but I did not quite understand some of the details, especially the rail-sticking. What causes rail-sticking? ilimzn made a post describing a digitally driven amplifier, but it was unclear to me what impact that had on rail-sticking. My amplifier will be FPGA/DAC driven, so if I use the FPGA to pre-switch the rails (or post-switch them on a falling edge) would that eliminate the rail-sticking, or is it caused by some inherent feature of the transistors?

Finally, all the schematics I have seen on here use almost entirely all discrete parts. What are the disadvantages to just using an op-amp and incorporating a voltage gain stage followed by a power gain stage in the feedback loop?

Thanks,

Jon

[Andrew Eckhardt]

The way you mention a bunch of them, a power bus, and size, it seems like you're going to put all these in one place. I don't know what you're going to do with around 1/4 MW of audio amplifiers but the biggest one I've ever seen was a 50kW tube amplifer designed to drive a shaker table that was mounted on a 20ft cube of concrete poured into the ground. When it ran it shook the whole building anyway. By the way, that amp took up two entire 6 foot tall relay racks. It's a pretty safe bet you're not going to het a whole lot of advice on designing equipment like that here. What you're doing is going to end up looking like inverters for electric cars, except even bigger, cuz it's gonna make some Heat. The opamps will be the least of your worries.

[Andrew Eckhardt]

Will you be contracting with the power company to install a 3 phase service or are you going to purchase your own diesel generator?

[opc]

Hi Jon,

As Andrew mentioned, I have no idea what you plan on doing with 32 8kW amplifiers, but Class H is definitely not the solution for you. I think it might be imperative to reassess how much power you really need, and whether or not your 200V bus can supply the necessary power to run all those amplifiers.

As for an amplification solution, you'd be embarking on a 2-5 year R&D project which would ultimately cost you a small fortune. At that power level, class D is your only real option given your criteria, and you'd need to run a full bridge off your 200VDC bus with several parallel fets to deal with the 40A of current. A full bridge would minimize bus pumping, but you'd still likely end up with a fair amount of noise/EMI making its way back to your main bus.

Honestly, I don't think you're going to find a realistic solution to those requirements. Maybe if you share what it is you're trying to do, and exactly what these amplifiers will be driving, then there might be a chance that someone here would have a suggestion for a more realistic way of doing it. The only application I can think of is some kind of industrial automation, but with those power requirements, I'm really not sure.

A few questions:

1. Is that peak current/voltage output, or will the amplifier be required to deliver say 190VDC at 40A for any length of time?

2. Does distortion matter?
3. What kind of load is this thing driving? Purely resistive? Some kind of motor? A bank of batteries?

4. When you say small in size, what exactly do you mean?
5. Is the 200V bus well regulated? Reliable? Noisy?
6. Is EMI a concern?

Sounds interesting anyhow, if not a little high reaching...

Cheers,
Owen

[Andrew Eckhardt]

30Khz bandwidth and high linearity at that kind of power must be a very interesting application indeed, or just a mistake.

[jonjara]

Our lab is wired for 250kW 480V 3 phase. The large number of them is for a much larger experiment which is slated to begin construction within the next year or so. The lab space for that has 500kW of power available. The heat issue is why we need to go with either class H or class G amps for the efficiency, and even then all of them will most likely be water cooled. Additionally, it is highly unlikely that we will be drawing a full 8kW through each amp, that is just an upper limit. My questions are not so much on how to handle power delivery but on the topology of H class amplifiers. Also, I didn't mention it above, but I am most definitely not driving any type of speakers or audio equipment with these, actually the intended purpose is to use it as a arbitrary waveform generator/programmable power supply.

Jon

[godfrey]

Originally Posted by jonjara

a) 0-200V, capable of sourcing ~40A at all voltages
b) DC - ~30kHz, the higher the bandwidth the better
c) reasonably small (eventually we will need 34 of them so size optimization is best)

It is to be powered from a 200V DC bus so I do not need to work about a rectifier.

Hi Jon

This sounds like some kind of industrial application, rather than audio. Does it require true AC output or just varying DC (i.e. single-rail supply with load connected to ground)?

Either way, wouldn't class D amplification be more appropriate? 40A drawn off a 200V bus means up to 8KW dissipation. That would require a huge bank of transistors and serious cooling if done with a non-switching amp.

btw: You might find this of interest, even though it's not nearly in the same power range you're looking for: It's cheap, it's N, it's dirty, it's.... The CIRCLOMOS!!!

Originally Posted by Elvee

It is a "lab workhorse", an all-purpose power amplifier, designed to drive all sorts of transducers (including loudspeakers), transformer, coils, etc, at various frequencies, from DC to ultrasonic. It also does stress-testing of components or subassemblies, to simulate load-dumping conditions f.e., or inject interferferences into power supply lines.
This is why it has to be extremely stable, robust and tolerant, without any network in the output which might degrade the accuracy.
And it has to cheap, in case things go really wrong.

Regards - Godfrey

[Andrew Eckhardt]

I'm trying to imagine to acoustic noise of a linear amplifier array that size playing 10KHz. You'd probably have to wear ear plugs to protect yourself from just the transistors.

[jonjara]

Hi,

I work on a plasma physics experiment, and one of our research interests is electrostatic flow control, which means that the load I am driving is the plasma itself. This presents a number of problems, which is why my requirements are both vague and high. While it is possible to model the plasma in terms of circuit elements, this will basically require extensive numerical modeling on our end, as we have a very unique geometry. What this means is that I need to overdesign them so they can handle "100s of volts and 10s of amps", the reason I chose 200V and 40A is that the DC supply powering them is a 200V supply (it has plenty of current to handle my needs, and I believe it is decently regulated), and 40A is a conservatively high estimate on my part.

As I mentioned in my second post (I think because I am newly registered all my posts have to go through moderator's approval first so it hasn't shown up yet?), this is most definitely not an audio application.

The circlomos is interesting, but as you said it is definitely not powerful enough.

As a PPS, the goal is to be able to have one unit which can be set to just output a DC level, do a transient square pulse, sine waves, etc... so that we can look at the transient behavior of the plasma.

EMI is not too concerning, I would classify our lab as electrically noisy. Spectral purity would be nice, thus my interest in a class H.

My understanding is that class D amplifiers cannot reproduce a DC level, is this correct?

Thanks,

Jon

[Andrew Eckhardt]

You shouldn't have any special problems with a class D amplifer holding a DC output. Really one of your larger problems is going to be the top end of the spectrum. If you can get away with 10 times less (or at least half) bandwidth you should definitely put away the calculator and start searching the net for shaker table amps, where that kind of power a channel is very common. Who knows, maybe these days they make them that fast too.