High voltage amplifier

[Pack34]

For fun, I wanted to see if I could make one of these using discrete components.

http://www.microchip.com/wwwproducts/Devices.aspx?product=HV264

It's a Microchip/Supertex high voltage amplifier. The idea is to take a 0-5V analog input and then gain it to an output 0-150V analog signal. I haven't worked with analog signals at that high of a voltage level and was wondering if anyone had advice on where to start.

Would it simply be an input opamp to prepare the signal (gain it to the appropriate center point) then use a basic common-emitter BJT to scale it to the input high voltage?

[f5r5e5d]

lots of options from full discrete - like "classic" audio power amps, particularly vacuum tube amps

or composite op amp and added discrete stages: http://cds.linear.com/docs/en/application-note/an18f.pdf


of course there are a few safety issues some might want to become knowledgeable about when working with higher than ~40-50 V of SELV than most start with

below the SELV limits you really have to try hard to hurt yourself except for burns, flying fragments from exploding parts - to get dangerous current flowing inside your body at such low Volts involves sticking wires through your skin, or wetting the skin, especially with electrolyte solution

above ~200 V the potential can easily puncture even dry skin making electrocution safety a bigger concern

[Kleinstein]

Safty at the high voltage is a point to look at.

Just the 0-150 V is relatively easy. The video output stages of old CRT TV is in that range. So it's relatively easy to get parts and circuit, if you don't need much power.

[daqq]

I haven't worked with analog signals at that high of a voltage level and was wondering if anyone had advice on where to start.

Calculate your heat dissipation before - a few milliwats dissipated whilst operating at say, 10V can easily jump an order or two of magnitude at 100V when they start being a concern. Other than that, well, safety, beware capacitive crosstalk between stuff you wouldn't want talking to each other.

[T3sl4co1l]

Trawl for audio amplifier designs -- they'll be overly simple (e.g., lacking common mode range, supply voltage range, output current limiting, efficiency, etc.), but many are designed for quite high voltage levels (almost 200V total supply) for driving very large speakers (kW).

The basic idea is, it's no different from any other op-amp, you just use high voltage transistors...

Add protection where needed (e.g., you aren't going to find a symmetrical junction 200V transistor, so your diff pair isn't going to be independently rail-to-rail; you'll probably want to clamp it with diodes or zeners instead), compensate for the relatively low hFE (high Vceo transistors generally have lower hFE, though not by too much e.g. MPSA42) by using higher bias in earlier stages, or more stages, and compensate for Early effect (may be important due to the high voltage range?) with cascode circuits.

If you don't need a complete differential circuit, but just some gain, you'll probably end up doing that anyway (for reasons of familiarity and precision), but maybe with customized ranges for the common mode, gain stage and so on.  (Example, the diff pair could be PNP, supplied by a lower voltage, and the voltage amplifier stage would be GND referenced.)

Tim

[GK]

Here is one that I did when I was a wee apprentice, wet behind the ears. It was uni-polar in and out but could be switched between either polarity.

[dom0]

Circuits like these can be used for an application like this.

https://www.mikrocontroller.net/topic/380229#4327236
( https://www.eevblog.com/forum/beginners/inputoutput-caps-on-simple-headphone-amp/msg785411/#msg785411 )

I built yet another version of this a few days ago, no digital schematic, but it's very similar, just that instead of using the OP supply current the OP drives a complementary emitter follower (220 ? between input-output) driving the output indirectly (as in the links above), and the collector currents of the follower transistors are used to control the dumpers. For thermal + Uce relief I used a cascode-like circuit for the followers. Works like a charm. The advantage of this "formulation" is that the EFs can drive larger currents, so you can linearize the dumpers better, even at high output voltage swings. It also requires less output current from the OP, which will increase bandwidth and available voltage gain for many OPs. Since the cascode midpoints are biased around +HV/2 and -HV/2 output swing is not reduced (but it doesn't really matter here anyway).


/E: GK, am I right to assume that your amplifier uses a cascode-like arrangement for the output transistors? 14 transistors stacked?


I always opt to drive this kind of amplifier in an inverting configuration. It makes compensation easier (/better) and, if a discrete or hybrid input stage is used, eases CMRR requirements a lot.

[GK]

/E: GK, am I right to assume that your amplifier uses a cascode-like arrangement for the output transistors? 14 transistors stacked?

Yes, though they are MOSFETs and the bandwidth requirement wasn't particularly high. That wasn't the final version either; just a simplified/bare-bones prototype proof-of-concept. Don't have details of the final design anymore - it was 15 years ago now.

Anyway, the OP doesn't have to resort to that kind of thing for his specified voltage levels. I was just showing off.
 

[GK]

Whoa there, just found this.

[SeanB]

Would make a really good magnetron driver that, though you would melt them pretty quickly at full power. Your perfect electric chair energiser, selectable as to how you want to fry, and you can go out to a song.

Derate it a lot and you can make a nice electrostatic speaker driver, you only need 20W of power there.

[calexanian]

That is not a particularly fast part. Look at the Rise time at large signal. not good above 25khz so it takes it out of suitability for audio and above. For that purpose you would want to build a simple discrete circuit anyways. 

[Marco]

https://www.mikrocontroller.net/topic/380229#4327236

I know it's an old school trick, but what advantage does using the power supply current as a driver have over something like this :

[calexanian]

Before LM3875 and 3886 were commonplace I used to build medium sized power amps with circuits like that. Maybe I will write an article about something like that someday.

[GK]

Would make a really good magnetron driver that, though you would melt them pretty quickly at full power. Your perfect electric chair energiser, selectable as to how you want to fry, and you can go out to a song.

Derate it a lot and you can make a nice electrostatic speaker driver, you only need 20W of power there.

Though it could only drive a couple of mA and the bandwidth is way too low for audio.

[colinhnu]

Thanks for the high voltage omp circuit you post. I think it very helpful for my theme. But the picture is not very clear due to uploading. would you mind send me a more clear one. so i can study it and build the circuit on my own. i  can also do some Pspice simulations. Thanks. 
My email  :93657140@qq.com

[Wolfgang]

Hi,

I built something similar to this one and it works up to a few 10 kHz. I used MJE340/350 in the output stages to get more power (50mA).