Development of 500Vpp Linear Amplifier

[GN89]

For some Plasma measurement I'm interesed in the development of an higher voltage linear amplifier. The basic requirement are the following:

• 500Vpp output voltage
• 100mA output current
• 1kHz maximum output frequency (maybe lower)
• Control voltage -10 to 10V

The design problem I face is how to stack mosfets and control the gates because there are few to no mosfets which have such a SOA so I need to stack them to evenly distribute the voltage.

There are some designs available on the internet but there are no real measurements and sometimes there is something obviously wrong with them. Is there any Literature for DC Amplifier design for such high voltages?

[mikeselectricstuff]

Apex probably make something that will do this
https://www.apexanalog.com/support/linear_search.html

[Vovk_Z]

Here is one example, with measurements. And I've seen more in Internet. You may use similar keywords from an article.
High-voltage amplifier uses simplified circuit
For 500 Vpp you don't need so much Mosfets. I guess you need a stack of two in a row (so four for all amplifier).
It's hard to achieve 10 kHz frequency range or more with that exact topology, but it works quite fine up to several kilohertz (up to 2-4 kHz).

[mag_therm]

I am using GeneSiC G3R450MT17D in linear amplifier at 14 MHz

See datasheet Fig 14 SOA. Plot your Vd vs Id swing range  as 2 points on there.

TI App Note sluaao2.pd chapter 5 gives an approximate  method of derating the SOA  to your steady state heatsink temperature

[Vovk_Z]

Here is exactly 500 Vpp amplifier (designed by Leonid Ridico, Leoniv). I used this his design and it works pretty well. I used another opamp - OP37.

[M0HZH]

TSP just published a video on high voltage amplifiers a few days ago, watched it last night, it's gold.

[Terry Bites]

Bootstrap. In thse circuits the opamp rides the HV rails. Scary but it does work.
www.edn.com/bootstrapping-your-op-amp-yields-wide-voltage-swings/
http://www.edn.com/contents/images/45890.pdf
You do need transistors thay can hande the full output swing though. P-ch 600V MOSFETs are avaialble.
Yes, that signal path video from the other day was great.

[Terry Bites]

Prices from £120 and up!

[GN89]

Thanks for so much hints so far. I looked into the APEX Amplifiers. They don't have the SOA for my application which is why I need to stack some mosfets. Even using the suggested G3R450MT17D needs more than one in series.

@mag_therm which design do you use for your 14Mhz amplifier and which voltage swing do you achieve?

I found the link to EDN before but that design isn't working at all because the schematic is wrong at some points though this design looks interesting because it seems simple enough.

The design by Leonid Ridico looks simple enough to give it a try in a simulator beause there are no floating operational amplifiers which do not work with most models. From what I understand in this design that the gate control is isolated from the operational amplfifier. The operational amplifier build a feedback controller and the LEDs are controlled by the supply rails. Positive output voltage means the upper coupler is on and negative the lower one is on. I think R13 controls the current through the LEDs.

The LED intensity directly controls the gate. Higher intensity means the gate voltage becomes higher because voltage over R20 is controlled. This means a positive output voltage of the operational amplifier results in positive output voltage of the amplifier stage (Same applies for the lower rail).

I do not understand what purpose R14, R17, R16 and C2 serves. I think that is some negative feedback to compensate the control loop. The feedback loop looks a bit strange to me.

VT1 and VT2 are current limiters.

What purpose do R7 and R10 serve?

Another think is what about lonk term drifts in the opto couplers or drift in temperature or the mosfets? Is there any chance this can drift into a short circuit when there is no current limit in the path?

What happens when the 15V supply fails? In my understanding both mosfets are near turned off.

[Benta]

@GN89, do you have a low frequency limit? I mean, does it have to operate down to DC. or is eg, 20 Hz OK?

[mag_therm]

Hi GN89, 14MHz is homebrew design ( qucs) and build. Vd is from a variable  DC converter 13.8V to 50 ~ 200 Vdc depending on Tx mode Topology is like linear class B push pull.
Drain current ave DC is 0.7 A ave , peak Id per side  is about 1.0 Amp. Case temp at continuous full power is about 68C, but in ham radio usage overall duty cycle is less than 50%

I think you will be able to find TO247 ( or similar) mosfets to do your audio freq job.
Regards

[David Hess]

500 volts peak-to-peak or +/-250 volts at 0.1 amps does not seem that demanding for a traditional class-AB audio amplifier design.  Before higher voltage transistors became available, an output stage like that shown below was used.

Now the problem becomes the high voltage VAS stage, but since that can be NPN or n-channel transistors, higher voltage parts can be used.

[langwadt]

https://eu.mouser.com/ProductDetail/Apex-Microtechnology/PA194?qs=TiOZkKH1s2S8Qje%2FmQorJg%3D%3D&_gl=1*e6560q*_ga*MTE5MTA0MTg5NC4xNzE3MjAxMjA2*_ga_15W4STQT4T*MTcxNzIwMTIwNi4xLjAuMTcxNzIwMTIwOC41OC4wLjA. ?

[Ground_Loop]

Kepco BOP 500M. This unit does +- 500 volts, up to 80mA, and up to 6 KHz depending on operating mode.

[CaptDon]

It would seem that nearly any audio amplifier of 100 watt RMS capacity driving a step up transformer would work for you. As for the control voltage that would take place at the signal generator. I have a Telulex SG-100 that I believe can be externally voltage controlled. At the very least the output voltage can be controlled through the serial data port. Since your plasma impedance will vary you may need a voltage sense circuit to control the absolute output voltage of your amplifier. Loading and impedance shift will cause the output voltage to sag even though the amplifier is 'linear'.

[GN89]

I need DC output for that application so audio amplifier even down to a few Hz with a transformer is not an option. I think the audio output stage looks familiar and might be from the audio power amplifier design by Douglas Self. In this case the VAS stage is also critical because it might require a high voltage current source and an npn transistor but this transistor can't be controlled directly by an opamp (maybe here an opto isolator would be possible). I don't really see how the voltage is split across transistors in the output stage.

[David Hess]

I think the audio output stage looks familiar and might be from the audio power amplifier design by Douglas Self.

Yes.

In this case the VAS stage is also critical because it might require a high voltage current source and an npn transistor but this transistor can't be controlled directly by an opamp (maybe here an opto isolator would be possible).

Hmm, yes, so normally a high voltage series transistor PNP stage is still required.

I don't really see how the voltage is split across transistors in the output stage.

The resistor divider R1-R2-R3, or R4-R5-R6, divides the output voltage between the output and supply rail roughly in two, and then drives the series transistor.

[GN89]

The resistor divider R1-R2-R3, or R4-R5-R6, divides the output voltage between the output and supply rail roughly in two, and then drives the series transistor.

Thanks I got that. A single VAS transistor is still required and to be perfectly safe with the SOA I would like to stack them but I'm not familiar with the topology of the VAS stage for higher voltages.

[magic]

There is one more topology which reduces HV electronics to absolute minimum. Main disadvantage is needing a floating HV supply for each channel, not a big deal if there is only one channel. Another drawback is that the supply's earth capacitance appears as a load on the output stage, but that should be tolerable at 1kHz.

Note that an alternative way of dealing with limited SOA is paralleling output devices. For instance, Leonid Ridico's amp seems to be OK with about 40mA limit, so three such MOSFET pairs would get you to 100mA.

[David Hess]

The resistor divider R1-R2-R3, or R4-R5-R6, divides the output voltage between the output and supply rail roughly in two, and then drives the series transistor.

Thanks I got that. A single VAS transistor is still required and to be perfectly safe with the SOA I would like to stack them but I'm not familiar with the topology of the VAS stage for higher voltages.

The VAS stage is a transconductance (current) output with shunt feedback (capacitor).  So add the second transistor in series like you would do with a cascode VAS, and then a voltage divider between the output and negative rail to control the base of the second transistor.  If the voltage divider needs to be buffered, then an emitter follower will have to be made with 2 series transistors since it will see the entire supply voltage also, and the emitter follower can buffer its own divider.

I am not saying this is the way to go, but that a high voltage traditional class-AB amplifier has been a solved problem.  As an alternative I might extend the example below to higher voltages, and add short circuit protection.  This design avoids bootstrapping the operational amplifier so input common mode range is not a problem requiring positive feedback.  The cascode and output transistors have to be doubled up in series of course.  Current limiting can be added with a diode across R3 and R4.

[ejeffrey]

Bootstrap. In thse circuits the opamp rides the HV rails. Scary but it does work.
www.edn.com/bootstrapping-your-op-amp-yields-wide-voltage-swings/
http://www.edn.com/contents/images/45890.pdf
You do need transistors thay can hande the full output swing though. P-ch 600V MOSFETs are avaialble.
Yes, that signal path video from the other day was great.

Bootstrapping won't work here because of the gain needed.  The amplifier needs to amplify 10 V to 250V so will have 240V from input to output.

[GN89]

I made a first attempt to simulate the suggested circuit and extend it to a higher voltage by stacking mosfets. I had no luck simulating it with ngspice in a DC analysis. The result seems odd and the convergence is bad. Can you guys look into the schematic and tell me if the topology is correct so i can build a simple prototype?

[David Hess]

Bootstrapping won't work here because of the gain needed.  The amplifier needs to amplify 10 V to 250V so will have 240V from input to output.

It can be done by adding a positive feedback loop operating in parallel with the negative feedback loop.  Doing this allows high gain while swinging the operational amplifier's common mode voltage to stay in range of the output.

Figure 6 on page 5/124 of the attached PDF shows how this works in the non-inverting configuration and gives the math.

[macaba]

There is one more topology which reduces HV electronics to absolute minimum.

I have used this topology before with good success. I haven't come across any source material on this (aside from being a favourite of SMU designers) - I would like to read into the background more. Does your image come from a book or article?

[langwadt]

There is one more topology which reduces HV electronics to absolute minimum.

I have used this topology before with good success. I haven't come across any source material on this (aside from being a favourite of SMU designers) - I would like to read into the background more. Does your image come from a book or article?

I seem to remember the name "TransNova"

I think it was QSC that used it with sziklai pair output, so the heat sink cn be grounded and the output transistors don't need isolation