Right now, for my project, I am trying to design a circuit for generating 10 kHz to 1 MHz AC signal with ~80 Vpp to 100 Vpp. Currently, I chose to the AD9850 chip for generating AC with Vpp being around 0.6 V. By using multi-stage OPAMP with large gain-bandwidth-product, I can successfully get high-frequency AC with Vpp being around 8 V.

Using similar methods, I am trying to keep adding stages using OPAMP for boosting the AC. However, the challenges are limited by the power rails of the most of OPAMP with large gain-bandwidth-product. For example, if I want to get 80 Vpp AC, meaning the gain is X10, then the gain-band-width-product I need is 10 MHz. The OPAMP of such requirement typically doesn't have high bandwidth, which doesn't seem very useful.

I am wondering if there any way for boosting the Vpp of such small AC signal to higher voltage over wide frequency band (up to 1 MHz)?

P.S. If there are ways for designing power amplification circuit, it will be awesome. However for my application, there is no need to deliver power, hence amplifying voltage will be more critical.

  • \$\begingroup\$ OK, so what do you plan to use this for? Problem is: 100 Vpp is power, because if you need to get your signal from A to B, then you'll have to have some system impedance, and: P = U² / R; U² = 10000 V², in your case. \$\endgroup\$ Nov 21 '18 at 22:27
  • \$\begingroup\$ How much output current? \$\endgroup\$
    – bobflux
    Nov 26 '19 at 18:01

First of all: what you plan is dangerous. 100 Vpp can kill you if there's any beef behind the source, and your friendly national radio frequency regulation officials will kick in your door if you happen to emit a couple 100 W at 1 MHz as RF radiation. You've been warned.

100 V peak-to-peak load at 1 MHz isn't really trivial to deal with. Remember: any two conductors can form a capacitor. So, your high-voltage line in parallel to a ground plane will happily couple over energy. Make sure it doesn't – that might mean you'll have to construct things using enough spacing, or small enough dimensions, or with classical RF transmission line design.

Now, when we're talking about voltage, we'll have to talk about impedance: 100 V into a 100 kΩ load is much easier than into a 50 Ω system. At these frequencies, you don't get around thinking about impedance matching – especially since things like reflections can lead to standing waves and dangerously high voltages >> 100 V in places you don't expect.

So, first of all, if all you need is an oscillator: Maybe build a high-power oscillator instead of amplifying a small one. That's quite possibly easier! There's a lot of oscillator circuits out there, and you can adjust a lot of them using some control voltage. You can build a simple control loop that compares the frequency (e.g. using a frequency counter from a microcontroller) to the should-be value and adjusts the control voltage accordingly.

Other than that: overpowered high-frequency resonant devices are the very classical domain of amateur radio designs. You can get power amplifiers for the 300 kHz - 3 MHz region relatively cheaply, even as military surplus equipment.

But: the easiest way is pretty simple: Build a transformer! You don't need amplification, i.e. an increase of power, you just want more voltage – if you can deal with a proportionate loss in current, then a simple 100:1 signal transformer will do your job; 1 MHz isn't hard for these, but 100 Vpp might be. Datasheet reading, it is.


A ferrite-core transformer might be a good choice if you don't need much output current.

A material such as TDK/Epcos N30 should be suitable. Eg. B65611W0000R030.

Designing the transformer is a bit much for this kind of answer, but I would roughly guess 25:250 turns so very easily hand wound. Do go through a proper design process before ordering cores and so on if you want to try this- Colonel Wm. T. McLyman's book is a good reference, and there are others I'm sure you can find in any engineering library.

You will likely need a power op-amp that can supply 50mA or more, and had a high slew rate as well as GBW product.


Can we provide an amplifier with small-signal transistors? using some cascading to handle the "high voltage"? [this use of FETs as high-voltage-shields, where "high" may mean 10 volts, is an old trick in IC design, where the "shield" device need be no more than another polygate stripe + the onchip biasing for that strip. The gate-as-capacitor of MOSFETs is what permits this.]

What is this topology? Fundamentally, it is a current feedback opamp, single-sided. With voltage shields added. Note Q2 not only shields, but minimizes the input capacitance, thus minimally loads the signal source.

Assume you need to drive 20picoFarads at 35 volts RMS (100 vpp). The power is F * C * Vrms^2, = 1MHz * 20e-12 * 35^2

Power = 1e+6 20e-12 * 1,000 (approx.)

Power = 0.02 watts

So lets consider this:


simulate this circuit – Schematic created using CircuitLab

Is this circuit to be slewrate limited? Assume we need to be 4X faster, to avoid distortion even in this feedback-controlled circuit. For 1MHz at 50 volts peak, the slewrate is 1MHz * 6.28 * 50 = 316 volts/second. Will our output node support that?

What is the load capacity? Assume 10pF in each of the 3 output devices (Cob of the PNP, Cgatebulk of the PFETs), plus 20pF Cload, for total of 50 picoFarad.

To slew 1 volt in 1 nanosecond across 1pF, we need 1milliAmp current. We are 3X slower, at 0.316 volts/nanosecond, but we need to drive 50pF. Thus we need 15ma (and we have not included the 4X no-distortion safety factor).

What is our pulldown current? 50 volts/47Kohm or 1 mA. Thus the low-going slewrate will not support 1MHz at 100 voltsPP. The high-going signal will be clean, what with the active pullup of the 2N3904. But the low-going shape will be badly distorted.

Can we simply add a constant pulldown of 20 mA? This in addition to the 47K pulldown? The power dissipation soars, from 4mA * 100volts ( 0.4 watts), to 24mA * 100 volts, with that extra 2 watts dissipated in the new pulldown (a current source?) and in the 3 existing pullup devices (Q1, M1,M2). All of a sudden we have the need to control temperatures, to practice heat-removal, possibly with moving air.

You might ask how audiophiles achieve high performance power amplifiers.


It can be done easily using a square waveform using RF MOSFETs, Ixys have very good ones that works up to 1000 V and 20 MHz or more.

ST has this one not as fast but can work at 250 V and 17 V with <20 ns rise/fall times: STD18NF25

You should use an isolated half bridge driver with floating power supply at the high side.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.