# How do I get high voltage (200-500 V) out of a DAC?

I am working on a project in which I need to amplify a 0-3.3 V DAC output to 0-500 V (or to 0-200 V if 500 V iss too much to ask) and I need to draw up to 100 mA current out of this output. I also need to mention that my DAC output signal will probably have a maximum frequency of 10 kHz.

I have never worked with high voltages like these, compared to digital levels, so I'm open to any suggestions, methods, parts, etc.

• DAC outputs a sine wave at 10 kHz or ... DAC samples every 100 us? Oct 8, 2022 at 6:33
• I think I would find a way to use some type of programmable bench supply. Oct 8, 2022 at 8:42
• Does the output need to produce a DC level as well as an AC signal <-- full details please. Oct 8, 2022 at 9:07
• drawing 100 mA from a 200 V source means you have 20 W output power. This implies a heavy load, and probably means you need to think about protection against reactive properties of the load. What is your load? This, as is, a relative harsh requirement, and the more specifically you can tailor the circuit to the load, the easier it might get! Oct 8, 2022 at 13:08
• @mkeith Even easier, I know Stanford Research makes some high-voltage power supplies that can be controlled with an analog signal--which basically makes them high-voltage, low-frequency amplifiers. Oct 8, 2022 at 14:34

SAFETY first ... High Voltage ...
Be aware that High Voltage BJT is not very usual. (to be replaced by MOSFET ...)

Here is an example of a such simple "analog" circuit (current limited).
NB: Must be checked deeply (All Voltages, Currents, Powers).
The labelled transistors shown here are NOT suitable ...
Tried with TIP47, should be ok. I added a BJT for current protection.
Here, the link to the new file.

Here is an example with MOSFETs.
Under the same conditions (all checking) as above ...
Voltage and current versus Voltage control (parameter Load: shorted -> 2k).

Powers dissipated by MOSFETs.

With some adjustments, TRAN analysis ...
NB: simultaneous stepping on Voltage control and Load R6 (working at same max current).
This shows different "slew-rate" of rising voltages.
NB: using BUZ41 MOSFET is 3 times faster. Power not checked.

• If this is for an experimental project where you can use salvaged parts: High voltage BJTs are quite common in cheaper, older SMPS and in most anything with a CRT in it. For an extra quirky solution, use a vacuum tube :) Be aware that even high voltage BJTs can NOT stand emitter voltage going significantly above base voltage (which can easily happen with capacitive loads).... Oct 9, 2022 at 12:00
• Which SPICE simulator have you used in your simulations above? Thank you! Oct 14, 2022 at 10:42
• @jose.angel.jimenez FREE microcap v12, I add in answer. Oct 14, 2022 at 13:07

A 50 watt DC coupled audio amplifier would do the job, although 200-500 volts would be rather special.

Apex makes some high power amplifier modules that might be just what you need:

https://www.apexanalog.com/products/linear_selector.html

• Ordinary audio amplifier feeding a vacuum tube audio output transformer in reverse. Done that once. Oct 8, 2022 at 18:50
• I had considered that, but the output of a DAC is usually DC, so a transformer would not work. I also considered using a tube amplifier, without the output transformer, which would likely have a B+ of at least 200 volts. Oct 8, 2022 at 19:51
• @PStechPaul DC vs AC is only a simplification of how to analyze circuits. A DAC outputs voltages at points in time. If those voltages are modulated as a sign wave over time, then the output of the DAC is better considered as an AC signal than a DC signal. (Theres also a matter of DC bias, but if you use a bipolar DAC, that can be ignored). This is a standard way of analyzing audio circuitry. Oct 9, 2022 at 2:23
• As an example of this, most PC sound cards use capacitors in series on all their line outputs after the DACs. Yet a capacitor in series blocks DC signals.This is because any change in voltage over time can be considered a transient superposition of AC signals, and so the capacitor conducts the change in voltages. Oct 9, 2022 at 2:27

What you need is an amplifier. It does not have to be a particularly good amplifier, as you can add negative feedback to make it accurate at the low frequencies.

Because your power level is quite high (50W), Class D amplifier would be a good choice, for its efficiency and low amount of waste heat. It works similarly to switching mode DC-DC converters.

Easiest way to make a class D amplifier is to take a 500 VDC supply voltage, a MOSFET half bridge rated to that voltage and an output filter made of inductors and capacitors. Instead of a DAC, most straight forward way would be to use a microcontroller PWM pin to drive the H-bridge. The duty cycle determines the output voltage, and you will want the PWM frequency to be about 10x the analog bandwidth you want at the output.

If you only have an analog input signal, a sigma-delta modulator IC can be used to convert it to PDM signal. If the output noise level is important, PDM can give less noise than PWM in a microcontroller implementation also.

After the half bridge, you have a square wave that alternates between 0 and 500 VDC, having the average level of your target signal. To filter out the high frequency square wave components, you can use 1 to 3 stages of LC lowpass filters.

• A class D amplifier will certainly be more efficient, but unless one can be found off-the-shelf, it is a rather major circuit design project. Using a high voltage power op-amp (class B) for 50 watts output is a simple solution, and should be able to achieve at least 70% efficiency, so it is only necessary to dissipate about 15-20 watts of heat. elprocus.com/class-b-amplifier Oct 9, 2022 at 21:50
• @PStechPaul That 70% efficiency is true only for sine wave at maximum peak voltage. For e.g. DC output at 5 VDC & 100 mA, the efficiency of a class B amplifier is only 1%, with 49.5 W dissipated.
– jpa
Oct 10, 2022 at 5:06
• Yes, that is true. But the OP indicated that the DAC would be providing an AC signal, and presumably at or near the voltages required (200 to 500 volts). If a lower voltage is required, the supply rails might be made adjustable. In any case, the maximum power dissipated by the linear op-amp would be 50 watts, easily handled by an adequate heat sink and fan cooling. Unfortunately, the OP @spinmaster has not provided additional feedback and the topic is being closed. Oct 10, 2022 at 8:18
• @PStechPaul I agree that there are a lot of details missing in the question. And it is true that 50W is easy enough to cool with a fan.
– jpa
Oct 10, 2022 at 8:31
• But the OP implies that they want to use a 0-3.3 volt DAC signal to get 0-200 or 0-500 volts, which would require DC coupling. However, the OP has not replied, so details are missing. Oct 11, 2022 at 6:35