minikv — regulated kilovolt power supply

Note that this project is not released yet! The text below is preliminary. Any offers to give you parts are lies. Any claims that it works properly are fraud.

90s-ass under construction gif

minikv is a small power supply for projects using around 5mA from 600V to 1200V. This makes it ideal for powering those tiny soviet crts.

The outer dimensions are 35 × 39 × 10 mm. Eventually, it will be integrated into a 12V-powered CRT module kit.

table of contents

be careful

minikv creates enough voltage with enough current behind it to kill you. This is NOT a project for people who saw some funny slapstick on electroboom and want to recreate it. Touch it at your peril. I'm not going to explain high voltage safety here — if you don't already know, this isn't the place to start. Sorry.

how it works

primary side

minikv uses a resonant Royer oscillator, also known as a Baxandall converter. C5 in parallel with the transformer forms a resonant tank, and then Q1 and Q2 drive it in push-pull mode, with a feedback winding providing the base drive. L1 stabilizes the current, keeping the waveforms sinusoidal.

An aside about windings: you may notice that the feedback winding is a half turn. This is created by wrapping halfway around the core, then returning around the outside of the core where the magnetic flux is minimal. This is done to reduce the maximum base voltage below the maximum negative Vbe for Q1 and Q2. Half turns often provide poor coupling, but for the base drive this is acceptable, or even desirable. Now, you might wonder, if half turns provide poor coupling, why do the primaries each have a half turn? The answer is that they don't, really. The primary is not truly two 2½-turn windings, it is a 5-turn winding with a center tap. It wraps nice and tight around the core.

The Royer oscillator is powered by U1, a switchmode buck converter. A feedback network consisting of R1, R2, R3, and R5 divides the output voltage down to a reference point of 1.24V, with R5 allowing a trim voltage to be injected to select the divide ratio. The final feedback voltage is compared against U1's internal 0.8V reference to regulate the output.

C3 and R4 provide some compensatory feedback of the output voltags into the feedback node bypassing R2, canceling phase shift around the control loop that would otherwise cause control oscillation. These are placed around the additional feedback resistor R2 rather than across R1 per convention, because this protects the frequency compensation components from the high output voltage.

secondary side

At full drive, each half primary winding sees about 42V peak-to-peak (Vpp). This is multiplied by the turns ratio of 48, giving a final output of 2016Vpp. Losses reduce this voltage substantially under load. A voltage doubling rectifier converts this to DC.

how to get it

I am not currently offering this as a purchaseable kit, but it is easy to assemble one yourself. All design files are provided here. You will need the two PCBs, a vulcanized fiber ("fishpaper") washer, and all parts in the bill of materials. PCBs should be made at a thickness of 1.0mm to fit in the transformer.

While a kit is not available, I probably have leftover PCBs, and I have lots of fishpaper as well. Just send me an email (alexisvl at this domain).

how to build it

TODO: should I make a video of this?

building the pcbs

  1. Solder surface-mount parts onto PCBs following the interactive assembly diagram (above, TODO). Be gentle when holding the large ceramic capacitors in your tweezers, as they are brittle and very easily chipped.

  2. The through-hole header connector should be soldered onto the primary winding board as shown (TODO), but use a very small amount of solder and deliberately avoid creating raised solder bumps. The secondary winding boards will be stacked on top with the pins going through all boards, and solder bumps will spread the boards apart.

applying protective coating

Both outer boards require protective coating applied, as the miniaturization required exceeding minimum open-air clearances for these voltages. In the aforementioned assembly diagrams, areas are marked where coating should be applied. These are minimum areas — feel free to spread it over a larger area if it's easier.

You can use almost any coating that is not hygroscopic. Hardware store polyurethane will do fine, as will many clear enamels. I use MG Chemicals 4226 "Super Corona Dope", but you probably don't have this, and it has some unpleasant chemicals.

making the insulating washer

An insulating washer is required between the PCBs.

  1. Get some vulcanized fiber ("fishpaper"). See how to get it as I may be able to give you some.
  2. Print out the insulating washer drawing, from the "Fishpaper" layer in the primary winding board layout.
  3. Lay this template over the fishpaper, and cut on the line with a craft knife.
  4. Fold it as shown (TODO).

assembling the boards

NOTE: You can test the circuit before final assembly with the secondary windings omitted, so that it cannot emit high voltage. See the test instructions first.

TODO: get pics of every step

  1. Place the "E"-shaped ferrite piece on your desk with the wings pointing UP.
  2. Place the primary board on your desk with the "B" (bottom) marker facing down and the ferrite wings through the cutouts.
  3. Place the fishpaper washer on top, with the folded wings pointing UP.
  4. Set the middle board ("lower secondary") on top of this, with the "B" marker facing down. Press it down onto the header pins if necessary.
  5. Fold the fishpaper wings out onto this board.
  6. Set the top board ("upper secondary") on top of this, again with the "B" marker facing down. Press it down onto the header pins as well. The fishpaper wings should be holding it slightly off the middle board.

installing the core

  1. Place the "I"-shaped (flat) ferrite piece over the top of everything, with the debossed channel facing UP.
  2. Install the clamps through the board cutouts and into the debossed channel on the core to hold it together.

NOTE: these clamps have been poorly stocked lately. If you can't find them, wrapping some electrical tape around the whole stack tightly to squeeze it together can work as an interim solution to get it tested, but don't depend on it to stay like that permanently.

TODO: picture

how to test it

before final assembly

You can test the circuit before final assembly with the secondary windings omitted, so that it cannot emit high voltage. This allows troubleshooting the majority of the build without risking electric shock. To do this, follow instructions for installing the core but install it over just the primary board, as shown (TODO). Plug the pigtail cable (TODO - BOM) into the header, and apply power to the 12V and GND pins (current limit should be set to around 250mA).

Using an oscilloscope probe, check that the circuit is oscillating between 50 kHz and 150 kHz. To feel like a lazy wizard, simply clip the probe ground onto the probe tip and hold it over the top of the transformer core — it will pick up the field like magic! Next, verify the feedback mechanism by shorting from pin 1 to pin 2 of the header (+12V to TRIM), and confirm that oscillation ceases and current draw drops to around 1mA or less.

how to adjust it

The output voltage is programmed via the TRIM pin. When this pin is floating, the output voltage is 1102V and the TRIM pin biases to 0.8V. Pulling it lower will raise the output voltage further; pulling it higher will lower the output voltage.

All calculations of output and trim voltages revolve around the following Kirchhoff's Current Law equalities:

Given a supply voltage and an external resistor:

    Vcc - Vref     Vref     Vout - Vref
    ~~~~~~~~~~  -  ~~~~  +  ~~~~~~~~~~~   =  0
    Rext + R5       R3        R1 + R2

Given a voltage directly applied (e.g. from a DAC):

    Vtrim - Vref     Vref     Vout - Vref
    ~~~~~~~~~~~~  -  ~~~~  +  ~~~~~~~~~~~   =  0
         R5           R3        R1 + R2


    R1   = 10000000 ohms
    R2   =   330000 ohms
    R3   =     7500 ohms
    R5   =    68000 ohms
    Rext = external resistor from TRIM to power or ground, to be calculated

    Vref  = 0.8 volts (feedback voltage of U1)
    Vcc   = the voltage to which Rext pulls
    Vrtim = voltage applied directly to TRIM
    Vout  = final output voltage

with a fixed resistor

If you want a fixed output voltage, the easiest way to get it is to connect a fixed resistor between TRIM and either GND (for voltages above 1102V) or +12V (for voltages below 1102V).

The following calculator can solve these equations, try entering values in any of the fields.


with a voltage

Here are the equations above, solved and simplified to give Vout as a function of Vtrim and vice versa:

             (R1 + R2) R3 Vtrim - ((R1+R2+R3) R5 + (R1+R2) R3) Vref
    Vout = - ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
                                    R3 R5

    Vout  = 1224.2 - 151.912 Vtrim

    Vtrim = 1224.2/151.912 - 1/151.912 Vout
          = 8.05861 - 0.0065828 Vout

A 5V DAC can cover the entire range, and a 3.3V DAC can cover everything north of 750V.

how to change it

minikv is public domain because intellectual property is fake. Do whatever you want with it, i can't stop you.

Last modified: Mon Nov 7 20:49:34 MST 2022. Converted from projects/minikv/
RSS feedBlogHomeUpIndex