I am making 3V to 500V DC converter for a GM (Geiger-Müller) tube kind of application. Basically the tube needs to see 500V across it. I read this relevant thread here : 5V to 160V DC converter and I have a couple of queries:

  1. Would the LT1073 circuit be suitable for this application.What would be the maximum voltage felt by the LT1073 at the SW1 pin ? SW1 pin MAX is mentioned as 50V. Is this independent of the supply voltage?
  2. Suppose I use the common low cost MC34063 , would 3V be the absolute minimum I could go down to? Suppose I use a flyback topology instead of a boost converter,would I be able to get by using the internal switch of the MC34063 instead of an additional external switch? I suppose the external switch is needed more for the HV rather than the current drawn.
  • 1
    \$\begingroup\$ So, how much current do you need at 500 V, and for how long? \$\endgroup\$
    – Telaclavo
    May 14 '12 at 1:42
  • \$\begingroup\$ Ditto on the previous comment. The current requirement and frequency or loading characteristic should be specified. Try exploring here as well: ti.com/ww/en/analog/webench/power.shtml \$\endgroup\$
    – shimofuri
    May 14 '12 at 14:19
  • 1
    \$\begingroup\$ Hi ,I just noticed this now. I was usually looking around on the bottom of this page :-).The current needed is about 120uA. Basically the tube conducts when irradiated.The maximum times this can happen is 5000 pulses per second. \$\endgroup\$
    – EmbSysDev
    May 14 '12 at 19:48
  • \$\begingroup\$ Refereing to the above circuit from TechLib, how much V am I going to read on the output terminal using an ordinary multimeter or oscilloscope Probe? I am trying to replicate this circuit, but getting about 70V output with 9V input. As for the transformer, I can see peaks reaching 400V (14us duration) with 4ms between peaks. (250Hz i believe). Thanks. Sorry i am unable to post a comment on that particular answer. \$\endgroup\$
    – Zubair
    Jan 2 at 4:10
  • \$\begingroup\$ There was an article on this here: changpuak.ch/electronics/High_Voltage_Power_Supply_MC34063.php \$\endgroup\$
    – Letshin
    Jun 9 at 16:13

Making a 500V supply capable of a few uA is actually pretty trivial:

enter image description here
From TechLib.com

The transformer can be any generic 1:1 isolation transformer, the phone isolation transformers you can buy at radioshack work quite well.

However, this power supply is not capable of supplying any real power. It works great for a geiger-counter, but if you have a load smaller then ~\$50M\Omega\$, you will begin to overload it.


A typical conservative recommendation for boost converters is not to boost by more than a factor of 6 (six) in a single stage. It's harder to make the feedback loop stable at higher boost factors. Going from 3V to 500V is a lot more than 6x.

Flyback topology could work. I've just done a design, which had a 12V to 150V 20W flyback. Here's an EDN article that describes an HV supply: 1-kV power supply produces a continuous arc (2004). It has a flyback followed by a diode/capacitor charge-pump multiplier. LTC1871 is used in the article, but other PWM controllers designed for low side MOSFET (boost, flyback, sepic) can do this job too.

A third possibility is a push-pull converter.

If you want to buy an HV power supply module, you can go to a place like EMCO.

I read this relevant thread here: 5V to 160V DC converter and I have a couple of queries:

  1. Would the LT1073 circuit be suitable for this application.What would be the maximum voltage felt by the LT1073 at the SW1 pin? SW1 pin MAX is mentioned as 50V. Is this independent of the supply voltage?

[N.A.: I think, this question is in the context of figure D1 on page 93 of Linear Tech's app'note 47, which was originally suggested by Zebonaut in 5V to 160V DC thread].

The circuit in the app note is a combination of a boost and a diode/capacitor charge-pump voltage doubler. The output is of the boost stage is half of the total (give or take a few 0.7V diode drops). Both stages are controlled by a single outer control loop. In the original figure, combined output is 90V, so the output of the boost stage is around 45V. SW1 sees the voltage within it's rating.

Zebonauts post was suggesting to change the feedback resistors so that the combined output is 160V. In that case SW1 would see 80V.
+1 to the O.P. for noticing the voltage limit on SW1.

Another way of increasing teh output voltage of the aforementioned LT1073 circuit is to add more voltage multiplier stages. Each stage can add up to 50V the output a voltage (equal to the output voltage of the boost stage).


A circuit to provide 500 Vol output from a few Volts DC will usually use an output transformer. You could achieve this with a single stage boost converter but dealing with stray capacitance (which tends to limit peak voltage achieved) becomes difficult and if things 'gang aglae' and the 500V gets into the input circuitry they will gang very aglae indeed.

The <= 220 VDC ouput Nixie tube power supply that I referred to in my '160V question' answer IS capable of extension to 500V BUT it was already layout dependant and the author recommended following his design & PCB. extending it to 500V would be substantially harder as energy storage in capacitors increases as V^2 so as (500/200)^2 =~ 6:1 layout becomes much more critical.

Adding a secondary winding as in the EDN 1 kV converter {see Accompanying article here } or with a MC34063 using eg figure 25 page 17 in the data sheet

Below is an "indicative only" somewhat modified version of the EDN 1 kV supply to show something that would work. See article above for details. I've removed the output current protection FET (and left the unused components in place) and removed the voltage tripler.

enter image description here

enter image description here

MC34063 startup voltage.

You asked

Suppose I use the common low cost MC34063 , would 3V be the absolute minimum I could go down to?

The datasheet page 7 table 8 says minimum startup voltage is 2.1 Volt **typical* with MC34063A and 1.5V typical with MC34063E.
This is limited by the oscillator star voltage and you'd want to look at output drive issues etc. If you really wanted minimum possible Vin with an MC34063 you could provide a local supply driven by its own output once it started running. You could probably run such a circuit from two cells (NimH or Alkaline or ...) with due design care.

  • \$\begingroup\$ Thanks for that.Actually I would be using a 3x1,5V AAA cell pack.There is a uC and an back lit alphanumeric LCD powered by a boost converter LM3578-5V.I am just trying to eke out the last coulomb from them batteries.Any generic component recommendations for the PNP transistor and the MOSFET ? \$\endgroup\$
    – EmbSysDev
    May 14 '12 at 9:07
  • \$\begingroup\$ Would the FET get enough Vgs at 3V especially since the diode drops another 0.7V ? Maybe it would be better to omit the PNP and drive the gate directly by the MC34063 ? I could not understand the need for the PNP,is it for discharging the FET gate voltage ? TIA !! \$\endgroup\$
    – EmbSysDev
    May 14 '12 at 9:17
  • \$\begingroup\$ @EmbSysDev - The MOSFET drive will be enough for a MOSFET chosen to suit the design :-). I'd look for a very low Vgth MOSFET - about 1V - so actual working voltage of about 2V is ample. I use the utterly superb CES2310 from Taiwanese CETSEMI for this sort of cct - but hard to get in West (a NZ supplier imports some). Other alernatives are available. BUT that is why I said that you could provide a local supply for the MC34063 etc so it runs on say self supplied 12V once it gets started. BUT good MOSFET is enough. The PNP is a pull down emitter follower to provide good turn off drive to ... \$\endgroup\$
    – Russell McMahon
    May 14 '12 at 13:44
  • \$\begingroup\$ McMahon I was trying to source CES2310,but it seems like they do not support sampling quantities even if purchased.I was wondering if it is feasible to parallel BSS138(Vgs 1.5V,Vds 20V,200mA) and get the required current capacity. \$\endgroup\$
    – EmbSysDev
    Jun 6 '12 at 11:32
  • \$\begingroup\$ @RussellMcMahon, how to make MC34063A work with 2.1V input voltage? Using diode-OR input and output voltage to Vcc will not work, since there will be a voltage drop.. I am trying to make a 2-cell Ni-MH to 5V,100mA boost converter with it. \$\endgroup\$ Nov 1 '12 at 12:42

I've not done one with that sort of boost myself, but I've seen designs of 5V to 400V converters using several stages of boost type DCDC architecture.
I understand that you have to be very careful about harmonics of the switching frequency of each stage affecting the next. Synchronising the stages helps.
You have the advantage that the GM tube takes very little current (10's to 100's of uA peak) at a high voltage, so a ladder type voltage multiplier hanging off the end of a flyback might be a better choice.


The LT1073 is a gated oscillator converter. The MC34063 is a constant period converter. Neither of these approaches build up a high voltage quickly. The duty cycle changes dramatically during the ramp from 0 to 500 V. A photo flash charger, such as


accommodates the large voltage range better. It delivers a constant energy per cycle in the shortest time possible, by detecting when the energy has been delivered. Discontinuous operation also eases the component stresses.

Flyback works well at these high voltages. Boost does not. Also the magnetics will need to be tolerant of the voltages.

Please consider safety in this design. What happens to the stored charge in the output when power is removed? What protection is used to prevent user contact with the high voltage nodes?

  • \$\begingroup\$ The voltage is used to charge a "small cap" ca. 0.1uF.The entire instrument is in a plastic box,so that no user would be able to touch the HV. \$\endgroup\$
    – EmbSysDev
    May 13 '12 at 22:06

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.