I need to design a boost converter that outputs ~20 mA DC and an unloaded voltage of around 68-100 V. The inverter will be part of a Zener diode analyzer.

I built it on the idea of the Joule thief (blocking oscillator). During breadboard experiments at some point the base biasing diode burned.

I suspect it is due to transients from the base winding. Since then I modified the design to try and suppress the transients and not exceed the Veb voltage of the BD139 (is -5 V).

Also the shut-off transistor(BC548) was getting hot. From simulation it looks like it's also from transients(high voltage applied to collector).

My schematic values are different from the actual ones and component annotations.

  • T1-main switch-----BD139-16
  • R1-base bias resistor----10k--->~1mA base current for the transistor
  • D1-base bias diode ----1N4148--->Vf~=0.65V
  • C1-filter cap for the bias network(release some voltage stress from D1)
  • R2-main switch base resistor(should help prevent high currents in D2,D1 and keep Vbe>-5V)----10nF
  • D2-T2 protection diode ( a path for current to flow trough R2 away from the T1 base and T2's collector)----1N4148
  • T2---switch off transistor(takes away the T1 base current when the output voltage is too high,the Zenner conducts in the reverse bias and turns on the optocoupler's led)
  • R3---T2 base current and fototransitor's current limit resistor.(since BC548 has high hfe can be a larger value ~10k)
  • D3----T2 base biasing diode--->Vbe T2~=0.65V
  • D4----T1 collector protection diode(stop the high voltage spikes)
  • D5----Output diode(any fast rectifier diode i'll use SF 28 G)


  1. Will this design will work and if not what improvements can be made?
  2. Will I need a higher base drive current if the collector inductor will have smaller values (~10-22 uH)?

I'm intrested about other other designs too, but my main focus is to keep the inductor's physical size low.

  • 1
    \$\begingroup\$ I'll be honest: don't try to base a controlled high-voltage generator on a really minimal, simplistic example circuit such as the Joule Thief. Also, without an actual schematic, it's very hard to imagine what your circuit looks like. Please use the built-in schematic editor (yes! this website has a schematic editor in the question editor!) to draw exactly your circuit. \$\endgroup\$ Commented Jan 12, 2019 at 16:29
  • \$\begingroup\$ However, 20 mA at 100 V is still 2W. I'm pretty sure that you can do better than a 1970's BJT like the BD139 for that. And honestly, since this is for a measurement device, you want all this to be properly regulated. Everything in here screams "Use an existing SMPS controller IC; it will be cheaper, lower in component count, safer, and better!". Also, albeit you forgot to mention your source voltage, 100 V definitively sounds more like flyback than unisolated blocking inverter would be the right architecture. \$\endgroup\$ Commented Jan 12, 2019 at 16:33
  • \$\begingroup\$ smaller inductors are reached through higher switching speeds. Since all the semiconductors you use are really old, they wouldn't allow for MHzes of switching speed. So, really, nope. Get a flyback DC/DC converter IC and start with the schematic in the datasheet of that. \$\endgroup\$ Commented Jan 12, 2019 at 16:34
  • \$\begingroup\$ Hi,sorry about the schematic.The forum's one seems to require membership.Pdf Schematic from my drive \$\endgroup\$
    – Andrei112
    Commented Jan 12, 2019 at 16:47
  • \$\begingroup\$ Thanks for the schematic! I've added it to the question itself. But no, using the built-in schematic editor doesn't need any special account. It just works with your normal Electronics.SE account :) \$\endgroup\$ Commented Jan 12, 2019 at 16:57

1 Answer 1


Your diodes and transistors are insufficiently rated for the voltages you need to block in this inverter. So, frankly, you need to beef up this thing significantly if you want to do 100V. Hell, that's 20 V above the maximum ratings for the BD139 – an absolutely wrong and obsolete choice!

Anyways, this system is really not what you want for a 100 V source:

  • unregulated/uncompensated: The voltage generated by the Joule thief was not a design target; it was just supposed to be "high enough". You're building a measurement device, so you want to know what your voltage looks like. You need a closed-loop control mechanism that takes temperature drift and component tolerances into consideration
  • low-speed switching: It's hard for me to estimate what this will oscillate at, if at all, lacking any information about the transformer. But: I doubt it'll be a high frequency. You said you want a small inductor, so you want high frequency. Replace the self-oscillation by something that forces the thing to work at a specific switching speed range.
  • Safety: well, nothing here is isolated, you're not fusing your output, so, if this can reliably deliver >20 mA, it is dangerous.
  • Transformer windings: you're not specifying, but this looks like a 1:1 transformer. That's kinda bad if you want to make large voltage jumps: a higher winding ratio would help you increase the voltage in itself!

Especially the last point highlights that this is not the switcher architecture you're looking for.

You probably want a flyback-based DC/DC converter. I'd recommend you go for something simple and robust, like good old TI LM25xx regulators.

In fact, I just went to ti.com and clicked through to the design tools and the power designer, said I had an input voltage of 10 V – 13 V and wanted 100 V at a max. current of 0.02 A, and was greeted with this:


This circuit has specified ripple, it's way more efficient that your circuit, and, considering you safe a lot of transformer cost by having a smaller one running at a higher frequency, probably even cheaper than your hand-tinkered one.

however, this was the design I've got recommended when I went for "minimum BOM count"; when you go for "minimum BOM cost", you might get something different, as you'll get something different when you go for "maximum efficiency" or "minimum space".

  • \$\begingroup\$ Thx Marcus.I think i'll use an IC then.As a fun fact i got it to work up to 130V and outputed ~30mA in short.As about the frequency it was ~20khz something....since i could hear the oscillation noise.As for the transformer i used an unknown PC PSU one......primary for collector inductor and 12v sec for the base feedback \$\endgroup\$
    – Andrei112
    Commented Jan 12, 2019 at 17:32
  • \$\begingroup\$ build a blocking oscillator on the centre tapped secondary of that transformer. take output from the primary. \$\endgroup\$ Commented Jan 13, 2019 at 3:16

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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