How to stop blowing TIP122s?

Thread Starter


Joined May 9, 2009
I am modifying a plasma ball circuit, to generate an arc of 10-15 kV (at very low current - barely tingles when you touch it.) The goal is to replicate the first radio transmitters invented - spark gap transmitters.

The circuit is very simple. A two transistor oscillator generates a square wave. This controls a TIP122 100V 5A Darlington which is on a heatsink. The darlington drives the primary of the transformer and a secondary has many turns. The result is about ~2kV 20kHz AC at a very low current which drives a standard plasma ball and makes it light up.

The problem is, 2kV isn't enough - I needed a bit more, around 10-15 kV, to get a big enough arc. So I hooked it up to my power supply and ran it off 12V instead of the 4xAA (6V) it uses. That worked fine and gave me enough output to be received on the other end using an AM radio about 2 metres away.

My power supply was current limited though; the circuit was pulling about 1.5A and the supply sagged to 9V (I had the current limit set at 1.5A - the supply is a CC/CV type.) I didn't make much of this because the transistor can handle 5A.

Then I plugged it into a 12.6V LiPoly battery, and it worked great - for about 2 seconds. The TIP122 then went up in smoke. I have replacements, I planned for this to happen, but obviously I don't want this to happen all the time.

So, the question is - how do I replicate the current limiting the power supply provided me with? Because that worked well and the TIP122 lasted forever using this method. I could build my own current limited supply powered from the battery, but that's cumbersome and the widget needs to be small. I was trying to think of a way to use resistors, but I'd require a 10-20W resistor and that would make the battery life awful and also require serious heatsinking.

Any ideas?


Joined Oct 26, 2011
There are a good many issues here:
What is the turns ratio?
What is that primary leakage inductance?
Snubber circuit on primary?
Does the transistor have adequate drive to saturate properly?
Adequate heatsink?
Exceeding sustaining current?
Safe operating area (SOA) issue?

Please provide sketch
Do you have a scope that can catch the turn-off transient?


Joined Feb 19, 2009
Did you change the base & collector resistor values with the change in supply voltage?

Are you at max gain on the transistor?

Are they real TIP122 or chinese fakes? Post a pitcher and that can be determined (though I think you know how to spot a fake).

I'm guessing 1 or 3.

Thread Starter


Joined May 9, 2009
Well.... I wonder if they are fakes...

I pulled one apart (not too tricky with two plyers.) And the die is *tiny*... in fact it's only slightly bigger than one from a TIP31 which I believe to be genuine.

Anyway, the circuit reached the point of being nearly impossible to fix, so I built one on a breadboard using a 555 timer. Now I'm not blowing TIP's any more, but I'm also getting a much reduced output. I suspect I need to operate it close to 35 kHz - like the original design did - and at ~50% duty. I'm at 75% duty @ 2kHz.

Thread Starter


Joined May 9, 2009
Okay, I've got the circuit working, but the TIP gets very hot, very quickly... I scoped the base waveform and it's turning on fast (<200ns) but it's taking ages to turn off :confused:.

I've heard Darlingtons have problems with turning off fast - so is there any way to work around it?

I've got a 220R base resistor (very low) and a 47n cap in parallel - the cap is my attempt to improve the charge/discharge and it does work, but it doesn't seem to be enough.

Will post a waveform later. Have found my new analog scope to be better for this kind of stuff compared to my old digital one.

I'm thinking this circuit might work better with an N-channel MOSFET. Unfortunately, I don't have any - only old P-channel devices with really high on-state resistance (0.2 ohms - say hello to 0.0012 ohms...)

Also is there going to be a preferred oscillation frequency for these transformers? I've heard plasma balls use 35 kHz so I've got mine close (38 kHz) but it doesn't seem much better.


Joined Mar 24, 2008
You could build a simple sweep circuit (with no power) to look for resonant frequencies. Just a thought.


Joined Jul 17, 2007
The TIP122, being a Darlington, is going to have a high Vce(sat). That's going to cause a good bit of power dissipation in it when you're getting lots of current flowing through the primary winding. And yes, Darlingtons turn off pretty slowly.

You could do something like use a high-current non-Darlington for the current sink, and source current to its' base using an NPN voltage follower with a current limiting resistor from the supply to the collector. That will drop your saturation voltage very significantly.

Be sure to use a base return resistor for the power transistor. Also be certain the Vce rating is high enough.

You will probably need to add a snubber circuit, or the transistor will get zapped by the spike when the current gets cut off. The more quickly the transistor cuts off, the more quickly the spike will build, and the higher it'll go.

Thread Starter


Joined May 9, 2009
Good point on the snubber, I'll add that next time I work on the circuit.

Also, I just ordered myself some IRFZ44N's, which looked like suitable fet. 55V, 18.5mOhm on resistance.

The lower voltage will probably mean a snubber is necessary; the original design probably relied on the fact that the 100V rating was enough. (It operated from 6V.)

Will the 555 timer make a good gate drive? It's got +/-200mA sink/source, so I think it's going to be good enough to drive the gate.