Help - boost circuit problem for 360v photoflash

[AleXYZ]

I have three commercially made 400watt portable strobes, all of which exhibit the same problem. They run on six NiCd cells (7.2v) and charge a photoflash capacitor to 360v. Normally it takes about 6-8 seconds to charge from 0v to 360v. However, at one point the output was shorted and part of the boost circuitry fried. The cap itself is OK.

I mapped out the entire circuit, but for simplicity I have cut down the attached schematic to the boost section only.

I have replaced both of the high power transistors (Q1 and Q2) with near equivalent substitutes from Digikey. The original part numbers and what I substituted are both on the diagram. The 30A transistors (Q1) were definitely fried, on at least one of the strobes Q2 was also bad. I eventually replaced D1, D2 and C1 as well.

When powered up, the strobe charges the photoflash cap to only about 8 volts (or 72 volts for one of the strobes), then it looks like the two transistors go into steady state (non-saturation) and stop oscillating. If the cap discharges about half a volt, the circuit kicks on again until the voltage reaches 8 volts. Far short of 360v !!

Fortunately I have a working strobe of the same model. A scope trace of the working strobe on the emitter of Q2 looks like a square wave with a rounded top, maybe about a 15% duty cycle. When I put the scope on a nonworking strobe (same place) it looks a lot like a sawtooth wave, sharp dropoff and an almost linear rise.

I'm mostly a digital elec. guy, my knowledge of reactive circuits like this is very limited. I'm at my wits' end! Any ideas? Suggestions? I can post pictures of the scope outputs if that will help.

 

[FvM]

Doesn't look like a text book converter circuit. An important point that's missing in your schematic is the winding polarity, but if
I understand right, it's a forward converter with current feedback from the secondary. It's not obvious at first sight, why
the circuit operation would fail. If the current feedback assumption is correct, at too low transistor current gain may be a
reason, but there could be also a defect of another part.

 

[AleXYZ]

Thanks for the quick followup. I considered low current to be a possibility, but the differences between a working and non-working strobe seem to say differently. I am using a high-grade power supply set at 7.2v and 0.50A current limit for these tests. The "good" strobe has no problem charging all the way to 360v even at the low current setting... though it takes quite a while. (The battery packs are 10,000 mAH NiMH and who knows what the actual current draw is!)

I believe the transformer polarity is basically how I've drawn it. Today I removed the transformer and measured it using very low AC from a signal generator and measured the primary and secondary windings with the scope... which is how I got the 50:1 approximation.

It is of course possible I sourced the wrong replacement transistors, but I was pretty careful to check ALL of the specs on both original and new parts before I ordered them. As a side note, Q1 is actually 2 pcs of this transistor in parallel.

For sake of clarity, I went ahead and took pictures of scope traces as I described in the initial post. The "good" strobe has a regular active cycle, with time between each rise shortening as the cap charges.

The waveform period on the "bad" strobe actually gets shorter as the cap charges to 8 volts. Then eventually it stops oscillating with the scope showing 7.2 volts flatline.


Does this make any sense?

 

[bongkk]

Your 2SA1962 substitute have a bandwidth in the AF whereas the 2SB904 works up to 120MHz. That could explain why the 2SA1962 saturates as the voltage on the cap rises to a certain voltage. For this type of inverter configuration the freq. increases as the cap charges up.

 

[AleXYZ]

Bandwidth isn't likely to be a factor unless somehow the bandwidth difference were to affect overall slew rate. The oscillation of a working strobe is in the audible range, and measured on a scope only goes up to about 10 kHz. Nowhere near 120 mHz.

After poking around a few more hours I finally found the transformer to be faulty -- maybe shorted winding on the primary. The transformer shows evidence of heat damage. And when connected to a tiny square wave signal from a signal generator, the scope trace is significantly attenuated and different (no classic inductor ringing) as compared to a working transformer.

So... the primary winding appears to be melted or maybe everything was sealed with epoxy. I tried removing the primary to replace it but it is pretty stuck. I don't think I can rebuild it, does anyone know where to buy transformer cores so I can make a new one?

The core measures about 32mm square, 11mm thick. Already checked at Digikey and Mouser, they don't sell basic transformer parts. I can't find anyone who does.