So this topology seems to works OK for something made out of a bucks worth of junk box parts, it charges a big cap up to 250 volts or so in a couple seconds:
But, bad ringing on the square wave edges. How to reduce the ringing?
Heh prototyping (please ignore superfluous parts):
386's ??? >:-} ...Jim Thompson
Sure, why not? Decent bandwidth, can put out some power, basically free...
Maybe diodes across the output devices to snub the ringing? The efficiency sucks anyway...
cap across the coil..
Jamie
You have some caps across the power supply to the driver chips? Maybe an RC across the transformer primary. Probably won't take a very large cap, maybe 100 pF and 100 Ohms in series, mostly to round off the edges a bit.
Jon
Use something cleaner than an audio transformer, and stronger than an LM386.
You were warned, yet proceeded regardless. (And now you've learned something. :) )
As for what's going on, I would presume it's complex RLC behavior in the secondary, reflected to the primary. At the first cutoff frequency, this looks like shunt capacitance, but at higher frequencies, myriad peaks and dips simply reflect a lower impedance (or actually a higher impedance, because it acts to reduce the turns ratio -- a neat effect, if you needed it).
And I would assume these wrinkles in the impedance response act against the LM386(s), which will have a modest (probably inductive) output impedance at those frequencies, thus allowing the voltage to ring like that. Correspondingly, terminating it into a suitable impedance (probably by an R+C across the primary) should help. You'll inevitably lose some efficiency in the process, of course (the C isn't infinite Z at the fundamental or harmonics, so the R inevitably loads the main waveform a little).
NTE chip? Big spender!
Tim
Warnings are for the weak...
Ah, like a Zobel network? Makes sense.
It was free. It's fo' me.
Hey, I bet if I flipped the transformer around it would drive a speaker pretty well!
The LM386 was made to work just well enough to drive a pocket radio's speaker from 9V. I'm surprised it can drive a transformer at all.
If you have some TO-92 transistors lying around in your junk drawer:
You might need emitter-base resistors on the output transistors. Probably not, but you never know what's in the junk drawer.
You also can't use a normal bridged rectifier at 20 KHz, or even 1KHz. The diodes are intentionally designed to be incredibly slow so that they don't generate AC line noise. You'll need four diodes made for switching power supplies.
Why not? If it can drive an 8 ohm speaker to a few hundred mW, I don't see why they can't drive a transformer with a higher reflected impedance. Since it's driving an unloaded cap on the other side the reflected impedance is actually theoretically infinite for much of the cycle, if I'm not wrong. That's likey part of the ringing problem...the reflected impedance is varying all over the the place in a complex way as Tim Williams mentioned.
They're good for driving small servos too - that application is listed right in the datasheet.
But the '386 already has a class B power stage internal to the chip. If I were going to use external transistors to drive it, I wouldn't even use a power amp at all. Just the external transistors and a regular op amp for feedback gain.
The bridge I'm using (DG105 or something) seems to be switching fast enough at 20kHz.
You asked.
The LM386 has an output impedance of 5 to 7 Ohms. Negative feedback corrects for that only at low output levels. Reduce the output to the normal operating range of 2 VAC and you'll see the signal clean up.
A line AC bridged rectifier will produce DC from 20KHz but it's going to momentarily short at each polarity change.
So - you have a little bit of natural ringing, momentary shorts from slow diodes, and a high impedance driver. That's why it rings. Ignore it or get better parts. Snubbing the ringing will only make it run hotter.
Have something to add? Share your thoughts — no account required.
Ask the community — no account required