Analog challenge exercise

[T3sl4co1l]

I haven't seen a blocking oscillator since the last time I took apart a VCR, I think.  (Aside from the ones I've made myself.)  They were strangely popular in consumer AV equipment for... who knows why.  Just one of those domain-specific memes, I would guess.  Also, somewhat more famously(?) the Apple II, wasn't it?

Often the standby/aux supply in ATX PSUs was one, also.

They've almost kinda made a return, but typically with the help of a little SOT-23-6 controller which does a better job.  Of Taiwanese or Chinese origin, the price is almost negligible; I wonder if parts count -- placement cost -- motivates the update?

Apparently there's versions for MOSFET and BJT flyback, and self-excited half bridge (a recent EEVblog video showed a resonant LED supply with such a beast).  You can't avoid the cost of a separate drive transformer in the latter case, but perhaps that's justified by the performance advantage of the architecture (which apparently was also implementing PFC somehow!).

Tim

[D Straney]

I love these sorts of challenges.

You know what a good source of blocking oscillators is?  Lightbulbs.  I only know this because I spent a good deal of time at a previous job doing some "competitor analysis" and reverse-engineering a bunch of cheap-ish but reasonably-quality LED bulbs (and CFLs).  [This also means these circuits are too instantly familiar...so I've got to come up with something completely different]  Whether buck, buck-boost, or isolated flyback, they mostly converged on the same 2-transistor design: a single power transistor driven by a feedback winding (like Yansi's circuit), but with an extra "peak current control" transistor, with its base on a current sense resistor in series with the power transistor's emitter, which pulls down hard on the power transistor's base once the current ramps up high enough to turn this extra transistor on.  Much more elegant than the wild variability of the "turn-off by running out of beta" blocking oscillators, or the lossiness of the "turn-off by saturating the power inductor" ones.  There were some minor variations, including 3-component primary-side output over-voltage protection / output voltage regulation: a diode from the feedback winding would conduct in the "inductor discharge" phase, and put its scaled version of the output voltage onto a series resistor and zener diode, which led to the base of that "peak current turn-off" transistor.  If the reflected output voltage was high enough to reach (turn-off transistor Vbe + zener voltage + feedback winding diode Vf), then it would start turning off the power transistor early.

These descriptions are kinda terrible without diagrams honestly.  Maybe I'll draw some up later.

Interestingly enough, one of the LED bulbs I opened used a classic CFL circuit (self-oscillating half-bridge, with a defined-saturation-point gate drive xfmr, driving an inductor)...but which fed directly into a bridge rectifier and the output cap.  Not at all the optimal solution for a consumer product as it had at least 3x the components of one of the single-transistor blocking oscillators.  Must've had one engineer, who knew CFL drivers and only CFL drivers, who heard "DC output" and thought "oh, I can just take this normally-AC output from one of my circuits and rectify it".

[T3sl4co1l]

No clue, the photos are too poor to tell if it's a 3-terminal switching regulator or what.

Tim

[T3sl4co1l]

Interestingly enough, one of the LED bulbs I opened used a classic CFL circuit (self-oscillating half-bridge, with a defined-saturation-point gate drive xfmr, driving an inductor)...but which fed directly into a bridge rectifier and the output cap.  Not at all the optimal solution for a consumer product as it had at least 3x the components of one of the single-transistor blocking oscillators.  Must've had one engineer, who knew CFL drivers and only CFL drivers, who heard "DC output" and thought "oh, I can just take this normally-AC output from one of my circuits and rectify it".

Even more interestingly enough(?!?), Sony used a more refined version for a long time, especially in Trinitron monitors.  Resonant half bridge oscillator with saturable drive transformer.  By biasing the transformer, it saturated early, increasing the switching frequency away from resonance, reducing output voltage.  Control is simply a TL431 feeding the control winding, which is wound cross to the feedback windings on a 4-leg core.

Tim

[Mazo]

My 20min go at it.
20-something part count,it sucks in certain aspects mainly the DC accuracy of the error amp but a TL431 will put the transistor count way up