Controller-less flyback converter

[dobsonr741]

I'm puzzled by a few Cosel flyback power supplies I found today: P50E-12 and PMC50-2. They do not have any controller IC, just a simple TO92 transistor marked DI302 and the switching MOSFET.  https://en.cosel.co.jp/product/powersupply/PMC/PMC50/PMC50E-1-XULA/

Given the company is a reputable one, how can they meet the specs on the data sheets with such a simple controller? Or perhaps the DI302 is more than a transistor? It's labeled as T11, BTW.

[Whales]

Have you seen the other side of the board?  It might have SMD parts.

[dobsonr741]

Nope, nothing on the other side.

[dobsonr741]

Little googling reveled the GT5011, and patents for 3 terminal primary side controllers: https://patents.google.com/patent/US7911814B2/ko

[T3sl4co1l]

You'll have to trace more of the circuit, but it's likely a blocking oscillator.  These were very common in the AV sector for some reason; I've pulled more than a few from VCRs, etc. back in the day.  It was semi-famously the choice for the Apple II, and perhaps arguably TVs arguably used a similar circuit, though continuous-wave to also deliver horizontal deflection and high voltage to the CRT, with a sync input to allow stable video display when a signal is received.  (The free-running horizontal sweep oscillator/driver can be seen as a SMPS itself, and often the flyback transformer afforded secondary voltages not just for the CRT (HV), but the video amps, vertical sweep, etc. as well.)

Flyback isn't used much above 100W or so, but this is a 50W module in question, so it seems plausible they would still do it that way.  It seems kind of a large unit to me, but maybe that's just because of the number of outputs.

The typical two-transistor blocking oscillator resembles a current-mode flyback controller, with the one transistor sensing current and shutting off the output (sinking base current, or gate voltage if MOS) at a fixed or dependent threshold.  Feedback is supplied from a winding, which both provides positive feedback for oscillation as well as local power, say to bias a control (opto) transistor so that output voltage can be regulated.  Usually a TL431 or similar provides error amp regulation on the secondary side.

Also it's plausible it's (2S)D1302, a fairly general-purpose type that would suffice here: https://alltransistors.com/transistor.php?transistor=17926

Tim

[magic]

AKA a "ringing choke converter". Very common in cheap switchers.

[D Straney]

I saw a lot of simple peak-current-controlled discrete-transistor oscillators like those, when taking apart LED bulbs as a "competition analysis" at a previous job - you can save a lot of money in volume by just adding an extra winding on your inductor and having no control IC.  One kind-of-rough way to do voltage feedback, when you don't need tight output regulation, is by peak-detecting the reflected output voltage on the primary side of the flyback through an auxiliary winding - see the first attached schematic for an example of an off-line supply that does just that.

There's also another simple off-line power supply I traced a while ago which is probably almost identical to yours: it does exactly what Tim describes (see second attached schematic).  You can see in both cases how they're arranged as a flyback, with a second transistor that has its base connected to the current sense resistor.  This terminates each switching cycle at a specific peak current, when the current sense voltage becomes high enough to turn on that second transistor.  The turn-on then happens again via the auxiliary winding of the flyback inductor, and a resistor & cap from it to the base of the power transistor: when the flyback current ramps down to zero and the voltage starts to ring in the opposite direction (the "L" of the windings & the Coss of the power transistor form a nice series-LC circuit), that voltage reversal on the auxiliary winding provides a turn-on voltage to the power transistor ...as well as some positive feedback so that once it starts turning on it turns on HARD, and vice versa.

To control the converter, the peak current threshold can then be shifted downwards by variable amounts, by doing some resistor-based voltage summing at the base of the peak-current-sense transistor.  The simplest way is to have a zener diode coming from the reflected-output-voltage-sensing, so that any conduction through the zener lowers the peak current threshold.  The sketchy supply actually uses a more accurate output voltage sensing (again, exactly as Tim describes) with a TL431 doing voltage regulation on the secondary side, and driving an optocoupler which provides variable current to shift that peak current threshold on the primary side.

Hope this answers your question.  Happy to explain more details if you're curious about exactly what each part of the circuit is doing - I got deep into this for a little bit years back both to compare the component-by-component optimizations that other lighting manufacturers were doing (some of these had pretty sophisticated temperature foldback & "dim to warm" color-temperature adjustments, implemented all with a few discrete transistors), and also partly because these types of circuits made for great "weed out the people not suited for R&D" interview questions at work, watching engineers try and work through how they functioned.