I had a go at designing the same controller around the MC34063 controller and I was pleasantly surprised to see that it pretty much has the same operation mode as the one I designed (most likely unintentionally...) where it seems to just gate the clock. Simulating it in LTSpice pretty much produces the same results as mine too. Are many controllers designed around clock gating or are most linear in the sense of the duty cycle is proportional to the output power currently required?
Certainly, the result has to be duty cycle control. Whether that's achieved by something as crude as clock gating, or something more nuanced like peak or average current mode control, doesn't matter over the long term.
Over the short term, it does matter how well those pulses are controlled, because excessive pulse lengths cause extra peak current in the components and cost efficiency, and if the pulse widths are constantly bouncing around, those variations show up as output ripple, requiring more inductance and capacitance for a given output noise level.
The difference between "short" and "long" is basically why designs with the 34063 have about three times the inductor as the rest of the components in the circuit, and I mean volume per volume, for a THT build! The oscillation frequency is already fairly low by modern standards, and the rate at which it is controlled is even worse, hence the inductor must be oversized by that much.
An example of "medium term" variation might be chaotic operation of the peak current mode controller: when inductor current doesn't return to zero each cycle, the duty cycle will bounce back and forth erratically -- in fact, it is chaos (mathematically speaking!). The inductor emits an audible hissing sound, and the output becomes much noisier (in the sense of true random noise, rather than periodic switching transients). The output voltage and peak switch current are still controlled -- these are guaranteed by the design -- so in the grand scheme of things, it can't be said to be truly unstable, but it's a decidedly suboptimal operating condition!
I have got to ask as well Tim, how do you know what is required for your really intricate designs?! I look at some of your schematics and obviously they work but I find it really hard to decode what each transistor is doing in the circuit etc.
Eh... they're not usually too intricate, and the ones that are, are only by repeated iteration and contemplation. I tend to go for high level design, which tends to cost in parts count. My Theremin project is probably the best example of that, using at least several transistors in each stage, with each stage built in its own isolated section; the total count is around 50 transistors right now.
My discrete switching circuits I think are neat, but still largely high level, if you know what to look for. Timer, switching, current sense, reference.
If you want to see intricate... look up some old radio or TV schematics. The average TV set from the tube-hybrid-early solid state era (1960s) are all perfect examples of this: not one tube or transistor is wasted; connections go back and forth all over the place! AGC this, amplifier that; they're a mess to follow. They also regularly used single side PCBs, and wasted few jumper wires in the process.
Tim
Home
Help
Search
Login
Register
