BLR121 battery for AVO

[Mark Hennessy]

If Alan has had this VC1 for 30 years I find it hard to believe that its a LT1650 chip based boost converter.

Obviously not the LT1615 - that's a relatively recent chip (perhaps 1998). That was just picked as a random example of the breed - there are thousands of boost converter ICs out there. A good example to pick though, as it's able to run from a single cell.

The boost converter topology has been known about for many decades, but the drive for efficiency and flexibility from battery powered devices has caused it to become widespread in recent times. Many devices running from rechargeable batteries starts with one of these. As well as boost, there is a step-down variant called "Buck". Another version produces the opposite polarity to the input, and the magnitude of the output voltage can be greater or less than the input - that is usually called "Buck Boost".

The drawback with these topologies is they require a PWM scheme to regulate the output. Or perhaps fixed on-time, or fixed off-time. Either way, it results in quite a lot of silicon being required. Not a problem today, but back in the '70s when using the valve-based mentality of using fewest transistors in a design was the usual design objective, such converters were comparatively rare.

Instead, most early DC-DC converters used blocking oscillators (or similar). You could implement those with just 1 or 2 transistors but it required a more complex transformer (whereas the buck/boost types use just a simple inductor). But back then, that was the preferable route.

Also, converters using a transformer could achieve isolation, whereas the buck/boost converters share a common connection - usually ground, but could be the positive input if desired.

For something like this, you have to engineer it to take the minimum current possible when no load is taken from the output. That's not impossible - certainly not "alien technology" - but it does require attention to detail. Ultimately though, the only energy required is to make the thing oscillate. It might work in "burst mode", so only oscillates briefly to charge up the output cap, repeating every few seconds or whatever to keep it topped up. When loaded, the negative feedback will see the drop in voltage and run the oscillator as needed to maintain a steady output voltage.

I've just fixed something like this in an old (mid '80s) Philips oscilloscope (not my image). It's a single BC548 driving a transformer. It takes about 30mA from a 12V supply and generates just over 100V RMS to drive an electroluminescent panel that back-lights the LCD readouts mounted above the volts/div and time/div controls. It self-oscillates at the resonant frequency formed by the transformer and various surrounding capacitances - somewhere in the 1kHz region - and is regulated by a series string of Zener diodes (where the PIVs add ) feeding back to the base of the transistor. It's incredibly efficient. And simple.