I live in a place where sometimes we get power cuts. It does not happen really often now, but when it happens, I hate having to re-program the memories and timer of my video cassette recorder! It looses memory as soon as the power outage lasts for more than a few seconds. But I have a large storage battery in my radio room, under constant charge, and given the low power consumption of a VCR, I decided to put it on the battery. I looked inside, and tried to inject a backup voltage into the CPU circuitry, but having no schematic diagram I missed the proper spots, and the machine always kept forgetting everything. So I decided to power the entire VCR from the battery, through the AC power input.
The most obvious approach would be to make a DC to AC inverter, having 110 or 220V AC output at 50 or 60 Hz, but this requires a rather large and heavy transformer. So it's better to use a high frequency, at least 25kHz, so a very small and cheap transformer can be used. But sending 25 kHz into a power supply designed for 50 to 60 Hz is no good! The input diodes are not fast enough, and the noise filter at the input would place a heavy reactive load on the inverter! So I decided to rectify the secondary voltage in the converter, and power the VCR with DC.
Here is the schematic.
You can get the full resolution version too.
The design is a simple saturation-limited push-pull converter. There is
no special reason to use PNP transistors; I used them simply because I
had a box full of them around. You may well turn over the design to use
The 2SC945 is a bias switch for startup. When applying 12V power, this transistor applies enough bias to the power transistors to get the oscillation started. Soon later, the 100uF capacitor charges up, the transistor goes off, and the power transistors self-bias into cut-off, such that cross-conduction is eliminated. After removing power, the 6k8 resistor discharges the bias timing capacitor, as otherwise the circuit would be unable to restart!
The secondary rectifiers are ultrafast diodes. These are NOT 1N4007! And the 220nF capacitors for the secondary filter are no typos; the diodes deliver almost pure DC, since the oscillation waveform is square, so only some noise filtering is needed. No electrolytics are necessary here.
Note the filters at both input and output, using ferrite cores. These are necessary to avoid polluting your environment with RF noise! Using these filters, and joining the input and output negative leads, this converter is very quiet and does not cause any problem in my combined HF, VHF and UHF station.
All ferrite cores (for the transformer and for the noise filters) are manufactured by Amidon Associates, and can be ordered directly from them in small quantities. Look for Amidon on the web. The 77-material core used for the transformer is less than ideal. A square-loop ferrite would work more efficiently! This one gets really warm, operating in saturation mode at 25 kHz. But it has worked well enough for two years now. The filter cores, on the other hand, are well chosen, so try to use the exact ones.
For all windings, the schematic states the number of turns. "7t" means 7 turns. As the transformer is quite small for the involved power, use as thick a wire as you can fit, leaving about half of the space for the 2x7 turns primary winding, and the other half for the secondary, while the feedback winding can be made from very thin wire.
The transistors do not need any heat sinks. They are large enough without, and they need to dissipate little heat!
As in the other
case, here is a full resolution version for
The 78S40 is a very old, but still useful power conversion chip. It contains a 1.25V reference, comparator, gating, output transistor, flyback diode and an uncommitted operational amplifier, which is not used here, and draws very little power. As the internal transistor and diode are good only for 1A maximum, I had to use external parts instead. Even so, the circuit is simple enough! A P-channel MOSFET was used as power switch, and an unknown toroid from my junk box was pressed into service for the filter choke. The circuit works very well, with an efficiency of better than 60%, which is superb for such a low voltage output!
I assembled the entire circuit, complete with power switch, in a small aluminum box measuring 4x4x6cm. The MOSFET and the diode were bolted to this box, for heat dissipation. The diode will dissipate around 2W at full output, so it would get too hot when free running. The MOSFET would probably have no trouble, as it dissipates much less power.
If you don't need adjustable output, and are happy with a fixed 1.25V output, you can eliminate the trimpot and resistors around it, and connect pin 10 directly to the output.
The circuit is fully short-circuit protected. The 0.075 Ohm resistor sets the current limit. If you have trouble locating such a low value, place the proper amount of higher value resistors in parallel!