Over the decades I have accumulated a few (!) old digital clocks and clock radios, none of which have an internal battery to keep the time when power bounces. Some are kinda cool looking, but a pain to reset. I also have a few small 6 V SLA batteries, lotsa old filament transformers, and plenty of 2N3055s and just about anything else needed. Seems to me it would be fairly simple to make a small UPS with a Royer oscillator driving a 12 V 1 A center-tapped transformer backwards. Estimated output power is 5 W - ish.

Yes, power MOSFETs would be much more efficient, but this is a use-existing-inventory project. Other than that, is there anything I should look out for? Anyone have direct experience with this approach?

Thanks.

ak

 

Yes it can work with a simple 50/60 hz pushpull oscillator using your ct transformer, you may find the output voltage will be lower than the mains voltage due to losses, as it will be pulsed with a square wave and may fluctuate if not monitored by feedback.

 

Do the clocks use the power line for the timing? If so then you will want an oscillator that is fairly accurate. I am not sure that a Royer will be accurate enough.

It might be good enough to just make sure that the clock gains time when power is lost. That way you won't be late for any important meetings.

 

A parallel thought is not a true Royer self-oscillating circuit, but driving a center-tapped primary with two-phase drive derived from a crystal. This might morph into a true online UPS, but I don't want to jump in that deep just yet.

ak

 

A parallel thought is not a true Royer self-oscillating circuit, but driving a center-tapped primary with two-phase drive derived from a crystal. This might morph into a true online UPS, but I don't want to jump in that deep just yet.

ak

That's how I would start. I am guessing that almost any reasonable RC oscillator is going to be more accurate than the Royer. (In fact that is exactly what I did for powering a string of LED Christmas tree lamps). I used a 74C14 for the oscillator and drivers into a couple of small MOS-FET's with a 60 Hz transformer. After writing that I realized that the 74C14 is probably not much more accurate than the Royer...

 

Yes it can work with a simple 50/60 hz pushpull oscillator using your ct transformer, you may find the output voltage will be lower than the mains voltage due to losses, as it will be pulsed with a square wave and may fluctuate if not monitored by feedback.

If so, how do the keep time when the power is out and they are running on the battery ?

 

You could drive the clocks from the oscillator circuitry all the time, with the mains charging the battery. Then use a Phase Locked loop to keep the oscillator synced to the mains. That way there is no switching problem and the time should keep relative accuracy. In fact, the mains is not a real short term accurate frequency source but over each day is it as the utilities fiddle with the frequency a little to keep the 24 hour cycle number exact. At lease, they used to. I assume that still goes on.
Or just use a good crystal locked divider to generate the 60 (or 50)Hz and not worry about the mains at all.

 

I used a modified solid state vibrator (tube radio) that I did eons ago and gave it with new resistance values and transformer selection. So push pull on a center-tapped xformer worked fine for a telescope motor for a co-worker. I used the 50% duty cycle formula. I actually used 2N3055's. This was the early 1980's.

I'll bet the frequency isn't going to be a real problem, because advancing the clock a few seconds is a LOT easier than setting the time from "zero".

I actually WANT to put one battery powered clock on a drop out relay, so I can tell when the power glitch-ed. If it goes out, my NAS will alert me with an email until I get the UPS installed.

 

I would be tempted to just run the battery power directly to the clocks own internal DC rails at a voltage somewhere close to what it uses and see if it keeps time being there no point in stepping the battery voltage up to line level then back down again only to have DC once more.

If by some chance the clock is line frequency dependent it will have some sort of tie in point between the driver circuitry and the internal AC power some place which if so that can be easily rerouted to get its line frequency reference from a dedicated and accurate astable clock IC or circuit.

Mostly when you ar running off batteries you want to try and be as energy efficient as possible so having multiple unneeded power conversion stages would be a severe and unneeded burden on the battery.