How can I get a 60Hz supply?

[kalee20]

Using an induction motor as a generator is possible, as davegsm82 says - it's an induction generator - but the frequency is different from the rotor speed (just as when used as a motor, there is slip so it runs below synchronous speed). So as a precision 60Hz generator, it's a no-no.

If you connect an induction motor to the mains, and then physically drive the shaft at a faster speed than its off-load speed, it will transfer power to the mains.

(Which brings the question, if you connect a capacitor across the winding and spin the shaft, as an isolated self-excited system, what frequency output does it in fact produce?)

[christge]

The basic motor-generator idea will work, but will make a lot of noise. We do use the direct coupled motor-generator on some aircraft to create 400hz from DC. Incidently, a 60hz clock will lose 10 min per hour at 50hz. When I first came over to Europe (as a poor serviceman) I used to to set my alarm XX hrs/min ahead based on the 10 min loss. I never missed work!

[Sean Williams]

We run a standard 50hz alternator on our gensets - the nominal speed is 1500rpm.

The same generator will also do 60hz - we just run them at 1800rpm

The voltage is controlled by an Automatic Voltage Regulator - this gains a supply from a permanent magnet generator that is mounted within the genreator.

We just turn the excitation voltage down, this reduces the output to 380v across phases, to comply with 60hz supplies.

All that being said, this solution would cost you about £250k, so the electronic version would probably be cheaper....

[Billy T]

Originally Posted by steve102

The following link provides a solution for 60Hz clocks http://sound.westhost.com/clocks/freq-changer.html

Looking at that site, I have to say that for the sake of producing a 50Hz (or 60Hz) drive signal into a transformer to supply a clock, it is truly an extreme case of creeping elegance.

It is easy enough to build a 50 or 60 Hz oscillator with adequate stability and output, then add a power amp (an audio amplifier, no less) to drive the step-up transformer (an old power transformer using the 5 volt or 6.3 volt windings as the primary, or both in series perhaps). Old electric clocks are not that good at timekeeping anyway, so I'm thinking that the concept in my previous post is in fact entirely practicable.

Cheers

Billy

[ppppenguin]

Originally Posted by Billy T

Old electric clocks are not that good at timekeeping anyway.....

Really? If it's got a synchronous motor then it will run at exactly the speed determined by the supply frequency. It can only run at that speed, unless it's stopped. There are other technologies for electric clocks such as electrically maintained pendulums and tuning forks but these are unlikely to mind whether the supply is 50Hz or 60Hz.

[julie_m]

Originally Posted by Billy T

Old electric clocks are not that good at timekeeping anyway

Yes they are! The principles of keeping time accurately were known long before electric clocks were invented. It is just that the mains frequency is more stable nowadays than it used to be.

The accuracy of any clock is determined ultimately by the escapement; which, in a mains-driven clock, is a synchronous motor which makes one full turn for a certain, known, whole number of cycles.

All the electric clocks connected to any power supply will keep time at the same rate. The only way they can possibly lose or gain time, is because the supply frequency is incorrect.

You're right about the "creeping elegance", though. A crystal and divider would have been better IMHO.

[richrussell]

I too would disagree that old electric clocks aren't good at keeping time. Most use synchronous motors, so long as the National Grid are doing their job right and having the correct number of cycles per 24 hour period - which they do.

I've got several 1920s and early 1930s clocks (mainly Ferranti) and they keep better time that all the quartz controlled modern clocks in the house. The only clock that keeps time anywhere near as well is my watch, which sets itself to the MSF atomic clock signal every night. But even that can be a second or two out by the time it synchronises.

Hence why the OP needs an accurate 60Hz supply. Ironically generating this from a crystal and audio amplifier is likely to still not be quite as accurate as it was getting on for a hundred years ago running off the mains, as crystals can and do drift slightly if you're not careful.

[Mike Phelan]

Originally Posted by Billy T

Old electric clocks are not that good at timekeeping anyway,

Apart from synchronous clocks' timekeeping being determined by the stability of the supply frequency, until the 1930s, the Greenwich Mean Time signal was derived by a couple of electric clocks, the Synchronome-Shortt slave and master design.

[Paul Stenning]

If I was designing something like this I would go for a crystal controlled divider, divider to 60Hz square wave, op-amp low pass filter circuit to make the waveform into something resembling a sine wave, power amp (bridge mode), and step up transformer. A bridge amp avoids the need for a split rail power supply or big output coupling capacitor.

I would possibly add some sort of feedback control loop to monitor the output voltage and adjust the amp gain to keep it about right to allow for loading. Or the simpler approach of a meter and a pot for manual adjustment.

I don't know what sort of power is needed but if 10-15VA was adequate than a 20W power amp IC as used for car stereos would do the job, with the whole thing powered from a 14.4V external power supply.

[GrimJosef]

Presumably you could keep the very high accuracy which comes from the UK grid by multiplying the 50Hz by 6 (maybe full-wave rectify to get 100Hz and then filter out the 3rd harmonic) then divide by 5 (simple logic) to get to 60Hz ?

Cheers,

GJ

[valveaudio]

I agree with Paul use a xtal and divider plus a LP filter or a Wien oscillator, these are stable if left on and temperature controlled. No one has asked about the power requirement, you could use one of the amplifiers made by ILP Electronics with a suitable reverse connected mains transformer if the requirement is not too high.

Trevor

[ppppenguin]

GJ, that's an interesting approach.

Otherwise Paul Stenning's method is simple enough - the bridge amp is a nice refinement which may well be available as a single chip. Proper sine wave synthesis with a DAC seems a bit OTT for this application. A compromise solution might be to use a 2 or 3 bit DAC (a few resistors in practice) with some very simple logic to approximate a sine wave. Then the low pass filter can be lower order and/or give a cleaner sine wave. I've no idea how electric clocks would behave with distorted waveform but I can't imagine it's very critical. Provided you round off the worst excesses of a square wave I'm guessing you'll be fine.

[julie_m]

When I tried to build a single-stage inverter using two TDA2030s bridged, I found most of my volts were appearing across the output impedances of the two amplifiers leaving only a pittance at the "primary" of the transformer. More here.

The eventual design (which is also on my blog) still pushes a square wave into the transformer; which is OK in this case, because the intended load (a compact fluorescent light bulb) isn't fussy about waveforms. What comes out the other side of it isn't so square, of course.

[Lucien Nunes]

If I were going to build from scratch, I'd say this is perfect territory for a PIC microcontroller, to give an output locked to the mains just as Rod Elliott's does. Top of head design gives the component count as 1x16 pin DIL and 14 passives, to give a 60Hz sine output clean enough to go into the power amp.

The reference can be taken from the secondary of the mains transformer almost direct to one input. The PLL routine compares the scaled counts and trims the clock division ratio to keep them equal, using a nice lethargic PID (the input isn't going to change quickly!). It wouldn't need a crystal, the PIC's internal RC oscillator would do fine as it would entirely self-correct.

Use a lookup table to generate the waveform across an 8-bit output port, 8 weighted resistors serve adequately as a D-A because the outputs swing solidly rail to rail, followed by an RC filter. You could get rid of seven resistors by using software PWM to generate a sinusoid from a single RC-loaded digital output pin.

But my style is more the motor-generator, I rather like the idea of a big set rumbling away in the background, just to power a clock! M-G sets were used in industry to alter the speed of machines where it was not practical to change the drive ratio from the motors. One pair of 75kVA sets I had hoped to salvage was designed to give a 66.6 Hz output from 50Hz, using an 8-pole alternator driven by a 6-pole motor. The idea was to increase the speed of some machinery at a car plant above the original rating, after some design improvements.

For clocks you'd need a 12-pole alternator and 10-pole motor. Or a DC motor and a speed control driven from a differential gear between two synchronous motors, one on the mains and one on the 60Hz output. All the above have been used commercially...

Lucien

[Paul Stenning]

One advantage of crystal controlled oscillator (or something free-running that doesn't synchronize to 50Hz mains) is that you can run it from a DC supply such as a car battery away from a source of mains power. Probably not relevant for this application though.

I think a PIC is overkill when the job can be done easily with couple of logic ICs (a 4066 and maybe something else to get the divide logic correct) and a suitable crystal (32.768KHz).

[AlanBeckett]

Originally Posted by Paul Stenning

..... and a suitable crystal (32.768KHz).

60 is not an integer division.
Alan

[kalee20]

I don't think the motor-generator approach is going to work - unless the motor is a synchronous type, driving the generator through a gearbox. Induction types slip too much. At work we have a rotary converter to generate 400Hz, the generator aspect is fine with a precisely defined frequency, but the speed its rotor turns at is dependent on an induction motor driver. Net result is an output frequency of 397Hz.

It also depends on what you mean by "accurate 60Hz". A crystal-controlled oscillator will give a good, stable, accurate 60Hz using the techniques outlined in other posts. GrimJosef's idea (which I like very much!) will give nothing near this accuracy short-term, as the mains frequency can vary by ±1%. But long-term - over a day - the accuracy will be much better as the CEGB slightly rev up or slow down the whole country's generators to give the right number of cycles over 24 hours.

So, for powering a record turntable, tie the frequency to a crystal. For powering a clock, tie the frequency to the mains.

[Billy T]

Originally Posted by ppppenguin

Really? If it's got a synchronous motor then it will run at exactly the speed determined by the supply frequency. It can only run at that speed, unless it's stopped. There are other technologies for electric clocks such as electrically maintained pendulums and tuning forks but these are unlikely to mind whether the supply is 50Hz or 60Hz.

That is the theory, but minor frequency variations do occur, in fact when working in India a couple of years back I logged the mains for 24 hours (voltage, frequency, surges, sags and spikes) and the frequency dropped below 48Hz on a number of occasions. You don't want to know about the voltage sags and spikes, they were horrendous.

However, do remember I was referring to OLD clocks, perhaps no longer in their prime. I also run a couple of vintage synchronous motor clocks here and they both lose time over extended periods. A sychronous motor can have some slip and as they age and mechanical losses increase they get worse. Mine need repair but the cost is too high so I settle for adjusting the time every so often.

A thought just popped into my head though, if I set up a variable frequency supply I could increase the frequency to compensate for the losses. Now there's a useful idea!

Cheers

Billy

[Lucien Nunes]

Synchronous motors don't really 'slip' as such, in the way an induction motor does. What must be happening is that it actually 'pulls out' momentarily. The difference is that with most types of induction motor including shaded pole, there is always torque in the correct direction, regardless of the rotor speed. The more torque you apply, the greater the slip and the slower the rotor, but it keeps running. With a synchronous motor, by the time the rotor has fallen one pole-pitch behind, the torque will have reversed and the motor will normally stall immediately.

My limited experience with electric clocks has been that if the movement becomes stiff the motor will stall, but if it then reverses, gets caught by the starting ratchet and then sets off forward again it may be able to overcome the movement and continue forwards. But more usually it will just stall.

Lucien

[richrussell]

Originally Posted by Lucien Nunes

gets caught by the starting ratchet and then sets off forward again

Bah - look at you with your new fangled self starting ratchets. My clocks just stop if the power goes and need a spin of the starter knob to get them going. Of course if you're reaching behind the clock without looking, it's easy to spin it the wrong way and have it running quite happily in reverse (I once left one like this when we had people round for dinner. It took a while before anyone noticed that it was earlier each time they looked).