Converting 110V/60Hz to 220V/50Hz

[MSM]

I have an old mechanical clock which uses main line's frequency for its accuracy. It consumes 2.5W from 220V/50Hz power line. I intend to use it in US and using a simple step-up converter will not be enough.

I tried to look for frequency inverters (which work through AC ? DC ? AC conversion) or application specific precise inverters. But all of them are too powerful and/or generally too expensive.

Is there any other (cheaper) way to provide a low current 220V~50Hz from 110V~60Hz?

[Seekonk]

Search for inverter driver board EGS002 "EG8010 + IR2110" driver module on ebay.  These are available for $7.  Add a transformer and two FET and you got it or 4 FET in H bridge.

[alanb]

The problem is generating the 50Hz with sufficient accuracy. You will need to lock the 50Hz oscillator to a crystal. In the UK although the short term mains frequency varies, over the period of a day the number of cycles is constant and therefore your clocks long term accuracy is ok.

[Ian.M]

The accuracy problem is extremely intractable.  A typical crystal will have a tolerance of +/-30ppm, which will result in a culminative timekeeping error of up to 18 seconds per week.  The grid frequency is stabilised with the aim of keeping the long-term total timekeeping error within +/-60 seconds, so over a year or so, the quartz driven clock will be less accurate than a mains synchronous clock.   See http://wwwhome.ewi.utwente.nl/~ptdeboer/misc/mains.html

A crystal oscillator can be trimmed to within a few ppm of its nominal frequency, but as it will drift with temperature, you'll be lucky to hold 10ppm in an ordinary shirtsleeves environment unless the crystal is precisely temperature controlled.
If you want to do better, the oscillator must be locked to an external timesignal, e.g. from a GPS receiver.

Additionally, most mains appliances that are likely to be used on inverters aren't too fussy about the frequency.  If they can tolerate +/-1%, there is little incentive to improve the accuracy of the oscillator unless the result can be upsold as a premium product.

[AF6LJ]

For the cost of this project it would be more reasonable to buy three of those WWVB synchronized clocks.

[alanb]

How about having a counter and disconnecting the supply for 100 cycles every time a count of 500 is reached. The display would be jerky if you have a seconds hand on the clock but would not be noticeable if you only have minutes and hours. I suspect that it would create more wear on the motor.

[Ian.M]

It would be possible to build a 50Hz sinewave oscillator and PLL lock that to either the US 60Hz mains frequency or the 1pps output from a GPS module, then use a class D audio amplifier module off EBAY and a 220V:9V mains transformer in reverse to boost its output up to 220V.   You might even have change from $100 after you've got it built and tested, but its a lot of money to drive a clock.

If your clock can be converted for low voltage operation, you've got a lot more options.  See http://sound.westhost.com/clocks/ocm.html
If your motor is similar to the one pictured and the coil is not a slip fit on the core, one could salvage a stack of laminations from a scrap transformer to make a core for the replacement coil, so you can keep the original coil intact.  If its a slip fit, slide it off and make a new former for the new coil.   Once you have a low voltage coil, driving it is merely a matter of some code running on an Arduino with an H-bridge shield + a GPS module for a timesignal to pace it against.


@alanb:  I suspect it would drift badly due to the limited acceleration available at the start, and the friction limited overrun at the end, of each burst.

[Seekonk]

The clock is likely just for show.  I have some of these IR half bridge drivers, they are just a big 500V 555 that can put out an amp.  Just voltage double the line and put the motor in a RC.  Selecting the right  will get you a boost in voltage.  So you have to adjust the clock every other day.  That is what people did for a couple hundred years.

[alanb]

  So you have to adjust the clock every other day.  That is what people did for a couple hundred years.

The problem is that the error will be 4 hours every 24 hours.

[StillTrying]

Just in passing I'd say...

Locking a 50hz oscillator to a 60Hz signal shouldn't be too difficult because they're in-phase 10 times per second, and 5, 2, 1.

[Ian.M]

Yes, but making a sinewave VCO with a center frequency of 50Hz is quite tricky without resorting to digital synthesis.

[StillTrying]

Is there any other (cheaper) way to provide a low current 220V~50Hz from 110V~60Hz?

It might be possible to change the motor if you're lucky.

These people supply bricks to run 60Hz clocks on 50Hz, and say: There is also a version to run an English clock in the USA.

http://www.electric-clocks.co.uk/index.html

Yes, but making a sinewave VCO with a center frequency of 50Hz is quite tricky without resorting to digital synthesis.

I'd use a square wave! A micro to produce 50Hz, and sample a zero crossing detector on the 60Hz, every 5th 50Hz cycle, and then add or subtract a leap uS to to the 50Hz oscillator's period depending on the ZCD.

[Zero999]

Search for inverter driver board EGS002 "EG8010 + IR2110" driver module on ebay.  These are available for $7.  Add a transformer and two FET and you got it or 4 FET in H bridge.

Yes, that will work. It's crystal controlled so should be fairly stable.

[Ian.M]

Yes, but making a sinewave VCO with a center frequency of 50Hz is quite tricky without resorting to digital synthesis.

I'd use a square wave! A micro to produce 50Hz, and sample a zero crossing detector on the 60Hz, every 5th 50Hz cycle, and then add or subtract a leap uS to to the 50Hz oscillator's period depending on the ZCD.

I'd be a bit leery of driving a shaded pole motor with a squarewave.

If you are going to use a MCU, take an enhanced midrange PIC or similar with ZCD, and full bridge motor control PWM,  code a sinewave lookup table to control the PWM duty cycle, step through the table at a rate that gives 50Hz, and trim the internal oscillator frequency using the OSCTUNE register to lock it to 5/6 the frequency on the ZCD input.    Add a full bridge MOSFET driver, four MOSFETs and a voltage doubling rectifier to get a 325V DC bus + a bit of filtering on the output, and it should be fine for a clock.

[vk6zgo]

300Hz is the LCD of 50Hz & 60Hz.
Make a 300Hz oscillator locked to 60Hz,either PLL,or just triggered every 5 cycles,retro-style.
Divide the 300 Hz by 6 & you have 50Hz.

The downside is that it will still be a square (or squarish) wave.
Converting it to a sinewave can be done with a LPF,but you still have the problem of amplifying that sinewave to a useful voltage.

Another alternative is to change the mechanical gearing  so the timekeeping is correct for 60Hz ,& just use a transformer to give the right voltage

[MSM]

Search for inverter driver board EGS002 "EG8010 + IR2110" driver module on ebay.  These are available for $7.  Add a transformer and two FET and you got it or 4 FET in H bridge.

Yes, that will work. It's crystal controlled so should be fairly stable.

I don't want to do any modifications on my clock. This solution seems to be reasonable. I'll try it and let you know about the accuracy of the clock afterwards. Thanks everybody!!

[Circlotron]

http://www.siliconchip.com.au/Issue/2016/May?res=nonflash&fullscreen=1
Article number 10.

[albert22]

Here you have a simple converter that can go from 50->60 or 60->50Hz, while locked to the incoming frequency.
http://www.romanblack.com/one_sec.htm
(scroll down to "50Hz to 60Hz converting inverter!")
IMO this is the simplest solution you could find.

[Circlotron]

Or maybe use a 4046 PLL and multiply 60/120 Hz up to a certain figure and then divide it down again to exactly 50Hz. Perfect long-term mains tracking.

[SeanB]

Use a 3VA transformer with a centre tapped secondary in reverse, and drive it with the 50HZ.  The transformer will do a good enough job of making the low pass filter, with a 1u 400VAC capacitor across the primary, and will run the clock fine. Supply of 12VDC and a transformer of 21-0-12VAC will do, the output will be around 180VAC which will be enough to run the clock. Drive with a SMPS driver ( SG3524) with the 100Hz clock applied to it.

[oldway]

Or maybe use a 4046 PLL and multiply 60/120 Hz up to a certain figure and then divide it down again to exactly 50Hz. Perfect long-term mains tracking.

I agree, generate 300Hz with PLL from 60Hz and divide it by 6