Hacker News Reader

Comment by olyjohn | Hacker News Reader

I have an old wall clock from Japan that runs on 110v/50Hz. It keeps time like all old clocks, using the frequency of power. I can plug it into a US outlet and it runs, but it runs fast, since we're 60Hz here in the US. To remedy this, I bought a 12v power supply, and an inverter from Japan that had the 50/60Hz selectable on it. I couldn't find any other inverters that had an option to run at 50Hz.

I get the feeling that the frequency wasn't checked for accuracy / stability, because the clock still eventually goes out of time. My KillAWatt shows something like 51 or 49Hz or something like that. Not good enough to run a clock.

Been looking for some other way to get 50Hz AC power... This seems like it could be promising... but I have no idea how stable the frequency will be from a project like this...

View on HN

Sooooo how much power does that clock use ?

Because simplest one would be:

* a cheapo chinese subwoofer amplifier * 12V wall-wart to power it * a quartz-stabilized 50Hz generator (soooo an arduino, with DAC, even simple R2R + some filtering). * transformer fitting subwoofer amp output voltage. Measure amp output voltage at near-max, connect amplifier to secondary and tweak the "volume" till it is right.

Sub amp is like $5, $3 for cheapest arduino clone, probably like $2 for transformer, and few bucks in proto board and other components

If you want to overcomplicate it you could put rPi into it and sync the 50Hz clock to NTP

> If you want to overcomplicate it you could put rPi into it and sync the 50Hz clock to NTP

That’s ridiculous! Buy a little GPS-disciplined oscillator and either scale the PPS output up to 50Hz or use a PLL to derive 50Hz from the oscillator output :)

Just use a pll to derive 50 from 60. Problem solved.

The GPSDO will also give you correct time of day and unwavering frequency. I used to have one set up at home to feed my test equipment a 10 MHz reference.

If you're powering 50/60Hz clock you won't be setting time remotely on that easily

But how do GPSDO work? Is GPSDO susceptible to metastability?

GPS is a system for distributing time first and a means of computing position second.

A GPSDO is typically a PLL that uses the recovered timebase as a reference input and will have an incredibly slow loop filter. It can be outrageously highly damped and stable.

Such clocks are the basis of the cell network and have made doing client handoff between towers almost easy. The hard part is operating through GPS denial (mostly due to weather).

Would you mind revealing where you would go to for a power transformer that cheap? Clearly they must exist, as a full AC/DC adapter costs barely more than that, but eBay is wont to give me listings for merchandise of a certain animated TV series...

Aliexpress would be my first bet. Random used junk second

https://www.aliexpress.com/w/wholesale-12V-220v-transformer....

I've literally used an audio amplifier to directly create 240V 50Hz in a professional capacity so your idea is completely reasonable.

Yeah I have prototype (working, sans transformer) somewhere on the pile of shame.

The (presumably) low power demand means you could do pretty much anything... you could digitally control the frequency of the power to this clock if you ran your own inverter. Like in software you would know exactly how many cycles had elapsed since previous time and how many need to elapse before the next to achieve synchronization with some NTP source. There are all sorts of ways you could sort it out, assuming the clock is not mechanically slipping relative to AC cycles.

It seems quite strange that the clock isn't compatible with both 50 and 60 Hz. Japan uses both frequencies in different regions.

Fun thing is that inverter is the clock in this case. The time keeping device. The wall clock is just a display.

You could use a mechanical 60 to 50Hz converter. Basically a motor connected to a generator. They tend to be very expensive but maybe there are options.

If I were making something for this problem I would make an AC-DC-AC converter with a PLL to divide the 60Hz input frequency to 50Hz to control the inverter.

> If I were making something for this problem I would make an AC-DC-AC converter with a PLL to divide the 60Hz input frequency to 50Hz to control the inverter.

This is the best way to do it, especially if the synchronous motor inside the clock is actually fed with a lower voltage from a transformer (which seems to be common in old radio clocks as they needed a transformer anyway to power the radio circuitry). If that is the case, it should be possible to bypass the transformer entirely and build a converter that operates entirely on low voltage; some quick searching suggests that this exact kind of project has already been done before in fact [1].

[1]: https://mitxela.com/projects/phase-locked_inverter

> If I were making something for this problem I would make an AC-DC-AC converter with a PLL to divide the 60Hz input frequency to 50Hz to control the inverter.

I'd put a $3 breakout board with any microcontroller and quartz... why would you want to sync to power network in the first place ?

> why would you want to sync to power network in the first place

It's very stable, because it was used/ok to use for time. Well, it was until very recently [1].

[1] https://www.usatoday.com/story/money/economy/2018/05/17/cloc...

Iirc, the power grid frequency is adjusted over the long term to give excellent long term stability. A crystal will drift.

It won't keep time because on an actual mains grid the operator tracks frequency variations over time and makes corrections in the opposite direction to keep the average very close to the stated Hz - one reason for that being mechanical clock sync.

For example if the grid ran at 59.87Hz during the hottest part of the day due to high load then they might run the grid at 60.06Hz all evening offset.

You could probably hack your inverter with a simple DPLL, to keep it in sync with the wall. I'm assuming you could interface with the clock generator (cap or resistor), if you wanted to keep it "analog" :).

Or, perhaps replace the existing clock with a microcontroller that you can sync with NTP or the 60Hz line.

>I have an old wall clock from Japan that runs on 110v/50Hz.

No, it's 100V, not 110. Japanese mains power is 100VAC, 50Hz in the east and 60Hz in the west. Your clock is from the east, made for Tokyo probably. US 110-120V is a bit high for the clock, but it's close enough that it probably doesn't cause any problems.

It sounds like the inverter you bought isn't terribly accurate, unfortunately.

>Been looking for some other way to get 50Hz AC power...

Actually, it's really simple to get highly-accurate 50Hz power to power your clock. All you need to do is move to Europe. Then you can use a 1:2 step-down transformer to convert the 240V/50Hz power there to 120V/50Hz power, and your clock will be accurate.

What's the power draw of your clock? 50Hz is within the range of most LFO circuits (Low Frequency Oscillators) of the kind used for modular synthesisers and guitar effect pedals. You could combine one of those circuits with a simple voltage follower (consisting of a power op-amp or BJT transistor plus a couple of resisters) to keep the frequency at a stable 50Hz under load, and finally a transformer to convert the signal up to 110V.

All in all, I would estimate that this could be done with a single IC providing a few op-amps, a handful of passive components and a transformer; probably under US$30 or $50 with a nice case and plug.

> 50Hz is within the range of most LFO circuits (Low Frequency Oscillators) of the kind used for modular synthesisers and guitar effect pedals.

i don't think those are designed for long-term frequency stability, either. not at the <0.01% level needed for a clock. rest of your comment is on track, but the original low-voltage low-power 50Hz signal needs to come from something that was designed for low long-term drift.

Good point, and thinking about this further makes me realise just how stable the frequency of mains power is! Also, in the museum at the Greenwich Royal Observatory, I saw an exhibit of a clock which was synchronised with regular pulses in the mains electricity, which I believe were in addition to the usual 50Hz AC. However, I don't know if the power grid still has such a pulse in Britain, or whether there was/is a similar system in the USA.

I love that this discussion is rapidly headed to sticking an XO on the circuit suppling current and frequency to control the old digital clock that lacks an XO. Hackery at its finest.

Are you positive it’s not meant for 100V? That’s the standard in all of Japan from what I know.

The voltage doesn't affect the time keeping capabilities. It's based upon grid frequency. I've got one of those US<->Japan xformers I use to run a very special toaster in my kitchen. Doesn't do anything for frequency, but that doesn't matter in my particular case.

Unless you mean for sentimental value, very special personally, can we have more information on the very special toaster?

I've literally just searched 'Japanese toaster', but is it perhaps a Balmuda 'The Toaster'?

Mitsubishi Electric TO-ST1-T

Reply on news.ycombinator.com