Bringing the Goblin Teasmade to America

[madsci]

Some time ago, a friend purchased a Goblin Teasmade from the UK along with a cheap Chinese 750VA step up converter. He's obsessed with this thing; it's his favorite appliance. I have modified it with a switch that allows him to mute the "tea done" signal as not to disturb his roommate. The alarm for a missing teapot works in either switch position however. (safety issue..no teapot = hot tea flowing everywhere!) The mains plug was also rewired and its strain relief put back into service. The 13A fuse was removed and a 5A was put in place.

We currently have two issues: One, I need to rewire the converter. The autotransformer appears to be of decent quality with good heft and nice leads but the soldering and insulation in this thing is *poor*. 240V mains is no place to be stingy with the solder or the heat shrink tubing. I placed some insulation where needed and redid several joints, including the earth terminal. (yikes!) A new chassis without the stupid (and nonoperational!) USB charger is in the works along with color codes that make sense and a thermal switch.

The second issue and the one I need input on is the clock. It uses a 50Hz motor that is supposed to turn at 250RPM. Needless to say, the clock is quite fast on 60Hz and there are no commercially available 250RPM AC clock motors for 60Hz. This makes since as my research indicates that they are synchronous motors with an integrated gear reduction.

I could in theory rig up a second gear reduction but fabricating that might be difficult and it's not something I'm familiar with fabrication wise. Custom brackets, gears, etc would need to be machined. There is a surprising amount of room inside the box for some circuitry. My idea was to get a small universal input SMPS, a small audio amp and a sine source on a board and use a transformer to step up the output from the amplifier to 240V. We only need a few watts of output so this seems feasible. The back plate of the clock is aluminum so that could provide heat sinking.

There must be a more elegant way to do this though.... A cycloconverter of some sort is in order.

[AndyC_772]

Would it be simpler to design a circuit with a counter and a power switch of some kind, so it runs the motor at 60Hz for a while and then briefly interrupts its power supply in order to correct the average speed? If the motor runs 10s on followed by 2s off, the speed difference will average out.

All you'd need to do that would be a PIC, a triac, an opto-isolator and a tap off the transformer secondary and a handful of passive components. No significant power handling required.

#include std_mains_safety_blah_blah.h

[mikeselectricstuff]

Maybe easier to replace the timer with an electronic one?
Failing that, a crystal controlled 60Hz supply wouldn't be too hard - mains transformer wired backwards, you can probably drive with a fairly square waveform - simplest probably a centre-tapped transformer with an N-channel fet in each leg.

[mikeselectricstuff]

Would it be simpler to design a circuit with a counter and a power switch of some kind, so it runs the motor at 60Hz for a while and then briefly interrupts its power supply in order to correct the average speed? If the motor runs 10s on followed by 2s off, the speed difference will average out.

All you'd need to do that would be a PIC, a triac, an opto-isolator and a tap off the transformer secondary and a handful of passive components. No significant power handling required.

#include std_mains_safety_blah_blah.h

Nice idea but I'm not sure how well that would work,as synchronous motors may lose some steps starting up, and may have some run-on when stopped. However if you can get some feedback from it to verify position, that may work. e.g. an encoder, or a magnetic gear-tooth sensor. Could also have some acoustic noise issues due to gear backlash.

[johnwa]

Maybe easier to replace the timer with an electronic one?
Failing that, a crystal controlled 60Hz supply wouldn't be too hard - mains transformer wired backwards, you can probably drive with a fairly square waveform - simplest probably a centre-tapped transformer with an N-channel fet in each leg.

Or, you could use a frequency converter phase locked to the incoming supply. I just remembered that there is a project for doing exactly this on Rod Elliott's site. Have a look at this:

http://sound.westhost.com/clocks/freq-changer.html

[madsci]

Electronic timer is right out; I already suggested that. He wants the original clock, hands and all, to be fully functional.

Adding a servo mechanism to verify start and stop position might be a chore on this. The clock on the later Teasmade is a module.

Having a crystal controlled supply to the clock motor seems to be the most direct route.

I was thinking about driving it with a sinusoidal waveform to make the motor quieter and to ensure accuracy. I could perform an experiment with square wave to see if it significantly increases noise (this thing is on the desk next to the bed, the harmonics may make objectionable noises) or changes the accuracy.

[SeanB]

Making a 50Hz oscillator and driving the clock with a H bridge will probably be easiest. Simplest will be to take an old PC power supply, and then remove almost everything on the secondary side and leave only the mains input side, high voltage side and the standby 5V section to power the electronics you will need. Then use a 50Hz clock generator to drive the half bridge transistors using the existing transistors but using some optocouplers to do the gate drive. You will need another pair of them to do the other half of the bridge, and connect the motor between the 2 centres, and drive it via the existing 680n 630Vcoupling cap to reduce the current to a pulse, and drive it with a large off period for the transistors. That will give you the required drive and will keep the motor from overheating, though it may be a little noisy. If so just add the PFC chokes from some older PSU units ( the iron laminated ones on the lower cost 5-8 year old ones without active PFC) in series with the output and extra 220n capacitors across the output.

Otherwise just buy a USA made mains clock ( they are available )that has the same output speed at 60Hz and do a transplant.

I did a similar thing for a USA made telephone answering machine, but there as it was electronic I made a board with some CMOS and a US TV crystal to generate the 60Hz required. Was not nice, or a good looking 60Hz signal, but it did the job, and drifted only about a minute a month.

[G7PSK]

Why not use a phase delay scr circuit. Perhaps old fashioned these days but is about 95% efficient.

[mikeselectricstuff]

Electronic timer is right out; I already suggested that. He wants the original clock, hands and all, to be fully functional.

Adding a servo mechanism to verify start and stop position might be a chore on this. The clock on the later Teasmade is a module.

Having a crystal controlled supply to the clock motor seems to be the most direct route.

I was thinking about driving it with a sinusoidal waveform to make the motor quieter and to ensure accuracy. I could perform an experiment with square wave to see if it significantly increases noise (this thing is on the desk next to the bed, the harmonics may make objectionable noises) or changes the accuracy.

In which case a small audio amplifier driving a transformer is probably the easiest approach - you may need some compensation on the audio output to keep the amp stable with the transformer's inductive load.
To generate the sinewave, simplest way would be a microcontroller PWMing a sinewave from a lookup table into a lowpass filter.

[amyk]

Electronic timer is right out; I already suggested that. He wants the original clock, hands and all, to be fully functional.

How about replace the synchronous motor with a stepper motor and driver configured to run it at the right speed...

[mrflibble]

Electronic timer is right out; I already suggested that. He wants the original clock, hands and all, to be fully functional.

Ah okay, he's the practical type. Well, simply move to a country with 50 Hz mains. Problem solved.  Fully authentic behaviour of the widget.

[fcb]

Recalculate time to cope with the increase in line frequency.  Easy to do on a spreadsheet.

[GK]

Probably not a lot of help to the OP, but the idea of a automatic tea making machine on an alarm timer (I googled it) gives me cause to groan 
It's not THAT long a wait for the kettle to boil if you flick the switch after getting out of bed! While waiting the 30 or 60 seconds or so you could even amuse yourself with something like this:

[madsci]

Recalculate time to cope with the increase in line frequency.  Easy to do on a spreadsheet.

Funny you should mention that....that's exactly how he manages to get tea in the morning. He simply sets the clock forward every night.

[KJDS]

Probably not a lot of help to the OP, but the idea of a automatic tea making machine on an alarm timer (I googled it) gives me cause to groan 

It is very crass to own one for two reasons.

Coffee in the morning, tea in the afternoon

and far more importantly, the servants should be sorting out the breakfast.

[envisionelec]

Some time ago, a friend purchased a Goblin Teasmade from the UK along with a cheap Chinese 750VA step up converter. He's obsessed with this thing; it's his favorite appliance. I have modified it with a switch that allows him to mute the "tea done" signal as not to disturb his roommate. The alarm for a missing teapot works in either switch position however. (safety issue..no teapot = hot tea flowing everywhere!) The mains plug was also rewired and its strain relief put back into service. The 13A fuse was removed and a 5A was put in place.

We currently have two issues: One, I need to rewire the converter. The autotransformer appears to be of decent quality with good heft and nice leads but the soldering and insulation in this thing is *poor*. 240V mains is no place to be stingy with the solder or the heat shrink tubing. I placed some insulation where needed and redid several joints, including the earth terminal. (yikes!) A new chassis without the stupid (and nonoperational!) USB charger is in the works along with color codes that make sense and a thermal switch.

The second issue and the one I need input on is the clock. It uses a 50Hz motor that is supposed to turn at 250RPM. Needless to say, the clock is quite fast on 60Hz and there are no commercially available 250RPM AC clock motors for 60Hz. This makes since as my research indicates that they are synchronous motors with an integrated gear reduction.

I could in theory rig up a second gear reduction but fabricating that might be difficult and it's not something I'm familiar with fabrication wise. Custom brackets, gears, etc would need to be machined. There is a surprising amount of room inside the box for some circuitry. My idea was to get a small universal input SMPS, a small audio amp and a sine source on a board and use a transformer to step up the output from the amplifier to 240V. We only need a few watts of output so this seems feasible. The back plate of the clock is aluminum so that could provide heat sinking.

There must be a more elegant way to do this though.... A cycloconverter of some sort is in order.

Audio amplifier driving a transformer. 10W should do it.
Inductively couple signal from the incoming mains wire using wire wrapped around the HOT conductor.

[David_AVD]

I've made a telephone ringer (for stage use) that used a small (~ 3VA) transformer backwards to generate the ring voltage.  Same sort of setup as Mike suggested, but mine didn't need the accuracy of a crystal.

It wouldn't take long to lash up something to drive the transformer with a square wave and test for mechanical noise.

[IanB]

Coffee in the morning, tea in the afternoon

Heavens, no! Tea when you wake, coffee at 11, tea in the afternoon, coffee after dinner.

[madsci]

Yes, I know it's trivial but that's not the point. There is occasionally method to my madness.

The point is to take on little projects like this, document them, place them on a website and then put that URL on my CV for prospective employers to see after I graduate.

I'm thinking about a PIC12 or 16F series driving a TI filterless class D amp IC. These are available in DIP or HTSSOP, the latter of which is no trouble for me to solder. Some friends have a vat of FeCl3 and I have a 1200dpi laser printer ready to print out some masks for the photographic process we use to place art on the boards.

I may make Mr. Teasmade write the damn firmware; he's a CSE.....or it could be my first MCU project!

[fcb]

I wonder if you could do something as simple as build a TRIAC driver that disconnects power to the clock mechanism every 6th mains cycle.

If you had something like a MOC3041 (zero crossing opto-coupled triac), then you might get away with a very simple 555 timer astable set at something like 80% with an f=10Hz.

Perhaps even a 4017B configured as a 6 step counter, driven from the line. A capacitive dropper on the mains would give you enogh juice to drive a small triac and the circuit.  Probably no more than 10 components (and no software...)

[David_AVD]

I wonder if you could do something as simple as build a TRIAC driver that disconnects power to the clock mechanism every 6th mains cycle.

Was mentioned (and discounted I believe) already.