How to time a watch winder

[flamer]

Hi guys, I'm going to attempt to make my own watch winder. the basic engineering of it is very simple, but I don't know where to start when it comes to the programming of it.

I already have an electric motor that can run in both directions (will run it about 1-3 rpm). I want to do this:
1) spin counter clock wise for 5 minute
2) turn off for 20 minutes
3) spin clockwise for 5 minute
..repeat forever


(the actual numbers don't matter I still need to figure those out exactly)

the commercial ones do exactly this, the DIY home built ones they just turn on them off and on manually which is not what I want. What component should I use to control these type of function?

TIA

 

[Old Steve]

flamer, you don't mention whether you have the experience and equipment, (fine-tip soldering iron etc), to put together a circuit on a veroboard from ICs, transistors, a DPDT relay etc, working from a schematic diagram, or whether you want to do it with pre-made timer modules.

Using ICs and so on, it's fairly straightforward using a 555 timer, a T-type flip flop, a few transistors and a relay. It would be a little trickier using pre-made timer modules.

 

[Old Steve]

Here's a 555-based circuit that would do what you want. If a 12V motor and relay are used, the +V supply for the ICs etc can also be 12V, making the power supply fairly straightforward:-



The 555 timer powers the motor 5 minutes on, 21 minutes off, and the 4013 reverses the direction at the beginning of every 5-minute period. There may be some variation in exact timed periods. That shouldn't be critical, but trimpots could be used to replace the fixed resistors for fine-tuning the periods.
A quick search on each chip will give you their pinouts. All unused pins on the rest of the 4013 should be tied to ground, except the Q and Q-bar outputs. They should be left open. (The 4013 is a dual flip-flop, with two F-Fs in one package.)
Edit: I forgot to label the 4013 ground pin, but it's fairly clear which it is.

*** Oops: The second BC548 should be replaced by a transistor suited to your motor's current. ie. A BC338 for about 0.5A, or a BD179 or similar for an amp or two.
A BC548 can only handle a max of 100mA.

 

[flamer]

perfect thank you! well yes I do have all the tools, I built RC stuff for a hobby but I don't usually have to get my hands this dirty! Ill have a look at all the components and start trying to put it together. thanks again!

 

[Old Steve]

perfect thank you! well yes I do have all the tools, I built RC stuff for a hobby but I don't usually have to get my hands this dirty! Ill have a look at all the components and start trying to put it together. thanks again!

Before soldering up a final version, I'd recommend that you buy a plug-in breadboard and assemble the circuit on it first. Then any bugs can be ironed out before you commit it to a permanent board. You might need to solder short leads on the relay pins temporarily to do this - it probably won't fit in the breadboard due to it's pin layout.:-



Note that I didn't include a power supply in the schematic, and also that I used no 'bypass' capacitors across the pins of the ICs. I just quickly added a couple of bypass capacitors to the schematic. Here's the updated copy:-



Assuming that it's a 12VDC supply, and that the relay and motor supply connect to each other and that they're close together, the 100uF cap will serve for both, otherwise possibly another cap from the relay +12V and ground might also be a good idea, maybe 10uF. All caps should be rated for twice the supply voltage. ie 25VDC for 12VDC. And the chips can't handle more than 18VDC.

 

[Old Steve]

Oh, and a final point. if you're planning to use fixed resistors and not adjustable trimpots for the 555 timing resistors, it's probably a good idea to use metal-film resistors with a 1% tolerance. That'll get you closer to the desired period, although the capacitor value could still vary considerably from it's rating. It's possibly best to use trimpots, really, a 200K to replace R1 and a 500K to replace R2. Then you could get spot-on 5min and 20 min with a bit of fiddling.

 

[flamer]

Oh, and a final point. if you're planning to use fixed resistors and not adjustable trimpots for the 555 timing resistors, it's probably a good idea to use metal-film resistors with a 1% tolerance. That'll get you closer to the desired period, although the capacitor value could still vary considerably from it's rating. It's possibly best to use trimpots, really, a 200K to replace R1 and a 500K to replace R2. Then you could get spot-on 5min and 20 min with a bit of fiddling.

Great advice yes I would use a breadboard first to test it out, okay I have used trimpots before, I guess that makes me sense as I will likely buy the cheap components so the resistors may vary a lot from their specs!! thanks so much.

 

[flamer]

Im struggling a bit with the timing, using

C1 = 4700 uf
R1 91k
R2 390k

Im getting:
Thigh = 26 minutes
Tlow = 21 minutes

Doesn't quite seem right!

 

[Old Steve]

Im struggling a bit with the timing, using

C1 = 4700 uf
R1 91k
R2 390k

Im getting:
Thigh = 26 minutes
Tlow = 21 minutes

Doesn't quite seem right!

Did you fit the diode across R2? If so, check that its' polarity is correct and that it doesn't have a bad connection. It should allow a duty cycle less than 50%, so must be the culprit.
Edit: Your figures are exactly what I would expect if the diode was left out - 21min low, 21min + 5 min high

 

[Old Steve]

To elaborate, the output is high while C1 is charging.
With the diode in place, shorting across R2, the current only flows through R1 to charge the cap, resulting in 0.7 x R1 x C1

The output is low while discharging.
The DIScharge pin goes low, the diode now doesn't conduct, so the cap discharges through R2, giving 0.7 x R2 x C1

Without the diode, the cap charges via R1 and R2, and discharges via R2, giving 26 min high and 21 min low.

 

[AnalogKid]

I bounced out to Digi-Key and looked at some large cap datasheets. Leakage current calculates out to around 500 uA, which is greater than the charging current through a 390K resistor at 12 V. Also, the NE555 has much higher input currents than the CMOS version. All of this could add up to the timing errors you are seeing.

Another way to do this is with a CD4060 (oscillator and 14-stage divider) or CD4521 (oscillator and 24-stage divider). With the 4060, a 5.5 Hz oscillator will not have the capacitor problems, the 5 minute ON period becomes 6.25 minutes, and can be decoded from the outputs with some diodes. The schematic shows both chip options, and assumes the same relay arrangement as in post #5.

Note: This circuit powers up with the motor off for 18.75 minutes followed by the 6.25 min. on time. To move the motor time to the start of the cycle takes one more transistor and one less diode.

ak

 

[Old Steve]

I bounced out to Digi-Key and looked at some large cap datasheets. Leakage current calculates out to around 500 uA, which is greater than the charging current through a 390K resistor at 12 V. Also, the NE555 has much higher input currents than the CMOS version. All of this could add up to the timing errors you are seeing.

If that was the case, the error would be in the charging time, and not the discharge period, wouldn't it?
The only timing error here is in the discharge time, the low period. The high period is exactly as calculated.
Leakage current shouldn't affect discharge, except to possibly speed it up, not slow it down. My bet is still on the diode.

Edit: AK, the discharge resistance is 91K, not 390K.

 

[AnalogKid]

Alternate output timing.

ak

 

[Old Steve]

@AnalogKid , how about a response to my post #12? Throwing multiple replacement circuits up isn't the answer.

 

[flamer]

Did you fit the diode across R2? If so, check that its' polarity is correct and that it doesn't have a bad connection. It should allow a duty cycle less than 50%, so must be the culprit.
Edit: Your figures are exactly what I would expect if the diode was left out - 21min low, 21min + 5 min high

Bingo that's exactly what it was!! Timing is now looking good now to start on the rest!! Thanks

 

[Old Steve]

Bingo that's exactly what it was!! Timing is now looking good now to start on the rest!! Thanks

Excellent. I'm pleased that it sorted out easily. (The 26min / 21min gave it away.)
Good luck with the rest of the project, and yell out if you need a hand.

I didn't think it could be leakage current. Besides the other considerations, it can start out high, but drops off quickly to a very small value as the capacitor charges.

 

[AnalogKid]

Wrote this yesterday, didn't hit Post.

The diode effectively reduces the terminal voltage for the charging capacitor by 0.6 V during the charging phase, by over 10% for a 5 V circuit, or by 5% for a 12 V circuit, but it does not reduce the voltage at the top of the 555 internal reference divider. Thus, it will take longer than 0.7RC to reach 0.67V+. But his results agree very well with an open diode.

ak

 

[Old Steve]

Wrote this yesterday, didn't hit Post.

The diode effectively reduces the terminal voltage for the charging capacitor by 0.6 V during the charging phase, by over 10% for a 5 V circuit, or by 5% for a 12 V circuit, but it does not reduce the voltage at the top of the 555 internal reference divider. Thus, it will take longer than 0.7RC to reach 0.67V+. But his results agree very well with an open diode.

ak

Your point about diode voltage is very true, and that could be taken into account when doing the calculations, but due to capacitor current leakage, and variations in capacitor value (and resistor value to a lesser extent), because of rated tolerances, there's not much point in that calculation.
A trimpot cures all.

All's well that ends well.

And there's still the rest of the circuit to build and test.....

 

[AnalogKid]

The 555 is a wonderful little critter, but I've never liked to use it with large timing caps. The early parts were weird with anything over about 10 uF, and 100 uF was trouble, and back then a 2.2 uF ceramic was real money. Of course the newer CMOS parts are way better, but old habits... Whenever a timer requirement gets up into aluminum electrolytic territory, I start thinking 4060, 4521, etc. The oscillator timing is not as temperature-stable as with a 555, but most multi-minute delays don't need millisecond accuracy. When they do, there's always the 32 kHz watch crystal.

ak

 

[Old Steve]

The 555 is a wonderful little critter, but I've never liked to use it with large timing caps. The early parts were weird with anything over about 10 uF, and 100 uF was trouble, and back then a 2.2 uF ceramic was real money. Of course the newer CMOS parts are way better, but old habits... Whenever a timer requirement gets up into aluminum electrolytic territory, I start thinking 4060, 4521, etc. The oscillator timing is not as temperature-stable as with a 555, but most multi-minute delays don't need millisecond accuracy. When they do, there's always the 32 kHz watch crystal.
ak

I agree as far as the watch crystal goes for accurate timing. I have a bunch of them here, and a gigantic collection of CMOS 4000 series chips, but as you know, I prefer to use a micro-controller these days. Getting lazy in my old age.

I actually considered recommending the CMOS version of the 555 to flamer, too, but figured the standard version was easier to get, plus no risk of static damage. Pretty hardy little things, NE555s, and despite your misgivings, good for timing from microseconds to hours, according to the datasheets. (I wouldn't personally use one for hours though. 30 min is about my limit)