AC capacitive dropper power supply for Disney mug warmer

[Peabody]

My favorite coffee mug warmer is the Mickey Mouse model from Disney.  It has enough wattage to actually work.  But I seem to go through about one a year.  They work fine when they turn on, but it becomes increasingly difficult to get them to turn on, and eventually they just won't.  There's a push-on, push-off button which turns on the heating element relay and lights up an indicator LED.  when you push it, the LED lights briefly, and you can hear the relay clicking, but immediately both turn off.  "It's dead, Jim."

I finally had the time to mess with it, so I took it apart and traced out the circuit - see attached.  I attached my scope, and found that when I press the button, the 5V supply to the daughterboard drops to about 2V.  So whatever is happening in the mystery chip is probably getting reset, and that's why it won't stay on.  I assume the mystery chip is a processor of some kind, and it seems unlikely that the chip is sinking enough current internally to cause this problem.  The relay coil measures 128Ω, so that doesn't seem to be shorted.

That would leave the 5V power supply as the probable culprit.  It is a transformerless capacitive dropper supply.  All of the bridge diodes test ok.  And the zener is obviously working.  So unless I'm missing something, it seems the problem must be one of the two capacitors - either the dropping capacitor or the smoothing capacitor.  Does anybody have experience with this type of supply?  If so, I'd like to get your thoughts about this before I start desoldering stuff.

[amyk]

The dropper capacitor degrades over time with things like mains surges, slowly reducing in capacitance until they don't pass enough current to keep the voltage up.

Related video from bigclivedotcom:

[cvanc]

The dropper capacitor degrades over time with things like mains surges, slowly reducing in capacitance until they don't pass enough current to keep the voltage up.

This.

[Peabody]

Yes.  Thanks very much.  I got out my Kill-a-Watt, and a brand new warmer shows about .5 watt when plugged in but turned off.  The old one only draws .3 watt.  If I understand the circuit correctly, the 5V supply always runs at the maximum current, which goes through either the zener or the load.  So the different watt readings are a good indication of how much the dropper capacitor has deteriorated.

It is marked CL21   105J250V, which is a metalized polyester film capacitor.  I was expecting maybe an X2.  Anyway, I can't find any that look like this one on Digikey or Mouser.  Any suggestions?

[amyk]

A 1uF 250VAC X2 will work.

This one should do: https://www.mouser.com/ProductDetail/Panasonic/ECQ-UAAF105S1

[Peabody]

That one is 20% tolerance, and the one I have is (was) 5%.  Anyway, it looks like I can get the CL21 by the handful on Ebay as long as I'm willing to wair.

[Zero999]

That one is 20% tolerance, and the one I have is (was) 5%.  Anyway, it looks like I can get the CL21 by the handful on Ebay as long as I'm willing to wair.

But 20% tolerance is sufficient, irrespective of what's already there and I'd recommend going to Mouser, rather than eBay. It will be an X-rated capacitor, which is designed to fail safe, reducing their capacitance, rather than going short circuit. A cheaper capacitor might not be a genuine X-rated capacitor and could fail dangerously, shorting out, blowing up and taking the rest of the circuit with it.

[Peabody]

I found in my junque box a very similar capacitor, but smaller.  It's a Panasonic marked 224K250V.  Could I add that in parallel with the original cap?  I'm thinking it might restore enough capacitance to get it working again, and it wouldn't be any more dangerous than the original product. 

[Gregg]

I have fixed my wife's Mickey Mouse coffee warmer with the same symptoms.  It is the electrolytic capacitor that dies due to the heat and probably the low quality of the original cap.  It isn't right here for me to open, but if you need more information, I can open it and take some pictures.

[Peabody]

I have fixed my wife's Mickey Mouse coffee warmer with the same symptoms.  It is the electrolytic capacitor that dies due to the heat and probably the low quality of the original cap.  It isn't right here for me to open, but if you need more information, I can open it and take some pictures.

Thanks very much Gregg.

Well that's a surprise.   I understand that the smoothing cap could theoretically be the problem, but it's never exposed to more than 5V, so it's hard to see how it could degrade in just a couple years.  Well, ok, there's the heat.  That could be the explanation.  But I can easily test that fix.  Probably should have done that in the first place.

I wish my meter measured capacitance.  Maybe I can use my old analog meter and known good caps of the same capacitance to at least ballpark it.

Film at 11.

[Gregg]

I've replaced that cap twice from whatever I found in my junk pile that fit and had greater capacitance than the original.  The last one was bigger diameter and I had to do some creative lead bending to fit it.  The capacitance isn't critical, it just needs enough to close the relay while powering the rest of the circuitry; additional capacitance could mean longer life.  I thought about drilling some vent holes and possibly a heat shield inside between the heater element and the ci9rcuit board, but I think it is far easier to just change the cap as needed.

[fzabkar]

A larger capacitance value would mean that the Zener and its series resistor would need to carry more current, in which case it might be prudent to choose a Zener with a higher wattage rating.

[Peabody]

I replaced the electrolytic smoothing cap with one of the same value from my junque box, and the warmer is now working fine.  The old one measures a bit lower capacitance with my analog meter than the replacement and one other 220µF cap I had, but not majorly so.  I also noticed when I was desoldering it that one of the leads may not have been making a good connection to the trace.  The lead didn't quite come all the way through the hole.  Obviously it worked when new, but may have worked itself into a bad connection over time.  Anyway, the electrolytic was indeed the problem.  I also checked the dropping capacitor while I had it out, and it behaves exactly the same as two other 1µF caps, so I think it was fine.

So it's working now - without any long delays or Mouser shipping charges.  Thanks very much for everyone's help.

[amyk]

The bad news is we all got the culprit wrong.
The good news is that your mains supply seems relatively free of spikes and surges.

[Zero999]

I replaced the electrolytic smoothing cap with one of the same value from my junque box, and the warmer is now working fine.  The old one measures a bit lower capacitance with my analog meter than the replacement and one other 220µF cap I had, but not majorly so.  I also noticed when I was desoldering it that one of the leads may not have been making a good connection to the trace.  The lead didn't quite come all the way through the hole.  Obviously it worked when new, but may have worked itself into a bad connection over time.  Anyway, the electrolytic was indeed the problem.  I also checked the dropping capacitor while I had it out, and it behaves exactly the same as two other 1µF caps, so I think it was fine.

So it's working now - without any long delays or Mouser shipping charges.  Thanks very much for everyone's help.

Great, I'm glad you fixed it and can't believe none of us thought the smoothing capacitor failing could be a possibility. Don't forget to bin the old one. I've had a bad experience before with a defective part finding its way back into my  box of assorted components.

[Peabody]

Great, I'm glad you fixed it and can't believe none of us thought the smoothing capacitor failing could be a possibility. Don't forget to bin the old one. I've had a bad experience before with a defective part finding its way back into my  box of assorted components.

Well, I offered both caps as possibilities in my original post.  But the dropping cap is the one connected directly to mains, and subjected to - whatever.   The smoothing cap never sees more than 5V, so there's really no reason for it to fail, except possibly the heat.  And based on the Big Clive video, the dropping cap is almost expected to fail over time.  But Gregg says he replaced the electrolytic twice, and now it was the problem with mine, and presumably in both cases the dropping caps were fine.  So maybe there's something in the design that's not obvious.  Or maybe it's just the heat.

I've saved the bad cap in case I ever get a capacitance meter - to see what it actually measures.  I only have a toy scope (DSO150), and was able to do the ramp on the 1µF smoothing cap, but I can't get that to work on the 220µF electrolytic.  Maybe I need a much lower value resistor.  Actually, I would bet  there are Arduino-based capacitance meters, and I have an extra Nano.

[Peabody]

Ok, I set up the Arduino capacitance meter and measured the other "good" 220µF capacitor from my junque box.  Recall that I found two caps of that value which behaved identically when charged on my analog meter, and which the meter showed clearly had more capacitance than the "bad" cap I removed from the warmer.  I used one of the good caps in the warmer, but still had the second good one.  Nothing is calibrated in this setup, but that remaining good cap measures 237µF on the Arduino meter.

The bad cap measures 190µF on the Arduino meter.  So I don't know what it really is, but it is down about 20% from the cap I consider to be good.  I don't know if that's enough to explain the problems I was having, and it may be that the bad solder joint was really the problem.  But if the good cap is actually 220µF instead of 237µF, then the bad one is actually 176µF.  That's not necessarily enough of a drop to produce failures, but it's a significant drop.   There's nothing on the bad cap to suggest what its tolerance is supposed to be.

By the way, the Arduino meter produces very consistent results.  Not necessarily correct results, but consistent results.  If you measure the same cap 10 times, the value won't vary by more than 1µF.  I guess if you had a known capacitor, it could be calibrated.

 

[Zero999]

Ok, I set up the Arduino capacitance meter and measured the other "good" 220µF capacitor from my junque box.  Recall that I found two caps of that value which behaved identically when charged on my analog meter, and which the meter showed clearly had more capacitance than the "bad" cap I removed from the warmer.  I used one of the good caps in the warmer, but still had the second good one.  Nothing is calibrated in this setup, but that remaining good cap measures 237µF on the Arduino meter.

The bad cap measures 190µF on the Arduino meter.  So I don't know what it really is, but it is down about 20% from the cap I consider to be good.  I don't know if that's enough to explain the problems I was having, and it may be that the bad solder joint was really the problem.  But if the good cap is actually 220µF instead of 237µF, then the bad one is actually 176µF.  That's not necessarily enough of a drop to produce failures, but it's a significant drop.   There's nothing on the bad cap to suggest what its tolerance is supposed to be.

And what was the ESR (Effective Series Resistance)?

By the way, the Arduino meter produces very consistent results.  Not necessarily correct results, but consistent results.  If you measure the same cap 10 times, the value won't vary by more than 1µF.  I guess if you had a known capacitor, it could be calibrated.

Whether that's good or bad, depends on the capacitor value.

[amyk]

Electrolytics are usually 20% tolerance.

[Peabody]

And what was the ESR (Effective Series Resistance)?

Yeah, I figured some EE was gonna ask me something like that.  First I gotta measure the capacitance, and now the ESR.  I shoulda just bought a new warmer for $14.  You even get a Mickey Mouse mug with it.

But working with what I have, I got a pretty good indication of what the ESR situation might be.  The EEs here will have to do the math, or even the maths.

The test oscillator on my scope is a 1KHz, 3.3V square wave, and when I connect two 2.2K resistors in parallel from the oscillator output to ground, the amplitude is cut exactly in half.  So I think that means the output impedance is 1.1K.  Removing the resistors, and connecting the capacitor directly to the oscillator in their place, I connected the scope probe (1X) to the positive end of the capacitor and took pictures of what showed up on the scope.

The Good Cap picture has the scope set at 10mV per division.  The Bad Cap picture has the scope set at 200mV per division, or 20 times the Good Cap.  I didn't include it, but if the scope is set to 200mv per division for the good cap, the result is essentially a flat line.

I'm not sure how the calculation goes, but just from looking at the scope traces, it seems that the ESR of the good cap is responsible for about 1mV of the voltage across the cap, whereas the ESR of the bad cap accounts for about 150 mV.  So it looks like the bad cap is a real stinker.

[Zero999]

When the voltage suddenly changes from 0V to 3.3V or back again, the only thing limiting the current is the resistance of the circuit. The capacitive element can be ignored, because it's massive at those frequencies. You now have a potential divider, with a supply voltage of 3.3V, with an internal resistance of 1k1, an unknown load, with an output voltage of 0.15V.


Vin = 3.3V
R1 = 1k1
R2 = ?
Vout = 0.15V

The potential divider formula:
Vout = Vin*R2/(R1+R2)

Rearrange of R2:
R2 = Vout*R1/(Vin-Vout)

R2 = 0.15*1100/(3.3-0.15) = 165/3.15 = 52.4R.

[Peabody]

Thanks very much.  That makes perfect sense.  And an ESR of 52.4R is huge, right?

Does this mean if I put a 47R resistor in series with the good capacitor, it would behave pretty much like the bad one?  If that's the case, then it's understandable that the bad cap wouldn't have been able to respond adequately to the relay being energized, with the resulting sag in the 5V rail voltage.

Edit:  I did try the 47R resistor with the good cap, and did indeed get the 150mV spike.  But after that, it goes horizontal.  It's just a 150mV square wave.  There's none of the charging ramp stuff that's visible with the bad cap.  So it's not just the ESR.  The bad cap is also behaving like it has much less capacitance - the Arduino's capacitance measurement notwithstanding.

[Peabody]

I'm resurrecting this thread to report that I've repaired my backup Disney Mickey Mouse coffee warmer, which had the same problem as the first one - the single electrolytic capacitor is bad.  Apparently all of these warmers suffer from that problem.  Anyway, I wanted to bring it up again because I didn't have a direct replacement for the 220uF 16V capacitor, so I used a 100uF 16V cap, and it appears to be working fine.  If I ever order again from Digikey, I will probably get a 220uF for it, but for now it seems to be working ok.  This cap is the smoothing cap for a zener "regulator" 5V supply (see schematic above), and the only significant demand on it is when the 5V relay is switched on by the microcontroller.  With the original cap, that would drag the voltage down, which would reset the MCU.  The fact that that doesn't happen with the 100uF cap will tell you how bad the original cap was.  It's an ESR issue as far as I can tell.

I don't know if the bad caps were a deliberate choice so the warmer would need to be replaced after a couple years.  I would hate to think that of Disney, but they continue to use the bad caps long after they must have known about the problem.

Anyway, for future reference in case anyone sees this, I just wanted to report that the smaller value cap works ok.  Aside from this problem, the warmer actually works quite well.  It's 27 watts, which is more than other warmers, and actually keeps the coffee fairly warm.

[Zero999]

What was the temperature rating of the original capacitor? It sounds like it needs to be higher.

Another option is a tantalum, but it will need a generous voltage rating to stop it from going incendiary.

[Peabody]

The original was 105C, and where it's located doesn't get particularly hot.  So I don't think it's a heat problem.  Attached is a picture of the capacitor.   I don't recognize the brand.  I think these are just junk capacitors.