Multiple thermostat failures

[sambe]

Hi, I'm aware that this may be classified as a beginner post, so feel free to move it to that section if it is more appropriate.

I had a wall thermostat fail in slightly strange circumstances. There is nothing visibly wrong with it such as burned or melted chips/capacitors. The fuse is fine and there was still AC at the socket. I am not aware of any power surges or lightning etc. at the time - no other equipment was damaged. I was originally putting it down to random failure, and a replacement thermostat has been working fine at the same socket. Then, a few days later, two more failed (identical make & model). I have not checked those as thoroughly, but e.g. the fuses are also fine. No others have subsequently failed (there are 4 more) in over a year. I'd really prefer to understand what is going on and repair rather than replace them. Now the weather is getting colder I have to either get it done or replace them. Photos attached.

In addition to the fuse and AC I also:

• checked the two obvious capacitors using a Bob Parker-designed ESR meter. I don't have the exact readings to hand at the moment but one was in-line with the grid on the front of the meter, and the other on the low side. I'm told that low is usually fine, as they can be low-ESR intentionally, and usually high ESR signifies a fault. I can re-test and note the readings if deemed important.
• attached two 9V batteries to the relay (data sheets for what looks like the same model put it at 18V minimum) and can hear it moving.
• looked around with a magnifying glass for broken tracks and physically damaged components and didn't find anything
• spot-checked a few connections with a multimeter. I don't feel I am able to "read" a circuit well, so not 100% sure I followed the right paths. However, nothing obvious failed.

What else should I be testing here/what are the likely failure modes?

[m3vuv]

how old were these?,the ones we get here have a lithium cell soldered to the motherboard,when that is spent they f**k up forcing you to replace the whole unit,built in obselecence!

[Gyro]

Replace it them with a bimetalic strip ones - they last for decades! 


EDIT: Or try replacing the capacitive dropper X caps, that have probably self-healed them to bu**ery as usual! (I know you checked the esr, but did you measure the remaining capacitance value?).
     |
... V what he said!

[nali]

IME normal mode of failure is the capacative dropper - the large grey 0.15uF component in this instance. Can you measure the actual capacitance?

Every time they get zapped & "self heal" their value drops a little bit until eventually they are too high impedance to allow enough current to drive the relay. I have a few motorised valves using the same, and if one goes I tend to replace the lot as a preventative maintenance as it only takes a few minutes.

If you do change them make sure you get a suitably rated part! (AC rated)

[sambe]

how old were these?,the ones we get here have a lithium cell soldered to the motherboard,when that is spent they f**k up forcing you to replace the whole unit,built in obselecence!

Only a few (max 5, I'd think) years, and no lithium cell.

[sambe]

IME normal mode of failure is the capacative dropper - the large grey 0.15uF component in this instance. Can you measure the actual capacitance?

I think I'd need to de-solder and get a full capacitance meter for that? Trying to avoid that if possible, mostly for cost reasons.

Every time they get zapped & "self heal" their value drops a little bit until eventually they are too high impedance to allow enough current to drive the relay. I have a few motorised valves using the same, and if one goes I tend to replace the lot as a preventative maintenance as it only takes a few minutes.

Wouldn't high impedance show as high ESR anyway?

[dietert1]

In this circuit the dropper appears to have a protection resistor, probably the big SMD resistor 270 Ohm labeled "271". In the photo that one looks suspicious, i mean the 2 has some shadow like it was burnt.
The protection resistor gets up to 320 V when connecting the device to mains. I would never use a SMD resistor for this, but something more sturdy. Did you have power breaks recently?

Regards, Dieter

[Gyro]

It's an easy job to check the 270R resistor in circuit, no need for soldering..

As for the capacitor, if you don't have a capacitance meter then the easiest way to see if it is the culprit is simply to replace it. ESR measurement (presumably at 100kHz) it a tricky indicator with low value capacitors - it doesn't really give a great idea of whether it has enough remaining capacitance to pass sufficient current to drive the relay at 50Hz. That's unless it is well enough calibrated to work out the capacitance from its impedance, or can compare against a known good one of the same value.

It's a low cost part, if it doesn't fix them then you haven't really lost anything versus the cost of replacement thermostats. Get proper ones from a distributor rather than ebay 'specials'.

P.S. 

• checked the two obvious capacitors using a Bob Parker-designed ESR meter. I don't have the exact readings to hand at the moment but one was in-line with the grid on the front of the meter, and the other on the low side. I'm told that low is usually fine, as they can be low-ESR intentionally, and usually high ESR signifies a fault. I can re-test and note the readings if deemed important.

I Just checked the Bob Parker ESR meter page - the grid on the front doesn't go below 1uF.   You need to confirm the value you were getting.

[nali]

Wouldn't high impedance show as high ESR anyway?

Sorry, I was talking about it's overall (mostly reactive) impedance due to it's lower capacitance, not ESR. These things use reactance as a way of dropping most of the mains voltage without excessive power dissipation, so usually comprise of something along the lines of a capacitor, bridge rectifier and a zener to clamp the voltage down to 20-something volts (and smoothing cap).

II Just checked the Bob Parker ESR meter page

So did I... now my eyes hurt

[sambe]

In this circuit the dropper appears to have a protection resistor, probably the big SMD resistor 270 Ohm labeled "271". In the photo that one looks suspicious, i mean the 2 has some shadow like it was burnt.
The protection resistor gets up to 320 V when connecting the device to mains. I would never use a SMD resistor for this, but something more sturdy. Did you have power breaks recently?

Good spot! I was going to mention that I saw this and "confirmed" it was just a scratch. Glad I didn't: I think you've got it. Turns out I neglected to measure it, and it's showing up infinite. I didn't notice any power breaks or other problems and, as I said, 3 of them seemed to fail at different points in time. However, I will go and check the others too.

What would you recommend to replace it? There's probably not a lot of space at the back, which may have influenced the decision.

[sambe]

It's an easy job to check the 270R resistor in circuit, no need for soldering..

I thought I had but in fact had not - sounds likely that this is culprit (per my reply to dieter).

As for the capacitor, if you don't have a capacitance meter then the easiest way to see if it is the culprit is simply to replace it. ESR measurement (presumably at 100kHz) it a tricky indicator with low value capacitors - it doesn't really give a great idea of whether it has enough remaining capacitance to pass sufficient current to drive the relay at 50Hz. That's unless it is well enough calibrated to work out the capacitance from its impedance, or can compare against a known good one of the same value.

It's a low cost part, if it doesn't fix them then you haven't really lost anything versus the cost of replacement thermostats. Get proper ones from a distributor rather than ebay 'specials'.

I Just checked the Bob Parker ESR meter page - the grid on the front doesn't go below 1uF.   You need to confirm the value you were getting.

Thanks. I think I sort of extrapolated the values a bit for that capacitor, and guessed that low values were generally good vs high ones. If the problem turns out not to be (only) the resistor, I'll re-measure.

[Gyro]

It's an easy job to check the 270R resistor in circuit, no need for soldering..

I thought I had but in fact had not - sounds likely that this is culprit (per my reply to dieter).

You mean it has gone open circuit... or it is still 270R?  It takes 5 seconds to check.

EDIT: Sorry, I re-read your reply to dieter.

[dietert1]

Maybe two SMD resistors 560 Ohm parallel of the same size fit in (side by side). Gives you another factor of two and some redundancy. I mean in case one fails, the circuit may still work.

Regards, Dieter

[Gyro]

I am a little surprised that sufficient high frequency energy is making its way through the 150nF capacitor to fry the resistor. It might be self-healing a lot or you may have a big noise source somewhere, which might coincide with the close failure times. Either that, or they may just have cut the design a bit too close.

I agree with Dieter, paralleling higher value resistors (by stacking) may give a higher spike withstand capability. Two 130R resistors in series would probably give a slightly better withstand - each element having twice the current capability for the same device dissipation and half the voltage stress. This would be more difficult to arrange however - you would need to partially tombstone each resistor and join them in mid-air, not ideal.

It is slightly possible that the failed resistor is a fusible type as a safety feature (quite common in through-hole designs) but from what I can see in the photo, there are no markings to suggest this.

[sambe]

I think a more complicated layout will be beyond my soldering capabilities. If I just replace with another 270ohm, should I look for high power dissipation rating/anti-surge properties? I measure the size at 6.4x3x0.5mm, which I think makes it a 2512/6432. RS have the 2W but not 3W version of this:

https://uk.rs-online.com/web/p/surface-mount-fixed-resistors/7551211/

in stock for small batches. Farnell have what appears to be the 3W version:

https://uk.farnell.com/cgs-te-connectivity/3522270rjt/res-270r-5-3w-2512-thick-film/dp/2476384?st=2512%20resistor

Mouser has some other brands but nothing strikes me as obviously more appropriate. Am I on the right track?

[Gyro]

I have to admit that I am rather worried about why the resistor blew. If the 150nF capacitor is ok and it is being fed with normal 230V AC mains, then the resistor dissipation ought to be below 100mW. Its size was probably picked for package voltage rating rather than power dissipation.

You're not going to find a resistor of that package size with a low wattage rating but I'd stay as low as possible and replace the capacitor too.

[sambe]

Oh, I thought several people were suggesting that the component was likely under-specced, so it'd need to tolerate more power. Besides redundancy, which property is improved by the suggestion to have two in parallel? Maybe I'm misunderstanding here...

Indeed I don't find anything under 1W in that package size. The examples I linked are rated to 250V (500V max overload). 2W is the minimum rated at 250V.

As for the cause: I guess there could have been power outages/surges that I did not notice, odd as it seems for them to go separately.

[Gyro]

As you say, it could just have been a bunch of surges [EDIT: The rating is a funny thing - at steady state it's sub 100mW but in the presence of lots of spikes and surges it's... well depends on the number of spikes and surges!]. Go with whatever is most convenient qty wise in that package size. I would probably stick a few X caps on the order though (you have plenty of data on the label). They're cheap an it saves another potential order and hassle.


P.S. Thick film resistors are more surge resistant than metal film.

[dietert1]

The resistor acts as a clutch, and you know that you want to release the clutch before starting to accelerate.
The most difficult situation arises for the resistor when the cap is empty and mains gets connected at a moment when it is at 320 V or -320 V. Then during 270 Ohm * 0.15 uF = 40 usec the resistor takes up to 1 A and 300 W. Once the dropper cap follows mains voltage it gets much easier, resistor dissipation will be a fraction of the power requirement of the relay coil.
That's why i was asking for power outages. Each time mains comes back, there is the risk of destroying the resistor.
Some engineers like building the circuit without any resistor at all, which will result in continuous loss of the dropper capacitance, as mentioned above.

Regards, Dieter

[sambe]

All three had the same failed resistor and are now working with it replaced. I bought same-spec capacitors but did not replace them in the end. I figure I will learn something if they do or do not cause the resistors to fail again. For informational purposes: the ESR reading on both the broken and working thermostats for that capacitor was 10ohm. The ESR reading on the replacement ones (out of circuit) was 12ohm.

Thanks for the help!

[Gyro]

Thanks for reporting the outcome (we often don't get to hear). 

[mikerj]

All three had the same failed resistor and are now working with it replaced. I bought same-spec capacitors but did not replace them in the end. I figure I will learn something if they do or do not cause the resistors to fail again. For informational purposes: the ESR reading on both the broken and working thermostats for that capacitor was 10ohm. The ESR reading on the replacement ones (out of circuit) was 12ohm.

Thanks for the help!

ESR is not the parameter that degrades in these self healing caps, it's actual capacitance.  This increases the impedance the capacitor presents at the 50Hz mains frequency and limits the current the circuit can draw.