I have made thermostats using arduino. I'm using the 24vac from the heating system to supply buck converter and bring down the voltage to 5vdc. So far so good. It is working fine except.... it has a hard time dealing with power failures. I think the back emf from the 24 volt transformer is killing them. The buck converters I used is this:

https://a.co/d/5GmodXK

I have 3 thermostats and in the last power fail, 2 of them had the buck converter blown out. I'm not sure how to protect them. A zenner diode? May be.. not sure at all. The only thing I find is how to protect back emf on dc motors. Pretty simple on dc; a diode can do the job. But on ac.... not sure what to do. Anyone can help me with this?

 

Is this US 120-0-120 VAC power feed with the transformer running off of 120 VAC?

There isn't much margin between the possible input voltage and the absolute maximum input voltage on the LM2596 regulator.
A 24 VAC heating system transformer might put out 26 VAC easily with low load which would result in the
LM2596 seeing 36+ VDC after the diode bridge and input filter capacitor.

During a power failure and recovery there could easily be spikes and overvoltages on the power line. Seeing double
this voltage seems possible.

The above assumes the LM2596 is the HV version (max input 57V vs the regular one max input 46V).
Or possibly not even a real LM2596. Is the switching frequency of the LM2596 really around 150KHz
or something else (indicating it's not a LM2596)?

 

Is this US 120-0-120 VAC power feed with the transformer running off of 120 VAC?

There isn't much margin between the possible input voltage and the absolute maximum input voltage on the LM2596 regulator.
A 24 VAC heating system transformer might put out 26 VAC easily with low load which would result in the
LM2596 seeing 36+ VDC after the diode bridge and input filter capacitor.

During a power failure and recovery there could easily be spikes and overvoltages on the power line. Seeing double
this voltage seems possible.

The above assumes the LM2596 is the HV version (max input 57V vs the regular one max input 46V).
Or possibly not even a real LM2596. Is the switching frequency of the LM2596 really around 150KHz
or something else (indicating it's not a LM2596)?

Not sure exactly how to check for the frequency. My knowledge of electronic is really basic. I'll try to find the way to hook up the scope and check it.

 

Your buck converters are being killed by voltage spikes when the 24 VAC transformer is switched on/off or during power failures. The ringing and back-EMF can push the voltage way above approx. 34 V peak your modules see normally. Since those LM2596-style boards have almost no surge protection, they don’t survive the transients. The fix is to add proper suppression parts before the buck: a MOV rated approx. 33 VAC across the transformer secondary to clamp surges, plus an RC snubber (100 nF film + 100 ohm in series) to damp ringing. Use a bridge rectifier and a decent bulk capacitor (470–1000 µF, >=50 V) so the buck sees stable DC.


On the DC side, add a TVS diode (like SMBJ36CA) across the rectified supply to catch any fast spikes that sneak through. A small resettable fuse (polyfuse) in series with one AC leg will protect against faults. Also make sure your buck converter is a version rated for 40–60 V input; some cheap LM2596 boards are marginal above approx. 28 V. With a MOV + snubber on AC, bulk cap + TVS on DC, and a properly rated buck, your thermostats should ride through power outages without blowing modules. Here is a simpler design of buck converter that you may want to check. https://www.pcbway.com/project/shareproject/Super_Power_Buck_Converter_4bf4de4c.html

 

MOV rated approx. 33 VAC across the transformer secondary to clamp surges, plus an RC snubber (100 nF film + 100 ohm in series) to damp ringing

That sounds like a good solution. Thanks

 

Add a PTC thermistor (the type for circuit protection) between bridge and TVS diode. It will probably add enough resistance to keep the TVS diode safe when a spike comes along.

 

That advertisement does not say if the IC is a LM2596 or a LM2596HV.
You have the lower voltage part. You could replace the IC with the HV version.
I just went through this. My 24 transformer is 26V rms. 26 x 1.414 = 36.76 then take off 1.5V for the diodes. The IC is seeing 35V for certain. Some of my transformers measure 28V with no load. I know the part should survive 40V.
I think we all are a little concerned about the input voltage. It is a little hot.

 

It's probably not an HV. But still, 45v should be ok. Not sure what a PTC thermistor is, but I'll ask.

I'm wondering how the cormercial thermostats handle this. By looking at the parts on their boards, it seems to be so easy

Thanks guys... I go to the shop tomorrow and get the parts.