Shelly devices 'locking up' or software freezing?

[Slartibartfast]

Hi,  I have around 40 Shelly devices around my remote property here in Australia.  The power in my area is 'end of distribution' and regularly experiences power outages as well as spikes and brown outs.  I invested in a Tesla Powerwall for backup but this does NOT provide UPS style protection against transient dips in voltage.  The issue I'd like to raise is that several of my Shelly devices seem to freeze or 'lock up' all at the same time (and frop off the WiFi and therefore appear 'offline' in the app).  This seems to coincide with bad weather conditions and so I'm making the deduction that it is due to a transient power disturbance.  It seems to also happen more to my older Shelly 1PM and Shelly 2.5 devices rather than the newer gen2 & 3 devices.

Has anyone any experience of this?  If so, do have you any advice!  The issue could also be related to my WiFi system which is a Ubiquiti system with four access points.  I'm thinking of powering those off a very small UPS just to ride through these very short disturbances.  FYI when the grid does go down (fully off) the Tesla Powerwall takes some time to switch over to battery.  I understand that this time can be up to a few seconds as it has to shut down solar inverters and it's own inverter to avoid overloads before coming back online.  The more I write this the more I'm convinced a small UPS on the WiFi/computer system is required.  However, If I do this it still leaves all the Shelly devices which are mains powered.  Do they have a low level watchdog function to reboot if they get it a disconnected state?

Edited July 31 by Slartibartfast

[Heinz]

brown outs and spokes are never good for any electronic device. 

If they are going on and on the whole time I would maybe set them to a static IP address that they don't need to obtain the ipaddress from the DHCP server when its online or offline. 

This might save the issue that the device called out for a Ipaddress and then does not get one and sits in limbo. 

[Slartibartfast]

I already have them set-up with a reserved IP address on the Unifi router.  I may try with one device with a manual IP set on the Shelly itself but this is not my preferred approach.

They only go off occasionally (a few times a year) and as I noted, this seems to coincide with very windy weather.   It's a bit of a pain as they are in a remote location and so reliability is a real requirement.  I wrote the original post a couple of weeks ago when it happened again and when I got on site they all reconnected fine with a reboot (power cycle).  Non of the devices were in AP mode, and 2 Shelly 2PM's frozen with the relay "on",  (IE 2 set's of lights were on 24/7).

I read another post that suggested NOT using "eco mode", I'll try turning his off to see if the behavior changes.

Edited August 10 by Slartibartfast

[Heinz]

ok this is a other situation with its related to bad weather.

this i am not sure about, with the IP reserved address from the router side this would help but only if the router boots up before the shelly. Most cases Routers take some time to finish the boot cycle.

There is a setting with the shelly if the device restarts it can then be set to the following.
return to previous status ( if it was on it will return on when it was off it would stay off: this is good but please make sure that its not connected to a fridge or system that needs to be on after a power outage)
switch on
switch off

 eco mode off that would be a interesting test to see if that changes the behavior of the device.

[TN22]

I have a similar problem. My Shelly Pro 4PM controls a hot water heater. When I have enough energy from my solar panels it turns on the water heater. When there is not enough energy from my solar panels it turns the water heater off. Sometimes I have no hot water. when I check the Shelly I see that it is blocked. The screen is white and unresponsive. The only thing that helps is to reset it by turning off the power. Does anyone have the same experience, or does anyone know how I can fix this?

[Olsche]

11 minutes ago, TN22 said:

Does anyone have the same experience, or does anyone know how I can fix this?

It can help by connecting an RC snubber parallel to the load. The type of load is not typical for the problem but it's worth a try.

https://kb.shelly.cloud/knowledge-base/rc-snubber

[Slartibartfast]

On 8/11/2024 at 1:02 AM, Heinz said:

There is a setting with the shelly if the device restarts it can then be set to the following.

I have done this on a few devices, I'll go through all of them and ensure all devices are set up to default to what I would hope.

No bad weather for a few weeks so no further issues.  I did a close up restart of one of the devices and noticed it is 'singing' a bit.  I have read elsewhere some complaints over the quality of the 2x 'mains' 400v 4.7uF caps and the 1x 16v 100uF capacitors.  I'm going to experiment replacing them.  A bit of a pain but if it works it will be well worth it.  It has further occurred to me that as the age of the devices has increased so have my issues.  This makes me think that perhaps a degradation over time may be an issue, I just have to ensure I don't read correlation with causation!  I managed to find a nice Panasonic polymer electrolytic which is actually 150uF.

https://au.rs-online.com/web/p/aluminium-capacitors/2237685

...but the Wurth 400v caps are a bit longer and I had too remove the reprogramming header to fit them in.

https://www.digikey.com.au/en/products/detail/würth-elektronik/860021373025/17830202

This isn't the device that is 'singing' I'm going to do that sometime this weekend to see if I can silence it.

Edited August 15 by Slartibartfast

[Slartibartfast]

OK, I got around to looking at the "Singing Shelly".  (some people have described this as a hissing sound which is also accurate, or perhaps inductor whine which I think is the component that is vocalizing the 'angry pixies').

I only replaced the 16V capacitor and the noise went away, as a result I didn't go ahead and change the 2x 400v caps at this stage.

I tested the capacitor I removed and it doe seem low on capacitance and high on ESR.

I've read more posts about changing this cap and some people recommend a 25v cap in this location, but my assessment is that heat is probably more of an issue.  The choice of a polymer electrolytic will be much more resilient to drying out and also much more efficient with a negligible ESR and as a result able to handle high ripple currents.  Anyway time will tell!

Here is a 20v 120uF version of the capacitor I used.

https://au.mouser.com/ProductDetail/Panasonic/20SEPF120M%2bTSS?qs=sGAEpiMZZMvR1wrj203KOKNYKQR8GUMBFrLXFDxoKrA%3D

[thgoebel]

Over more than two years, I did renew the capacitors of some hundred Shelly 2.5. Have some types of capacitors in stock (with the correct dimensions) - and a type of polymer („dry“) dielectric capacitors.

I‘m very happy to offer a replacement service and a delivery of capacitors at net costs. Please drop me a PM if interested…

[Slartibartfast]

Thanks very much for the offer, given my location in Australia I think the return postage would be more than the devices cost!  Good to see you are providing a service to those who maybe don't have the patience or skills to complete the swap outs.  Out of interest, what cap did you opt for as the replacement for the 2x 4.7uF 400v?  I was struggling to find a smaller good quality cap of the right size and the Wurth ones are a slight compromise due to their length... and they are still electrolytic which will dry out eventually!

[thgoebel]

Using 3,3μF/400V with 230V AC mains grid is an alternative. This parts are available with 6,5mm diameter and 11,5mm length. Unfortunately NOT from  well known brands. The high voltage capacitors are in more than 66% of all faulty Shelly 2.5 not defective - the 100μF/16V is the culprit!

[TheRegister Unique]

I was able to source 4.7/400 caps from China (AliExpress) of the required size, the same "brand" as used originally by Shelly. They are dirt cheap. I however wouldn't recommend using them unless you are OK to replace the caps every other year or so :-). For low-voltage caps (16V) it is possible to find polymer type which is much more reliable.

[thgoebel]

As I mentioned before: The two HV capacitors are not the culprit - it’s the 100μF/16V. This capacitor loses capacity (values from 400nF to 50μF can be measured). With such a capacitor the step down converter cannot work constantly but falls in an intermediate mode, which causes the noise in the inductor. 12V DC are not stable and with very much ripple…

The two HV capacitors are part of the DC link prior to the step down converter. DC input voltage to the converter isn’t crucial: The step down converter is running from approximately 22V DC up to 340V DC.

With each repair, I do an ESR test with the HV capacitors. In more than 30% of all cases they are good.

Edited August 18 by thgoebel

[TheRegister Unique]

You are absolutely right - it is 16V caps that usually exhibit drastic degradation of capacity while changes in 400V ones are much more subtle. However, I've seen more than once that 400V caps replacement was required to regain full stability (in particular in i3 devices).

[Slartibartfast]

Thanks to you both, this is all good information.   I think I'll be working through the Shelly 1PM and 2.5 with the aforementioned low voltage DC capacitor replacement and, where necessary using the Wurth capacitors on the mains side (seen as I now have a small quantity of them)!

I have actually bought a small Qty of the 20v 120uf polymer type and I'll report back in a couple of weeks to say if it fits OK.  It's probably overkill (from a voltage point of view but if someone is looking for an alternative).

 

Edited August 21 by Slartibartfast

[Slartibartfast]

OK, this will be my last (planned) post on this subject.  I received the 20v 120uF polymer caps yesterday and fitted it to a Shelly 2.5, along with the Wurth 400v 4.7uf caps.  The HV caps are a tight fit with components butting up against each other, but the case sill goes on ok.  The 120uF poly cap is a shorter/fatter design (than the 180uF 16v version) and is a great fit for this application.

I have a final question regarding the ESR of the HV caps.  I notice that most datasheets for these caps do NOT give an ESR rating, why is this?  I can measure it on my meter, but without a specification how do I work out a pass/fail criterion except by comparison of new to old?

 

Size comparison of the 2 low voltage caps.

p/n 16SEPG150M = 16v 150uF = 5mm Dia x 9mm L x 2.0 lead pitch

p/n 20SEPF120M = 20v 120uF = 6.3mmDia x 6mm L x 2.5mm lead pitch