Shelly devices

[michalk-k]

I'm currently measuring the following devices:

• Shelly 1
• Shelly 1PM
• Shelly 2.5
• Shelly Dimmer (v 1)
• Shelly i3
• Shelly 3EM
• Shelly EM
• Shelly Duo GU10
• Shelly Pro 3EM

Removed Shelly UNI and Shelly RGBW2 from the list since both are DC-powered only.

[bramstroker]

I just got the Shelly Pro 3EM delivered btw, so also measured it using Zhurui PR10, before installing it.
Measured 90% of the time 1.04W, jumping to 1.05 every now and then.
Curious what figures you get.

[bramstroker]

I will also check with mini PM gen3 tomorrow, got that device yesterday as well.
Did you only connect C clamp or also A and B?

[michalk-k]

Yeah, all 3 clamps are connected.

This is interesting; consumption depends on enabled interfaces (200 samples / sleep 0.5 sec).

eth+wifi_ap+wifi_clnt+bt: 1.379W
eth+wifi_clnt+bt: 1.3395W
eth+bt: 0.979W
eth: 0.930W
eth+wifi_clnt: 1.3175W
wifi_clnt: 1.0225W

What are we going to do with this?

Let alone, that it's expected that consumption will depend on scripts running too.

[bramstroker]

We could just provide 7 sub profiles where the user can choose from.
https://docs.powercalc.nl/library/sub-profiles/
I'm planning to connect via Eth only, so that profile will be applicable for me.
What do you mean with scripts?

[michalk-k]

All gen2 and gen3 devices can run JavaScript scripts inside. It is taxing on the device's CPU as any computing. Hence higher consumption.
While higher consumption is obvious, I have no idea how much.
For example, scripting is needed on gen2 to make BLE gateway from them. In gen3 the BLE gateway is built-in and may be turned on/off.
Also, have no data on how many people are using scripting.
There are more use-cases for sure

Ahh and I expect TLS will increase consumption for sure :)
Covering all possibilities is almost impossible. We are doing the best approximation. But let's not overdo that.

[bramstroker]

Yes, I would say the best we can do is provide some profiles for the common use cases.
I highly doubt people will bother with custom javascript besides the built in BLE proxy. When you have HA at your hands to do anything you want with the device regarding automations.

[michalk-k]

I think we need another device_type. it would work like power_meter but it's not. It doesn't measure anything. Also it has no relays. It's just a controller. I'm thinking about Shelly i3 and i4. Those have only inputs performing some actions based on them (sending events, switching scenes etc). What might be more challenging is, that it must not have any associated entities. It might generate events only.

For such cases would be comfortable to select a device instead. A tricky is, that device cannot exist without entities, lol.

[bramstroker]

I have added this device_type last week which is a perfect fit imo: https://docs.powercalc.nl/library/device-types/generic-iot/

[michalk-k]

Hahaha. you are so foreseeing

[michalk-k]

BTW Is there any benefit of using power_meter since we can use generic_iot ? Especially in context of devices like Shelly 3EM / Pro 3EM ?

[bramstroker]

No, but it's easier to find in the library to have it categorized.
No functional benefit besides that.

[Scales82]

Can I ask what purpose does adding the shelly devices to powercalc do? I already have the power being reported by each shelly module (in applicable PM models)

[bramstroker]

It'a about the self consumption of the devices.
They provide a power meter, but it would only report the power of connected appliance.
That's what the powercalc profiles for the smart switches provide, to also have a power sensor for the self usage.
It's not a lot but it will add up when you have tens of them, which a lot of smart home enthousiasts have.

[bramstroker]

Btw I have also added an option to exclude certain device types from discovery.
When you are not interested in the above you can simply exclude it from being discovered.

https://docs.powercalc.nl/library/discovery/#excluding-device-types

Not released yet, but you can use this after next release of Powercalc.

[michalk-k]

The self-consumption is not as negligible as one might think

[mdb17]

I added the Shelly Duo RGB and when the light is in standby it reports 0.65 watts and 4.8 watts when turned on. Is this correct? The bulbs own power monitoring already reports 8.6 watts when it is on.

[bramstroker]

Which "own power monitoring", I don't understand.

[michalk-k]

He probably means monitoring (sensors) provided by the bulb itself.

@mdb17 could you please provide link for this bulb (best from kb shelly pages). I cannot see Shelly Duo RGB. There is RGBW.
https://kb.shelly.cloud/knowledge-base/shelly-bulb-duo-rgbw

It might be interesting. It seems that this bulb appearing with different sockets (GU10 and E27) are identified with the same identifier. As the table from KB shows, they have different power ratings and brightness

When measuring GU10 version I took the identifier reported by the bulb to core HA integration.
Also, GU10 doesn't report the power while providing an endpoint for it.

@mdb17can you open page http://your_bulb_ip/shelly ?
GU10 is identified by SHCB-1. You will see how your bulb is identified in the response.
Also please paste here the name of the device reported by HA on the device page.

[bramstroker]

@michalk-k Thanks for chiming in. Probably SHCB-1 is not unique enough, curious which output @mdb17 gets on the API endpoint.

[michalk-k]

IMO it's a normal sensor created by HA. it's also available in the rest API under ip_address/status
In the case of my GU10, HA has created the power sensor but makes it hidden by default. And this sensor didn't report any value (I think it stay 0).

Now I'm looking at REST API and indeed, I can see the power reported. Seems I have to re-check it in HA.
Once mdb17 tells us the identifier of his bulb and confirms its exact model, we may start thinking about what to do next.

[mdb17]

I apologize for the confusion and late response I have been slammed all day. Here is my identifier information (with the mac address removed just to be safe):
{"type":"SHCB-1","mac":"","auth":false,"fw":"20230913-111548/v1.14.0-gcb84623","discoverable":false,"longid":1,"num_outputs":1}

I am using the E27 bulbs which can be set inside the web gui as the bulb itself doesn't know what it is and must be set when setting it up. One thought I have would be to treat it like same as say a 1PM and leave the device power as a fixed amount with the assumption that the processor consumption does not increase when the bulb is on.