Support for Shelly 3EM

[goodkiller]

Hello

#11909

Seems like Shelly 3EM is not supported yet. Shelly 3EM is 3 phase energy monitor. It's very similar as regular EM and 2.5.

Unfortunately both templates does not work currently well. 3EM is using ADE7880 chip.

Here is some information what I already digged out.

A terminal is connected to ADE7880 > pins 7 & 8
B terminal is connected to ADE7880 > pins 9 & 12
C terminal is connected to ADE7880 > pins 13 & 14
N terminal is connected to ADE7880 > pins 15 & 16

Some library to communicate with ADE7880
https://github.com/NazimBL/ADE7880

ESPhome component
https://github.com/michaelpiron/esphome/blob/dev/esphome/components/ade7880/ade7880.h

Many thanks!

[vtainc]

Any news about support this device? Is it possible or will be possible it the feature? I also have this device and want to go to tasmota firmware.

[plewka]

ADE7880 looks interesting from datasheet view.
"Does not work well"...what does this mean?

[sfromis]

It means that goodkiller just tried templates for very different devices, and found them to not be working for this one.

Getting this device supported is up the usual question of someone with the right combination of hardware, interest, skills and time sits down to do the work, an submit a PR.

[plewka]

Ok, I'm interested in that one, too.
Would be nice to know if the ADE is at least found on I2C bus. It could be useful to follow
the i2c tracks on the PCB...ADE-Pin37/38 to the ESP...

[goodkiller]

Hello. Thanks for showing up interest about this device. Those PCB tracks are so tiny, hard to trace it down. I suggest to get one of this device, how else you can try to dev with it :)

[plewka]

You're right. I'd take a modern multimeter ;-)
I'm fluctuating between a low cost version like that and a version which is matching a solar controller...
Tasmota is preferred of cause.

[goodkiller]

Can you share which one is low cost version of it?

[plewka]

Shelly is anounced ~€100 until tomorrow in their shop.

That's quite cheap with this compact form including the transformers and really easy to handle.
Shelly maybe is not a "premium manufacturer" but I was very happy with them up to now and the
controller does full sampling of faster harmonics.
If you look for compatible and useful meters for Kostal they are at €360+ if I remember correctly.
One or two of these are supported in Tasmota as far as I remember.

[michaelpiron]

Hi @goodkiller ,
Thanks for opening this discussion. I see you referenced my ESPhome code for the ADE7880 (the energy monitoring chip that’s inside the Shelly 3PM), thanks for that.
Please note that this code is not tested yet.
I created the code by modifying the existing ADE7953 (Shelly PM single phase) integration for ESphome, and taking into account the ADE7880 datasheet specifications.

The reason I started to develop this integration is the fact that I have a 3-phase grid connection without neutral (3x230V, 230V between phases). This is not supported by standard Shelly firmware, however the ADE7880 supports it. Basically, a specific register inside the ADE chip has to be set to a specific value. (Datasheet: https://www.analog.com/media/en/technical-documentation/data-sheets/ADE7880.pdf )

Currently I’m still stuck with some issues in setting up the integration. If someone could help me out with that, it would be a win-win: not only my ESPhome integration would be ready then, it would also be a solid base for a Tasmota integration.

The issues I’m still facing:

• Which are the connections between ESP and ADE chip? I have to know which GPIO pins of the ESP are connected to the IRQ’s of the ADE. I can’t see this visually on the PCBs without desoldering the top and bottom PCB, and even then I’m not sure if it can be deduced from the PCB tracks.
• physically: which type of pins/headers do I need to flash the 3EM. The connection header is rather small.
• There are quite some calibration constants to figure out. Anyone who is familiar with this or who could reverse engineer it from the Shelly 3PM firmware binary?

[michaelpiron]

Setting the GPIO linked to the ADE’s IRQ in a correct way is important, as wrong configuration can lead to heating of the device and potential fire hazard! Refering to this related issue: #7991 (comment)

[Axel830]

I took a series of pictures of the PCB an stiched them together. Additionally I measured som connections of the ADE7880:
ADE Pin 29 (/IRQ0) -> ESP8266 Pin 12
ADE Pin 32 (/IRQ1) -> ESP8266 Pin 24
ADE Pin 36 (SCLK/SCL) > ESP Pin 9, via R36 to VDD
ADE Pin 37 (MISO/HSD) > NC
ADE Pin 38 (MOSI/SDA) > ESP Pin 10, via R35 to VDD
ADE Pin 39 (/SS/HSA) > via R46 to VDD

Perhaps this helps.

[michaelpiron]

Fantastic! Thank you @Axel830 ; based on these pin mapping I can continue my effort on an ESPhome integration/component

[sfromis]

I don't have the hardware.

1. You can usually use a multimeter in continuity mode (beeping) to trace what's connected to what.

2. That 6-pin header row looks a lot like what Shelly is using on other devices. It usually works with something like the wires from a solid core Ethernet cable. On Shelly 2.5 (very similar headers) I used the metal parts of a Dupont female header, twisted to push into the small holes. Worked perfectly.

3. Calibration constants should be derivable from the specs (did not read them), and using a known good meter for calibration. The latter is the usual calibration procedure for Tasmota, having the user input the correct value, and let the driver calculate the constants to match.

[sendingstrings]

I've got a 3EM in hand. Handy with a meter. How can I help?

[barbudor]

Identify the circuit / draw the schematic of all connections between the ADE7880 and the ESP

[goodkiller]

Any progress with it? Community, lets pump it up! :)

[barbudor]

Would you give me a unit ?
No plans to spend 120€ on it just to please the community for now 😂

[boehla]

I would give you one, if you want to do it.

[barbudor]

I don't have much time now but would be happy to help in March.

[AndreKR]

I have ordered one with the intention of dumping the I2C communication, but it's on backorder and won't arrive until next month.

[degreal]

Shelly is advertising this product that it can detect leaks and energy theft using optional neutral clamp but failed to deliver. I'm starting to think Tasmota is the answer, Shelly has not responded to this issue since July.

[barbudor]

I'm not sure that Tasmota would be the solution either.
Tasmota model would be to mesure the power on each phase.
I'm not sure how to detect a leak/theft unless all your equipment is mono phase and distributed on the 3 phase and using neutral. In that case the sum of powers on 3 phase should be equal to the power on neutral, wouldn't it?
A 3 phase equipment such as a motor would be connected to the 3 phases but not to neutral. Would it be considered a theft?

[AndreKR]

I got my device and did a little bit of research.

Obligatory safety warning: Do not try to repeat this without the necessary equipment. There are voltages inside the Shelly that can kill you. I powered the Shelly off an isolation tranformer, ran the USB through a USB isolator and put the whole setup into an IKEA SAMLA box before plugging it in.

Note that I had the current transformer flipped, so everything is reported as negative. Oopsie, sorry for that.

First some more photos of the PCB.

Overview front and back:

The core part magnified:
The backside is mirrored and aligned, so you can flip between them to follow the traces through vias. It seems to be a 4-layer board though. On the back side photo I took the RTC battery out so we can see what's underneath.

Pinouts for reference:

Datasheets: ADE7880, PCF85363A

I took three captures, one while having the ESP externally powered by 3.3V (the power rails of ESP and ADE are not directly connected, but it seemed to answer anyway), one with an incandescent light bulb (242 V, 0.11 A, PF 100, 27.9 W) and one with a CFL lamp I had lying around (242 V, 0.09 A, PF 52, 11.5 W).

Here are the original captures, they can be opened with Saleae Logic 2:
captures.zip

In general the communication looks like this:

1: After powerup the ESP will first talk to the PCF85363A RTC (7-bit I2C address 0x51) that is also on board. Ignoring this here.
2: Following a low-pulse on IRQ1 (possibly triggered by that) it will talk to the ADE7880 (I2C addr 0x38).
3-7: After that there's a pulse on IRQ0 once every second that seems to trigger some communication. The first two conversations (3 and 4) are different, all of the rest look very similar to each other.

As described in the datasheet on page 76 the ADN follows a register read/write pattern for all(?) I2C communication, so I have interpreted the I2C communication in terms of register reads/writes like this:

rel_time addr dir   reg    data
    7.092 0x38 read  0xE503 0x01A08000
    7.093 0x38 read  0xEA01 0x35
    7.093 0x38 read  0xE9FE 0xE503
...

i2c.txt

Those only show the communication with the ADN and you should be able to easily identify the conversations from the relative timestamps in the first column.

The datasheet (p. 79) describes a burst mode to read multiple registers. I believe in my file those would simply show up as a long read, but the longest read I observed was 4 bytes, so it doesn't seem to use the burst mode. (Of course those 4-byte reads might be multiple 1-byte or 2-byte registers, I didn't check the register lengths.)

[arendst]

Just ordered one and will try to write a driver based on yuor findings.

[arendst]

Based on the above information I wrote the driver for the Shelly 3EM (ade7880). It is default disabled (USE_ADE7880) so you'll need to compile it first.

I haven't tested it as I do not have the hardware yet so if someone is willing to, be my guest and report back.

Remember to make a firmware copy first before installing tasmota.

[arendst]

Latest PR results in this:

Still need to do calibration and GUI change like Sonoff SPM:

[goodkiller]

Did you published a branch with this example?

[arendst]

As always in development

[arendst]

Did some research and for now it's very important to make a 4M firmware dump BEFORE you install tasmota with ADE7880 support.

The Shelly 3EM firmware dump contains calibration parameters in a JSON file called calib.dat.

{
  "state": 0,
  "rms": {
    "current_a": 3166385,
    "current_b": 3125691,
    "current_c": 3131983,
    "current_n": -1474892307,
    "current_s": 1756557,
    "voltage_a": -767262,
    "voltage_b": -763439,
    "voltage_c": -749854
  },
  "angles": {
    "angle0": 180,
    "angle1": 176,
    "angle2": 176
  },
  "powers": {
    "totactive": {
      "a": -1345820,
      "b": -1347328,
      "c": -1351979
    },
    "apparent": {
      "a": 214966,
      "b": 214961,
      "c": 214949
    }
  },
 "energies": {
    "totactive": {
      "a": 8752,
      "b": 8751,
      "c": 8751
    },
    "apparent": {
      "a": 40496,
      "b": 40513,
      "c": 40524
    }
  }
}

The parameters in this file currently need to be added in to file xnrg_23_ade7880.ino. See below:

#define ADE7880_AIGAIN_INIT     3166385      // rms, current_a
#define ADE7880_BIGAIN_INIT     3125691      // rms, current_b
#define ADE7880_CIGAIN_INIT     3131983      // rms, current_c
#define ADE7880_NIGAIN_INIT     1756557      // rms, current_s !!
#define ADE7880_AVGAIN_INIT     -767262      // rms, voltage_a
#define ADE7880_BVGAIN_INIT     -763439      // rms, voltage_b
#define ADE7880_CVGAIN_INIT     -749854      // rms, voltage_c
#define ADE7880_APHCAL_INIT     180          // angles, angle0
#define ADE7880_BPHCAL_INIT     176          // angles, angle1
#define ADE7880_CPHCAL_INIT     176          // angles, angle2
#define ADE7880_APGAIN_INIT     -1345820     // powers, totactive, a
#define ADE7880_BPGAIN_INIT     -1347328     // powers, totactive, b
#define ADE7880_CPGAIN_INIT     -1351979     // powers, totactive, c

Once done the possibly negative power values turn to the correct values like

As I mounted all three current sensors on the same power line one can see I have an issue with L3 constantly showing wrong values.

I plan to release this driver on a 4M tasmota image allowing to use our filesystem to strore the calib.dat file.

[arendst]

NB. latest template is

{"NAME":"Shelly 3EM","GPIO":[1,1,288,1,32,8065,0,0,640,8064,608,224,8096,0],"FLAG":0,"BASE":18}

using the former link led also for relay state.

[rossi75]

.> {"NAME":"Shelly 3EM","GPIO":[1,1,288,1,32,8065,0,0,640,8064,608,224,8096,0],"FLAG":0,"BASE":18}

using the former link led also for relay state.

I have a question regarding the second LED, is it possible to adress it? Then it would make sense to use one LED for the relay, the other one for WiFi link (and mqtt processing).

[charly181274]

Das würde mich auch sehr interessieren wie man die zweite LED benutzen kann.

[rossi75]

Das würde mich auch sehr interessieren wie man die zweite LED benutzen kann.

I tested all gpios. But none of them lit the red led.

I assume that this led will be set and unset via the new ade7880 chip...

[rossi75]

Das würde mich auch sehr interessieren wie man die zweite LED benutzen kann.

The 3em pro version has also a led connected via the ade Chip. So i assume that without any special feature request, the red led won't lit any further with tasmota Firmware

[michaelpiron]

Great progress, thank you @arendst . With my limited, rusty coding skills this would have taken me ages... ;-)

Would it be possible to support different wiring configurations?

Standard setting

With the current/standard register settings, the ADE7880 is capable of measuring three single-phase loads or a 4-wire wye load:

Possibility to measure 3-wire delta load

I would like to use the Shelly 3EM in configurations with a 3-wire delta load (some homes have a 3-wire connection without neutral, a delta configuration). This can be enabled by setting a specific register of the ADE7880:
According to datasheet ADE7880 page 45:

Pg 50:

Looking into the AN-639 Application note:

[pieterdt]

Are you sure?

[michaelpiron]

I believe you should look at the references to the right of the “Attenuation network” blocks. The left side is the outside of the Shelly

[pieterdt]

Yes @michaelpiron , but those voltage of A and C are measured against B, due to lack of N, hence I think L2 should be connected to both B and N. Otherwise, with B unconnected, those voltage will not measure anything meaningful.

[michaelpiron]

My interpretation is that in 3phase3wire mode, the chip will measure two voltages:

• voltage between Vcp and Vn. This would be the voltage between our wires L3 and L2.
• voltage between Vap and Vn. this would be the voltage between our wires L1 and L2.

that’s why I don’t think a connection to terminal B is needed.
I would connect:

• wire L1 to terminal Va on the Shelly
• Wire L2 to terminal Vn
• Wire L3 to terminal Vc

however, I don’t think it does any harm to bridge L2 wire to terminal Vn and Vc. It’s just not necessary

[Mousketeer]

@michaelpiron @pieterdt I'm very much lost on how to connect the wiring in 3-wire delta (no neutral) mode. I've compiled a tasmota 13 with these changes (hopefully).

The config you suggested (L1-Va, L2-Vn, L3-Vc) appears to have reasonable values, except that the voltage on L2 is shown as 0V - this is not unexpected, as that sensor is disconnected, but has the effect of all the amps producing zero watts ;-).

Bridging L2-Vn-Vb together has the effect of the L2 grid frequency going wonky (96 to 198Hz, in a 50Hz country..). The voltage is also still 0.

Bridging L1-Va-Vb together gives plausible values for voltage, frequency and current, but the L2 active power (and power factor) goes negative. These negative values are subtracted from the total. This is with the clamp in the correct orientation and no power generation going on. (I guess I could invert the clamp??)

(I can't remember if I tried a L3-Vc-Vb config, I'll try again if need be).

Lastly, I tried L1-Va, L2-Vb, L3-Vn with external 3.3V DC power supply. Frequency & current appear correct. But the voltage is shown as 138V, thus the power calculations are way off. The power factor is currently also showing weirdly low values, but potentially this is just my specific load. I'll go around the house with a resistive load later to test.

So all in all no correct configuration thus far.
I see two more options:

1. Use the suggested L1-Va, L2-Vn, L3-Vc, but somehow guesstimate the voltage - the phase shift is constant and the voltage doesn't differ much between phases.
2. Perhaps somehow use ground (the literal earth connection, so this is not the neutral as I don't have one) as a reference to the configuration with the external 3.3V source? This surely violates some safety codes, but as long as there is no current...

But I wouldn't know where to start... Any help welcome :-)

[goodkiller]

Well.... thats funny now. I installed my 3EM to electricity enclosure and did proper installation. After trying to install custom firmware, device resets itself. This happens after applying template ({"NAME":"Shelly 3EM","GPIO":[1,1,288,1,32,8065,0,0,640,8064,608,224,8096,0],"FLAG":0,"BASE":18}):

00:00:00.001 HDW: ESP8266EX
00:00:00.050 CFG: Loaded from flash at F8, Count 12
00:00:00.055 QPC: Count 1
00:00:00.057 CFG: CR 340/699, Busy 0
00:00:00.062 ROT: Mode 1
00:00:00.071 SRC: Restart
00:00:00.073 Project tasmota - Shelly 3EM Version 11.0.0.3(QW-3EM-TEST)-2_7_4_9(2022-03-09T13:24:15)
00:00:00.226 WIF: Checking connection...
00:00:00.227 WIF: Attempting connection...
00:00:00.563 WIF: Connecting to AP1 IoT Channel 6 BSSId 32:23:03:E0:D9:98 in mode 11n as tasmota-D68BD5-3029...
00:00:02.426 APP: Button1 multi-press 1
00:00:02.427 WIF: Checking connection...
00:00:02.428 WIF: Attempting connection...
00:00:03.427 WIF: Checking connection...
00:00:03.428 WIF: Connected
00:00:03.680 HTP: Web server active on tasmota-D68BD5-3029 with IP address 192.168.1.193
00:00:04.432 WIF: Sending Gratuitous ARP
00:00:04.433 NTP: Sync time...
00:00:04.481 RTC: UTC 2022-03-09T11:29:18, DST 2022-03-27T02:00:00, STD 2022-10-30T03:00:00
12:29:18.001 RTC: Synced by NTP
12:29:18.206 RSL: INFO1 = {"Info1":{"Module":"Shelly 3EM","Version":"11.0.0.3(QW-3EM-TEST)","FallbackTopic":"cmnd/DVES_D68BD5_fb/","GroupTopic":"cmnd/tasmotas/"}}
12:29:18.211 RSL: INFO2 = {"Info2":{"WebServerMode":"Admin","Hostname":"tasmota-D68BD5-3029","IPAddress":"192.168.1.193"}}
12:29:18.222 RSL: INFO3 = {"Info3":{"RestartReason":"Software/System restart","BootCount":5}}
12:29:18.230 RSL: RESULT = {"POWER":"ON"}
12:29:18.233 RSL: POWER = ON
12:29:20.441 QPC: Reset
12:29:22.409 APP: Boot Count 5
12:29:22.670 HTP: Main Menu
12:29:23.267 CFG: Saved to flash at F7, Count 13, Bytes 4096
12:29:23.382 RSL: STATE = {"Time":"2022-03-09T12:29:23","Uptime":"0T00:00:11","UptimeSec":11,"Heap":28,"SleepMode":"Dynamic","Sleep":50,"LoadAvg":30,"MqttCount":0,"POWER":"ON","Wifi":{"AP":1,"SSId":"IoT","BSSId":"32:23:03:E0:D9:98","Channel":6,"Mode":"11n","RSSI":76,"Signal":-62,"LinkCount":1,"Downtime":"0T00:00:05"}}
12:29:35.348 HTP: Main Menu
12:29:36.894 HTP: Configuration
12:29:37.454 WIF: Checking connection...
12:29:37.826 HTP: Configure Module
12:29:39.032 HTP: Configuration
12:29:39.132 HTP: Configuration
12:29:56.815 HTP: Configure Other
12:29:57.425 WIF: Checking connection...
12:29:58.861 RSL: RESULT = {"Reset":"Reset and Restarting"}
12:29:59.206 CFG: Use defaults

What makes it odd is that before installation (using only one phase) works well...

[arendst]

I think the I of IO needs to be connected.

[arendst]

Any phase will do

[goodkiller]

I don't. I is Line input for contactor control. It's isolated. 3EM gets it's power from V(A/B/C).

[arendst]

You're right.

What is the output of command i2cdriver?

[goodkiller]

19:08:41.657 CMD: i2cscan
19:08:41.676 RSL: RESULT = {"I2CScan":"Device(s) found at 0x38 0x51"}

[arendst]

Something different. As Tasmota currently cannot calibrate the device it's important to start the Shelly firmware at least once.

Then add the device to your network and access webpage http:///calib.dat. Store the information in a safe place as it is needed again when Tasmota is installed.

[AndreKR]

https://shelly.cloud/knowledge-base/devices/shelly-3em/#tab_624733fb153bf

[kopfkraft]

thanks for the info!!! will give it a try!

[AndreKR]

@kopfkraft As usual please disconnect mains power before connecting anything to the programming port.

Also remember to download calib.dat from the original Shelly firmware before you flash Tasmota.

[pieterdt]

@sfromis , which one would be gnd? the one on the side of the sensors or the side of the line connections?

[pieterdt]

nvm found it by googling for pinout

[meidlinga]

I flashed a build with these flags enabled via OTA, applied the template, the calibration data as Rule and enabled Option 21. (many times :))
It startet to work only, after I went to Configure module and saved it once.

I am not sure why, but wanted to leave this here for people struggling at the same point.

[goodkiller]

How did you get calibration data. Any instructions?

[barbudor]

As written above:

Also remember to download calib.dat from the original Shelly firmware before you flash Tasmota.

[meidlinga]

@goodkiller This part is also described here: https://templates.blakadder.com/shelly_3EM.html

[ArturZurawski]

Do I have to anything else than what's described in https://templates.blakadder.com/shelly_3EM.html ?
Run SetOption21 1 in console.
Flashed with http://ota.tasmota.com/tasmota/tasmota.bin
Can't see any readings :(

[muebau]

Well I am not sure if I understand it correctly. I used the Shelly 3EM with the original firmware for the last ~6 month and got positive values every second.
Now with Tasmota (with calib.dat) I get negative values without changing the clamps. I guess this means the clamps are installed 'correctly' (according to the original firmware).

As I need the data for a scientific project I thought to go back to the original firmware as I was not able to set the update interval to 'every second' like it was before with the original firmware (TelePeriod 10).

I have a normal setup (no solar). I will check the direction of the clamps (even if it was fine with the original firmware).

[AleXSR700]

I will look up the tool to revert to Shelly's stock and post it either tonight after work or this weekend.

What I could imagine is, that Shelly changed the logic and that is why for some devices the clamp needs to be installed according to instructions and for some the opposite direction. And tasmota always uses the same logic.

But that is pure speculation and why I reocmmended to flip them and then attach a know device to ensure that the measurements themselves are accurate afterwards.

P.S.: Kind of a stupid question, but when you copied the values to the compilation options and to the rule, you checked that the numeric signs of your original calb.dat and the template are the same, right? Really just covering all bases here :D

[muebau]

Thank you lot. Yes I checked everything a few time as I also expected me for this mistake ;-). I think I will check with the original firmware and than with Tasmota again to make sure I don't miss some stupid mistake.

In the end I need to figure out how to get a full picture (power, power factor, voltage ...) every second as it was with the original firmware. If I am lucky I get the harmonics working to.

[AleXSR700]

As I said, I would try the clamps first because tasmota does give you a lot of options on how to handle your data.
You could even filter what you publish via MQTT and probably even publish as quickly as you like (basically real-time) using rules.

You may also be able to get real-time data with something like SetOption108. But I cannot test myself right now, so maybe that is not the one.

If you decide to stick with Tasmota, the discord channel also offers excellent expert support. :)

[AleXSR700]

https://github.com/yaourdt/tasmota-to-mgos

[hagenbuch]

It is very unlikely that the meter provided by the grid company would not balance the phases. It does! If you have an OBIS / SML meter it will sum up consumption in cells 1.8.x and the production in 2.8.x.

So if you consume 1000 watt on phase 1 and 2 each and produce 5000 watt on phase 3, it will be counted as 3000 watt production.

Tasmota's set of counters work differently: The main counter goes up and down according production and consumption, but then there are the "ExportActive" counters that go only even up (integrate) but integrates only at moments where there is a net export. By some calculations, you could infer the official 1.8.0 and 2.8.0 counters, too.

[ThiloGa]

The main counter goes up and down according production and consumption, but then there are the "ExportActive" counters that go only even up (integrate) but integrates only at moments where there is a net export.

My 3EM works fine with tasmota, beside one little issue:

export active is always 0.000kwh, even engery today is going down.

Any hint?

[muebau]

Did somebody try to get the harmonics part of the ADE7880? It might be interesting to play with this information.
I have my shelly 3 EM not flashed yet as it just arrived.

[muebau]

I read the source code and now the harmonics part of the datasheet. It looks like you cat get just 3 of the harmonics at a time. I am not sure if it helps if one would like to get the full spectrum (all 63 harmonics) for all the phases at once.

Does somebody know if there would be a way to switch trough all the phases with all their harmonics to get the full picture lets say every second?

EDIT: As the ADE7880 needs 500ms to settle after a new harmonic is selected the maximum rate is 3 harmonics every 500ms. This way a full picture would need ~95 sec (>1.5 min).

[muebau]

What would be a good strategy to implement a simple read of one harmonic and give it to MQTT?

Is there some code in the project where I could read and understand how something like this could be implemented in Tasmota?

[silentrain23]

Hi guys. I am trying to use a bigger CT with a shelly 3EM. Standard ones are 120A/40mA. The ones I am trying to configure is 400A/40mA. The readings should be 3.33 times bigger but I am not able to callibrate the device with the new CT. Any idea?

[yettiz]

Hi,
I have Tasmota on the Shelly 3em, and it also works,

Volage, Frequency are correct, the Current I dont now
but Active Power, Apparent Power and Reactive Power are wrong, at least at phase 3

The wiring must fit because everything was displayed correctly before the switch to Tasmota.

My rules3:

rule3 on file#calib.dat do {"rms":{"current_a":3019559,"current_b":3362201,"current_c":1329104,"current_n":1219168,"current_s":266669599,"voltage_a":-700535,"voltage_b":-695400,"voltage_c":-687116},"angles":{"angle0":156,"angle1":168,"angle2":168},"powers":{"totactive": {"a":-4279903,"b":-3932160,"c":-7019040}}} endon

How can I fix it?

[yettiz]

OMG, the Shelly 3em run 2 Weeks, today I found out that the shelly is no longer running, the fuse has blown and it has dismantled the shelly. My Heatpump wasn't running.
I guess I'm not the first this has happened to.
I have something to do again tomorrow, expand and send in.
Oh great.

[AndreKR]

You're not trying to switch a heatpump with the 3EM, are you? The built-in relay is intended to switch a contactor which then switches the load (e.g. a heatpump).

[rossi75]

OMG, the Shelly 3em run 2 Weeks, today I found out that the shelly is no longer running, the fuse has blown and it has dismantled the shelly. My Heatpump wasn't running. I guess I'm not the first this has happened to. I have something to do again tomorrow, expand and send in. Oh great.

I had the similar issue. It was running with stock firmware. Sent it back to the vendor and received a new one

I assume Heatpump and 3em were on the same breaker, the Heatpump wasnt switched through the 3em

[yettiz]

You're not trying to switch a heatpump with the 3EM, are you? The built-in relay is intended to switch a contactor which then switches the load (e.g. a heatpump).

My Heatpump has an low power switch, so that the energy supplier (EVU) can switch off the heat pump at off-times. And i can manage my Heatpump, and i can switch my Heatpump via IO-Broker. So i dont need the Switch of the Shelly. I need only the power measurement data from the Shelly.

[yettiz]

I got a refund, so i can buy a new one.
But I still don't know if I should buy the 3em again or 3em Pro, because the 3em Pro probably doesn't have this problem anymore.
However, Tasmota is currently being developed:
#18125

[001oc]

Hi,

I am trying to flash my 3EM. I have compiled (vscode, pio) the latest tasmota version with #define ade7880. After that I try to upload it to via my ftdi adapter connected to the 1,27mm header at the side of the 3EM. I have, of course crossed tx and rx. I get constantly "A fatal error occurred: Failed to connect to ESP32: No serial data received." Of course I have changed the tx and rx but with straight tx to tx and rx to rx nothing happens at all, not at least "No serial date received". I believe I cant set 3EM into bootloader mode. I have tried to have GPIO0 connected to GND while booting or connect RESET to GND while having GPIO0 even connected to GND. Nothing puts the 3EM into bootloader mode. Is there a specfic baud rate the 3EM is communicating. When I open the serial monitor (vscode) I get solely rubbish but no readable infos.

Maybe I missed something but I have not found any specific hints how to flash an 3EM via usb adapter and interface at the side of the 3EM.

Thanks

[sfromis]

Looks like you are trying to use it as ESP32. The Shelly 3EM is an ESP8266 device, needing a binary to match, and making sure not to "suggest" ESP32 when trying to flash. Baud rate of Tasmota during boot is 115200, but initial messages from the ESP8266 are on 74880. I've flashed other Shelly ESP8266 devices using the connector, and the usual pins, tx/rx/vcc/gnd/io0, without trouble.

Seeing "FTDI" sounds like you may be using one of those common fakes (usually a red board), not to be trusted for reliable flashing.

[001oc]

@sfromis Thanks for your reply.

I think I have compiled it for ESP8266 but I will check that again. My "FTDI" board is a red one but I use it successfully for some years now. I will check that too. My platformio_override.ini has #define upload_speed = 115200. I will try the 74880 this evening too.

[sfromis]

Failed to connect to ESP32

And completely useless to try upload speed 74880, that number is solely for output in the initial boot phase. Anyway, not having failures yet with a borderline board does certainly not mean "all ok".

[001oc]

Well, my fault :-). Rarely the is a bug in tasmota, mostly it is because of not reading the massiv documentation or doing such failures I did lately. As usual tasmota rocks. I had messed up the configuration abit - with the hints from @sfromis I was able to compile and upload with ease. I still had to reset more than once to get into the bootloader but after that upload was working and 3EM is running with tasmota now.

Thanks for the fast and competent help.

concerning the donated cents ... over the years I have gotten many useful help from the tasmota crew and I am gladly being ready to spend some test devices to the tasmota crew but dont know how to do that safely.

[Jason2866]

if we got a cent donated for the sentence "it worked before" Tasmota crew could by many test devices ;-)

[kpfleming]

FYI: After noticing that the most recently Shelly firmware has a 'Clamp Calibration' function, it became clear to me that at least some of the values in calib.dat are related to the clamps themselves, and not the 3EM board in use. I'm going to contact Shelly to find out which of the values are dependent on which parts of the equation, because I don't remember the clamps even being labeled A/B/C when I got my units (and I probably mixed up clamps between units too as I installed multiple units at once).

[hagenbuch]

My clamps (which i bought approx. a year ago in Germany) are clearly labeled A-C - but I could not tell if that had been true for earlier purchases. I'ts a bit weird because if you have the exact same number of windings on all the clamp coils you would not need individual calibration to my knowing. It could however be that they don't measure with a coil but with a magnetic sensor, then it's understandable.

BUT: My clamps have only two contacts in their plugs (of four possible), this clearly indicates a coil.

[kpfleming]

I checked some of mine, and some are labeled A-C and others are not, so it appears to have changed at some point.

As far as calibration... I suspect there are minor variations in manufacturing and that results in the need to calibrate each coil for the phase angle.

[hagenbuch]

Right! I think no matter what technology they might be using now or in the future, it is always a good idea to make everything configurable, especially 100% of the chip's config registers.

[mof0rlive]

Hi, I'm using a Shelly 3EM with Tasmota version 13.1.0 for all my 3 phases (one with a solar energy plant).
Used the template from https://templates.blakadder.com/shelly_3EM.html

I did not notice any errors when importing the calibration.

Does anyone have the problem that the value (Energy Today / Total) becomes negative when the solar system overproduces?

My solar system is on L1 and if its consumption value is negative I can watch the total value decrease until it is negative.

Is it possible that negative consumption values are not included in the total? Ty.

[AleXSR700]

One might not need to ignore the negatives but work with abs() instead if one would want to monitor the total consumption.
Right now Total would be Grid Delta in your setup.

[mof0rlive]

Ty for your answer. Can you further explain what you mean with abs().

[rossi75]

abs means the absolute value of a value: abs(-7)=7

[mof0rlive]

Is this possible in Tasmota Config (and how)? By editing the rule or the template?

[jossie67]

I did someting stupid , I flashed my Shelly 3EM ota with the file shellyem3-esp8266-0.4.6.zip that worked fine.
Filled in the templete
" {"NAME":"Shelly 3EM","GPIO":[1,1,288,1,32,8065,0,0,640,8064,608,224,8096,0],"FLAG":0,"BASE":18}"

Then I upgrade to the latest version 13.2.
Now I can see any reading any more

I reset the device with Reset 5 and started over but still no reading.

_```
00:00:00.001 HDW: ESP8266EX
00:00:00.038 CFG: Loaded from flash at F9, Count 13
00:00:00.044 QPC: Count 1
00:00:00.065 Project tasmota - Tasmota Version 13.2.0(tasmota)-2_7_4_9(2023-10-19T09:02:07)
00:00:00.555 WIF: Connecting to AP1 Home Channel 1 BSSId BA:FB:E4:91:46:8B in mode 11n as tasmota-DCA4AA-1194...
00:00:01.803 WIF: Connected
00:00:02.003 HTP: Web server active on tasmota-DCA4AA-1194 with IP address 10.10.0.108
20:11:54.010 RSL: INFO1 = {"Info1":{"Module":"Shelly 3EM","Version":"13.2.0(tasmota)","FallbackTopic":"cmnd/DVES_DCA4AA_fb/","GroupTopic":"cmnd/tasmotas/"}}
20:11:54.015 RSL: INFO2 = {"Info2":{"WebServerMode":"Admin","Hostname":"tasmota-DCA4AA-1194","IPAddress":"10.10.0.108"}}
20:11:54.025 RSL: INFO3 = {"Info3":{"RestartReason":"Software/System restart","BootCount":4}}
20:11:54.033 RSL: RESULT = {"POWER":"OFF"}
20:11:54.037 RSL: POWER = OFF
20:11:57.445 QPC: Reset

 What do I do wrong

[AleXSR700]

I am not sure what you did, but if I were you I would perform reset 0 and then enter all information again.
But you must power cycle the device if you reset, otherwise it will not behave as expected.

If that does not work, post status 0 :)

[rossi75]

I even more suppose you flashed a Version without the adc driver.

What the 3em needs is a special compiled Version of tasmota.

Have a look for the channel bitbastelei at YouTube for a good explanation of this

[jossie67]

Yes I did, I flashed it with the mini and then with version 13.2.
I tried to flashed with the bin file from the shellyem3-esp8266-0.4.6.zip, that failed.
Is there a bin file that I can uses to flash it?

[AleXSR700]

Version 13.2 from where?

@rossi75 meant that you need a custom build for the Shelly 3EM. The tasmota stock firmware that you can download from tasmota github/servers does not include the drivers for the Shelly 3EM power monitoring.

I did not find the link from my mobile with all the info.

But essentially you either need to build your own firmware with the options set (search the Shelly 3EM github thread) or find a precompiled one.

It is described in the video that @rossi75 mentioned at ~28 min (but that is in German).

You need to compile using #define USE_ADE7880

[jossie67]

Thanks, I will use TasmoCompiler > https://github.com/benzino77/tasmocompiler
Add
#define USE_ADE7880
in user_user_config_override.h

[fpdragon]

Hi!
I flashed and calibrated the Shelly according to the instructions and so far everything seemed to work with the light bulb test. Now I've connected the 3EM to my 3-phase heat pump and I'm confused. All 3 voltage phases and current clamps checked again. Current clamps point from the house to the power plant. Nevertheless, L1=979W L2=-337W L3=-694W. So two out of 3 phases are negative even though there is no PV system and the three current clamps are all mounted in the same direction. What is this going on? Thx.
here is my calibration:
rule3 on file#calib.dat do {"rms":{"current_a":3200723,"current_b":3248620,"current_c":3226184,"current_n":758768,"current_s":1639104,"voltage_a":-699675 ,"voltage_b":-718111,"voltage_c":-713908},"angles":{"angle0":184,"angle1":184,"angle2":192},"powers":{"totactive":{ "a":-1392083,"b":-1388311,"c":-1384500},"apparent":{"a":617944,"b":617931,"c":617921}},"energies" :{"totactive":{"a":25098,"b":25097,"c":25097},"apparent":{"a":63174,"b":63212,"c":63257}} ,"freq":0} endon

[hagenbuch]

Are you sure that the voltage for L1 is measured at the same phase that the clamp L1 belongs to - and so on for L2, L3? Have they all been calibrated that way?

BTW in a three phase AC system, negative voltages make no sense. Is your "N" properly connected? Do you measure three phases or even current through N? Frequency = 0 looks weird.

[fpdragon]

Are you sure that the voltage for L1 is measured at the same phase that the clamp L1 belongs to - and so on for L2, L3? Have they all been calibrated that way?

BTW in a three phase AC system, negative voltages make no sense. Is your "N" properly connected? Do you measure three phases or even current through N? Frequency = 0 looks weird.

Thanks for your answer!

I connected 4 cables: L1, L2, L3 and N and double checked each. I even dropped the braker for each L on their own to make sure that the order is correctly.
Voltage is correctly displayed.

For the clamps: I also double checked the clamps. L1 -> A clamp, L2 -> B clamp, L3 -> C clamp. And for all clamps the arrow shows reverse from the house to the power plant.

This is what I get:

Another strange thing is that the sum of all three phases is always about 0 W.

[001oc]

I had the problem with one of my hall-sensors too. I checked if there is no reverse current and after that simply turned the sensors arround. After that I checked with a multimeter that current is (nearly) the same to the 3EM. Since then it works for me ... I am quit sure my sensors are mounted "backwards".

[fpdragon]

Thanks, but are they all 3 backwards or just 1 or 2 of them?

The strange thing is that one measurement seems to be positive and the other two are negative.

[001oc]

I have two of them in the "right" direction and one in the "wrong" direction. I am quit sure, that one of my sensors is mounted in the wrong direction in the factory. I thought about to exchange the sensor but I am living with my "solution" for a couple of months now and it is working ok for me that way.

[JayBraker]

Hi, sorry I am still at a loss combing through this gigantic thread.
So I have successfully flashed Tasmota 13.3.0 (tasmota-4M) onto my Shelly 3EM.

I am suprised to see the relevant Driver is not part of the firmware yet (it was developed >1 year ago and mentioned in like release ~11.1). So I guess I will need to compile it myself.
I am not sure which custom parameters are needed to support the 4MB of flash as well?

I tried using TasmoCompiler, there I selected the Energy Sensor feature as well as the SD Card/LittleFS (ufs?) feature and added the line #define USE_ADE7880 to custom parameters - the build output still mentions 1MB flash.

What is the minimal needed setup to compile tasmota to support my Shelly 3EM including the energy sensors as well as the available flash chip?
I don't mind using alternative build tools but this is my first time dealing with the compilation and I am really confused why the device isn't fully supported by mainline tasmota yet.

Really looking forward to your input, maybe I am just missing the obvious part?

Kind regards

[Th0maz]

You can select "Wemos/NodeMCU" on top of TasmoCompiler to get a 4MB image.
I think I used the following lines in my userconfig override file for the Shelly 3EM:

#define USE_I2C
#define USE_ENERGY_SENSOR
#define USE_ADE7880

[rossi75]

I am suprised to see the relevant Driver is not part of the firmware yet (it was developed >1 year ago and mentioned in like release

This is the reason why I stayed with the 3EM on the initial flashed firmware

[JayBraker]

Okay, I tried 2 ways with tasmocompiler:

• ESP8266 Generic
• Web Interface
• Timers
• Rules

#define USE_I2C
#define USE_ENERGY_SENSOR
#define USE_ADE7880

• ESP8266 Wemos/NodeMCU
• Web Interface
• Timers
• SD card/LittleFS
• Rules

#define USE_I2C
#define USE_ENERGY_SENSOR
#define USE_ADE7880

both succeed in building but
a) I cannot find any mention of ADE7880 in the build log
b) Even after I applied the template [0], the sensors are not displayed correctly - if I try to configure the template manually (GPIO05, GPIO13) I cannot find ADE7880 in the drop down

So I am afraid the driver is never included somehow?
Kind of weird

[sfromis]

If the driver is included, output from the command I2CDriver should include index 65 for ADE7880. And the command I2CScan should list address 0x38. Of course, then you should also see the ADE7880_IRQ pins in the config dropdown.

Maybe check the build timestamp on the Information page to verify that you really have upgraded to your build.

[JayBraker]

Okay, conclusion: I got it working today

1. Fired up TasmoCompiler
2. Selected Wemos/NodeMCU to utilize 4MB Flash
3. Default features
4. Custom Parameters [0]
5. Build

The build log still doesn't mention ade7880 but the Driver was included any way, when I upgraded the firmware.

Flashing process:
I tried the way described here utilizing the ota-functionality of shelly + a special prebuilt binary, curtesy of #13515 (comment) but I ended up soft-bricking my 3EM (I guess I fd up the ota url somehow). So I ended up manually flashing via serial

1. Connected serial probes (.1mm pitch for the pins.. tiny! I used needles as I didnt have such small pins), pinout taken from here
2. Bridge gpio0 and gnd, do not connect to 3.3v
3. Sacrifice power cord, connect strands to N and A
4. Attach tty adapter to gnd, tx, rx
5. Power up 3em
6. Follow tasmota installer instructions

[0]:

#define USE_I2C
#define USE_ENERGY_SENSOR
#define USE_ADE7880

[saulous]

Hello together,
i have a Problem with the Tasmota Flashed Shelly 3EM.

I have Flashed the Shelly 3EM with this Instruction : "https://www.congenio.de/infos/shelly3em.html"
I don't have self compiled the Tasmota Firmware. On this website the Firmware can be downloaded.
(At first Minimal Firmware, then tasmota.bin.gz 14.0.0 )

Both Firmware Upgrades i have installed at the Shelly.
Tasmota has been startet up.

Actually i can switch the relais in the shelly. But no Measurement will be shown. (see Picture)

i2cscan in the console :

01:36:13.351 CMD: i2cscan
01:36:13.376 MQT: stat/tasmota_755DC5/RESULT = {"I2CScan":"Device(s) found at 0x39 0x51"}
01:36:22.278 CMD: i2cscan
01:36:22.305 MQT: stat/tasmota_755DC5/RESULT = {"I2CScan":"Device(s) found at 0x39 0x51"}
01:36:23.154 CMD: i2cscan
01:36:23.179 MQT: stat/tasmota_755DC5/RESULT = {"I2CScan":"Device(s) found at 0x51"}
01:36:23.717 CMD: i2cscan
01:36:23.744 MQT: stat/tasmota_755DC5/RESULT = {"I2CScan":"Device(s) found at 0x39 0x51"}
01:36:24.282 CMD: i2cscan
01:36:24.307 MQT: stat/tasmota_755DC5/RESULT = {"I2CScan":"Device(s) found at 0x51"}
01:36:24.782 CMD: i2cscan
01:36:24.806 MQT: stat/tasmota_755DC5/RESULT = {"I2CScan":"Device(s) found at 0x51"}
01:36:25.183 CMD: i2cscan
01:36:25.208 MQT: stat/tasmota_755DC5/RESULT = {"I2CScan":"Device(s) found at 0x51"}
01:36:25.691 CMD: i2cscan
01:36:25.716 MQT: stat/tasmota_755DC5/RESULT = {"I2CScan":"Device(s) found at 0x39 0x51"}
01:36:26.621 CMD: i2cscan
01:36:26.645 MQT: stat/tasmota_755DC5/RESULT = {"I2CScan":"Device(s) found at 0x51"}
01:36:27.027 CMD: i2cscan
01:36:27.053 MQT: stat/tasmota_755DC5/RESULT = {"I2CScan":"Device(s) found at 0x51"}
01:36:27.450 CMD: i2cscan
01:36:27.474 MQT: stat/tasmota_755DC5/RESULT = {"I2CScan":"Device(s) found at 0x51"}
01:36:27.851 CMD: i2cscan
01:36:27.877 MQT: stat/tasmota_755DC5/RESULT = {"I2CScan":"Device(s) found at 0x51"}
01:36:28.251 CMD: i2cscan
01:36:28.276 MQT: stat/tasmota_755DC5/RESULT = {"I2CScan":"Device(s) found at 0x51"}
01:36:28.704 CMD: i2cscan
01:36:28.730 MQT: stat/tasmota_755DC5/RESULT = {"I2CScan":"Device(s) found at 0x51"}
01:36:29.053 CMD: i2cscan
01:36:29.078 MQT: stat/tasmota_755DC5/RESULT = {"I2CScan":"Device(s) found at 0x39 0x51"}
01:36:29.498 CMD: i2cscan
01:36:29.523 MQT: stat/tasmota_755DC5/RESULT = {"I2CScan":"Device(s) found at 0x51"}
01:36:29.906 CMD: i2cscan
01:36:29.931 MQT: stat/tasmota_755DC5/RESULT = {"I2CScan":"Device(s) found at 0x39 0x51"}
01:36:30.321 CMD: i2cscan
01:36:30.347 MQT: stat/tasmota_755DC5/RESULT = {"I2CScan":"Device(s) found at 0x39 0x51"}
01:36:30.779 CMD: i2cscan
01:36:30.805 MQT: stat/tasmota_755DC5/RESULT = {"I2CScan":"Device(s) found at 0x39 0x51"}
01:36:31.132 CMD: i2cscan
01:36:31.158 MQT: stat/tasmota_755DC5/RESULT = {"I2CScan":"Device(s) found at 0x51"}
01:36:31.534 CMD: i2cscan
01:36:31.560 MQT: stat/tasmota_755DC5/RESULT = {"I2CScan":"Device(s) found at 0x51"}
01:36:32.789 CMD: i2cscan
01:36:32.814 MQT: stat/tasmota_755DC5/RESULT = {"I2CScan":"Device(s) found at 0x39 0x51"}

The ADE Chip will be found for a short time. Is this normal ?

i2cdriver in the console :

09:58:50.830 CMD: i2cdriver
09:58:50.842 MQT: stat/tasmota_755DC5/RESULT = {"I2CDriver":"7,65"}

rule0 in the console :

01:39:20.309 CMD: rule0
01:39:20.320 MQT: stat/tasmota_755DC5/RESULT = {"Rule1":{"State":"OFF","Once":"OFF","StopOnError":"OFF","Length":0,"Free":511,"Rules":""}}
01:39:20.324 MQT: stat/tasmota_755DC5/RESULT = {"Rule2":{"State":"OFF","Once":"OFF","StopOnError":"OFF","Length":0,"Free":511,"Rules":""}}
01:39:20.330 MQT: stat/tasmota_755DC5/RESULT = {"Rule3":{"State":"OFF","Once":"OFF","StopOnError":"OFF","Length":300,"Free":211,"Rules":"on file#calib.dat do {"rms":{"current_a":3166385,"current_b":3125691,"current_c":3131983,"current_s":1756557,"voltage_a":-767262,"voltage_b":-763439,"voltage_c":-749854},"angles":{"angle0":180,"angle1":176,"angle2":176},"powers":{"totactive": {"a":-1345820,"b":-1347328,"c":-1351979}},"freq":0} endon"}}

Can anyone help me with this Problems ?

[JayBraker]

Which exact tasmota.bin.gz 14.0 did you install - where did you download that?
I had to compile the firmware myself (13.5) because the default build, available as a Github Release, did not include the necessary drivers (see #13515 (comment)) I am unsure what the config for Github releases is and why back then the driver was not configured
But from your observations I anticipate that the default build still doesn't include USE_ADE7880

So head over to tasmocompiler and build a custom firmware image:)
That should do the trick

[JayBraker]

I am unsure what the config for Github releases is and why back then the driver was not configured But from your observations I anticipate that the default build still doesn't include USE_ADE7880

Had another look, USE_ADE7880 is still not part of the provided builds. You need to compile yourself to use Shelly 3EM properly
(Build configurations: https://github.com/arendst/Tasmota/blob/6673fb906bdc755eddbfc71d85e3eb9b9d11cd91/BUILDS.md)

So you can simply follow the steps I described right in the previous Thread: #13515 (reply in thread)

[saulous]

Hello, thank you for your Help. I have compiled and uploaded the Firmware in the 3EM. The result is the same like before :(
The Tasmota Version now is 14.1.0.

[rossi75]

Can you try uploading the firmware via serial? Be aware to not upload it via serial If 230v connected in any way

[madduck]

This thread seems to be the only resource for the 3EM, so I hope it's okay that I turn here with a question… two actually.

I had a spare 3EM and so decided to do a total house measurement. It's wired up right behind the "official" utility meter. I saved and restored calib.dat, of course. But the meter's readings are off. See the following graph, which shows the "official" meter in red, and the 3EM in blue over a period of a few days with measurements at random points in time. They shouldn't diverge this much in just a week:

My first question is: Is there a way to calibrate the meter in its current position, using the data from the official meter?

Second, when I try to use EnergyTotal[123] …kwh… to set the total energy measurement to match that of the main meter, that's not idempotent. If I run EnergyTotal[123] to get the current readings, and then I set the total to those readings, something else seems to get written. A new call to EnergyTotal[123] will produce a different number. Similarly, it's impossible for me to actually make the meter display e.g. 88,456 kWh, the current house reading, because if I enter this, or a third for each phase, then I get a much smaller total.

Can someone shed some light on this, or maybe refer me to a better forum (I have already posted this on Discord, but nobody replied…)? Thank you!

[rossi75]

To your second issue....

I have in mind that there was sth with a max value, 65535.....? Hence it either goes into negative or will just count upwards.
Can you share the logs as soon as you enter a new value...?

To which tasmota firmware did you compile?

To your first question, If the value is about 10% less than the official one, i simply would try to multiply the values given from the dat with 1.10.
Maybe only one Phase is affected, then you need to figure out this one phase

[madduck]

I have in mind that there was sth with a max value, 65535.....? Hence it either goes into negative or will just count upwards. Can you share the logs as soon as you enter a new value...?

Sure:

10:32:56.464 CMD: Backlog EnergyTotal; EnergyTotal1 101193087; EnergyTotal2 101187399; EnergyTotal3 101117501
10:32:56.489 MQT: stat/tas-house/RESULT = {"EnergyTotal":{"Total":[101193.109,101187.914,101117.602],"Yesterday":[6.035,5.596,1.882],"Today":[1.697,1.826,0.800],"ExportActive":[0.000,0.000,0.000]}}
10:32:56.720 MQT: stat/tas-house/RESULT = {"EnergyTotal":{"Total":[101194.805,101187.914,101117.602],"Yesterday":[6.035,5.596,1.882],"Today":[1.697,1.826,0.800],"ExportActive":[0.000,0.000,0.000]}}
10:32:56.961 MQT: stat/tas-house/RESULT = {"EnergyTotal":{"Total":[101194.805,101189.742,101117.602],"Yesterday":[6.035,5.596,1.882],"Today":[1.697,1.826,0.800],"ExportActive":[0.000,0.000,0.000]}}
10:32:57.214 MQT: stat/tas-house/RESULT = {"EnergyTotal":{"Total":[101194.805,101189.742,101118.398],"Yesterday":[6.035,5.596,1.882],"Today":[1.697,1.826,0.800],"ExportActive":[0.000,0.000,0.000]}}

So it turns out that the values aren't completely off anymore, since I figured out that EnergyTotal reports kWh, but actually wants Wh when setting. That is arguably a bug.

To which tasmota firmware did you compile?

I've compiled this for 12.5.0.

To your first question, If the value is about 10% less than the official one, i simply would try to multiply the values given from the dat with 1.10. Maybe only one Phase is affected, then you need to figure out this one phase

There are 22 different values in the DAT file. I wouldn't know which ones to modify. And wasn't that the entire point of saving the calib.dat file? Or was the device not properly calibrated when new?