Adding support for PZEM004T energy monitor for 3 phase - request

[roblad]

Hi,

Is it possible add USE_PZEM004T and add support for PZEM004T Energy monitor for 3 phase by using USE_SERIAL_BRIDGE and report each phase through that functionality ? May be there is possibility to use 3 phase config for USE_PZEM004T as it is in that project:

https://github.com/apreb/eNode

[rt400]

thanks.. i run it with 3 PZEM04 v3 and igot in console " PzemAc error 8" and error 7 , and still got all 0

[adityama]

This is a great news. I will try it at the earliest. I guess we still need to give correct addresses to the modules before connecting them together.

On Sun, 15 Sep, 2019, 8:52 PM Theo Arends, notifications@github.com wrote:

I changed both -004T and modbus versions so all should be supported.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/arendst/Sonoff-Tasmota/issues/2315?email_source=notifications&email_token=ACEHVVIK7IB5745KIXHXMLTQJZHJXA5CNFSM4EYK56K2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD6XS5MA#issuecomment-531574448, or mute the thread https://github.com/notifications/unsubscribe-auth/ACEHVVMVQL3WPF6PV2UQMHLQJZHJXANCNFSM4EYK56KQ .

[rt400]

@arendst i try to use with this , all model allready set to 0.0.0.1 , 0.0.0.2 , 0.0.0.3 connect the rx tx to nodemcu model , isee in web that i have 3 model but they going down to 2 and stay in 1 . in web log i can't see anything...so strange

[rt400]

i found my problem...the gnd was little bit out . now amazing its works...

[localhost61]

I follow this thread with great interest, as my goal to monitor on same modbus an SDM230 and two SDM120 is getting closer now ;-)

[jziolkowski]

Since the latest builds are not that common, I suggest changing the new command to something more specific, like PzemAddress in case more similar commands will appear in the future

[meingraham]

Or ModuleAddress <type>, <address> where e.g., <type> = PZEM to avoid a whole slew of new commands that are similar in function.

[klotzma]

What is the command now to change the address of the modules?

[jziolkowski]

No change, since Theo didn't reply with any decision nor any new commits pushed in this regard

[klotzma]

He added the command "Add command ModuleAddress 1/2/3" or not? Add command ModuleAddress 1/2/3

[jziolkowski]

Yes, it is in the commit and in the changelog

[klotzma]

So can I give the module the address with the command "ModuleAddress 1"?

[adityama]

@yuval I am at least couple of days away from testing the new code. Could you help test one specific scenerio - let's say one of the 3 phases is down, so you get the readings from remaining 2 phases. Would you be able to test this please.

On Tue, 17 Sep, 2019, 5:11 PM jziolkowski, notifications@github.com wrote:

Yes, it is in the commit and in the changelog

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/arendst/Sonoff-Tasmota/issues/2315?email_source=notifications&email_token=ACEHVVM3KOAZLLKRYLJMTALQKC65BA5CNFSM4EYK56K2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD64HUOY#issuecomment-532183611, or mute the thread https://github.com/notifications/unsubscribe-auth/ACEHVVLXNK4NNTGHS2CZVEDQKC65BANCNFSM4EYK56KQ .

[jziolkowski]

@klotzma yes, but you don't need to, since all PZEM modules have address 0.0.0.1 until specifically changed. So in practice you need to connect the module for L2 only and set it to address 2, and then only the module for L3 and set it address 3.

[jziolkowski]

@adityama I'm almost done building my 3 phase setup and could test that scenario for you.

[adityama]

Your setup looks great. I need to find a similar enclosure 😊

On Tue, 17 Sep, 2019, 5:27 PM jziolkowski, notifications@github.com wrote:

@adityama https://github.com/adityama I'm almost done building my 3 phase setup and could test that scenario for you. [image: image] https://user-images.githubusercontent.com/11555742/65039659-0cb1c280-d953-11e9-9193-346a809763ed.png

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/arendst/Sonoff-Tasmota/issues/2315?email_source=notifications&email_token=ACEHVVOH5XO266CBZU7JQSTQKDA4PA5CNFSM4EYK56K2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD64I2WQ#issuecomment-532188506, or mute the thread https://github.com/notifications/unsubscribe-auth/ACEHVVJ6GQYVAMRN55BFONLQKDA4PANCNFSM4EYK56KQ .

[jziolkowski]

@arendst @adityama so I've done some testing.

Assembled the whole device, configured ModuleAddresses and started all. Then in TDM I've configured autotelemetry with 2-second interval and watched the readouts. As expected, when all 3phases are connected, telemetry has all three readings, but when I disconnected one device from AC while tasmota was online, the readings were inconsistent, as in: around 7-8 out of ten readings were correct (correct V, correct V, 0), but every now and then the telemetry returned [0,0,0].

2nd test wa booting the device with only 2 phases present on PZEMs. Device returns telemetry only for two phases, bt this time without 0 for the missing phase. Reconnecting the "missing" phase doesn't make the readings appear in the telemetry, because I've found that if PZEM is not connected to AC voltage during boot/sensor initialization, then it will remain inactive (doesn't return any readings and I'm 50/50 sure it even accepts commands, but the latter wasn't tested thoroughly by me).

I hope that helps. Let me know if any further testing is needed.

[arendst]

Thx for testing.

• 0,0,0 when one is lost is currently expected. I will need to split the watchdog.
• Adding a sensor on the fly is currently not supported; sensors need to be connected at power on. no conneected address is disabled and not probed after this. The reason is that every probe takes 5 seconds. Every sensor needs 1 second for probing.

I'm currently offline. Will follow up this weekend.

[jziolkowski]

Solving the first "issue" will be great; if someone expects that some phases might be disconnected as a normal scenario, having consistent telemetry output is important.

2nd I understand. However, under some circumstances (when user has one phase offline) the PZEM will be offline as well, until next re-initialization of a fully connected PZEM. But it's not "our" fault, it's PZEM issue.

Thanks again for adding three phase support. Grafana charts will be glorious :)

[roguestreak]

(Pre-script; just realised there are two PZEM threads, so some may relate to the other thread) Thought it would be best to discuss/include in this thread for future user reference;

• does an installed but disconnected CT need to be short-circuited? ie. If installed on a wire but temporarily not connected to the PZEM should it always be short-circuited first? Thinking of using switching audio jacks connected to the CTs, so if you don't have the plug from the PZEM installed it automatically short-circuits? If short-circuiting should there be an inline resistor?

• has anyone done the Pin 51 hack to see if the PZEM can flag current flow direction? ie. if exporting to the grid is current 180 degrees out of phase with voltage, and does this pin toggle in such instances as the datasheet indicates? Obviously it would be nice if future PZEM versions had this built-in and included in their data outputs, but I'm yet to see verification that this does actually work (see notes here: https://github.com/apreb/eNode) My v3 should be arriving soonish, but not sure when / if I'll be able to get an electrician in to install the CT clamp on my mains connection to test. Another reason to ask about the short-circuiting, in case I need to have it installed before the rest of the system is ready.

PS. I'm only on single phase, but greatly interested in using multiple CTs in future to monitor all circuits in the house; has anyone tested more than 3 addresses yet, purely from the data perspective rather than the web interface? I could query each address directly via NodeRed

[jziolkowski]

I've read that CTs should not be installed without termination, especially on high-current lines. Maybe it's my imagination, but when I was installing the CTs today, I first installed them on the wires which then were pulled live and when fiddling around with the cables I felt a small sting from a CT installed on a ~2A current wire. But it makes sense that a transformer can be damaged if the energy isn't being used.

No idea about the hack, but AFAIK PZEM always shows positive flow, irelevant of the current flow. IDK if it's in the chip nor is it configurable.

3 addresses is the newly expanded limitation of tasmota. It doesn't matte rif you dont need the webui interface: tasmota won't poll any more PZEMs than 3. Therefore, for monitoring multiple curcuits you would have to use a tasmota device per each three circuits. Also take note of the above tests I've did; if a PZEM is not connected to live AC during tasmota boot, it won't be recognized and its data will be ignored.

PS can't you get a PZEM with open clamp?

[roguestreak]

Thanks @jziolkowski I just found that this was in fact the one thread, just that many of the posts were collapsed; wondering whether using @adityama's serial method with NodeRed can address more than 3 units?

RE: current export detection, I believe the PZEM-004t doesn't have Pin 51 of the main SDIC SD3004 chip connected to anything, hence why it can't detect/specify direction, but the datasheet for the chip itself says the following of pin 51: "Goes high when phase difference between voltage and current is greater than 90 degrees. REVP updates its logic state when a CF pulse is issued." The THEORY is that if you have this pin expanded out, and feed to a GPIO, you could use that as a flag so you can change the sign of the power data from the PZEM, eg. the 2kW it's reading is actually 2kW export rather than import. But I've yet to see verification that this pin does in fact toggle when you are exporting power.

I can get a split CT clamp, and have ordered my v3 with one, but while I'm happy to play with power on extension cords etc. behind safety switches and RCDs, I'll leave it to the professionals when opening up my meter box and installing inside, especially on the supply side where there's no "off" switch. Heard a few horror stories of literally smoking corpses where the mains fuse hasn't blown, not something I want to subject my kids to (or myself for that matter)

[Jason2866]

You have a very small analog voltage to measure, adding additional stuff or make the wires longer of the CT will result in issues.

[jziolkowski]

I've extended the CT wires with ~20cm of .14mm2 wire. Hope that's not an issue

[Jason2866]

20 cm should work...

[adityama]

@arendst Fantastic. Thank you. This fix makes the solution perfect.

[ascillato2]

Closing this issue as the feature has been added by Theo.

Thanks to everyone for working on this :+1:

[klotzma]

Thanks for the great work.

I got it to work when I wired it as shown in the picture. Is it correct that way ? Or do I have to wire it differently?

Because it is indeed a Modbus, must there now be a terminal resistor?

Improves me.

As shown here, it did not work: Uart Bus issue comment

[jziolkowski]

You definitely need protection diodes. See drawings earlier.

[Jason2866]

https://github.com/arendst/Sonoff-Tasmota/issues/2315#issuecomment-384643624

[klotzma]

So I had connected it before and it did not work

[localhost61]

@klotzma , did you use fast shottky diodes? Did you respect the polarity? The schema shows an "open collector" principle, the bus is pulled high by the pullup resistor and pushed down when a slave is sending a Low level on its Tx pin. The Modbus resistors are used for line impedance adaptation, to avoid signal reflections that cause spikes, this is totally different.

[Jason2866]

It does work @jziolkowski has builded as described. If your setup does not work you have done something wrong.

[jziolkowski]

@klotzma initially I didn't use diodes and resistors and it wasn't working. I guess that people who said it's required knew what they were talking :)

[rt400]

Must connect diodes otherwise it just doesn't work. From experience.

[klotzma]

Yes, I used the Bat43 diodes! I checked everything twice and three times. But it only worked like in the picture from my previous post.

[klotzma]

With this setting

[localhost61]

Isn't it #6585 issue solved 3 days ago by Theo? All PZEM004T devices have their own address? Can you describe your issue with more details, what doesn't work?

[klotzma]

Danke für die tolle Arbeit.

Ich habe es zum Laufen gebracht, als ich es wie im Bild gezeigt verdrahtet habe. Ist es so richtig? Oder muss ich es anders verkabeln?

Da es sich in der Tat um einen Modbus handelt, muss es jetzt einen Abschlusswiderstand geben?

Verbessert mich

Wie hier gezeigt, hat es nicht funktioniert: Uart Bus Problem Kommentar

It works as pictured in this picture, all pzem have their own address, from 1-3.

But if I put the Bat diodes in between, then no data comes back

[klotzma]

I had used the 6.6.0.15.

I will try it later with the 6.6.0.16.

[localhost61]

6.6.0.17 now ;-)

[klotzma]

Ok, I got it, the diodes were labeled wrong. Or in the plan the diode is wrongly drawn. The cathode was not where the line is, but on the other side.

When I turned the diode, it worked.

[klotzma]

Can I reset all the energy to zero with tasmota?

[localhost61]

Yes, it should be allowed since commit v6.6.0.15 20191003. Try SetOption72 0 to select software counters Then EnergyReset3 0 (Didn't try myself)

[rt400]

@localhost61 @klotzma i have the same issue. i try all the command but still i got the "Total" from the PZEM even with SetOption72 0

[localhost61]

@rt400 Hum, yes I see your last comment on issue #6561

[klotzma]

@localhost61 @klotzma i have the same issue. i try all the command but still i got the "Total" from the PZEM even with SetOption72 0

Does not work for me either.

[arendst]

At least with version 6.6.0.16 and SetOption72 0, EnergyReset3 0 will reset the total counter to 0, then add the energy from today.

But if you want to reset the total counter to 0 now, you will need to execute first EnergyReset1 0 followed by EnergyReset3 0.

[rt400]

@arendst but its not work.. look in my issue #6561

[arendst]

See #6561 for latest change.