1p/3p switching via external device

[ecfmau]

I would like to suggest a new feature: 1p/3p phase switching via an external relay (e.g. https://www.reichelt.com/de/en/shelly-pro-3-three-phase-3-channel-relay-shelly-pro-3-p335405.html, maybe this one would not be supported by evcc but you get the idea).

Background
Most cars and wallboxes need a minimum of 6 A to start charging. With a 3-phase wallbox and a 3-phase capable car, this means a minimal power of 4140 W. But most wallboxes can be wired - and will work well - with only 1 phase, and to my knowledge all cars can charge with 1 phase.

Problem
For small PV installations (4 - 10 kW) or on less sunny days or during winter, there is a significant amount of marginal hours where the PV surplus is higher than 1380 W (1 phase, 6 A) but lower than 4140 W (3 phases, 6 A). The energy produced during these marginal hours is lost (not quite lost, it is fed back into the grid, but it would be better to load the EV battery with it).

Example
Here is the production of my 7.2 kW PV installation on a mildly sunny Friday:

With my current setup (elli wallbox, 3 phase, 6 to 16 A), here is the amount of energy I was able to load in my car:

With a software-configurable 1p/3p wallbox (unfortunately the elli wallbox is not), here is the energy I would have been able to load in my car:

Feature request
Add a possibility to disconnect and reconnect phases 2 and 3 of a wallbox via an external switch and ignore / set to zero any data reported from the wallbox for phases 2 and 3 while running in 1-phase mode.

[premultiply]

This is unsafe and will not be supported by evcc.

[premultiply]

You may install an external manual load contactor ("Lastrennschalter") like the Hager HAB304 to switch L2+L3 together.
You may even use the remaining contact to use it for phase configuration signalisation purposes to evcc.

[AtariAndre42]

The Shelly is the same thing, only controllable through Ethernet, Bluetooth or Wifi so why use the Hager when it's a dumb switch. The Shelly is also a switch but smart controllable. Electrically there is no difference.
The only risk is that the Shelly could switch "uncommanded" during charging. It would be wise to wire the realais contacts so that it goes to a "safe" state when the shelly loses power. But then, what is "safe". I'd say going from 3 to 1 phase would be "safe" because this could also happen in real world by other situations (grid failure) and any hardware (charger, car) should be prepared for such a scenario. The other way round (going from 1 phase to 3 during charging), now that's a thing that would never be possible to happen in real world situations so careful consideration how the Shelly is hooked up should be done.

[AtariAndre42]

I ordered a Shelly Pro 3 and consider it not to be suitable for continues 16A (x2) loads. The screw terminals are not suitable for thick installation wire either. I bought Tongou switches rated at 63A instead.

[AtariAndre42]

This is unsafe and will not be supported by evcc.

It would be great if it was explained why this would be unsafe as you claim.

AFAIK the norms that apply to EV charging and chargers require chargers to simply ignore the extra phases when they would be applied "in error" during a 1 phase charging session. The EVSE should simply ignore L2 and L3 and not activate its relays towards the EV. Even modern chargers that can switch between 1P/3P require the session to be stopped and started again in 3 phase mode, simply because it has to run safety checks before being allowed to activate the phases towards the EV.

Even EVs themselves should ignore any extra phases that would be activated during a charging session.

The other way around (phases being switched off uncommanded) can happen in real life to (in the grid or in your electricity box). Both EVSE and EV should have protections for this.

So....what is unsafe about this idea?

[ecfmau]

Hey guys! I successfully added 1p/3p switching capability to my Elli Wallbox with the help of a Shelly Pro 3 switch. I added the phase switching logic in the script evcc calls to measure my solar production, so it is executed every 10 seconds. I make calls to the evcc api to read some values and to configure the loadpoint's active phases (1 or 3).

Here is how it works.

Set delay timers

if (available_solar_surplus > 4140W) {
    reset 3p1p_delay_timer
} else {
    reset 1p3p_delay_timer
}

Phase switching logic

if (evcc_mode == 'now') {
    switch to 3 phases (if not already)
} elseif (evcc_mode == 'pv' or evcc_mode == 'minpv') {
    if (1p3p_delay_timer > 60s) {
        switch to 3 phases (if not already)
    } elseif (3p1p_delay_timer > 120s) {
        switch to 1 phase (if not already)
    }
}

Actual phase switching

read the actual evcc_mode and save it for later
set evcc_mode to 'off'
wait until (evcc_mode == 'off' and charge_power == 0) or timeout_expired
if (evcc_mode == 'off' and charge_power == 0) {
    set phases 2 and 3 on Shelly switch to off|on
    set evcc active phases to 1|3
    revert evcc_mode to its previous value
}

Some additional considerations:

• It actually works very well, mainly because the Elli wallbox is a 3-phase device which is able to work with only one phase wired.
• I think that the critical part is to avoid disconnecting the phases 2 and 3 while under load (i.e. while charging is active), that's why I added a safety loop in my code.
• I'm not sure if this would work with any 3-phases-but-1-phase-capable wallbox, the Elli wallbox doesn't seem to bother when phases 2 and 3 suddenly disappear, some other wallboxes might be less tolerant to this trick.

Anyway, if you want to replicate this, do it at your own risk. Don't blame me or anyone but yourself if you happen to fry your wallbox, your electric installation, your car, or set your house on fire. I will happily provide the PHP source code of the above described pseudo-code if interested.

Here are two pictures of the Shelly switch installed within my Elly wallbox. It is a VW-branded ID. Charger Connect (the meter was not included in the wallbox, I added it afterwards).

[oberraeuber]

I am intrigued by this. Do you use this as a switch between 1p and 3p only? Or are you using 2p aswell?

[ecfmau]

Yes, I use it to switch between two modes : 1 phase (L1) or 3 phases (L1 + L2 + L3). The wallbox internal circuits are powered with L1 and it is designed to work with one phase or 3 phases. I never heard of two phases charging and I wouldn't try it because it is most probably not supported by the wallbox. Maybe it would ignore the phase 2 and fallback to 1-phase charging, or maybe it would detect a faulty grid and would not charge at all.

[IonasElate]

I'd love to take a look at your code. I did not find any way to contact you. my email is jonas.boetsch@me.com
it'd be great if you would get in touch.

Regards
Jonas

[tarelda]

Yes, I use it to switch between two modes : 1 phase (L1) or 3 phases (L1 + L2 + L3). The wallbox internal circuits are powered with L1 and it is designed to work with one phase or 3 phases. I never heard of two phases charging and I wouldn't try it because it is most probably not supported by the wallbox. Maybe it would ignore the phase 2 and fallback to 1-phase charging, or maybe it would detect a faulty grid and would not charge at all.

AFAIK SAE J1772 doesn't define singalling of available phases. Thus probably it works like this:

1. EVSE goes from state B to state C.
2. EVSE opens relays for the used phases.
3. EV detects available phases and uses provided through CP current limit to adjust charging.
   So it is solely at the discretion of wallbox manufacturer how to handle the situation. Probably most of them stops the charging when phase goes missing (at least, my Alfen WB goes into fault mode when one phase goes missing and has to be restarted).

@ecfmau I think that they don't support scaling down phases from 3 to 1 probably because phase relays are driven through common driver or there is hardware circuit that detects phase current imbalance. Summing it all up, I think your approach is safe since this is equivalent to starting WB in 1 phase or 3 phase mode. Critical thing is that switching occurs on GRID side and in OFF state.

DISCLAIMER if you don't understand everything in my comment, don't consider this as safe pass to try it.

[Knuppel1983]

Some additional considerations:

• It actually works very well, mainly because the Elli wallbox is a 3-phase device which is able to work with only one phase wired.

I added a normal household Hager relay, don't trust the Shelly's. I hooked up an ESP that also detects if power is flowing through the wires. In HA i have a manual switch and a sensor so i never switch when the car is charging. I can also see the if the Tesla is charging via esphome-tesla-ble, so i can also add an extra failsafe there. I only switch manually if i have to charge the car fast.

For me the problem with 3-phase charging is that the steps are too big. 35230 = 3450w, 35230 = 4.140, etc. Thats 690w per step. Also if it's hovering on the threshold between 1 and 3-phase charging i wouldn't know what to do. The solar ouput is never linear. So i'm not bothering to automate this and by default only 1-phase charge. I have another EV-charger that's also 1-phase and together they take 7,2kw, which is more than my PV can supply (6,5kw). I'm looking at purchasing a home battery to take the PV output if one of the cars is not home.

[andig]

Mehrfach diskutiert und wontfix für uns. Ohne physikalische Absicherung gegen klemmende Relais etc besteht da ein hohes Risiko das Fahrzeug zu zerstören. In Software kannst Du das nicht absichern. Wer das machen will kann den Code des custom Chargers leicht selbst anpassen.

[andig]

Insb. wie man beispielsweise eine komplexere Wallbox wie die Heidelberg um so eine Funktionalität "erweitert".

Das geht nicht. Technisch nicht und auch sonst würden wir das nicht anbieten.

[monnimeter]

Meine Einschätzung war nicht pauschal gedacht - ihr habt sicherlich Recht, dass es Fahrzeuge gibt, die bei solchen "technischen Spielchen" intolerant reagieren. Das stelle ich nicht in Abrede :-)

Aber OK... wenn's in EVCC technisch nicht geht, dann muss ich das Feature woanders implementieren... all good.
Ich kann die Wallbox-Schütze über meine SPS ansteuern, die genauso wie EVCC bidirektional an meine openHAB-Instanz angebunden sind.
D.h. ich muss nur etwas "glue logic" (analog der Lösung von @ecfmau) implementieren und bin im Rennen.

Danke Euch!

[andig]

Noch einmal fürs Protokoll: Sicherheit = Verriegelung des Schützes gehört in Hardware. Alles Andere ist Lotto.

[AtariAndre42]

Ich halte das Risiko, dass das Fahrzeug durch sowas "zerstört" wird für ausgeschlossen

Das ist deine persönliche Fehleinschätzung. In der Realität gibt es bei einigen OBC-Designs beim Umschalten ohne geeignete Maßnahmen und Signalisierung einen fetten Kurzschluss und die Kontakte des internen Phasenkoppelrelais verschweißen. Wer sowas schon mal von der eigenen Autohändlerei reparieren (=Ladesystem tauschen) lassen musste wird dir von 4 bis 5stelligen Spassprämien berichten.

If this happens in a charger or EV then the manufacturer of the charger of EV has not followed the international norms. Kia MY20 cars were one of those cases.

[premultiply]

There is no international standard, German standard or other specification for phase switching during charging with alternating current.

The only safe way is therefore to ensure that phase switching can only be performed during an interrupted charge on the EVSE side, including complete shutdown, disconnection/fault sequence on the EVSE side, offline phase switching and start sequence. This must be done within the EVSE and should be supported by safety contacts and other hardware-based measurements.

[IonasElate]

-deleted-

[tarelda]

Learn how Hyundai and Renault/Smart OBC systems are designed for example. Again: adding phases after initial grid sensing by the vehicle without proper interruption sequence may cause serious and expensive damage to some vehicles. Believe it or not.

@premultiply Provide either sources or at least some stories with confirmed diagnosis.
So you confirm now that there is no harm in adding phases to EVSE supply when contactors are not energized?

@IonasElate why didn't you try to bind your guarding circuit to contactor enable line?

[IonasElate]

If i am really implementing this in the future i have to think and test about the use of the current detection relais.
In the drawing i uploaded previously (and deleted it short time later because it had a flaw) l2/3 could only be switched on if l1 current is low. this would result to l2/3 being switched off if l1 is used during 3 phase charging so exactly something we would not want to happen.
right now i have other priorities but i will report back once i want to implement it or i have a working (and tested) analog solution

[tarelda]

If i am really implementing this in the future i have to think and test about the use of the current detection relais. In the drawing i uploaded previously (and deleted it short time later because it had a flaw) l2/3 could only be switched on if l1 current is low. this would result to l2/3 being switched off if l1 is used during 3 phase charging so exactly something we would not want to happen. right now i have other priorities but i will report back once i want to implement it or i have a working (and tested) analog solution

IMO, current is wrong approach here.
@IonasElate What do you mean by analog solution ;)? You want to toggle additional phases switch with some analog signal that is AND gated with negated boolean signal from current detector?

[IonasElate]

something like that, yeah

[tarelda]

@IonasElate two things :

• EVSE (aka wallbox) sets current limit to 0A, thus effectively suspending charging, but contactors are still closed and you might attach additional phases directly to onboard charger. If it uses internally coupling multiple inputs into to the one, you will definitely trip overvoltage protection and overcurrent or RCD detection if EVSE is properly designed (phases voltage add up due to Kirchoff law and thus current increases due to Ohm law).
• also as I mentioned before, your charger might not be using phase you are measuring.

[mateo-1337]

I have found that Elli combined with ID4 support disconnecting and reconnecting L2 and L3 even during charging as long as the charging starts in 3PH mode. The phases can be disconnected and reconnected and evcc manually switched to either 1ph or 3ph mode. However if you start in 1ph mode the car wont accept the additional 2 phases. It's weird, probably not meant to work like this, but it works. I have yet to check this on Tesla. Elli apparently does not even monitor the L2 and L3 phases and just blindly goes with what you throw at it as someone mentioned it is powered by L1 and it just clicks with one relay on all three phases and just works.

Edit: As my Tesla was in warranty :) I tried the switching during charging. Start off with 3PH when you disconnect L2 L3, it keeps charging, throws a warning, and does not let itself return to 3PH charging after reconnecting the PH2,3. ON/OFF cycle is required to restore 3PH charging. Not a physical disconnect though, which is kind of nice ? Edit2: Tested on M3 2021.

[walter-ve]

I do have a digital energy meter. But what is a P1 Port?

What country are you in?
In belgium/Netherlands the digital meters have a digital putput port to which you can connect a device... called a P1 port.
You can connect the wallbox via a dongle on that port and it reads directly from your meter.
One constraint; that port only gives values every 2 seconds.

[KneterPeter]

I do live in Germany, I Have a conexa 3.0 Smartmetergateway with a HAN output. In the App from the ABL Pulsar Wallbox it says I can connect it to a P1 Port. That looks like the easiest thing but it cost 200 € silently crying

[walter-ve]

Didn't know it is so expensive... A P1 reader like from Homewizard with good HA integration costs 25 Euros...

[Knuppel1983]

Edit: As my Tesla was in warranty :) I tried the switching during charging. Start off with 3PH when you disconnect L2 L3, it keeps charging, throws a warning, and does not let itself return to 3PH charging after reconnecting the PH2,3. ON/OFF cycle is required to restore 3PH charging. Not a physical disconnect though, which is kind of nice ? Edit2: Tested on M3 2021.

The problem is not the car not accepting the switch. The problem is your relay contacts arching and welding together because you keep switching during load.

[mateo-1337]

Edit: As my Tesla was in warranty :) I tried the switching during charging. Start off with 3PH when you disconnect L2 L3, it keeps charging, throws a warning, and does not let itself return to 3PH charging after reconnecting the PH2,3. ON/OFF cycle is required to restore 3PH charging. Not a physical disconnect though, which is kind of nice ? Edit2: Tested on M3 2021.

The problem is not the car not accepting the switch. The problem is your relay contacts arching and welding together because you keep switching during load.

Yes and no, i use contactor rated for 25A to switch it + there are relays in the wallbox that operate as a charger usually operates. Sould the contactor weld for some reason, the circuit still works as "regular charger". As per arcing and possible fire i believe that would create a substantial voltage drop for the car to fault out if not, well i will have to get a new paint job on my concrete wall and a new wallbox :)

[AtariAndre42]

What is the status about this? Since the OP a number of chargers that support 1/3 phase switching have surfaced and evcc supports them. Those chargers have relays that can also become “klemmende relais” so electrically there is no difference.

[premultiply]

These are commercial products for which the manufacturer is liable.
The manufacturer would therefore be well advised to have thoroughly tested the advertised features and, in case of doubt, to be liable for any damage caused during proper use.