Finally! Dynamic Power Management for wall connector

[mongo]

I have one of the 1938241-00-A devices from Amazon. Actually I have two, one for tinkering and one that will get installed. A little glitch has come up. The CTs don't fit around the input bus bars on my year old SquareD SC3042M200PF meter panel. I'm assuming CTs will not work correctly unless the split side fully closes. I'm pondering if PG&E would let the CTs be connected around the service entrance wires upstream from the meter, in the utility sealed section of the panel. I think it's likely they don't exceed the 0.75" limit, and the meter panel is customer owned, but assume the wires are utility owned and maintained. I'm not sure who to contact at PG&E support for a question like this. I'm super happy to see Tesla Wall Charger support dynamic load management, and perhaps they will read this and for future customers use CTs with a larger hole. The CTs for my Enphase solar fit just fine, I think they have a 1" square hole.

The device talks RS-485 modbus RTU at 115200,8,1. I unfortunately don't know what the modbus register map is but am putting together some stuff to make good traces of the communication to the Universal Wall Connector. I looked at it on a scope and the Wall connector polls every second and the Neurio responds. The serial decode said the RTU request was read 10 holding registers at address 0x88 and the sensor responds back with 20 bytes of payload, which I don't know the meaning of yet. There may also be some startup handshaking, like asking what model of load controller it is. When I get good tracing working I'll try to vary the power parameters and see how the returned data changes. I see if the Wall Connector looses connectivity to the modbus sensor it retards back to 6A charging. I don't yet know what it does when the current limit is reached.

I don't have things installed yet, and am currently collecting parts and verifying everything fits and works. I haven't decided for sure if I will do the actual install or get an electrical contractor. My brain says DIY, my not so young back says hire an electrical contractor. Either way, I do want to be sure all the parts will work out. I'm guessing a contractor would not have figured out the CT doesn't fit until installation was started, so discovering this beforehand was fortunate. I don't know of a different load manager for a Tesla Wall Connector.

In an ideal world, what I probably want is a little gadget that talks ZigBee to my PG&E electric meter (I have a Rainforest Automation Eagle that does this now for years) and constantly reports power consumption to some local energy manager. Ideally, the Wall Charger could pull the power limit over WiFi. A little less ideal, a little box, like maybe a modbus to wifi gateway ($38 on Amazon) could bridge communication to the local energy manager, and respond to the modbus requests from the Wall Connector, avoiding the need for another CT sensor. My meter panel already has CT coils to my Enphase solar controller, so it's a bit silly to need a second set of CTs for the EV charger. The first CT's are silly because my meter already talks ZigBee reporting instantaneous power consumption (well delayed a second or two). I'm probably just going have a straight install of the Wall Connector and load manager, but figure while it's not yet connected I should record what it does for some future home automation company or DIY. Tesla could just make this really simple and perhaps could even make load management of the mobile connector plugged into a NEMA 14-50 outlet.

If you can determine the transfer function of the CT you might be able to find one with a larger opening.

 

[jcbottorff]

If you can determine the transfer function of the CT you might be able to find one with a larger opening.

I've done some research on CTs and there are factors other than the ratio, like the expected burden resistor value. It seems like the sensing circuit may need to be calibrated to the exact model of CT. This CTs also use a 4 pin connector and has an eeprom holding calibration data. It's some sort of 1-wire protocol, but have not yet decoded the data. The Generac patent describes some of the possible data needed. The load controller CTs say 264A max and 176mA output, so think they are 1500:1. The CT design has two patents by Neurio and Generac (the calibration eeprom and the curved shape that's causing my problem), so don't see replacing the CTs with anything generic.

I suppose one option is I could purchase a different brand of charger. I'd rather use a Tesla charger with my Tesla car, but Tesla has chosen poorly in their load management sensor. I've heard there are other chargers that support dynamic load management. A different brand of modbus meter may have CTs with a larger hole. A good software option in the Wall Connector for Tesla to add would be some parameters to set which registers to read over modbus so then you could use other modbus power meters. I suppose there is a way to do modbus translation, so I could install brand X of modbus meter and the translator could convert the register values to emulate the Tesla Neurio device.

I need to verify my assumption that a split CT will not work correctly (accurately) unless the split is totally closed. I suppose it's possible if I squeeze the CT shut with a cable tie or something similar, the plastic shell will distort enough to fully close. I'm opening a support conversation with SquareD to get them to tell me the exact dimensions of the bus bar and I can make a little wood or 3D printed model with the same dimensions and see up close how the fit is not quite right. I'm reluctant to try anything fancy on the actual live bus bars. Since the bus bars are upstream from the main breaker, there is no way to deenergize it without getting PG&E to come out.

I could just live without load management, and would likely have to derate the maximum charge rate, perhaps significantly. Since I may be living with this charger for 10-20 years, I'd really prefer it works as well as possible.

 

[Jughead135]

I’m just getting started looking at this, very intrigued.

One thing I’m scratching my head about: Why do I need a wired data connection? These Neurio boxes appear to have WiFi capability; and the Wall Connector certainly does. Why require another hardwire?

Trivial enough to run a #22 wire along with the power for an initial install—not so trivial after the fact….

 

[mongo]

I’m just getting started looking at this, very intrigued.

One thing I’m scratching my head about: Why do I need a wired data connection? These Neurio boxes appear to have WiFi capability; and the Wall Connector certainly does. Why require another hardwire?

Trivial enough to run a #22 wire along with the power for an initial install—not so trivial after the fact….

Determinism and latency. If there is a step change in house load, the WC (and vehicle) need to reduce the charge rate fast enough to not trip the main breaker.

 

[Jughead135]

Determinism and latency. If there is a step change in house load, the WC (and vehicle) need to reduce the charge rate fast enough to not trip the main breaker.

Ah. Yes, of course <*slaps forehead*>. Fairly obvious, now that you say it. Thanks for being gentle.

I guess these boxes have a more generic monitor function for non-Tesla applications? With the latency question in mind, what good is the WiFi-only connection?

And: does the Tesla version maintain the WiFi capability? (For monitor or control purposes, say?)

 

[mongo]

Ah. Yes, of course <*slaps forehead*>. Fairly obvious, now that you say it. Thanks for being gentle.

I guess these boxes have a more generic monitor function for non-Tesla applications? With the latency question in mind, what good is the WiFi-only connection?

And: does the Tesla version maintain the WiFi capability? (For monitor or control purposes, say?)

I'm guessing on implementation, no deep insight
The boxes themselves could be handy for any power monitoring application.

 

[Visvism]

Any chance we see these updated to work with group management or with more than one unit setups? I gave two Wall Connectors in group management mode and eould love for them to do dynamic management AND adhere to max group management amperage simultaneously.

 

[Icarus]

@jcbottorff - If you make any progress decoding the RS485 on the Gen3, there's a project that could use your help -

GitHub - ngardiner/TWCManager: Control power delivered by a Tesla Wall Charger using two wires screwed into its RS-485 terminals.

Control power delivered by a Tesla Wall Charger using two wires screwed into its RS-485 terminals. - ngardiner/TWCManager

github.com

TWCManager evolved as an open-source community software project that controls the Tesla Gen2 wall charger via RS485. They have now implemented a range of modules to support many solar inverters and home automation controllers. That includes the Enphase CT modules amongst others -

TWCManager/lib/TWCManager/EMS at main · ngardiner/TWCManager

Control power delivered by a Tesla Wall Charger using two wires screwed into its RS-485 terminals. - ngardiner/TWCManager

github.com

There's a very long but interesting thread on how the original Tesla RS485 protocol was reverse-engineered. If you have some time, it might have some valuable gems for the Gen3 -

Tesla Wall Connector load sharing protocol

Not sure if this has been discussed in the existing HPWC/WC thread yet, but for the sake of discussion (and people being able to find this info afterwards) I'm opening up a new thread to discuss this. The new 80A single-phase (US) / 32A three-phase (EU) wall connector has an option to daisy...

teslamotorsclub.com

 

[mself]

My charger is behind my Powerwall Gateway (whole house backup), which already has current sensing for the load drawn from the grid. It would be great if Dynamic Power Management could use that existing capability rather than needing a new device.

The Tesla is already adjusting the charging rate based on "excess solar", and this is very similar. The Charge on Solar feature works over Wi-Fi without needing a serial cable. But, in that case, the only thing that happens if it reacts too slowly to changes is that you may draw a little power from the grid (rather than potentially overloading your system?). Tesla adds some buffer where it won't start charging unless there is at least a minimum amount of excess solar. This mostly eliminates cases where it draws more than the available solar.

The same approach might work here. In my case, I would be happy to set a low charging limit (even 30A below the max panel load). That's because the nominal "load calcs" are incredibly overstated. The load calc for my house comes out to 112A, but the utility data show that the highest 15-minute period in the last 2 years was only 45A (I have a 32A charger, but I'd like to increase it and electrify more stuff). Going by the load calcs, there is no room to add anything. But in reality I could add more as long as the EV charging backed off in the "worst case" scenario.