openinverter forum

Often when I do support calls we spend a whole lot of time getting a laptop close enough to the inverter for the wifi to actually work. It seems better with Prius and Leaf inverters but for the Tesla ones it's pretty bad.

So at least for the Tesla boards I consider making the board a bit cheaper and omitting the onboard ESP32 module. As an option the new Wifi CAN module can be bought along with the board. Omitting the ESP32 allows me to switch to economic PCBA and produce the board cheaper.

The CAN module is actually more reliable as it doesn't suffer from the baud rate switching issue but of course it's a bit less plug&play as you need to power another board and connect CAN.

What is everyones take on this?

I like the new version because I can access all OI devices with a single, stable network connection. The adapter is normally inside the car
next to the notebook. I also have a 20m CAN bus extension cable that I can use to easily get from the garage to the office.

Thanks for the votes so far. Leans to omitting. Keep the votes coming ( and comments too!)

Personally, the fewer boxes the better but I have struggled with poor comms so do see the benefit of a more robust communication system.

As a UK based customer finding a way around customs charges would offer about a 20% saving ( Damien seems to manage but I don't know if that's due to the luck or if he has a system?)

Another option is to keep the physical space on each inverter board, but move to a D1-Mini footprint,
ESP32 D1 carrier boards are $4ea, and can be fitted much like the previous olmex module was.
Still keeps the legacy ability to fit, without much added inventory (just keep a few modules in stock and upcharge if needed for those who desire it)

I'd almost suggest the same for the CAN board, Change to be a carrier board for the D1-mini which should allow the price to drop considerably to get away from the added charges for the board; That is unless you're thinking about adding RTC functionality, as I believe the Cell holder would require the more expensive board assembly.
Added benefit to the carrier boards is the addition of USB, so it makes recovery/loading slightly easier.

Example of the ESP32 D1-Mini (First google popup, but you can find cheaper with bulk discounts) https://www.aliexpress.com/i/32816065152.html

Downside is that you may deal with some quality issues on the cheaper module boards, but the Olmex modules weren't much better in recent times

Yes worth thinking about. Unfortunately I can't edit the Tesla design files myself and am reluctant to change them yet again. Unless of course someone wants to chime in with JLC tested manufacturing files

It's a pick your poison type of thing. You'll run into problems either way.

I've never heard anything good about the Olimex boards you use, and have heard nothing but good things about the D1 Minis (which I use).

I believe Brattitude was making or going to make Olimex --> D1 converter adapters because of how frustrated everyone was with how shit the Olimex's were. Maybe 2 years back? Never heard if he ended up making any. There should be a thread here I think.

Brattitude sent me one of his boards to try on an early zombie and while the D1 was better it still wasn't great. I initially had issues connecting with with the ESP32-WROOM-32 on Johu's latest mini main board but it seems that was due to me having the current limit on my ( cheap Chinese so don't treat these numbers as gospel)power supply set too low; I found it needs at 200mA to start working even though the LEDs flash at a lower limit and it settles down to 110-115mA - this is all on the bench I haven't put it back in the casing yet

Will you be able to do real time logging with the Wi-Fi module communicating over can? I seem to remember that was a con of the can bus Wi-Fi module.

We could also do our own D1 mini type design for jlc and add a beefy (aluminum?) capacitor to the board to keep the 3.3v stable.

That same module could plug into either a main board or a can bus Wi-Fi adapter. That way you could bench test, log, then move it over to the can bus Wi-Fi.

jrbe wrote: ↑Wed Jan 17, 2024 2:50 pm Will you be able to do real time logging with the Wi-Fi module communicating over can? I seem to remember that was a con of the can bus Wi-Fi module.

That's indeed the only feature we don't have. I tend to think it is mostly important for developers who need real time plots of the high speed control code.

jrbe wrote: ↑Wed Jan 17, 2024 4:26 pm We could also do our own D1 mini type design for jlc and add a beefy (aluminum?) capacitor to the board to keep the 3.3v stable.

That same module could plug into either a main board or a can bus Wi-Fi adapter. That way you could bench test, log, then move it over to the can bus Wi-Fi.

Probably quite pricey compared to using a solder-on esp32 in the first place. The cap can be offboard.
The CAN boards look super stable and easily make it from my shed to my router (30m and 2 walls)