Idea: simplified inverter firmware

[johu]

The current inverter firmware is actually VCU+Inverter with stuff like cruise control. Of course this results in many parameters that tend to confuse people. Also the two variants SINE and FOC tend to create confusion.

The idea is to create a firmware that implements only inverter functionality, no VCU things, no cruise control, no contactor control. Just like most OEM inverters. All other tasks must be handled by a separate VCU like Zombieverter.

So I want to run this idea by you:
- Reshuffle the firmware to only accept torque commands via CAN
- Keep the battery voltage and current deraters
- Remove all other parameters
- Merge sine and FOC firmware or better still run async motors with FOC
- Integrate the various auto tuning and test modes

What are your thoughts? Will this reduce the support burden?

[MattsAwesomeStuff]

Who's support burden? Presumably yours?

Is that what you see most of the support questions about?

- Won't most people want cruise control and contactor control? So now they just have to get a VCU on top of the inverter?
- Does splitting the hardware and software into two packages solve that?
- I don't understand how you merge SINE/FOC firmware, they're different. You mean just bundle both and have the user toggle which they want to use on the inverter screen?
- Is forcing CAN helpful to people?
- Integrating things sounds like it would be helpful.

This is an interesting challenge for you, because you have to put yourself in the shoes of someone too dumb to understand as much as you understand. It can be easy to step off this path, presume dumbness in the wrong way, and end up putting effort into something that is a side step on usability. Overly dumb in ways no one needs, and still too complicated to be solved by a beginner in others. Sometimes I feel that way about parts of Damien's tutorials. Usually good, good, good, Oops he said "connect it", and didn't say what "it" was or where to connect it. So if you can't figure that out, now you're stuck. It needs a lot of polish, because, a tutorial that has 1 part not well enough explained is like code that won't compile because of just 1 error. To someone who wrote the code, it's 99.99% of the effort. But to a novice, 1 error is equivalent to 100, either way they can't get through the process.

Just for example, how many times are people confused about what the wifi name and password are? And that's something you've written down and told them. If people get stumped there, you really have to dumb it all down to that level.

As a novice, the parameter list is basically just alphanumeric soup to me. The only reason I'm touching anything is because some tutorial told me to change something, or someone troubleshooting why I can't get it running is telling me to change something. Generally, I find the process to be not well explained or documented, or at least not usefully documented in a way I'm able to figure out on my own. I'm throwing darts in the dark if I try any of it on my own. I don't know what I'm changing or if it's having an effect.

I think a "Quick Start Wizard" process would be helpful, instead of walking noobs into the cockpit and telling them to feel free to flip any switches they need to. I'm not sure what this would look like, something like a merging of the initial setup tutorial and the inverter firmware. It hand holds you through the process, just asks you questions or gives you instructions one step at a time. Perhaps difficult to pull off because of the size of the interface file? Or it could be hosted, Wizard completed, and then file uploaded to the inverter, but then you'd lose the ability to read things on the inverter during setup.

This could also combine a bit with a parameter database. There's not THAT many options for motors, no point in reinventing the wheel. But right now the parameter database includes very specific settings from other people's builds that aren't relevant to yours, and will mess yours up. Different throttle pedals or battery voltages or whatnot.

Having parameters available certainly isn't a bad thing, but, perhaps hidden unless asked to be revealed, with the more idiotproof settings up front?

...

Backing way up...

A beginner shows up with a motor and a battery and an inverter, and some OI hardware. ... what do they do? How do you get them from that point, to "Your inverter is now set up", and what is the best way to get them there?

We currently have a tutorial for this, why doesn't that tutorial work, where do people get stuck, how is it incomplete, and would any proposed changes avoid that?

[johu]

Thanks for your input, much appreciated.

Who's support burden? Presumably yours?

Questions anywhere, forum, email, youtube comments. Or lets not call it burden but part of the "making openinverter more accessible".
To be honest it's also thought from the developers end. If you can concentrate more on the core task of a device becomes easier to maintain and polish.

- Won't most people want cruise control and contactor control? So now they just have to get a VCU on top of the inverter?

My assumption here was that most people will use a VCU anyway. With all the CAN controlled stuff it seems difficult nowadays to build an EV without a VCU.

- Does splitting the hardware and software into two packages solve that?

I'm hoping to shorten the path from unboxing to first spin. The total setup work is of course the same because some parameters just wander off to the VCU.

- I don't understand how you merge SINE/FOC firmware, they're different. You mean just bundle both and have the user toggle which they want to use on the inverter screen?

Yes, that. Select or autodetect

- Is forcing CAN helpful to people?

As above, I presume in 90% of modern conversions it's already present. But I could be wrong.

Agreed some sort of "wizard" might achieve the same. Most stuff can even be determined without talking to the inverter. So it could be hosted at say openinverter.org/wizard with all the beautiful GUI stuff the web has to offer and what comes out is a configuration file that you upload to your web interface.
The wizard could offer a commissioning parameter set with all the deraters out of the way for example.
Of course motor characterization can't be done this way but the "VCUish" stuff could.

We have many loose ends already there, e.g. the parameter database.

[MattsAwesomeStuff]

Questions anywhere, forum, email, youtube comments.

Yeah I was wondering to what degree what you see as support differs from what I see as support. I really only see forum stuff. Was thinking about "What takes up time out of your day that I might not be seeing."

My assumption here was that most people will use a VCU anyway. With all the CAN controlled stuff it seems difficult nowadays to build an EV without a VCU.

Ahh, not everyone's converting a 30+ year old car.

I don't even know how that's possible then. Like, if you're trying to convert a more modern vehicle, who's doing all the reverse engineering to decypher the CAN streams and get all the vehicle features working? That's going to be a vehicle-specific task, and, basically no one who would need a tutorial would be capable of doing that work.

I kinda figured it was just a lost cause to try to convert something modern.

As above, I presume in 90% of modern conversions it's already present. But I could be wrong.

Honestly I probably start to tune out of people's builds at that point, but I haven't seen a lot of that. So many loose ends.

The wizard could offer a commissioning parameter set with all the deraters out of the way for example.

Yeah, I mean, ideally you'd hardly even need to log into the inverter page. For all most beginners care, you could leave it mostly as-is if there was a more intuitive, guided way of establishing parameters.

What motor are you using? Leaf, Prius, Lexas, Tesla, Other.
Which inverter are you using?
What battery pack voltage are you using?
Etc etc.

Of course motor characterization can't be done this way but the "VCUish" stuff could.

Still some handholding is needed for actual hookup. Like, "Have your battery mostly charged up. Measure your current battery voltage with a multumeter. What is the value?" and then it can compare what it's reading to the voltage and set those parameters up for you, instead of the user working out the math themselves.

A more automated synchoffs procedure would be a fancy addition too.

[P.S.Mangelsdorf]

To offer my thoughts, for however much they're worth:

The all-in-one nature of the OI control board is part of what drew me to Open Inverter in the first place. I'd used a simple-as-dirt Soliton for a DC motor conversion project in college, and the idea that I could run a Tesla drive unit with a board swap and just 12V wiring was extremely appealing.

Obviously, as the build went on, other systems have required/allowed me to learn CAN and add other control boards, to the degree that I expect my next build will utilize a VCU. But I still think the one board nature of the core OI and other drop in boards is far more accessible to the average car guy that is trying to find their way into EV conversions. Effectively, it lets you bite off as much or as little as you want. If your inverter and motor can all be controlled with 12V signals, there are plenty of chargers and DC-DC converters out there that run the same way, and that's really what you need to get started.

I think its hard for those of us that have been doing this for many years to step back and see what a wall of knowledge there is between the average car guy and EV conversions. Places like Open Inverter are really good at creating doors in that wall, and that is something I am deeply thankful for, and something you (Johannes) should be extremely proud of creating.

I've spent a lot of time this summer with the '40 Chevy at traditional car guy events, and have spent more time talking to non-EV guys about EV conversions than I ever could have imagined. A lot of them were really interested in how everything worked, and how someone learns how to do a conversion. Frankly, so much of the American auto enthusiast media makes it out to be this complex witchcraft that requires a super knowledgeable shop to do it.

The status quo, at least with the Tesla drop in boards, really lets you learn as you build the car. You don't need to have everything planned out in advance or purchase everything at once. There's a lot of great things about a VCU based conversion, but I don't think simplicity is one of those things. It may be the more rational way, but asking someone to learn not just HV wiring and inverter control, but also CAN bus and VCU set up and so on seems to be a bit much.

To put it more simply, I'm planning on using a VCU in my next build, and I'm still very nervous and confused about how to set it up, despite having been playing in the OI world for 4 years at this point and with EVs for 8 or 9. I think the all in one aspect is a huge strength of the OI approach, not something to be eliminated.

[arber333]

Hm...
Would it be possible to change the ID of the torque request CAN telegram? Would ESP32 interface still remain?

I would suggest you add a NOTAUS telegram as well. The VCU would repeat some CAN info which would enable or disable inverter torque output.
This could also be used to disable inverter when charging.

On the other hand; how do you think different motor parameters would be determined? We would need a good database and a simple way of delivery.

[celeron55]

Is it possible to modularize it so that it's possible to make a minimal build, but also possible to make a build that adds the VCU functions and contactor control capability as an extra layer but uses the exact same code for the actual inverter implementation as the minimal build?

Aside from VCU functions and contactor control, connecting the throttle directly to the inverter is something Tesla is known to do. That particular thing is not necessarily a bad design as nothing else in the vehicle really cares about the throttle. Cruise control could be similar, and then you need brake input also. Of course brake input is a good idea with throttle too, for safety.

Contacor control is a more difficult question. I think generally it would be best left to a dedicated contactor controller which accepts requests from the inverter, charging system and the BMS and controls the contactors based on those. It would implememt the precharge sequence. It could also implement a PWM economizer. In my conversions I have integrated the contactor controller into my BMS but that's a relatively unpopular choice. A popular choice is having Zombie or OI do it.

[jrbe]

I see the inverter firmware as is to be simpler for someone new to this to end up with a drivable vehicle. Every extra piece steepens the learning curve and likely lowers the chances of success. I don't mean people who would never be able to do a conversion should be the target audience. Just to consider what the path ends up looking like for someone new. Sometimes it's hard to see how steep it is from the top.

Splitting out the functions makes sense evolution wise but you need rock solid communication, hardware, and info to support them.
I think part of this is a discussion of using separate critical and lower priority can lines. Otherwise you might hear something like I finally found my illuminated Mercedes emblem is what was screwing up my inverter.
A 2nd hand example. A friend had a cheap / noisy cell phone charger that was scrambling can info in their ~2019 Subaru. It was a few hours of diagnostics to find that was the cause of their dashboard Christmas tree lights. Unplugging the cell phone charger got rid of the issue.

[jerrykco]

I for one am still finishing my 1970 MGB build and won't have a VCU. My next build (already have it is a 1964 VW crew cab pickup). It also won't have VCU.

Its been awhile on the MGB, medical issues and life issues, but I am back at it again. But as you noted as a newbie I had lots of issues and questions about things I did not understand. Johu helped and many others also helped. I think the wizard idea however has great merit.
KISS applies here.

[Bratitude]

hmm.

i think extra function be a separate menu would clean things up. its very nice having a board that runs the core ev system. no extra boxes. but to simplify things, i agree.

[johu]

Great stuff, thanks for all your input and reassurance, too.

It looks like the general vote is quite obviously, do not change the structure, just keep polishing the current one. Sometimes it's worth visiting another country to find out your own is actually not bad either.

So yes, when pulling up the web interface something less intimidating than the 60 parameters and 60 spot values could appear. The friendly wizard, at least until you disable that and have to legacy page back.

Driving 270 km to Mannheim now, might not be here much this weekend. Keep it coming anyway.

[Zieg]

I'm glad the OI can do contactor control and interface with my throttle, it means I don't need a separate VCU. That's a nice savings in cost and complexity for me.

[jerrykco]

BTW et al, I bought a Charger for my MGB from Thunderstruck. A TSM2500 Output: 312V (Adjustable) with CAN and adjustable ID, 1 X $555.00 USD
Wolf from Wolftronix is writing the interface to talk to it via CAN bus (the PackSniffer2 has two can bus ports and can continue to run the Leaf BMS at 500kbit, and the Thunderstruck TSM2500 Charger would be 250kbi). With that and the currant interface he has to the Leaf BMS I'll be set.

[E46Driver]

To offer my thoughts, for however much they're worth:

The all-in-one nature of the OI control board is part of what drew me to Open Inverter in the first place. I'd used a simple-as-dirt Soliton for a DC motor conversion project in college, and the idea that I could run a Tesla drive unit with a board swap and just 12V wiring was extremely appealing.

I'm glad the OI can do contactor control and interface with my throttle, it means I don't need a separate VCU. That's a nice savings in cost and complexity for me.

I'm in full agreement. My first conversion was the Soliton/DC world. And that was ported over to my 2nd conversion --- and now swapped to a Leaf with O.I. The parameters seem daunting at first. But anyone performing a conversion should "read" and "learn". There is a magnificent amount of good information on this forum - along with a magnificent group that is always trying to help. Once I had a core base of knowledge, I could approach the parameter setting with more confidence. Some things are still confusing, but I'm still learning.

I also like the "simplicity" of a drop-in board which can fundamentally operate my entire EV. Yes, the vehicle has several different CAN buses --- but that can programmed/handled with other methods. When I first started, I was able to run the Leaf with nothing more than a few 12V signals and a throttle. Absolutely a win and thrilling sensation to know that I could perform surgery on an OEM unit and end up with full control over it!

So my vote is to (at a minimum) keep the contactor control and throttle pot. Maybe an external regen signal. And, of course, the basic safety/protection protocols already in place. But, please don't "over-simplify" and "over-complicate" with CAN control only.

[marcexec]

I think the web interface is a bit intimidating (not that I've actually worked with it...). Maybe hide a few of the more obscure settings behind an "Advanced" toggle? E.g. the all the charger bits when looking at https://openinverter.org/wiki/Parameters