The VCU change discussion

[New Electric Ireland]

From a purely economical view point this is probably true.

No doubt about it. We have several ~1000 vehicle fleet conversion projects in the works and had hoped to use Open Inverter in an 800 electric Bus project in Ireland. In the middle of our discussions with the bus vendor a radical change in both the VCU hardware and software architecture was announced without any consultation. While you are obviously free to do whatever you like with your hobby it does have ramifications for those of us who are trying to take open inverter to the next level.

It's also worth considering for a moment all those people who have purchased VCU's to date who now find themselves with an unsupported product after investing considerable effort in getting them working (not an easy task for many).

[mdrobnak]

From a purely economical view point this is probably true.

In the middle of our discussions with the bus vendor a radical change in both the VCU hardware and software architecture was announced without any consultation.

Can you be more precise as to this change? The only changes recently that I've seen have been due to parts availability changes as I understood it.

[johu]

In the middle of our discussions with the bus vendor a radical change in both the VCU hardware and software architecture was announced without any consultation

While I understand your frustration I want to ask: Are any of the people involved NEI employees? If not, why should they consult with you?

[New Electric Ireland]

Can you be more precise as to this change? The only changes recently that I've seen have been due to parts availability changes as I understood it.

The only parts shortage is from your chosen PCB vendor. All the parts for the 'old' VCU are widely available and AFAIK none are EOL today (and even if they were we could have undertaken a LTB that would have kept us busy for years).

[New Electric Ireland]

While I understand your frustration I want to ask: Are any of the people involved NEI employees? If not, why should they consult with you?

I thought this was a community project. That's why we donate so much cash, parts, and time.

[johu]

Yes but you have to understand that the outcome of our development may or may not be ready for commercialization. If I were to discuss every design change with everyone using the previous version it would totally kill innovation.

If you want to commercialize you have to make sure to employ someone who will maintain your platform of choice. Because its original authors have the complete liberty to do something else any day.

Thats why the following sentence is in the header of every open source file

* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.

[mdrobnak]

Can you be more precise as to this change? The only changes recently that I've seen have been due to parts availability changes as I understood it.

The only parts shortage is from your chosen PCB vendor. All the parts for the 'old' VCU are widely available and AFAIK none are EOL today (and even if they were we could have undertaken a LTB that would have kept us busy for years).

I'm confused. The designs are open, the code is open. I'm sure if you hit into a specific problem someone would be able to help.

[tom91]

WOW guys this thread is clearly veering of track, please take this else where.

[New Electric Ireland]

Yes but you have to understand that the outcome of our development may or may not be ready for commercialization. If I were to discuss every design change with everyone using the previous version it would totally kill innovation.

We understand this is a hobby for you and don't expect you to commercialise the products. We have a lot more experience doing that ourselves anyway. However, by radically changing the VCU hardware and software with zero consultation you undermine our ability to deploy solutions based on your hardware and even bring into question our promotion of your solutions to the wider community.

Remember that all the existing VCU users are now on their own having invested considerable cash and effort getting them working (not an easy task). A much more reasoned approach would have been to engage with the community when you first started having PCB supply problems. Maybe another solution could have been found if you'd asked us. Lots of us have PCB's made.

I also see nothing that prevents you from making another arbitrary choice when JLCPCB stop supporting your current processor. All this does is strengthen the case for using a commercial solution which is stable and supported.

[New Electric Ireland]

I'm confused. The designs are open, the code is open. I'm sure if you hit into a specific problem someone would be able to help.

We're trying to convert thousands of vehicles not ones and twos.

I also wonder which version of the hardware and software you think we should deploy... the 'old' version that's no longer supported but tested over a couple of years in a few vehicles on the road, or the 'new' version that's incomplete and untested

[johu]

Remember that all the existing VCU users are now on their own having invested considerable cash and effort getting them working (not an easy task). A much more reasoned approach would have been to engage with the community when you first started having PCB supply problems. Maybe another solution could have been found if you'd asked us. Lots of us have PCB's made.

Their VCUs don't magically stop working. They just might not see further development. But there is a fundamental difference between a proprietary product becoming unsupported and an open source one: anyone can pick up development in case of the latter.
As an aside nobody else complained thus far.

We understand this is a hobby for you and don't expect you to commercialise the products. We have a lot more experience doing that ourselves anyway.

I also see nothing that prevents you from making another arbitrary choice when JLCPCB stop supporting your current processor. All this does is strengthen the case for using a commercial solution which is stable and supported.

Now thats a bit of a contradiction. The first sentence seems to express that you have well understood that you can't use an open source system "as is" when scaling up. While the second sentence expresses that you are expecting an open source system to be "stable and supported" i.e. usable as is.

It is your job as a commercial vendor to stabilize the open source system.
EDIT: what this means: you prefer the old design? Then use it and get someone to customize it for you.

[New Electric Ireland]

Their VCUs don't magically stop working.

Unfortunately they only 'just' work now and will require a lot more effort to get them usable. In reality they will be forced to replace the VCU's in most cases because they lack the skills to support them going forward.

As an aside nobody else complained thus far.

That's because they're working really hard to minimise the burden put on the developers. Just because you're not aware of the effort doesn't mean it's not happening. We have a group of 20 people from our course who are actively helping one another. Most do not post here.

[johu]

First off, I split the discussion because it was unrelated to the CCS topic.

I see the problem now. They started their conversion with an unfinished system trusting it would see further development. Like Leaf drop-in board customers were when FOC software was still under development. That's always a gamble of course.

Edit: actually it is quite a bit better then the Leaf board pre-FOC

[New Electric Ireland]

I see the problem now. They started their conversion with an unfinished system trusting it would see further development. Like Leaf drop-in board customers were when FOC software was still under development. That's always a gamble of course.

That may be true but they are still your customers and many are not overly tech literate so can't support themselves. Indeed, this years courses have an even higher number of 'end users' and 'mechanics' which is a challenge because we now face a dilemma about how much we can promote Open Inverter (in the past we did a lot of OI promotion on our courses).

[johu]

That may be true but they are still your customers and many are not overly tech literate so can't support themselves. Indeed, this years courses have an even higher number of 'end users' and 'mechanics' which is a challenge because we now face a dilemma about how much we can promote Open Inverter (in the past we did a lot of OI promotion on our courses).

Well I'd say don't promote anything you consider "just" working. You know which systems are proven over years, right?

[New Electric Ireland]

Edit: actually it is quite a bit better then the Leaf board pre-FOC

Unfortunately it doesn't work like that out of the box

Fiachra, James, and Nick all had a lot of work to do to get this working. We have another ~5 people from our course in the same boat. This is difficult.

Edit: just one example;

viewtopic.php?f=14&t=396&p=21654#p21654

[New Electric Ireland]

Well I'd say don't promote anything you consider "just" working. You know which systems are proven over years, right?

None? They all seem to be in flux and unsuitable for end users. What do you think is stable today?

[celeron55]

There are many kinds of customers for these products. Many people are very happy with an outdated open source board. They would have customized it anyway and do not need any support. What they need is just a stable board which reliably runs code and has the right electrical interfaces. They might even have considered just ordering their own batch from JLCPCB, but still ordered from Johu or Damien for convenience or for supporting development.

[New Electric Ireland]

What they need is just a stable board which reliably runs code and has the right electrical interfaces.

Absolutely!

[johu]

None? They all seem to be in flux and unsuitable for end users. What do you think is stable today?

The "stm32-sine" based systems (Tesla SDU, Tesla LDU, Leaf drop-in, Prius Gen2 & Gen3 boards, forgot any?). They are pretty much plug&play and used in commercial products (Zero EV, Stealth EV etc.). Especially known motor/inverter combos work well as they can be (and some are) completely pre-parametrized.
The Tesla charger board also works well.

[johu]

Speaking of community effort: if more people submitted their parameters, we could improve plug&play-ness
https://openinverter.org/parameters/

How can we make this better known?

[New Electric Ireland]

Tesla SDU, Tesla LDU

AFAIK they have new versions which are untested in any number of deployments. Is this incorrect?

Leaf drop-in

OK

Prius Gen2 & Gen3 boards

Which versions? We seem to have several.

They are pretty much plug&play and used in commercial products (Zero EV, Stealth EV etc.).

They make their own versions and control the firmware release. Also not available to end users only as part of a kit.

The Tesla charger board also works well.

AFAIK they have new versions which are untested in any number of deployments. Is this incorrect?

[johu]

AFAIK they have new versions which are untested in any number of deployments. Is this incorrect?

Are you trying to suggest a system that receives evolution updates is unstable? Yes the SDU received a new voltage regulator, so did the charger board. These changes are completely backward compatible and the new parts operate well within their specs. They also receive systematic logic-level testing. Of course no 1000 hour burn in tests on a million dollar testbench.

Yes Zero EV did just what a commercial vendor should do: stabilize. And profit from the groundwork laid by the open source project.

Maybe back to the VCU. I wasn't involved in the design change but I would still defend it. Moving to the STM32 + libopeninv platform eliminates many redundancies and will speed up development. Sorry about folks that have the old version and *might* not receive updates but it was their gamble (or whoever suggested it to them). And it's not the end of a world to order 20 new boards off JLCPCB to jump on the new system.

[New Electric Ireland]

Yes the SDU received a new voltage regulator, so did the charger board. These changes are completely backward compatible and the new parts operate well within their specs. They also receive systematic logic-level testing. Of course no 1000 hour burn in tests on a million dollar testbench.

So, we have new hardware and firmware releases both of which are untested in any significant number of vehicles today. I don't think many people would regard this as stable and that's a significant problem for anyone who's promoting or using open inverter solutions.

Sorry about folks that have the old version and *might* not receive updates but it was their gamble (or whoever suggested it to them)

We're in the EU and that means a 2 year warranty for goods that don't work as advertised. I'm sure you'll say 'tough' but don't you have a moral obligation to help your customers?

[johu]

So, we have new hardware and firmware releases both of which are untested in any significant number of vehicles today. I don't think many people would regard this as stable and that's a significant problem for anyone who's promoting or using open inverter solutions.

Why don't you suggest Firmware version x and hardware version y which is more proven? New firmware comes out every month, people try it, report problems or success, revert to old firmware if all else fails. Or just stick with the firmware they are running if unsure. Yes the users are also the testers, this project would not exist otherwise. And it seems 99% of the people are aware of this. If you are more comfortable suggesting 10000€ off the shelve drive trains like back in 2010, then do that.

Don't you have a moral or legal obligation to support customers who purchase your products?

Apart from the fact that this is not "my" product; yes I do. I feel such a deep moral obligation that I even hand out the blue prints so they can use my product without giving me a single penny. Can we all get some fault tolerance in return??