V12 Cutout tracking

[Mrd777]

I’m planning on testing my V12 Saturday when I’m back in the city, however this entire run around seems pointless and silly as we are always going to lack confidence in our wheel if we pass the test. Trust and confidence is everything in this sport. Why is Inmotion not just doing a total recall on all of our control boards?  Case closed, no monkey tests.

[spitfire1337]

Some interesting damage I found on one of the phase wires at the controller caused by the small heatsink in the top board 🤔

[TiguriusRides]

Just did the test tonight, killed it after only 3 tries.

 

[Sympul]

47 minutes ago, TiguriusRides said:

Just did the test tonight, killed it after only 3 tries.

 

Oh man... that's shitty. And it looks like you added a bunch of upgrades as well.  

[Waulnut]

15 minutes ago, Sympul said:

Oh man... that's shitty. And it looks like you added a bunch of upgrades as well.  

He just needs to have it repaired... it isn't like its on fire and melted down at least.

[rolis]

Thank you everybody for your help and sharing your test results. I have collected data from telegram groups, discord, facebook and this forum. I will continue doing this for the next week. please contact me on telegram: https://t.me/roliwe or facebook: https://www.facebook.com/roliwe/ if you dont see your test result on the page
https://spurious-tea-df7.notion.site/Inmotion-V12-MOSFET-Test-statistic-7786b4fdde50416eb7dc1a4674121ec7

Personally, I have canceled my order on the v12 because of this.

Have a nice day

[Khazik]

@rolis Nooooo, come suffer with us!

Quick! He's climbing outta the bucket, grab 'em! 

*Edit* Really, ty for doing the site to track the results, it's great. 

 

Edited January 6, 2022 by Khazik

[Djal]

Same, cancel v12 order replace by v11.

[Planemo]

7 hours ago, Mrd777 said:

I’m planning on testing my V12 Saturday when I’m back in the city, however this entire run around seems pointless and silly as we are always going to lack confidence in our wheel if we pass the test. Trust and confidence is everything in this sport. Why is Inmotion not just doing a total recall on all of our control boards?  Case closed, no monkey tests.

Totally agree, sod that. No way I could put trust in the V12 at the moment. I wouldn't even want a go on one.

So people have been advised to run the test 10 times. What if it happens on the 11th?

[Paul A]

Could Inmotion use V11 control board(s) to replace the V12 control boards?

Is that possible, would it work?

 

[Rawnei]

11 minutes ago, Paul A said:

Could Inmotion use V11 control board(s) to replace the V12 control boards?

Is that possible, would it work?

 

No.. it's not the same.

They could probably use the boards from the new HT model.

[Chriull]

9 minutes ago, Paul A said:

Could Inmotion use V11 control board(s) to replace the V12 control boards?

Is that possible, would it work?

 

No, the V11 is an 84V wheel, the V12 a 100.8V wheel. The used dc/dc converters to supply cpu, leds, etc could maybe work over the whole range, but the big buffer capacitors and mosfets at the inverter likely cannot take the higher voltage, as a commonly used voltage rating above 84V is 100V which leaves a negative safety margin if used with 100.8V....

So it's rumoured that exactly such a reuse could be the/a culprit...

[terlikaa]

That's mine:

https://www.facebook.com/groups/inmotionv12/permalink/499504478148478/

It passed. 

[Paradox]

I do hope that when a V12 passes this test that it is safe to ride.  I do not hope we start hearing of wheels that passed this test only to fail in real world riding.

[terlikaa]

The way I did the test, I don't think that in real world I'll push it that hard. It should be fine. 

[Chriull]

1 hour ago, Paradox said:

I do hope that when a V12 passes this test that it is safe to ride.  I do not hope we start hearing of wheels that passed this test only to fail in real world riding.

Cozld not find a prayer and knock on wood emoji... ;(

1 hour ago, terlikaa said:

The way I did the test, I don't think that in real world I'll push it that hard. It should be fine. 

With every lean were the wheel has to accelerate/decelerate the whole riders body mass the burden is magnitudes higher as with such a "lift spinning test". 

During riding a wheel just the burden changes could be slower?

However - just did some reading - "normal" avalanche effect above breakdown voltage should not be the primary thread, as infineon states avalanche voltage beeing about 1.2x higher then breakdown voltage.(1) Avalanche energy "taken" by the mosfet IPP023N10N5 seems nice ?high? with 1166 mJ - so "seldom" spikes could not deem a real threat.

Just such low energy repetitive avalanche will destroy the mosfets over time by micro damages (1)

Much more troublesome seems the SOA diagram ;( For drain source voltages near this 100V ridiculous low currents or low duty cycles are allowed!

So imho the main threat cozld be high burdens at lower speeds, like doing pendulums.  Doing the lift spin test one should try to initiate the burden (spin direction change) once the wheel decelerated to low speed. Changing the direction while the wheel spins fast could render this test useless?

Edit: Seems i just had a total brain breakdown regarding mosfet SOA and how they are used in inverters... ;(

3 hours ago, Chriull said:

big buffer capacitors

They are nicely rated 160V rated according to ecodrifts teardowns of v12 from batch 1 and 2.

 

(1) Infineon AN_201611_PL11_002 "Some key facts about avalanche"

Edited January 6, 2022 by Chriull

[supercurio]

2 hours ago, Chriull said:

With every lean were the wheel has to accelerate/decelerate the whole riders body mass the burden is magnitudes higher as with such a "lift spinning test". 

During riding a wheel just the burden changes could be slower?

The synthetic stress test has the advantage of triggering peak power near max speed (then max voltage), as the controller does its best  effort to apply the maximum torque to change or even reverse the wheel rotation.

I guess it's good to simulate what was triggering the cutouts, also brief intense spikes on bumps or curbs.

Hard braking and accelerations are several seconds long instead of a few milliseconds, about 100-1000 times longer.

We don't have the firmware source to check the limits of what's allowed to peak for a few milliseconds vs seconds so it's hard to tell if there's risk for hard acceleration and braking.

Bumps, especially bumps in succession during hard acceleration or braking sounds like the nightmare scenario.

[Richardo]

@OldFartRides  @Tawpie

What kind of stuff would you look for in a root cause analysis to make you comfortable that the issue is fully understood and solved?

[Tawpie]

1 hour ago, Richardo said:

What kind of stuff would you look for in a root cause analysis to make you comfortable that the issue is fully understood and solved?

I'd want to have a decent sized sample of parts de-epoxyed so the actual failure could be seen in a microscope and you can better guess (it's a guess) as to the actual cause. But it certainly is looking like the part just isn't up to the job as hinted by its negative design margin as a 100V part so it's absolutely valid to skip the microscope and jump to that conclusion.

The obvious first step is to redesign the driver board with MOSFETs that are rated (eg: selected at time of manufacture) for the use case. This may require redesign of the drive circuit as well, because the drive circuit is not independent of the MOSFET. Then you retest to determine where the statistical limit is and decide if you have sufficient margin to reduce the risk of a cutout to an acceptable level. "Test to destruction" is on the table here, and should be for ALL drive circuits... at what point do the wings actually break off the airplane?

But I would never accept/ride/keep the V12 as it's currently designed... something is badly wrong and while I enjoy casinos, when the potential for rehab is involved I do prefer having the odds more in my favor than what we're seeing here. Replacing a poorly designed board with another of the same design isn't a solution.

It's time for a general recall. <-- my opinion!

Edited January 6, 2022 by Tawpie

[rolis]

10 minutes ago, Tawpie said:

But it certainly is looking like the part just isn't up to the job as hinted by its negative design margin as a 100V part so it's absolutely valid to skip the microscope and jump to that conclusion.

This. 

 

And just to confirm, a quote from Cecily from Inmotion:
"Theoretically it’s should be better to solder higher rated mosfet, but in reality there are a lot more to consider to make sure the selected model a good fit for the system, so we don’t recommend to do that at this moment"

Either they find a suitable replacement for the mosfet, or they will redesign the boards for the next batches.

[rolis]

I have also got a response on what Inmotion is pushing to sellers as a reason for failure:

Quote

So far they said that the issue came from the change of the MOSFET provider for their production. The defective ones will have to be replaced and they will probably take measures to fix the problem during the production of the next batches. 
That's all we know so far.

Since we know that they are using the same MOSFETs as on their 84V wheels, it is very unlikely they would have another provider for the exact same product {same manufacturer, same model)

[Paul A]

Inmotion - about us.  Our story.

Inspired by their win at the RoboCup 2007 Competition in China, INMOTION was founded by a technical team with experience in robotics and vehicle dynamics. 

_____

How is it possible that the engineers of Inmotion haven't thought in the same manner as Tawpie's post above......especially in having sufficient margin to reduce the risk of a cutout to an acceptable level...testing to destruction....

They've been making EUC's for years, accumulating experience all the while.

 

Yes, replacing a poorly designed board with another of the same design, not really a solution at all.

[rolis]

1 minute ago, Paul A said:

How is it possible that the engineers of Inmotion haven't thought in the same manner as Tawpie's post above.....

I would love to answer all of these technical questions, but I am just a electrical engineering student. In theory you put more mosfets together to act as a big one to either handle more current, or higher voltage.

The circuitry has to be pretty precise in such configurations so it does not happen that at one moment, a single mosfet has to bear all of it. I have seen from pictures that the V12 has 12 Mosfet? I have no idea what configuration they are in, so its hard to tell. 

Problems like these are the reason why I love electrical engineering  you can really dive into it and get into a rabbit hole full of information

[Chriull]

17 minutes ago, rolis said:

I have seen from pictures that the V12 has 12 Mosfet? I have no idea what configuration they are in, so its hard to tell

That's the bldc driver (inverter) with 3 H Bridges - 6 "switches". Each switch is made of 2 paralleled mosfets.

Imho paralleling is "easier" to bear more current as putting in series to bear more voltage. Especially as higher voltage models are easily available...

[nosamplesplease]

2 hours ago, supercurio said:

The synthetic stress test has the advantage of triggering peak power near max speed (then max voltage), as the controller does its best  effort to apply the maximum torque to change or even reverse the wheel rotation.

I guess it's good to simulate what was triggering the cutouts, also brief intense spikes on bumps or curbs.

Hard braking and accelerations are several seconds long instead of a few milliseconds, about 100-1000 times longer.

We don't have the firmware source to check the limits of what's allowed to peak for a few milliseconds vs seconds so it's hard to tell if there's risk for hard acceleration and braking.

Bumps, especially bumps in succession during hard acceleration or braking sounds like the nightmare scenario.

This stimulated a thought:

Initial testing, almost everyone said the braking was NOT strong enough and a new firmware "fixed this".

Sudden high stress directional changes seem to be the cause of these failures. High stress directional changes like braking?

Was the initial firmware intended to limit this failure condition?