How to overlean every wheel

[alcatraz]

Single engine aircrafts are better than your ferrari example but they would have to be flying at very low altitudes to be comparable with eucs.

Because the crash then comes almost immediately after the loss in power. Probably still faster on the euc but... 

[soulson]

2 hours ago, alcatraz said:

Heck, at "only" 57% of free spin I'd expect plenty of power reserves too.

I think this might be where you and mrelwood are not connecting. Motor speed and motor power consumption are not directly related; being at "only" 57% of free spin speed does not imply that there is any power in reserve, nor anything else about power at all.

2 hours ago, alcatraz said:

The hardware should in fact do a lot more. They limit it to protect the hardware. The performance limit caused a crash that otherwise wouldn't have happen, or it would have happened later. Few talk about that.

What led you to believe that there is a software performance limit? Was it something in the log data?

Edited March 3, 2023 by soulson

[alcatraz]

43 minutes ago, soulson said:

What led you to believe that there is a software performance limit? Was it something in the log data?

Yeah, I believe so. The wheel hit 100A for the first time, ID(A) went from floating around 0 +/- 1 to -25 in an instant. The forward angle hadn't started its forward fall yet, but it started just after this point in time. Amperage went down to ~90A instead of staying at 100A despite the forward angle increasing.

Yeah I know what you mean about 57% and not connecting. I'm not trying to be rude about it. I just mean that the motor is not in any zero torque range yet.

If the guy did input an insane sudden lean, I would have still wanted to see a progression in the log to be more convinced of that. Not a single log entry sudden change. 

I could be wrong. I just like to be able to investigate. Why did the amperage go down by 10% even though it was in an overlean?

[soulson]

I could be wrong too, but that wasn't my interpretation of events from the log. I see two possibilities:

1. The wheel was simply overpowered. The controller is trying to keep Id as close to zero as possible while also stabilizing the wheel by controlling Iq, but keeping Id close to zero and continuously adjusting Iq both take power. When the power ran out and that wasn't enough to set Id to zero and Iq to whatever its target was... they rapidly deviated from their targets. Remember that Id and Iq exist in a rotating frame of reference. The 3D-space vectors that represent the axes along which Id and Iq are calculated are spinning pretty quickly. If the controller fails to "keep up" at any point, loss of balance will be rapid.
2. The controller malfunctioned. It's very strange that both Vd and Vq latched at 20852 (or 43.63 in half-precision floating point) for half a second. I don't know enough to jump to conclusions about this, though. It could be that the values shown are not actually half-precision floating point – maybe this is some kind of control code. It could also be that once the controller knew that it was overpowered, it stopped trying to correct the stator currents via Vd and Vq and the latched values are meaningless. But, it could also be some kind of defect. I just don't know.

Also, the controller is adjusting Vd and Vq to correct the error in Id and Iq thousands of times per second (4kHz to 30kHz are common field-oriented control frequencies) but this log data is sampled at 10Hz, so we're not seeing anything close to the whole picture of what happened by looking at this log. By the time we see the Id of -25, the system has probably been out of control for thousands of frames.

[mrelwood]

4 hours ago, alcatraz said:

I'd be curious to compare the other overlean logs too. Don't get me wrong. I'd plot all the overleans by their % of free spin and the momentary motor power.

And that would tell us a whole lot more than just the free spin speed. Especially if you add battery temperature and parallel battery currents into the picture.

4 hours ago, alcatraz said:

The performance limit caused a crash that otherwise wouldn't have happen, or it would have happened later. Few talk about that.

It could be that without a possible hard current limit (20A higher than tilt-back) the rider would’ve crashed at 84km/h instead due to a fried board. That would’ve made the crash quite a bit more violent (cutout vs overlean), at a higher speed, and the rider wouldn’t have been able to ride home. So a worse situation in all accounts.

4 hours ago, alcatraz said:

I bring up the Master doing 90 to point out that the V13 isn't close to its hardware limits by comparison.

You seem to imply that the only hardware limit is the one related to the free spin speed. How about battery temperature and current, or the mosfets’ capabilities?

4 hours ago, alcatraz said:

Heck, at "only" 57% of free spin I'd expect plenty of power reserves too.

And there were. But the large reserves still aren’t limitless. Like I said, I have overleaned all my wheels at 10% of their free spin speed. Shouldn’t the power reserves have been much more plentier then?

4 hours ago, alcatraz said:

why not focus on the fact that all that beefy hardware is at the mercy of software? Riders might think they're safe when they really aren't.

Because those apply to all wheels. The only reason we can even stand on the wheel is at the mercy of the software. Every overlean is a situation where the rider thought the wheel would be able to do more than it did. These are not news.

4 hours ago, alcatraz said:

I'm trying to spread a bit of caution to those looking at the hardware and extrapolate certain power reserves based on previous experience.

A worthy goal. No point in talking only about the V13 then. Again applies to all wheels.

 

4 hours ago, alcatraz said:

Euc might better be compared with single engine airplanes.

I don’t know much about airplanes. But I guess there are similarities to a situation where a new airplane model advertises a steeper climbing angle. The pilot then yanks the airplane pointing straight up and tries to reach the stratosphere “as fast as possible”. Without surprising anyone else, the engine still overheats, and the explosion prevention mechanisms stall the plane only at 10k km.

 During the aftermath a guy at the forums is perplexed that the plane only reached 10k km of height before stalling despite all the “empty promises”. He also references to older and weaker planes that were able to climb (slowly) to 15k km.

 Despite stalling at 10k km, the new airplane was more powerful, and was able to reach steeper climbing angles than the previous models. But dear lord, not at 90 degree climbing angle! All other pilots were successful in climbing up to 20k km on the new plane. They just used a reasonable climbing angle, which they decreased further the higher they got.

[alcatraz]

2 hours ago, mrelwood said:

And there were. But the large reserves still aren’t limitless. Like I said, I have overleaned all my wheels at 10% of their free spin speed. Shouldn’t the power reserves have been much more plentier then?

They were enough for the controller to draw 100A = 10kw just before the crash.

Sure, it could have been a different limit other than the motor. Such limits could have been ignored (except maybe fuse dependant limits) to save the fall from happening. It is of course fine to impose limits if riders are aware of the manufacturers safety philosophy and take great care when comparing performance between different models.

Why would the wheel not draw full amperage as it was going down? I suspect it's what was hinted at earlier. Perhaps a time limit on the pack had been exceeded risking to blow the fuses. Has Inmotion come out with a crash report?

Why was there no progression of the input lean if the log is in 10Hz? Looking at the overlean moment as an example, the overlean angle took maybe 4-5 log entries from start to end so that's around 0.5s then. A clear progression.

Looking at the logs I'd say that the wheel has (with current software) a 90A sustained draw limit that can only be overridden momentarily. The rider crossed 90A for too long and the controller protected the wheel.

I'd love to understand soulson's analysis better. I'll read that again. Yes, I also found the 20852 fixed for several log entries odd. 

So going to the plane analogy.  You're landing in fog and your radar malfunctions. A mast appears in front. You ask for 100% engine power and 2 seconds later it goes to 90% and boom. I know terrible analogy but still.

[Chriull]

18 minutes ago, alcatraz said:

Why would the wheel not draw full amperage as it was going down?

If one looks at the log numbers it's just a normal overlean - max torque limit for the speed reached.

Voltage sag over the internal resistance and at the motor coils assuming about 140km/h lead to very possible figures.

Can easily be reproduced with the formulas in my first post - don't know if i kept my calculations...

There is no valid reason mentioned till now to assume or speculate any artificial limits or power reserves.

[Mono]

1 hour ago, Chriull said:

If one looks at the log numbers it's just a normal overlean - max torque limit for the speed reached.

Voltage sag over the internal resistance and at the motor coils assuming about 140km/h lead to very possible figures.

Can easily be reproduced with the formulas in my first post - don't know if i kept my calculations...

There is no valid reason mentioned till now to assume or speculate any artificial limits or power reserves.

I agree that overall it's not surprising that the wheel runs out of torque. Still, do you have an explanation for the observed values of VQ, VD (they become exactly equal just before the wheel drops forward) and of ID (which assumes a large negative value just before the wheel drops forward)? The evolution of these values should IMHO be better explained.

[Chriull]

1 hour ago, Mono said:

I agree that overall it's not surprising that the wheel runs out of torque. Still, do you have an explanation for the observed values of VQ, VD (they become exactly equal just before the wheel drops forward) and of ID (which assumes a large negative value just before the wheel drops forward)? The evolution of these values should IMHO be better explained.

These values are internal computed values, too. As @soulsondid some interpretation of these values they seem not helpful in identifying a specific problem but this overlean.

As the other log numbers.

IMho the only problem is that the V13 was once marketed as overlean proof, which is stupid and not possible with actual technology.

And people are puzzled now that after some time of strong acceleration the wheel overleaned.

[UniVehje]

I just realised that I haven't seen anybody asking what was the gradient on the recent V13 over lean. Sure it looked like pretty flat but even a small uphill will limit the max speed. How likely it is that the road was exactly flat or slight downhill? What if it was 4 degree uphill which would look like flat?

It looks like once a company sets a speed limit, it's considered as a promise. It is not the max speed V13 can do, it's the speed where tiltback is forced no matter what. Max possible speed of V13 is much higher, depending on rider weight, gradient, wind etc. It can also be much lower. Nobody's expecting V13 to go 90 km/h on 40 degree uphill. 90 km/h is just the highest speed allowed before forced tiltback. Not a promise. 

Fast acceleration, uphills, heavy riders etc. are totally different scenarios. In these scenarios there is a different protection, a tiltback triggered by motor current (or whatever). In this case it is fair to call for clearer tiltback or better implemented alarm sounds. But the over lean itself has nothing to do with max tiltback speed or max possible riding speed or max no-load speed. Those are only possible in certain scenarios, namely flat ground (or downhill), normal weight rider, normal wind, normal acceleration etc. 

[mrelwood]

4 hours ago, Chriull said:

IMho the only problem is that the V13 was once marketed as overlean proof, which is stupid and not possible with actual technology.

The main problem with the V13 marketing even going near the word “cutout”, was that most people would either remember or interpret it completely wrong. Or both. You included.

”No cutouts is what we strive for”

Unless I’m completely mistaken, that was all that V13 marketing has ever said about cutouts or overleans.

Cutouts are synonymous to overleans for some people, and add a little bit of bad memory and people indeed think that Inmotion promised that the V13 will never cutout or can’t be overleaned.

Edited March 3, 2023 by mrelwood

[novazeus]

running over a toad at top speed could cause an overlean.

a toad a turd, any t's.

yes, the cows poop on my road.

easiest thing to remember, "speed kills". e=mv squared

Edited March 3, 2023 by novazeus