Speed-to-voltage scatter plots for few popular EUCs

[Seba]

8 minutes ago, Chriull said:

The datapoints above tiltback speed are missing in these graphs? There were (not too many) in your speed over voltage diagramm!

Or they few points are just invisible by the 

52 minutes ago, Seba said:

data filtering, binning and clustering

?

They were filtered, because wheel speed and GPS speed wasn't correlated. In speed/battery diagrams I used less strict filtering criteria, so it was possible that plot could contain data points recorded after crash, with wheel lifted up (in both cases wheel rotating without load) or with incorrect GPS speed measurement (poor sattelites geometry or lot of signal reflections, like in dense urban areas with tall buildings).

These plots has been created almost by-hand, just for curiosity. I plan to create entire statistics & research page at euc.world so we will be able to get detailed analysis.

[andress]

It would be really cool to be able to compare your own data points (from all rides/ last ride) with these overall ones giving you rough idea how close to limits you were riding

[Aneta]

8 hours ago, Seba said:

Ok, here are speed-to-current and speed-to-power scatter plots for King Song KS-16X, KS-18L and KS-18XL.

Strange that we don't see any envelope at slow speeds and high currents, too few points? Is it because how power delivery of KS wheels work? I can get a high-current beep on my GT16 (more than 40A) almost any time I start from standstill. Here, we see about 1A from the start - how is this possible?

Makes me question KS's telemetry once again. We need to come up with some method of validating the telemetry. 

[Chriull]

9 minutes ago, Aneta said:

Strange that we don't see any envelope at slow speeds and high currents, too few points? Is it because how power delivery of KS wheels work? I can get a high-current beep on my GT16 (more than 40A) almost any time I start from standstill. Here, we see about 1A from the start - how is this possible?

Makes me question KS's telemetry once again. We need to come up with some method of validating the telemetry. 

Imo some points adding up:

KS has some good current current limiting for low speeds.

High needed phase currents at start up are transformed to low battery currents. With a "good" motor control algorithm this seems to work out.

My KS16S graphs show ~10A at about 0 speed. @Seba s graph show just some sparse point in this range. Afair he saves/transmits "only" 1 sample per second vs 5 samples per second per "original" wheellog. And at ~0 km/h driving least time is spend. Most time at this speed is standing and waiting without current needed. So high burden starts at this low speed need more data collection to really show up in the graphs?

@Seba s binning/clustering shows somehow "real" frequencies of how often samples occured by leaving seldom areas sparsely populated. While in my graph some spots are cluttered up but are compared to other areas in reality very sparse?

Maybe in a year this graphs look very different whith much more data collected?! 

[Aneta]

28 minutes ago, Chriull said:

KS has some good current current limiting for low speeds.

High needed phase currents at start up are transformed to low battery currents. With a "good" motor control algorithm this seems to work out.

My KS16S graphs show ~10A at about 0 speed.

Strange. Even with phase current limiting at 100A, the battery current can be about 35A from standstill:

https://www.ebikes.ca/tools/simulator.html?batt=cust_84_0.2_20&cont=cust_50_100_0.1_V&wheel=18i&frame=cust_1_0.01&hp=0&blue=Lbs&motor=M3540&cont_b=cust_100_200_0.03_V&batt_b=cust_84_0.2_20&wheel_b=22i&frame_b=cust_1_0.01&hp_b=0&grade_b=30&eff_b=97&gear_b=2&mass=100&mass_b=110&autothrot=false&autothrot_b=false&throt_b=100&grade=0&throt=100

(move the dashed black line to low speeds - this will imitate the wheel accelerating from standstill at max thrust) People have been burning controllers doing pendulums (esp. Z10 is prone to that?). Aggressive start from standstill should create big spikes in current, but we don't see that in these graphs. Perhaps, graphs for other brands will show them? 

 

[Mono]

59 minutes ago, Aneta said:

Strange that we don't see any envelope at slow speeds and high currents, too few points?

Look twice and don't focus on the dark red but on the light red. I can see the max speed envelope clearly lowering with increasing currents, like 15km/h@60A, 20km/h@46A, 40km/h@34A, 50km/h@19A.

[Aneta]

5 minutes ago, Mono said:

Look twice and don't focus on the dark red but on the light red. I can see the max speed envelope clearly lowering with increasing currents, like 15km/h@60A, 20km/h@46A, 40km/h@34A, 50km/h@19A.

Yes, I saw them, but nothing in the <5kph, >20A area. And way too few points. With people riding in cities, with constant accelerations from standstill at stoplights, I'd expect a clearly visible, dark blob of dots in this area, too.

[Aneta]

BTW, @Seba, I think it would be more intuitive if these plots were speed as X-axis, current or power as Y. But you probably have a good reason to reverse them?

[Mono]

8 minutes ago, Aneta said:

Yes, I saw them, but nothing in the <5kph, >20A area. And way too few points. With people riding in cities, with constant accelerations from standstill at stoplights, I'd expect a clearly visible, dark blob of dots in this area, too.

You can have high current at low speed only for a very short time  because the wheel accelerates away in this case. I don't find it very surprising that only few riders apply maximal force immediately from standstill. I, for example, almost always do a roll-in-mount, that is, I have super low torque requirements at <=3km/h. The only exception would be idling, but even there I rarely demand even remotely close to maximal torque.

Edited December 5, 2019 by Mono

[Seba]

9 hours ago, Chriull said:

Afair he saves/transmits "only" 1 sample per second vs 5 samples per second per "original" wheellog.

Yup, that's the reason. In fact, update interval is variable from about 2 seconds to even several seconds. But what is important to point is that it's not sampling rate but transmission rate. EUC is sending data frames continuously aboyt 5-6 times a second. But sampling rate is much higher. This already causes some aliasing errors and artifact so there is a need to apply low-pass averaging filtering. This is why extremum values are removed from the graphs just at the application level.

As for high current values at startup, they are very short. It usually takes less than second to accelerate to at least few km/h and for current to drop. As I used 15% opacity in the current/power plots to better see data distribution, some extremum points may be hard to see.

When I will gather more data (especially from CSV logs that are "raw"), I will be able to create more detailed plots and data analysis that won't be masked by antialiasing.

[Blueblade]

16 hours ago, Aneta said:

Strange that we don't see any envelope at slow speeds and high currents, too few points? Is it because how power delivery of KS wheels work? I can get a high-current beep on my GT16 (more than 40A) almost any time I start from standstill. Here, we see about 1A from the start - how is this possible?

Makes me question KS's telemetry once again. We need to come up with some method of validating the telemetry. 

Yes it is interesting how the highest currents occur in the 10-20kmh band, centered around 15kmh.

Also interesting is the half circle shaped cluster in the negative current side, you can see a sort of sweet spot around 20kmh/-5A, probably relevant for that other regenerative braking topic, i would guess riding that model wheel downhill @20kmh or so would maximize regenerative energy, although that is strictly looking at current, not factoring descent time, which will be longer at a slower speed.

[Lefteris]

On 12/4/2019 at 1:46 PM, Seba said:

[...] Unfortunately you won't find any Gotway wheels here. This is due to the fact that these wheels do not send any data to the application allowing for their unambiguous identification[...] 

Why not add an option for gotway users to name their wheels? 

Also can I have a plot for Z10 riders and also be able to compare my own rides so can too check If I'm putting my wheel enough or not enough pressure. 

[Seba]

12 hours ago, Lefteris said:

Why not add an option for gotway users to name their wheels? 

That's the plan for nearest future, but for now I can only count on EUC identification data.