Energy consumption and regen on mountain rides?

[Aneta]

On 11/16/2019 at 12:17 AM, Seba said:

This means that during hard braking regenerative current may go quite high - during rather normal braking I constantly gets regenerative currents of 10 A and more.

So, this confirms then that even during hard braking, there's no energy drawn from the battery - i.e. it's always "normal regen", using Eddy currents.

[Mono]

51 minutes ago, Aneta said:

So, this confirms then that even during hard braking, there's no energy drawn from the battery - i.e. it's always "normal regen", using Eddy currents.

How so? It could also mean that the switch away from regen braking is taking only at relatively high voltages and/or currents.

[Chriull]

On 11/16/2019 at 9:17 AM, Seba said:

This means that during hard braking regenerative current may go quite high - during rather normal braking I constantly gets regenerative currents of 10 A and more.

 

58 minutes ago, Aneta said:

So, this confirms then that even during hard braking, there's no energy drawn from the battery - i.e. it's always "normal regen", using Eddy currents.

Any "positive" log/observation can just harden a hypothesis a bit more - but not confirm it. They are just no disproofs.

I searched for one of my quite old logs of a KS16C/?S? ride:

Although the KS16C/?S? in these times did not report negative(regenerative) currents, one sees here nicely braking maneuvers which draw energy from the battery - voltage drop while speed drops.

So at least this wheel had non regenerative breaking implemented.

 

... and here a second one with almost 40A regenerative breaking with the same wheel/same ride. Which is an quite crazy peak charging current for a 4p battery...

 

[Aneta]

17 minutes ago, Chriull said:

 

Any "positive" log/observation can just harden a hypothesis a bit more - but not confirm it. They are just no disproofs.

I searched for one of my quite old logs of a KS16C/?S? ride:

Although the KS16C/?S? in these times did not report negative(regenerative) currents, one sees here nicely braking maneuvers which draw energy from the battery - voltage drop while speed drops.

So at least this wheel had non regenerative breaking implemented.

 

... and here a second one with almost 40A regenerative breaking with the same wheel/same ride. Which is an quite crazy peak charging current for a 4p battery...

 

It's a strange graph - voltage is going up when accelerating and down when braking.

[Chriull]

5 minutes ago, Aneta said:

It's a strange graph - voltage is going up when accelerating and down when braking.

It's a quite normal graph for EUCs - the sampling rate is not really satisfying. Although the KS with ~5 samples a second are already the "better" wheels. But still the "spikes" from self balancing are just statistical samples of what's happening and voltage/current values not necessarily samples at exactly the same point in time...

Also Li Ion cells approximation equivalent circuit diagram with a fixed voltage source and just an ohmic internal resistance is not describing reality - they show "delayed" voltage sags under burdens which are "equalized" again delayed once the burden decreases...

Such voltage sag/rise delays in regard to burden(current) could also occur/be amplied from the capacitors at the motor driver since it is told that KS reports battery current and not phase current?

I can see just some voltage sag reliefs once acceleration gets less but no "disturbing" voltage rise while acceleration. The trend with voltage going down steadily while the two longer braking maneuvers in the first graph is quite clear.

[Marty Backe]

Here's an interesting observation from a ride last week, and it changes how I thought regenerative charging worked.

With a fully charged wheel, I had to travel a distance down a steep street. When I rode the wheel slowly (fast walking pace) the wheel had no issues. But when I rode the wheel fast (low teens) down the hill I was immediately kicked off by the overcharge warnings. This is opposite of what I previously assumed was the case; if you want more recharge while going down hill, go slower and brake more.

So it turns out that the faster you travel downhill, the more regenerative charging occurs.

Edited November 21, 2019 by Marty Backe

[Planemo]

I actually thought that it worked the way that you didn't

I had always assumed that slow braking down a hill could be taken care of by 'shorting' (if that's the right term) the mosfets via duty cycle to give the required resistance, whilst more aggressive braking would require the regen to come into play.

It's the reason why, on a full battery, I have always gone down hills at walking speed on my Z10. I haven't been kicked off yet so assumed I was doing something right!

[Aneta]

13 minutes ago, Marty Backe said:

Here's an interesting observation from a ride last week, and it changes how I thought regenerative charging worked.

With a fully charged wheel, I had to travel a distance down a steep street. When I rode the wheel slowly (fast walking pace) the wheel had no issues. But when I rode the wheel fast (low teens) down the hill I was immediately kicked off by the overcharge warnings. This is opposite of what I previously assumed was the case; if you want more recharge while going down hill, go slower and brake more.

So it turns out that the faster you travel downhill, the more regenerative charging occurs.

Yes, I've had exact same situation and same observations. I was able to very slowly descend the hill on fully charged battery without exceeding 4.2V/cell. My theory is that since at low speed back EMF is low (Vemf = Vbatt*(Speed/NoLoadSpeed)), the generated phase current I=Vemf/Rwindings is just enough to maintain balance/braking, none of this current can be "diverted" into battery. It's akin to disconnecting phase wires from the controller and shorting them all together.

Edited November 21, 2019 by Aneta

[Chriull]

9 minutes ago, Marty Backe said:

Here's an interesting observation from a ride last week, and it changes how I thought regenerative charging worked.

With a fully charged wheel, I had to travel a distance down a steep street. When I rode the wheel slowly (fast walking pace) the wheel had no issues. But when I rode the wheel fast (low teens) down the hill I was immediately kicked off by the overcharge warnings. This is opposite of what I previously assumed was the case; if you want more recharge while going down hill, go slower and brake more.

So it turns out that the faster you travel downhill, the more regenerative charging occurs.

The motor driver acting as  boost converter is least efficient with transforming (very) low motor voltages up to the bit more as battery voltage to recharge the battery. Could even be that for too low voltages (speeds) this "power" (non-regenerative) braking is applied.

For too high speeds regeneration dimishes again since air drag helps braking more and more...

 

[Mono]

12 hours ago, Marty Backe said:

So it turns out that the faster you travel downhill, the more regenerative charging occurs.

What you suggest seems to be that increasing speed increases the charge current. That is not at all surprising. It doesn't tell us though which speed maximizes the overall battery charge when you reach the foot of the mountain.

Here is a funny thing: what if you brake further and approach (and reach) zero speed? In this case, obviously, the current must have been reversed somewhere during the process of slowing down, because we know that to keep the wheel at standstill at a slope consumes energy (like the additional power needed to go the hill up backwards, namely, a force of sine(slope angle) x weight must be created). That's an(other) example of non-regenerative braking.

Edited November 22, 2019 by Mono

[Mono]

3 hours ago, Aneta said:

It's a strange graph - voltage is going up when accelerating and down when braking.

It seems to me that the voltage going up under acceleration can be explained as recovery from a previous voltage sag.

[Struck]

22 hours ago, Seba said:

For example KS-14 and KS-16 report only absolute current values

This has never been the case in my wheel (KS-14S), and I have had several firmwares
Both the original wheellog and your EUC.World show negative values for regenerative braking

Best,

[Struck]

On 11/14/2019 at 9:03 PM, Planemo said:

does the motor actually use power to brake?

This question drives me mad. And I couldn't see it answered. Meaning, you are going fast, you want to brake, you use the motor to "eat" the energy. There must be a limit at which point the wheel should decide to put energy in the motor to brake, to accelerate in the "other direction".

Braking is negative acceleration, but continuous negative acceleration at one point means shift speed sign. When does it happen? Can the braking be so hard that the wheel uses power to brake?
This looks like a sharp shift that I can't grasp

[Planemo]

There may even be three states:

1. Light braking - motherboard 'shorts' the fets to provide braking.
2. Medium braking - motherboard uses regen to disperse energy to prevent the fets from melting.
3. Hard braking - wheel actually forces negative power into wheel (as if trying to run it in reverse).

If there are indeed multiple states, the motherboard must be able to switch between them extremely smoothly as I will be buggered if I can notice.

Like you, unless I have missed something, I have yet to see the answers. I guess only the manufacturers know the real situation.

I find it all very clever I have to say. It does remind me just how complicated our wheels are, despite how simple they appear.

[Mono]

28 minutes ago, Planemo said:

If there are indeed multiple states, the motherboard must be able to switch between them extremely smoothly as I will be buggered if I can notice.

True, on the positive side, electrons don't have any (relevant) mass which would induce a delay or momentum under switching direction. We have one record, IIRC of a Ninebot One, which suggest that the switching did not go smoothly enough and lead to a fall. I also believe I could feel the switch on my old Gotway MCM2s. If the battery is large enough, switching at high speeds should not be necessary anymore: more than 1kW of braking power, which can keep 100kg at an angle of 45º under braking, should be rarely needed.

Edited November 22, 2019 by Mono

[Planemo]

21 minutes ago, Mono said:

True, on the positive side, electrons don't have any (relevant) mass which would induce a delay or momentum under switching direction.

I agree, although it's not so much the electrons I was thinking of, more the actual decision making speed of the motherboard in deciding which way to share out all this braking activity, not to mention the unbelievable speed control of the motor in order to maintain perfect balance whilst doing so, with a rock-solid pedal under all conditions. I find it truly staggering tbh!

[Seba]

1 hour ago, Struck said:

This has never been the case in my wheel (KS-14S), and I have had several firmwares
Both the original wheellog and your EUC.World show negative values for regenerative braking

Best,

Good to know! Maybe this is related to firmware version. When I was testing some older KS-16 it didn't reported negative values.

[Chriull]

1 hour ago, Planemo said:

There may even be three states:

1. Light braking - motherboard 'shorts' the fets to provide braking.
2. Medium braking - motherboard uses regen to disperse energy to prevent the fets from melting.
3. Hard braking - wheel actually forces negative power into wheel (as if trying to run it in reverse).

Number 3 is the same state like accelerating. The rotating magnetic field generazed in the stator coils is just made slower than the rim with the permanent magnets -> braking. If the controller makes it faster again the wheel accelerates. That's the selfbalancing happening all the time to keep the rider upright.

As with accelerating the magnitude of deceleration is controlled by the pwm duty cycle. (and/or the "difference" angle of the generated magnetic field?)

So that should be just one algorithm.

But with both non regenerative breaking methods EUCs would not be possible - they burn too much power. With braking by "shorting" one could use some power resistor (like the heating element of a cooking plate) to dispose the braking energy as heat.

So regenerative breaking with the battery taking the energy is the only solution - although the li ion cells get often tortured by real fast charging.

Number three (or just the "normal" control algorithm) is needed too, because it's the only technique that can generate a significant braking force near standstill.

1 hour ago, Planemo said:

I agree, although it's not so much the electrons I was thinking of, more the actual decision making speed of the motherboard in deciding which way to share out all this braking activity, not to mention the unbelievable speed control of the motor in order to maintain perfect balance whilst doing so, with a rock-solid pedal under all conditions. I find it truly staggering tbh!

Once was mentioned that the control loop runs about 100 times a second. That's no real challenge for available microcontrollers. There is no top state of the art controller needed for this...

[Planemo]

5 minutes ago, Chriull said:

But with both non regenerative breaking methods EUCs would not be possible - they burn too much power. With braking by "shorting" one could use some power resistor (like the heating element of a cooking plate) to dispose the braking energy as heat.

So regenerative breaking with the battery taking the energy is the only solution - although the li ion cells get often tortured by real fast charging.

Number three (or just the "normal" control algorithm) is needed too, because it's the only technique that can generate a significant braking force near standstill.

Once was mentioned that the control loop runs about 100 times a second. That's no real challenge for available microcontrollers. There is no top state of the art controller needed for this...

So you are saying that irrespective of braking force required, regen is always being provided to the batteries and there is no other method of energy dissipation used? I can see it would make things far easier, I am just surprised that the battery can accept the energy generated by slowing say 145kg of human and wheel from 40+mph in the stopping distances we are talking about. That said my physics is too rusty to do the math lol.

Oh I am sure that the control loop speed is nothing like some machines use (fighter jets? spacecraft?) but I am just still amazed at how smooooth our wheels can control balance and how quickly (un-noticeably) they can react to terrain/user differences.

[Mono]

3 hours ago, Planemo said:

I agree, although it's not so much the electrons I was thinking of, more the actual decision making speed of the motherboard in deciding which way to share out all this braking activity, not to mention the unbelievable speed control of the motor in order to maintain perfect balance whilst doing so, with a rock-solid pedal under all conditions. I find it truly staggering tbh!

It is staggering, I agree, though also the motherboard decisions are just a few electrons moving around, I think. After all, the speed of electron(ic)s is just so much quicker than gravity and momentum and that helps a lot

1 hour ago, Planemo said:

So you are saying that irrespective of braking force required, regen is always being provided to the batteries and there is no other method of energy dissipation used?

I think one other method used is heating up the motor. The need of energy dissipation is not unique to braking. To my understanding there must be a lot of energy dissipation going on when, for example, climbing a mountain slowly, that is, if a large current is flowing while the wheel is only slowly or not at all moving the energy needs to go somewhere else as well.

Edited November 22, 2019 by Mono

[Aneta]

3 hours ago, Mono said:

If the battery is large enough, switching at high speeds should not be necessary anymore: more than 1kW of braking power, which can keep 100kg at an angle of 45º under braking, should be rarely needed.

At deceleration of 1g (45 degree backward lean) the inertial force will be 100kg (980N), which at 36km/h (10m/s) means a mind boggling 10kW of braking power! (power = force * speed) Not 1kW! TEN KILOWATTS! Where can all this power go? Small amount (something like 300W) is taken by air drag. If the battery is 1500Wh, it can take 1.5kW at 1C. Suppose it takes 2C, which is not good for battery health (fast charging fully in just half hour!) We still have 9800W - 300W - 2*1500W = 6.5kW of power that needs to be released somewhere! The only remaining places where it can happen are motor windings and MOSFETs.

Thoughts?

P.S. Since deceleration 1g is greater than what tires can provide (max deceleration due to friction is k*g, where k is the coeff. of friction, typically about 0.7 for rubber on asphalt), let's assume 0.5g. Then we have much less dramatic numbers, but still we have 4900W - 300W - 2*1500W = 1.6kW of power for 2C regen current or 3.1kW for 1C that needs to be released somewhere.

Edited November 22, 2019 by Aneta

[Mono]

8 minutes ago, Aneta said:

At deceleration of 1g (45 degree backward lean) the inertial force will be 100kg (980N), which at 36km/h (10m/s) means a mind boggling 10kW of braking power! (power = force * speed) Not 1kW! TEN KILOWATTS! Where can all this power can go? Small amount (something like 300W) is taken by air drag. If the battery is 1500Wh, it can take 1.5kW at 1C. Suppose it takes 2C, which is not good for battery health (fast charging fully in just half hour!) We still have 9800W - 300 - 2*1500W = 6.5kW of power that needs to be released somewhere! The only remaining places where it can happen are motor windings and MOSFETs.

Thoughts?

Good point, I conveniently ignored the speed  so removing the non-regen braking switch from the code is not an option. I don't see why it shouldn't be the motor windings heating up. I am too lazy/busy  to re-compute this now, but last time I did the order of magnitudes seemed to add up.

Edited November 22, 2019 by Mono

[Aneta]

Re: the thought that during hard braking the battery must be a source of power, not a drainage (aka regen) - if we take the above example with 100kg braking at 0.5g at momentary speed of 36km/h, then about 4600 watts need to be dissipated in the EUC. If battery works in regen mode and consumes 1500W, then 3100W must be dissipated in the motor windings and MOSFETs; if the battery works in "drive" mode and supplies 1500W, then 6100W must be dissipated in the motor windings and MOSFETs.

Thoughts?

Edited November 22, 2019 by Aneta

[Mono]

If it's the motor windings, the power might not be an issue at all, but we should look at the energy, i.e. power times time, that effectively needs to be put away.

[Aneta]

5 minutes ago, Mono said:

If it's the motor windings, the power might not be an issue at all, but we should look at the energy, i.e. power times time, that effectively needs to be put away.

For 100kg with 10m/s initial speed, that's a total of KE = m*v^2/2 = 5000J, or less than 1.5Wh. With specific heat of copper roughly 400J/kg/C, if there are a couple of kilos of Cu in the motor, that's less than 10C temperature increase. You're absolutely right! Motor should be able to swallow this braking energy despite the huge peak power. As long as the MOSFETs don't burn from high peak currents...