Regeneration on Full Battery, bad news :(

[Cloud]

I might be saying something stupid, but Instead of a capacitor, Wouldnt it be easier to just prevent the battery from being charged when going downhill on a full battery? Granted the energy would be wasted but wouldnt this be the simplest solution to protect the battery? Some kind of logic to interrupt the charging circuit? I am not sure how its actually done, but  i just have a feeling it's easy to do:) after all, when charging the wheel at home the charger stops charging after it's full. Can the same mechanism be used to cut off the charging when going downhill? 

[Gimlet]

How many amps of excess are we talking here? I mean could one or more 100w car headlight bulbs burn it off?

[Jason McNeil]

Esaj measured 3KW peak braking regen, enough to run a small home for a fraction of a second!

[Jag_Rip]

When on the charger, as far as I understood, the BMS signals it cant take more so the charger stops sending energy to the Battery. While running downhill and regenerative braking, there is no way to stop the influx of excess energy from the wheel, unless you switch over to powerbraking or stop going downhill.

It also seems to me that the cokebottle of a capacitor directly intervenes with the preference of a light wheel, judging by the picture thats more than a Kilo of additional weight on a EUC. So its either a +15kg wheel everyday or you cant start downhill.
Cant grasp the amount of energy but I guess its more than a 100w bulb. Someone in another thread wrote, 800m downhill recharged the bot by 5% (thats not a quote and unconfirmed) So I my wild guess its a lot more than a 100w bulb.

 

[esaj]

Wouldnt it be easier to just prevent the battery from being charged when going downhill on a full battery? Granted the energy would be wasted but wouldnt this be the simplest solution to protect the battery? I am not sure how its actually done, but  i just have a feeling it's easy to do:) after all, when charging the wheel at home the charger stops charging after it's full. Can the same mechanism be used to cut off the charging when going downhill?

Not that easy as it sounds, the energy cannot just disappear, but has to go somewhere. It's a bit different with the charger, as the wheel doesn't have to balance you while charging, and the motor acts as a generator while braking. See for example here for more discussion about regulating voltage/burning the excess energy off:

http://forum.electricunicycle.org/topic/605-firewheel-custom-battery-pack/?do=findComment&comment=11438

 

How many amps of excess are we talking here? I mean could one or more 100w car headlight bulbs burn it off?

If the Gotways data can be trusted, we're looking (momentarily) above 50A:, and the voltage can be something like 68V (or even above) on more full batteries, so to be safe around 3.5kW or more? 35 100w bulbs...   In this graph, the batteries were more empty, so "only" 60V / around 3.3kW, but that's a very momentary spike.

 

 

[Cloud]

Not that easy as it sounds, the energy cannot just disappear, but has to go somewhere. It's a bit different with the charger, as the wheel doesn't have to balance you while charging, and the motor acts as a generator while braking. See for example here for more discussion about regulating voltage/burning the excess energy off:

http://forum.electricunicycle.org/topic/605-firewheel-custom-battery-pack/?do=findComment&comment=11438

 

 

Ok, i'll probably make a fool of myself again, but why not install a smaller additional built in rechargeable battery then that will only be charged during regenerative braking and not during normal charging with a charger? It will be empty when you start riding on a full charge but will only be charged when you need to spill over the extra?  You can then reuse this energy when you are running low...forgive my ignorance in electrical, i really dont know what im talking about

[esaj]

Ok, i'll probably make a fool of myself again, but why not install a smaller additional built in rechargeable battery then that will only be charged during regenerative braking and not during normal charging with a charger? It will be empty when you start riding on a full charge but will only be charged when you need to spill over the extra?  You can then reuse this energy when you are running low...forgive my ignorance in electrical, i really dont know what im talking about

That could work, of course the "extra" battery must be kept between safe limits (above 2.5V and below 4.2V per cell), so during long descent, it too could become overcharged (or if it's not charged for a longer while, it could go below 2.5V, after which it apparently becomes dangerous, as small copper "shunts" start to form inside the battery, meaning it short circuits, starts to lose charge even faster and could blow when charged again). Probably could be done with "smart" battery management to keep it around the nominal 3.6-3.7V per cell? "Large enough" pack can suck in a lot of charge if "half full", don't know if the high current can become a problem, but there are single-pack wheels that use regenerative braking. Do you think the generic that burned in Norway could have been due to overcharge during descent? It was using some cheap chinese cells...

I'm sure there is a solution to this, even just burning off the excess energy should be possible via large enough heatsinked power-resistor(s), as long as it doesn't overheat itself or any other components, or melt the wires, the wires are thick enough for the high currents and the heatsink is placed so that it can't burn the rider   My posts in the Firewheel custom battery -topic talk about modifying existing mainboard or battery wiring for something similar, which is hard and beyond my grasp, but probably could be made to work if the board & firmware is done with that in mind in the first place.

Electric cars got off easy, I've understood that the brake controller just decides whether to use regenerative braking, mechanic brakes or a combination of both to prevent overcharging the batteries. Unfortunately, mechanic brakes aren't exactly suitable for self-balancing devices...

[Jag_Rip]

I think you'd need to program something to control the choosing of the excess-energy battery. It should empty itself first but charge last (or only with regen braking). But its additional resources during development to circumvent a quite unlikely event. Thats a "nice to have" feature that gets thrown overboard at the first sight of hitting any limit during development (cost and/or time).

[Gimlet]

Not that easy as it sounds, the energy cannot just disappear, but has to go somewhere. It's a bit different with the charger, as the wheel doesn't have to balance you while charging, and the motor acts as a generator while braking. See for example here for more discussion about regulating voltage/burning the excess energy off:

http://forum.electricunicycle.org/topic/605-firewheel-custom-battery-pack/?do=findComment&comment=11438

 

If the Gotways data can be trusted, we're looking (momentarily) above 50A:, and the voltage can be something like 68V (or even above) on more full batteries, so to be safe around 3.5kW or more? 35 100w bulbs...   In this graph, the batteries were more empty, so "only" 60V / around 3.3kW, but that's a very momentary spike.

 

 

That's a lot higher than I was expecting. Frightening to think that we are charging the batteries at 50 amps when even the high end chargers only put in 5 amps. Just wondering why our batteries aren't going up in smoke at the moment?

Are you sure the decimal point is in the right place?

[esaj]

That's a lot higher than I was expecting. Frightening to think that we are charging the batteries at 50 amps when even the high end chargers only put in 5 amps. Just wondering why our batteries aren't going up in smoke at the moment?

Are you sure the decimal point is in the right place?

If the decimal point was off by one, then I could climb a sand/gravel hill starting from 10 degrees and going up to 25 degree at steepest with around 260W peak power (around 150W average) and 11.5km/h average speed, which sounds quite low..? Or decimal is correct, but GW gives way too high readings? Or maybe they're both off then?  Off by one decimal and Gotway gives then too low values (I doubt the current measuring is very precisely calibrated)? Don't know, have to check it again. I don't have anything to compare it against really, as I can't get an amp meter there while riding   Btw, the ampere-display in my Firewheel shows values over 30-40A during fast climbing (not that steep hills) and fast acceleration, but I don't know how much off that meter is either, as I have nothing to calibrate it (I do have a multimeter, but I don't have anything I could use as load to calibrate the high-end readings toward 50A).

Moreover, we're talking a short term peak pulse, both brakings were around 1.5 second long, first one averages around 1.3kW regenerative (26.85A) and the latter around 1.7kW (28.32A) over the peak... Don't know how well cells can handle short-term charging pulses (well, at least they didn't explode or catch fire, if the values even are correct ), but they definitely don't like high charge current for longer term.

[Jason McNeil]

Esaj's data is pretty consistent with some of the readings I've taken with the in-line power meter.

[MetricUSA]

I thought the eus, had over voltage protections in it, at least the packs themselves have it in the circuits...

[robca]

I thought the eus, had over voltage protections in it, at least the packs themselves have it in the circuits...

But all of that energy has to go somewhere, can't just disappear. So the only choice manufacturers have, is to dump all the energy into the battery pack and hope for the best. The alternative would be to have a circuit that can somehow use 3kW, and that would take a big resistor or other mechanical/physical contraptions

[cg]

The image is from here, read for details:  http://batteryuniversity.com/learn/article/how_to_prolong_lithium_based_batteries

Edit: Further it refers to this:  http://www.sciencedirect.com/science/article/pii/S0378775302003051  , don't know if it's available for free somewhere.

 

thanks for that link esaj.

 

I thought the eus, had over voltage protections in it, at least the packs themselves have it in the circuits...

yes, thats the thing that will faceplant you

[Jeffrey Scott Will]

Thanks for the info guys.  I go down a 400ft hill at the start of my commute every day, so this is a concern for me.  The 67v safe limit before degradation is helpful. Since I can't get esaj's data logger to record on my KS, I'll use their app to watch the voltage as I head down the hill to see if I exceed 67v. 

[Chriull]

...

Moreover, we're talking a short term peak pulse, both brakings were around 1.5 second long, first one averages around 1.3kW regenerative (26.85A) and the latter around 1.7kW (28.32A) over the peak...

Should be in the right scale. From one article (*) about regenerative braking the Power for braking was calculated as follows. For 20 km/h=5,5m/s and 90 kg (moving weight) give an kinetic energy of 1/2*5,5²*90=1390J. "Using" this energy within 1,5 seconds needs an average power of 1390/1,5=925W.

Don't know how well cells can handle short-term charging pulses (well, at least they didn't explode or catch fire, if the values even are correct ), but they definitely don't like high charge current for longer term.

A couple of posts capacitors where mentioned as buffer before the batteries to take the peak currents. But imho the capacity of the supercapacitors is not really usable for wheels: In the same article he used supercapacitors - 8 banks of 58F - each costs about €100 and is 4cm x 8cmx 20 cm big and has 0,6 kg  .... But so he could take more than enough current for breaking. Imho he finished that not all 8 banks where really necessary (he tested with an electric Trike with about 60kg - so about comparable to a wheel)

And also a problem was that "pushing" the charge from on brake from the supercapacitor to the battery with 10A took about 4-5 minutes - so one is not allowed to break too often

So i assume for nowadays its better to divide the current on as many as possible accu packs.

*)  http://docs.lib.purdue.edu/cgi/viewcontent.cgi?article=1005&context=techmasters

[esaj]

While the spikes during more extreme situations can be quite high (and pretty much all the graphs I've so far shown have been such situations), in reality the power usage by average is much, much lower, at least during more or less normal riding. I tried today to ride vee's MCM2s with the app as similarly as I do with the Firewheel (but of course with Firewheel my average and top speeds would be much higher ), to see what kind of powers were talking about.

This is from one of the recordings I did today, a 5.7km long run (Gotway trip-meter) mostly off-road (there's a point between about 12 and 17 minutes where I first ride down a paved hill, then back up it, then a little while on straight and then back to the hiking paths, you can distinguish it pretty easily):  Click for larger version

Again there are some pretty big spikes, as there are lots of up- and downhills in the hiking paths, ranging between 10 degrees to around 25 degrees. Average speed of 16,97km/h (Gotway app-speed, so about 10% lower in reality), maximum speed 24.98km/h. The minimum (highest regeneration) power was 2.54kW, and maximum power (climbing, using power) was 2.65kW. But average power across entire route was 299.89W. So while I don't know how bad the spiky charging and discharging is for the cells, the average power usage is easily within limits. For the record, my Firewheel original batteries still charge to 268.x Wh, so above the nominal 264Wh, and they've got something like 400 cycles of hard riding, plus I've always apparently overcharged it going down that very first hill on all of my trips.

Here's another one, this one's over paved route only (well, there is maybe a 100 meter path I took to cut a little towards the end, but other than that), 8.2km long.

Around the mid-part I climbed one of the steepest paved hills around here (don't know the angle, maybe closer to 20 degrees?), and didn't even drop much speed. Average speed 19.43km/h, max 25.1km/h. Maximum regenerative power was 1.58kW, maximum output 1.69kW. Average power was 279.11W.

Btw, considering I usually rode the battery of the Firewheel to pretty much empty in about an hour, the average power must have been around 270W on that too, so in that sense the power usage seems realistical.

EDIT:

Here are the csv-files, if anyone wants to do... something with them:

 metsa_20150915_204344.txt

 asfaltti_20150915_210726.txt

 

[Mono]

I'm guessing this test was done on a near-perfectly smooth, level surface...

Though if overall energy use from balancing is pretty low in the general case that's quite interesting.  Riding a non-electric unicycle is much more tiring than riding a bicycle at equivalent speed over the same distance from what I hear.

This can be mainly explained by the inefficient riding position and the rather suboptimal single speed of a unicycle. Also, steering a unicycle involves more muscles than steering a bicycle, even though this effect should be small for a trained rider. 

There would be another simple "test" for EUCs: if balancing would consume significant amounts of energy, going at lower speed would reduced the maximum range, because lower speed requires a longer balancing time for the same distance to cover. I cannot remember that anyone reported that reducing speed reduces the range, it seems that rather the opposite is the case, at least for usually driven speeds. Hence, balancing seems not to be one of the more relevant energy sinks. EDIT: the "test" is apparently not that simple. 

[Chriull]

Here it seems, that also when riding with quite low speeds you still have back currents into the battery (charging)?

Cause today i noted the battery capacities for my 2,6 km track with about 100 m height difference. Going up i started with 100% and arrived with 82%. The battery quite quickly recovered to 87% and i shortly put it on the charger, but then decided not to charge it more  So the app showed 90% when i started my decline - after 10 m going straight i checked again and had 88%. When i went down the track the capacity was at just 90%...

May first Idea would have been, that i went to slow after all this discussions and there was not enough back voltage generated to charge the battery. But in your graph you still have charging cycles while you are in the 10 kph range..

Or the Ninebot One E+ does almost no regenerative breaking at all? I hope the bluetooth protocoll of the ninebot will be reverse engineered too soon

At least i know now, that i can start with full batteries - at least i need it for my track. 500m from home the 10kph tilt back warning began... ;(

[esaj]

Here it seems, that also when riding with quite low speeds you still have back currents into the battery (charging)?

Two things are causing it: I'm going down a slightly declining street, and on those regenerative charges, I'm braking a little bit (if you look at the speed curve, it's going down slightly).

Cause today i noted the battery capacities for my 2,6 km track with about 100 m height difference. Going up i started with 100% and arrived with 82%. The battery quite quickly recovered to 87% and i shortly put it on the charger, but then decided not to charge it more  So the app showed 90% when i started my decline - after 10 m going straight i checked again and had 88%. When i went down the track the capacity was at just 90%...

May first Idea would have been, that i went to slow after all this discussions and there was not enough back voltage generated to charge the battery. But in your graph you still have charging cycles while you are in the 10 kph range..

The battery was also nowhere near full when I recorded that, the voltage is around 60V (67.2V on full), that's four (174Wh?) packs on the 680Wh MCM2s. With the Firewheel, and with current 3 192Wh packs (576Wh), starting from around 65-66V (I had to let the batteries run a bit emptier for it to be safe to brake ), it took about 30km before the voltage was around 60V. It certainly starts to drop faster the closer to empty you are.

Or the Ninebot One E+ does almost no regenerative breaking at all? I hope the bluetooth protocoll of the ninebot will be reverse engineered too soon

At least i know now, that i can start with full batteries - at least i need it for my track. 500m from home the 10kph tilt back warning began... ;(

I'll try to get around to reverse-engineering the Ninebot protocol too, but can't make no promises when that's going to happen. I'll need data captures from NB1 owners, and the app is obfuscated, so it's going to take a lot more work to reverse-engineer it than with Gotway or IPS Xima. Also if it's Bluetooth LE, building the communication is going to take a lot longer, as I have nothing to test BT LE with (and even if I had, it's going to work entirely differently than with the classic BT, it's lot more complicated to emulate the real device), so I'll just have to rely on others for testing. "Does it work now?", "No". Couple of days later: "How about now?" "No". Week later: "Could you run this test software and send me back the results...".  

[MrBump]

There would be another simple "test" for EUCs: if balancing would consume significant amounts of energy, going at lower speed would reduced the maximum range, because lower speed requires a longer balancing time for the same distance to cover.

Nope.

[Mono]

Nope.

Why not? 

[MrBump]

Why not? 

Going faster requires a great deal more energy in itself (more than what is in linear proportion with speed).  I also see no reason to believe that balancing consumes more energy at lower speeds - balancing is harder for the user, but that is due to lateral instability at lower speeds - balancing along the axis of travel is a different matter. 

For example, I saw on another thread that energy used at speed increased by two thirds when tiltback kicked in - not that that is the same as simple balancing, but it's a huge energy output that is not going into forward motion.

That said, I don't know how much energy goes into balancing as opposed to movement generally - it seems from this thread like it comes in lots of intense spikes, but with not very much area under the power curve.

It feels a bit counterintuitive to me that energy use due to balancing is so low, but I've not thought about it in great depth tbf.

[cg]

Here are the csv-files, if anyone wants to do... something with them:

 metsa_20150915_204344.txt

 asfaltti_20150915_210726.txt

 

I guess someone have to poke around those files

I have plotted the power as a function off the acceleration, I guess the acceleration unit is change in kph pr second.

I had to smooth the parameters to get some trends:

this is the plot where you where driving on asphalt, seems like you used the most power when going uphill, and lost a bit speed. And also, seems like you gain bigger chunk of energy when decelerating, than accelerating.

and this is the offroad one, it looks a bit like a fish... But I think it is interesting that the power spikes in both figures seems to be when the acceleration is not so rough...

here's the script for plotting: (using octave)

------

X=real(dlmread("metsa_20150915_204344.txt" , ",", "emptyvalue", 0));

# Acceleration defined as speed_sample(n)-speed_sample(n-1)]
S=X(:,2);
ACC = [0; S(2:end)-S(1:end-1)];

# smooth acceleration and power
s_size = 11;
S_ACC = conv(ACC, ones(s_size,1)./s_size, "same");
S_PWR = conv(X(:,3), ones(s_size,1)./s_size, "same");
plot(S_ACC,S_PWR,'.');

------

Notice, I use real(...) on the output from the loading function since it for some reason returns the numbers as complex! strange....

[esaj]

I have plotted the power as a function off the acceleration, I guess the acceleration unit is change in kph pr second.

this is the plot where you where driving on asphalt, seems like you used the most power when going uphill, and lost a bit speed. And also, seems like you gain bigger chunk of energy when decelerating, than accelerating.

and this is the offroad one, it looks a bit like a fish... But I think it is interesting that the power spikes in both figures seems to be when the acceleration is not so rough...

I'd guess that's because more power is required going uphill than on level ground, even if the speed stays the same or even lowers a bit. When I was riding & recording those, I actually thought that you could to some extent determine whether the wheel is going down or up a hill or straight based on speed & power change, like more power, speed doesn't change or drops -> going up hill, less power, speed doesn't change or raises -> going down hill. If both change, then it's hard to say