Analysis of regen on an ebike

[solarbbq2003]

bvillion wrote regard more regen when battery is discharged

2) At a given pulse width, a low voltage battery will take more current than a high voltage one, since the voltage differential will be higher between back EMF and battery.

I think that might be part of it, but battery resistances i'm quite sure decreases as battery is discharged, I guess is possible to put in ( measure) numbers and work out difference in regen current battery can accept as it becomes more discharged.
I guess also justins % graphs are based on current measurements via shunt? Would it be correct to assume the % graphs given will be slightly overestimating the actual % stored in the batts? if yes any ideas on size of adjustment to get actual %
Changing voltage and resistance of the batteries between fully charged to fully discharged are going to be infuencing the rate the current that can go into the batts. I've noticed using an analog meter while riding that as soon as regen is activated ( on a standard regen controller, not proportional type ) that there is a very quite current spike which can be quite high but very quickly drops down to a lower current which remains relatively constant as the bike comes to a stop.
example: 48v 15ahr cobalt pack using gldual motor in low speed winding typically gives a very brief peak of about 10amps ( for a fraction of a second), then drops down to about 2.5amps where it remains constant until bike speed gets quite slow.
This current spike is quite interesting, I'm wondering why it spikes then drops down to a quite low level of current.

 

[ZapPat]

Of course even better would be having relays that would dynamicaly switch batteries from series to parallel configurations, and even motor windings from delta to wye for a truly efficient system! This is in the dream real for now, but I just added it to my own controller features wish list!

ZapPat, i'll get the 3PDT relay soon!! the delta to wye config WILL be on my bike.. and also.. i already have the serie/parallel contactor on my bike for 100V 23Ah or 50V 46Ah... so that setup will be the best to try that!
I should receive a 5306/03 motor with two winding per phase for serie/parallel winding configuration.. all the possible situations on my setup for the begining of spring 2009!
Oh.. in the last minute i received a PM from John that have a direct contact with Crystalyte manufacture. See what he pm'd me :

I am talking to Crystalyte now. Do you want to try the Delta/Wye or the series/parr windings? either way, they are making the custom axles to get the 2.5mm Teflon wire out the hubs. Just need to know what way you want it. I think the price will be the same either way, around 400usd, but they need a 5 motor min order to make the custom axles so i will take two, we need to sell the other two extra if you want just one. thanks john

So there will be 2 availlable motor to complete the order.. any interested? I suggested to ask to build custom motor of Delta/Wye instead of serie/parallel cause the ratio difference is just 2 for s/p config or 1.73 for the Delta/Wye config.. thatDelta/Wye seems more interesting cause it only require one 3PDT relay and 6 phase wire out of the hub instead of two relay and 12 or 10 phases wires...
Justin, i wonder what would be the best to allow live change on the relay while riding without blowing the mosfet of the controller caused bt spikes when switching between Delta or Wye ?
Doc

Well, Doc, you sound like the perfect tester for a special edition controller! You were already on my list of potential beta testers, so I would just have to make a special one for you to test out on your electronic four speed bike setup. It will beat doing the "shifting" manually, and you won't have to worry about the relays switching at the wrong time. If you do decide to test your delta/wye switch manually though, just be sure your controller is outputing nothing at the time (no throttle).

This type of setup will get you more efficiency for sure, but I wonder how it compares with mecanical shifters (using both together would make a monster though!). Your motor poles won't be moving any faster, but you will have better matching between the battery voltage and the motor's BEMF. I think this will help regen efficiency quite a bit, since the current spikes in the motor will be reduced in amplitude, thus reducing R * I^2 losses both in the motor windings and the controller. It will also give you four possible full throttle speeds, which will also help because switching losses are almost eliminated at full throttle (if not in current limiting that is). Regen efficiency at low speeds will be much improved, which will also help reduce motor heating that happens when doing regen at very low duty cycles.

I can't join you guys for the special Xlite motor purchase, since this would break the bank right now. I have to put all my ressources into trying to go forward with my controller project, and I'm not sure if I'll be able to convince the local bureaucrats to give me a financial hand with this. They are generally in another world, and mostly have no realistic vision of the future. I guess I'll see, since I have a meeting on the 6th with a couple of them about this. I must say I'm positive about it, since one of the guys has even seen "who killed the electric car", which surprised me a lot!

 

[ZapPat]

At a given pulse width, a low voltage battery will take more current than a high voltage one, since the voltage differential will be higher between back EMF and battery.

I think that might be part of it, but battery resistances i'm quite sure decreases as battery is discharged, I guess is possible to put in ( measure) numbers and work out difference in regen current battery can accept as it becomes more discharged.
[...]
Changing voltage and resistance of the batteries between fully charged to fully discharged are going to be infuencing the rate the current that can go into the batts.

All these statements are true, but not complete, since for one battery behavior really varies a lot depending on what you are using (chemistry, quality, temp,...). Many other factors will also influence your maximal regen current, including controller resistance and motor resistance. I personnaly would not use just any old battery with a regen setup, but would choose a lithium with fairly low internal resistance (good C rate). These usually have a fairly good charge rate acceptance over most of the charge range (like 10-90% of the state of charge). Most importantly, make sure you choose your battery so your maximum regen current does not exceed the batterie's maximum charge current. Even my 20Ah Ping battery will accept 20 amps of regen current, which would probably be quite suitable for many people. Headways would be better yet of course.

I guess also justins % graphs are based on current measurements via shunt? Would it be correct to assume the % graphs given will be slightly overestimating the actual % stored in the batts? if yes any ideas on size of adjustment to get actual %

Lead and NiMH don't have very good charge efficiencies, so yes, there would be a fair difference in this case. However, if fairly good lithium cells are used then the real recaptured energy vs the displayed recaptured energy would be much closer together, since lithium has very high charge efficiency. You would also have to count in your batterie's internal resistance as a loss for this case, so better batteries (lower internal R) and lower charge currents would help here.

I've noticed using an analog meter while riding that as soon as regen is activated ( on a standard regen controller, not proportional type ) that there is a very quite current spike which can be quite high but very quickly drops down to a lower current which remains relatively constant as the bike comes to a stop.
example: 48v 15ahr cobalt pack using gldual motor in low speed winding typically gives a very brief peak of about 10amps ( for a fraction of a second), then drops down to about 2.5amps where it remains constant until bike speed gets quite slow.
This current spike is quite interesting, I'm wondering why it spikes then drops down to a quite low level of current.

Any links to technical details about the controller you are using? The split second current spike you describe sounds like an artifact of the way your controller does it's regen, and I don't think this is "standard" regen behavior.

 

[Doctorbass]

ZapPat I would be glad to test your controller.
Since i appear to become a new stuff tester and i like that. If my lab can help someone for getting data or stat on any ev stuff l'll be glad to help for that!


About the battery resistance and SOC, after many charge and discharge test i've done, i would say that the lowest resistance a lithium battery can have is in the 10-90% SOC range.

In other words, that happen when the curve is pretty flat and not when the slope is changing..


PM me ZapPat :wink:


Doc

 

[Doctorbass]

Hey, Justin! :wink:

I would thanks you for the gift i just received yesterday, that will help me alot with some experiments on regen when i'll work on my 4 electrical speed setup!

That was a very nice moment for my wife and me to receive you at our home in Quebec during your trip.

Your new posted data about regen are very instructive tpo me and i can match that to the paper at the Shwan restaurant your wrote about that for me when you was here! :wink:

thanks!

 

[JCG]

Justin, could you improve the regen capture by using an ultracapacitor as intermediary, with an ultracapacitor such as JCG uses here:
viewtopic.php?f=6&t=7511&start=0&st=0&sk=t&sd=a&hilit=JCG

So even if an ultracap would allow me to do a maximum regen current of say 20 amps. Yes, I would be able to do more regen and come to a faster stop, but no, I would not get any more energy back. In fact, I would recover quite a bit less energy than had I come to a more gradual stop at 6 amps regen using just a battery to absorb the energy. So the idea that an ultracap could help 'improve' the percentage of recaptured energy is quite false in this case.

With respect, I couldn't disagree more. When you plug that battery pack in at night (regardless of its type), you won't be using a constant current of 20 A to charge it, and there are lots of reasons why you wouldn't. I think that people are forgetting that the battery is an electrochemical device, the emphasis being on chemical. Don't forget that when you charge or discharge a galvanic cell, there are chemical processes being carried out in several places within the cell, each of them occuring at a finite rate and according to a specific set of electrochemical kinetic rules which are extremely complicated and difficult to describe in equation form without making several assumptions. I'm not going to preach equations here, but all of us knows that the actual energy put back into the cell is quite dependant on charging current (among other things) for a reason.

ZapPat brings up an important point that rings true.

battery behavior really varies a lot depending on what you are using (chemistry, quality, temp,...).

The chemical reactions at the electrode surfaces can’t happen instantly and there will be a chemical concentration gradient between the electrode surfaces and the bulk electrolyte established involving things like boundary diffusion layers and natural convection (even in “gel” electrolytes). The chemical conversion boils down to two main steps.

1) Charge transfer. This is the faster of the two processes, involving very short transport distances at the electrode/electrolyte surface and surface electron transfer.
2) Mass transfer. This is the diffusion of ionic and molecular species to or away from the electrode surface (either the supply of material for the electrode or the exhausting of reacted chemical species). This is a much slower process than bulk diffusion, and mass transport in an electrolyte is often more than one order of magnitude slower than charge transfer on the surface.

There are a whole host of effects to consider alongside the (dis)charging process depending on the cell chemistry. Intercallation in Lithium cells, electrode crystallization or filming, gas production, etc. Even the often quoted “charging efficiency” can be a tough number to rely on. Personally, I consider the concept of charging efficiency itself to be weak, I’ll refer you to the following discussion:

http://www.smartgauge.co.uk/chg_eff.html

There are efficiencies associated with cell charging under specific conditions (trickle, constant current, constant voltage, pulse, float, and burp charging, and so on), each with their assumed or calculated time constants. Typical charge transfer time constants might be less than a minute, and a mass transport time constant could be several hours for a cell with plenty of Ah. Certainly, this is the case with the large capacity batteries in ebikes discussed here.

All of these things lead to the real reason people talk about damaging batteries with regen. Using regen from your ebike is best described as “random charging.” Constantly varying potential and concentration fields within the battery, and variable electrode process environments. But it’s not all about damage. It's not even all about "internal resistance." Simply put, the electrode and electrolyte processes will be continually varying in efficiency.

solarbbq2003 hits the nail on the head here:

I guess also justins % graphs are based on current measurements via shunt? Would it be correct to assume the % graphs given will be slightly overestimating the actual % stored in the batts?

For the reasons I mentioned above, the Cycle Analyst will not by itself give an accurate accounting of the real, (re)usable energy put back into a galvanic battery by regenerative braking. It's not enough to measure the current and multiply by the individual slices of time. And forget about motor winding losses for the moment. Your battery, which even has charging losses when plugged into the wall on a 0.1C slow charge, will have a higher charging loss if you quick charge at 0.3C. Things will get worse with a fast charge at 1C. If you really want to kill the charging effectiveness, ride around with it and subject it to constantly varying charging and discharging loads!

I'll close with this - an ultracapacitor, while containing chemicals for things such as the dielectric layer, does not involve a chemical reaction at the electrode surface. No reaction diffusion time constant, nor is charge transfer between surface species is required. If I send 10 A for 14 s to my 140 F capacitor, I will expect (neglecting the very low resistances involved) an increase of one volt in the capacitor, with the associated energy storage that depends on the squares of the initial and final voltages. Send in 20 A for the same time, and it's two volts. 40? Give me four. The best news is that I can expect the Cycle Analyst to accurately track power and energy coming and going, nearly regardless of the rate of electron supply or withdrawal.

An ultracapacitor would absolutely help improve the percentage of recaptured energy when compared to a galvanic battery.

 

[ZapPat]

Justin, could you improve the regen capture by using an ultracapacitor as intermediary, with an ultracapacitor such as JCG uses here:
viewtopic.php?f=6&t=7511&start=0&st=0&sk=t&sd=a&hilit=JCG

So even if an ultracap would allow me to do a maximum regen current of say 20 amps. Yes, I would be able to do more regen and come to a faster stop, but no, I would not get any more energy back. In fact, I would recover quite a bit less energy than had I come to a more gradual stop at 6 amps regen using just a battery to absorb the energy. So the idea that an ultracap could help 'improve' the percentage of recaptured energy is quite false in this case.

With respect, I couldn't disagree more. When you plug that battery pack in at night (regardless of its type), you won't be using a constant current of 20 A to charge it, and there are lots of reasons why you wouldn't. I think that people are forgetting that the battery is an electrochemical device, the emphasis being on chemical. Don't forget that when you charge or discharge a galvanic cell, there are chemical processes being carried out in several places within the cell, each of them occuring at a finite rate and according to a specific set of electrochemical kinetic rules which are extremely complicated and difficult to describe in equation form without making several assumptions. I'm not going to preach equations here, but all of us knows that the actual energy put back into the cell is quite dependant on charging current (among other things) for a reason.
[... more interesting discussion of a cell's electrochemical functions ...]
Personally, I consider the concept of charging efficiency itself to be weak, I’ll refer you to the following discussion:
http://www.smartgauge.co.uk/chg_eff.html

This page refers to lead-acid cells, typicaly being very pathetic in their chemical charge/discharge characteristics. This is why no one with any sense would ever use lead in a modern EV that makes heavy use of regen (or pretty much any EV in my opinion). Someone switching to lithium from lead-acid (even without using regen) realises the huge difference between the two chemisties. I would like to see this kind of analysis applied to modern LiFePO4 though, that would be great!

I think Justin's point about the overall regen efficiency actually being better at 6 Amps when compared to a more intense regen of 20Amps for the same situation is quite valid. The point that using ultra caps would not really change much in this scenario is true, since much of the loss of energy happens elsewhere than in the battery anyways. Your extra 10% or so loss during charge/discharge of a lithium battery is not really much compared to motor and controller losses.

Of course the fact that ultra-caps are electro-static devices would give them significant advantages for high-power stops and starts. The real problem is that they are not yet suited to real transportation needs because of their energy density (until we have high-Voltage EESTORs or the high-Capacity MIT nanocarbon tube ultracaps at least). The real problem with using them coupled to a battery pack is that a bi-directionnal DC-DC converter has to be used, because straight paralleling of these two devices would leave the electrochemical device bearing almost all the current anyways. To partially get around this, you could have a contactor to switch between battery pack and ultracap pack when going from acceleration/breaking to regular cruise speeds. Either way, you would need some method to adapt the much different voltage vs charge curves of the two devices for the caps to do any real work.

There are efficiencies associated with cell charging under specific conditions (trickle, constant current, constant voltage, pulse, float, and burp charging, and so on), each with their assumed or calculated time constants. Typical charge transfer time constants might be less than a minute, and a mass transport time constant could be several hours for a cell with plenty of Ah. Certainly, this is the case with the large capacity batteries in ebikes discussed here.

All of these things lead to the real reason people talk about damaging batteries with regen. Using regen from your ebike is best described as “random charging.” Constantly varying potential and concentration fields within the battery, and variable electrode process environments. But it’s not all about damage. It's not even all about "internal resistance." Simply put, the electrode and electrolyte processes will be continually varying in efficiency.

But with appropriate lithium cells, this "varying efficiency" will still be quite high, as long as you are staying inside the cell's charge and discharge specifications. Same goes for your "random charging" comment - just keep the cell voltages, currents and temperatures at what the cells like and there should be no reason that "random charging" will have any negative effect on them (other than regular ware).

solarbbq2003 hits the nail on the head here:

I guess also justins % graphs are based on current measurements via shunt? Would it be correct to assume the % graphs given will be slightly overestimating the actual % stored in the batts?

For the reasons I mentioned above, the Cycle Analyst will not by itself give an accurate accounting of the real, (re)usable energy put back into a galvanic battery by regenerative braking. It's not enough to measure the current and multiply by the individual slices of time. And forget about motor winding losses for the moment. Your battery, which even has charging losses when plugged into the wall on a 0.1C slow charge, will have a higher charging loss if you quick charge at 0.3C. Things will get worse with a fast charge at 1C. If you really want to kill the charging effectiveness, ride around with it and subject it to constantly varying charging and discharging loads!

I understand what you're getting at here, but I think the wording at the end is kind of misleading. You start by talking about the error between current put into the battery vs the actual current that is stored and can be re-used, which could be called "charging effectiveness" . However, you then state that subjecting the battery to constantly varying charging and discharging loads will "kill" the charging effectiveness - what?? Your accumulated in-out error % will get bigger, but the battery is just as happy with small charge/discharge cycles, and likely even happier according to most of what I've read about LiFePO4. People here already have enough false negetive bias towards regen as it is, so my point is to watch the way we term things relating to it!

I'll close with this - an ultracapacitor, while containing chemicals for things such as the dielectric layer, does not involve a chemical reaction at the electrode surface. No reaction diffusion time constant, nor is charge transfer between surface species is required. If I send 10 A for 14 s to my 140 F capacitor, I will expect (neglecting the very low resistances involved) an increase of one volt in the capacitor, with the associated energy storage that depends on the squares of the initial and final voltages. Send in 20 A for the same time, and it's two volts. 40? Give me four. The best news is that I can expect the Cycle Analyst to accurately track power and energy coming and going, nearly regardless of the rate of electron supply or withdrawal.

Yes, the caps are more predictable since they are simpler devices and their SOC is so easy to know. BTW, what kind of internal resistance does that capacitor bank you use have, JCG?

An ultracapacitor would absolutely help improve the percentage of recaptured energy when compared to a galvanic battery.

True for an added reduction of maybe 10-15% in battery-related losses, but not without some additionnal circuitry to make it practical. However, controller and motor losses are still the same for identical currents, and these losses can be quite a bit bigger than charge efficiency (battery) losses especially when using higher currents.


JCG - I read all through both your supercap bike threads and wish you luck with your cool project. You like these caps so much, you must dream of the new developments going on in this field! Hey, if I thought that these newer high density super caps would be becoming available to the public in the next few years, I would be adding a nice bi-directionnal DC-DC converter to my controller project fo sure! But as it's going now, I don't think we'll be able to get our hands on any for a while yet...

Ciao!
Pat

 

[JCG]

Sorry that original post was so long, Pat. I found this (very interesting) thread pretty late in the game.

This page refers to lead-acid cells, typicaly being very pathetic in their chemical charge/discharge characteristics. This is why no one with any sense would ever use lead in a modern EV that makes heavy use of regen (or pretty much any EV in my opinion).

The principles are the same for any secondary galvanic cell. I should have focused on a line in the middle of that page to be more clear. Here it is:

"The problem is the 'Charge Efficiency.' It is something that can only be measured for a certain discharge and recharge cycle. It cannot be calculated. It cannot predict anything. The effect of it can be measured. But change the discharge loads, or their duty cycles, or the charge rate, and the "charge efficiency" will come out different every time. "

I think Justin's point about the overall regen efficiency actually being better at 6 Amps when compared to a more intense regen of 20Amps for the same situation is quite valid. The point that using ultra caps would not really change much in this scenario is true, since much of the loss of energy happens elsewhere than in the battery anyways.

That point is true for batteries, not caps. We need to make sure that we're comparing apples to apples. Take two otherwise identical bikes with either a 48V UCap module or a 48V Li-ion pack, and the caps will capture any braking current with equal or greater efficiency vs. the galvanic battery. I wanted to show the chemical reasons for that. Batteries should capture small currents more efficiently than large ones. My goal was to say that the caps are OK with either high or low regen currents (and all in between), and should be able to efficiently store any current that is provided. Batteries and caps may indeed capture small currents equally efficiently - but the caps will cover the whole current spectrum more efficiently.

Of course the fact that ultra-caps are electro-static devices would give them significant advantages for high-power stops and starts. The real problem is that they are not yet suited to real transportation needs because of their energy density (until we have high-Voltage EESTORs or the high-Capacity MIT nanocarbon tube ultracaps at least). The real problem with using them coupled to a battery pack is that a bi-directionnal DC-DC converter has to be used, because straight paralleling of these two devices would leave the electrochemical device bearing almost all the current anyways.

There has been some promising work on this; a while back NREL did a study just connecting the batteries and ultracaps in parallel, to "share the load" as their impedances would allow them to. No DC-DC converters, no switching. It's old, but I think you'll really enjoy this presentation:

http://www.nrel.gov/vehiclesandfuels/energystorage/pdfs/2002fl.pdf

In summary, the caps were able to protect the batteries from the worst surge currents. The same should go for sending current the other way.

But with appropriate lithium cells, this "varying efficiency" will still be quite high, as long as you are staying inside the cell's charge and discharge specifications. Same goes for your "random charging" comment - just keep the cell voltages, currents and temperatures at what the cells like and there should be no reason that "random charging" will have any negative effect on them (other than regular ware).

Not quite - see the next comment below.

I understand what you're getting at here, but I think the wording at the end is kind of misleading. You start by talking about the error between current put into the battery vs the actual current that is stored and can be re-used, which could be called "charging effectiveness" . However, you then state that subjecting the battery to constantly varying charging and discharging loads will "kill" the charging effectiveness - what?? Your accumulated in-out error % will get bigger, but the battery is just as happy with small charge/discharge cycles, and likely even happier according to most of what I've read about LiFePO4. People here already have enough false negetive bias towards regen as it is, so my point is to watch the way we term things relating to it!

There is a price to pay with variable current to or from a galvanic cell. It unfortunately takes energy and time for the electrode-electrolyte concentration fields to set up to any given charging or discharging current. A lot of current variation from one to another will cause the need for those chemical changes to occur and will waste energy in your bottom line.

Yes, the caps are more predictable since they are simpler devices and their SOC is so easy to know. BTW, what kind of internal resistance does that capacitor bank you use have, JCG?

Maxwell quotes the ESR (Equivalent Series Resistance) of the individual cells (which are in series to make the module) as less than 0.3 mOhm. They call it "ultra-low..." but you make the call. I guess it doesn't sound like much, but I don't know how much progress a number like that represents vs. the rest of the field.

JCG - I read all through both your supercap bike threads and wish you luck with your cool project. You like these caps so much, you must dream of the new developments going on in this field! Hey, if I thought that these newer high density super caps would be becoming available to the public in the next few years, I would be adding a nice bi-directionnal DC-DC converter to my controller project fo sure! But as it's going now, I don't think we'll be able to get our hands on any for a while yet...

I apologize if my previous post sounded edgy or combative - not my intent. I just didn't like hearing that the caps "wouldn't do squat" and thought I could provide a second opinion... NREL and Maxwell are banking a lot on the fact that they'll do far more than squat. I have a bunch of resources available on the cap-battery tandem in hybrids - anyone can send me a PM and I'll send them along.

Thanks for the kind wishes - luck is actually pretty useful to this kind of work. I've had so much fun doing this project I've kind of forgotten about the work that comes later with a full size car. My ultimate hopes for the caps lay in the talk of their "projected low cost," abuse tolerance, and lifetime. I am really excited by what I've seen so far with the bike, and I know we'll get a paper out of it. It's nice when you can really get excited about work (no matter what you do). All I know is I'd be stuck in the mud without these forums.

 

[solarbbq2003]

One time i was doing a long ride ( 60km, well its a long way for me! ) and had two cobalt 48v 15ahr packs, one pack had pretty much run out of juice climbing a fairly long hill ( quite a few km and quite steep), instead of changing to the other pack, I decided to use regen on the discharged pack when coming down the mountain ( big hill really) and see how much distance it returned to the pack, results were very poor, amount of regen capture was extremely low I was quite surprised I was expecting that a discharged pack would capture at least enough to continue some distance after the mountain. Sorry no data to go with it was some time back and I had no monitoring equipment apart from analog display of current.
The reason i'm mentioning is that puting regen energy back into a discharged pack is one way of getting some way to measure how much energy has gone back into the pack. I've been using regen braking with cobalt packs for a long time and seem to be little effect on longevity of the packs. Though I'm a bit worried about using regen braking with nimh packs and effect on longevity.
but I am a supporter of regen braking, it makes a hell of a brake ( not having to squeeze a brake on hard to stop is a good thing)
I was of the impression that ultra caps would make a significant difference in energy that can be captured ( but no experience with them unfortunately). Though have been shown a simple ultracap set up used with bike dynos for bike lighting ( done by kerry from ktronik australia : small lithium batt and ultracap used in parallel...I think), and I was surprised at how simple they were to use at least on that type of setup.

 

[Knuckles]

Turns out the Infineon has regen braking.
It was there all along. Works great too.
Testing ... Testing ... Testing ...

It has regen (down to dead stop) on the eBrakes.
It also has regen on the throttle control down to about 15% of full speed.

These values are all set by using the software flash to program the MCU.

http://endless-sphere.com/forums/viewtopic.php?f=2&t=8317

-Knuckles

 

[ZapPat]

Turns out the Infineon has regen braking.
It was there all along. Works great too.

Knuckles - I checked out your tech info threads to look at what can be configured on the infinion, but I couldn't find a current limit for the regen. We usually need to keep the max regen current lower than the max drive current for most batteries. This is doubly important since the batterie's BMS can not cutoff regen current itself because of the internal diode in the MOSFET that automaticaly conducts on reverse current.
Thanks,
Pat

 

[Knuckles]

I have not yet measured the Infineon regen current in the 72V range.
It appears to be software regulated and a fraction of the programmed drive current.
(Another reason for a home-made dynamometer perhaps?)

The regen works very well, however, on the flat during road testing.
I like the idea of "braking by wire". Very cool feature.

 

[Roy Von Rogers]

I had wondered if it is possible to engage regen in the same manner as a throttle via hall effect. But since I have never experienced how hard regen kicks in, I'm not sure if such a device is needed. Such a device could be engaged with the brake switch mechanism and the harder you pull the brake lever, the more the regen would operate. I know this would bring about more complexity, but was just an idea I had pondered about.

Roy

 

[ZapPat]

I had wondered if it is possible to engage regen in the same manner as a throttle via hall effect. But since I have never experienced how hard regen kicks in, I'm not sure if such a device is needed. Such a device could be engaged with the brake switch mechanism and the harder you pull the brake lever, the more the regen would operate.

Regen braking could easily controlled like this via a second hall-effect throttle, if a controller would exist that had a second input for this (in fact I'll add this regen control option to my long term controller build wish list!).

Something close to this setup does exist though:
- When your brake lever switch (on/off) is triggered, you controller is now in regen mode and the throttle now acts as a variable brake instead of a regular throttle.

Other ideas for regen input are:
- Dual direction throttle (center return): One way controls drive current and the other way controls regen current.
- Snap/Roll throttle: Uses regular style throttle, but in a different way; Quickly letting go (snaping back) lets you coast, while slowly rolling back the throttle engages regen and adjusts it's strength.

 

[Knuckles]

The Infineon has Throttle regen also.
The throttle regen is an option in the software interface.

I have the throttle regen turned off for now but it works very nicely.
Same amount of regen braking at higher speed.
You can ride (accelerate and decelerate) by throttle alone.

As the bike slows down, however, throttle regen turns -off.
The eBrake regen is continuous to a dead stop.
The throttle regen disengages at maybe 15% of full throttle.

 

[fechter]

That kind of 'throttle regen' is similar to the MCIPC24 controller my Zappy has. I did not like it since it does not allow for coasting unless you hold the throttle just right.

Having the regen activated by the brake lever is much more intuitive.

The old Xooter EX3 had a bi-directional throttle that operated regen in one direction. I don't think it would be easy to implement this kind of feature on the Infineon, but it would be a nice feature.

Using the regular throttle as a regen control during braking could be a problem if you were on full and suddenly let off the brake (it would go from full brake to full throttle). It's also a bit akward to operate a throttle and brake lever with the same hand.

 

[ZapPat]

The old Xooter EX3 had a bi-directional throttle that operated regen in one direction. I don't think it would be easy to implement this kind of feature on the Infineon, but it would be a nice feature.

Any idea where we could get such a dual-direction twist-type hall throttle (center return)?

Or maybe a lever-style hall throttle (linear) to be used as a regen brake? This might be easy to make with a couple magnets, a metal strip and a linear hall sensor (as used in the twist throttles - A1302 from Allegro, digikey #620-1022-ND) mounted to an old brake lever. The chip's output changes depending on it's position between the north and south magnetic poles. Might be a fun project if nothing exists out there...


On/off type regen braking (meaning single strength) is probably usefull but also kind of limited, although if "pumping" (ramping up/down of regen power) implemented in the controller's software it would be more acceptable since we could then modulate the regen power this way.

Using the regular throttle as a regen control during braking could be a problem if you were on full and suddenly let off the brake (it would go from full brake to full throttle).

I was wondering about this myself, and have read that a number of controllers force you to stop moving before changing modes (golden motors), but then force you into the permanent "engine brake" style drive/regen. No coasting. Does the infinion do regen via throttle in this manner? Knuckles showed us where we can activate it in the software interface, but I think there is also an input that can enable/disable all regen on the fly. It seems to me that to make this type of regen control safe, the controller should only do the switch to/from regen mode when the throttle is droped to zero.


I wish I could do more tests outside with both controllers, but it's about -20oC here and there's tons of snow! I'm looking into a front ski conversion and screws in the back tire to take care of this problem soon!

 

[StevenR]

Please excuse the excursion from theoretical to practical.

My motor is a C'lyte 5303 and the battery is a Ping v2 48V20aH with a Crystalyte 4840 controller. I enabled regen and saw 5 to 10% regen on the Cycle Analyst over the course of several hundred miles. I disabled it when I blew up the caps. I am considering upgrading the caps on my backup controller and re-enabling it, given this discussion.

When regen was enabled the phase wires to the motor got warm - perhaps even hot. They never seemed at all warm when regen was not enabled.

My question for this wise group is this: Does the C'lyte regen actually work? Am I pushing juice back into the battery or am I just heating up the wires because the voltage is below battery voltage?

On a related note: Can I enable regen on the ebikes.ca model C7248SI 36-72V 48A Start Immediate Digital Controller withIRFB4110 Mosfets in the same way I do it on the standard C'lyte controller (I would put the ebrakes and cruise control on as well)?

If the C'lyte does not work, what is a good regen controller? Curtis? Kelley? Other?

 

[Knuckles]

It seems to me that to make this type of regen control safe, the controller should only do the switch to/from regen mode when the throttle is dropped to zero.

The Infineon regen appears to be "smart". It follows an algorithm of sorts.

So no problem with WOT while braking and then releasing the brake.

Also no heating effect of motor, controller or harness. Just very nice braking with reverse current to the bats.

The regen works great. I just use it.

Very cool also. I tap on my rear brake handle (right side) and my front DD hub slows me down.

 

[jantorch]

Hi, First time post here. I found your article on regen braking very interesting but incomplete as you didn't explain how the current flows worked to recharge the battery. Essentially, regen braking works by shorting out the windings of the motor according to whichever one of the 3 bottom mosfets is turned on at the time and goes through near as dammit 1 &1/2 windings currents worth at any one time. This is actually quite power inefficent and just heats up your motor and can cause internal heat build up and can seize your motor at some point. especially if overvolting. ( W= VI).

The answer is actually very simple and occurs when no pulses at all go to any of the mosfets in the 3 phase bridge. This situation is true and applies to both controllers with and without regen braking but no freewheel and is actually a good simple test of whether your motor has a freewheel or not and works like this all the time, regardless of whether you've got a regen controller or not.

Remove the battery and connect a voltmeter to the battery leads from the controller or turn the lights on if they are inbuilt.

If you get a voltage or your lights start to work then your motor and controller is acting like a permanent magnet 3 phase alternator, like the one in a car. The reason for this is that with no pulses going to the mosfets, they are turned off and all that matters is the built in protection diodes. With every phase change, alternate higher and lower opposite pairs of diodes will conduct and store power in the controller's capacitor. When the voltage gets high enough, it will transfer current back to the battery if there is no blocking diode in the BMS. However, your battery will not recharge unless your wheel is spinning at a high enough RPM to overcome the voltage of your battery. Quite fast most of the time.

If you really want to recharge your battery at a lower RPM or dump the power back into your battery when braking or going downhill, then what you have to do duplicate the 6 protection diodes circuit with the input going from the 3 phase wires and the output to a capacitor and a DC-DC step up convertor with an protection bypass diode for when the generator voltage is high enough on its own to charge the battery, and can be controlled via semiconductor switches or relays via your throttle or pedelec or brake voltages or a simple switch and/or in conjunction with your regen braking system or preferably not.

Obviously not for everyone but If you are really serious about getting as much mileage as possible out of your battery, then this is a better way to go, as it dumps any surplus kinetic energy into your battery rather than your motor and may prevent it seizing up. Sealed hub motors with built in controllers seem to be prone to heat build up if operated out of spec or in the rain from what I've read.

What else are external disc brakes for except to dissipate heat as rapidly as possible while slowing you down as quickly as possible?

Here's a link to EE Times - understanding regeneration part 2 and has got some diagrams of how the current flow works when the motor is being used as a generator with mosfets.

http://www.eetimes.com/document.asp?doc_id=1274137

I hope I've given a few people some ideas - Cheers!

 

[fechter]

In a properly implemented regen setup, the motor windings are shorted only breifly at the PWM frequency. During the time the windings are shorted, the current builds up in the coils. When the short is released, the magnetic field collapses, creating a very high voltage pulse. The body diodes in the FET bridge will circulate this pulse back to the batteries (or hopefully the FETs are actually on at the right time to reduce the voltage drop in the diodes). In this scheme, it works like a switching boost converter. You can get current back to the battery at speeds way below the no-load speed of the motor. You can get strong braking down to a few mph.

Any motor spinning faster than the no-load speed will regen back to the batteries even if the controller doesn't have a regen feature.

 

[MACchine]

AMAZING information and interesting analysis but you lost me when you said the following ...

Now a freewheeling setup still has to overcome the motor drag when the motor is running, so the actual results in a realistic comparison are significantly more favorable than this. If for instance the motor was being used 50% of the time, then we would have gained over 4 times more energy from regen than we could have saved from having a freewheel instead. But even for people who only use the motor on the occasional hill, they have more to gain in terms of total energy consumption by having an always engaged motor and regen than they would save from perfect freewheeling.

Any motor will have drag regen or not when powered, but here you get to a 4 X advantage by pretending that there is no drag in the regen but there is drag in the non-regen. Charts you show later even show equal efficiency between regen and non-regen powered.

So here is how I see it. A regen bike will be ridden differently from a non-regen bike, a regen bike will be ridden like a motorcycle where you power most of the way to a red light and then release the throttle and let the engine drag bring you nearly to a stop before you brake, a non-regen bike as soon as you see the light turn red you release the throttle and coast all the way to the light. This means that the non-regen bike coasts almost always twice are far as the regen bike when approaching a red light, sometimes 3 times as far. This completely negates your not 4X advantage but 2X advantage of the regen bike and actually makes the non-regen bike a bit more efficient, but this is just analysis, it is NOT PROOF!

Thus, I believe based on the above revised analysis you have not proven which one wins. I believe the ONLY way to really get to the bottom of things is to build to nearly identical bikes one with regen the other with freewheels and run these bikes on the same path at the same time riding together in the way most efficient for either bike and see what wins.

I may get a chance to do this in the next couple years, we will see.

Thanks again for the analysis and info it is quite impressive, you covered all the bases.

 

[MACchine]

Support at Kelly Controller just told me that when going down hill with regen on, REGEN WILL USUALLY TURN OFF, because if the bike is going too fast down hill the current becomes too great possibly damaging the batteries so regen turns off.

Kelly is constantly advising me against using regen, yet the controller they are trying to sell me DOES REGEN -- NOT a good sign for regen.

 

[amberwolf]

No, it's just a not a good sign for *Kelly controllers* if they arent' designed to handle regen under common conditions.

 

[Punx0r]

There is something amusing about a braking system that disables itself only on very steep hills