Regen Braking Booster FOR LOW SPEED REGEN CAPABILITY

[Doctorbass]

I had an ides few month ago about the way to solve the probelm with regen difficulty at low speed or when using high voltage battery.

The main problem is that to regen and brake, the voltage that the motor produce MUST be higher than the battery voltage.. or in other words, would only work if the speed you have is higher than the max speed you can reach with throttle wide open..

Let say you run max at 55kph. that mean that to regen, the speed you will need must be 56+kph.

The reason is that if the voltage you have to run 55kph is 48V, so when you go downhill and reach 54kph, you will still have no regen.. but when tou will reach higher than 55kpg, the voltage produced by your motor will be higher so the current will flow to regen.

To regen, you must have reverse current between your battery and motor. to have current, you need a positive delta V from the motor to the battery.

The only way to regen or be able to brake using this method is to boost the voltage of your motor in regen mode.
DCDc have input-output isolated so you can add the output vltage of those to their input in serie to boost the voltage and then get current to regen.

This exemple show this booster idea i had:

Doc

 

[Tiberius]

Hi Doc,

That seems a very elaborate way of doing it, and it doesn't give you a smoothly variable brake control.

Why not just make a boost circuit with variable control? It should be possible to use the motor windings as the inductance. If I get a chance I will post a trial circuit later.

Nick

 

[ZapPat]

Hi Doc,

That seems a very elaborate way of doing it, and it doesn't give you a smoothly variable brake control.

Why not just make a boost circuit with variable control? It should be possible to use the motor windings as the inductance. If I get a chance I will post a trial circuit later.

Nick

You like these isolated DC converters, don't you, Doc? Well, I can't blame you! (I am ordering two artesyns and a bunch of relays to try a variation on your great balancing charger idea). I just received two nice vicor 48V 12.5A DC power supplies to play with too!

But I have to agree with Nick that I don't think this is a good solution to efficient, controlled regen.

IMHO, all we need to use for good regen is an appropriatly designed controller working in 4 quadrant PWM switching mode. Any synchronised buck converter is also a boost converter when seen from the output side, after all (check a schematic and you will see). It's more of a question of making a SMART controller that will regen in an efficient, controlled way.

The golden motors regen controller is one step towards this, but I have yet to know exactly what way it works... but it is definetly lacking intelligence to control it's regen (sounds like a fully manual process that has to be controlled by the user). Maybe I'll be getting one, but I hesitate between it and a crazyman (regen or more power?? :? ).

 

[Doctorbass]

I found something Nimbuzz posted about the idea of boosting voltage of the BEMF to get regen braking.. and i totally agree with the engeneer that replyed that.

in this thread http://endless-sphere.com/forums/viewtopic.php?f=2&t=2803&st=0&sk=t&sd=a&start=45

"The problem with the scheme that you propose is that you want to have braking currents flow when the back-emf voltage from the motor is LESS than the supply voltage of the battery. To do that you need a boost converter with an inductor to step up the voltage from your motor (which typically may be like ~20V or so) up to the voltage of the battery which needs ~42+ volts to be actively charging. The setup with a DC controller in reverse won't do that.

A proper 2-quadrant regen controller really is the right approach, no hacks required, something like this:
http://www.elmomc.com/products/piccolo-main.htm"

So that would confirm what i think about regen braking right?

So normal crystalyte controller with "regen" feature would only work if the back-emf voltage from the motor is higher than the supply voltage of the battery.... wich would need to run faster like downhill thand the max speed that the controller can give with the battery...

i'm not sure to understand the 4 quadrant explanation..

Doc

 

[Tiberius]

Doc,

All you need is a basic boost circuit:
The inductor can be the motor winding. I haven't drawn the three windings and the three phase rectifier, but they can be the same as in your drawing above. Provided the PWM frequency is high compared to the commutation rate it will work.

The PWM Mark-Space ratio controls the voltage step up ratio. So that will vary the speed at which the motor starts putting current into the battery. Below that speed there is no braking, above that speed the braking will come on quite rapidly.

The basic circuit will therefore act a variable speed limiter. That would be ideal for lowering a vehicle down a hill, but most people want the feeling of a normal brake. The way to do that would be to put a current sensing circuit on the motor side of the boost circuit and use that for closed loop control of the PWM Mark-Space ratio. That turns it into something that would give a constant braking force - and that braking force could easily be controlled by a lever.

As ZapPat and other people have said, its similar to the normal forward power control, but operating in reverse. In fact, its simpler, because it doesn't need to do the commutation. You don't really need a synchronous boost circuit, or two or four quadrant complexity; the efficiency gains over the simple circuit shown are only tiny.

Nick

PS. Sorry about the quality of the diagram. My schematic package won't export a gif or jpg, so I have to paste into something else first.

 

[fechter]

Just like when driving the motor, you can use the motor windings as your inductor for a boost converter.

You briefly (microseconds) short the motor windings and the current builds in the coil. When you release the short, the collapsing magnetic field will generate a very high voltage spike which will be circulated back to the batteries by the body diodes in the controller. Even at very low speed, the collapsing field will make enough voltage to charge into the batteries.

 

[ZapPat]

This is a simple buck circuit using one FET and one diode:

(Vg here would be the battery, and Vo the motor's Back EMF)

This is a simple boost circuit using one FET and one diode:

(Vg here would be the motor's Back EMF, and Vo the battery)

Notice the differences, but notice also the similarities between the two? Now replace those diodes by FETs (they're synchronous rectifiers now, with an actively controlled FET instead of a passive diode). If you observe well, you see that these two circuits are the same when using synched FETs, just look at it from the opposite end of one circuit to see the other!

The only thing here controlling which way the current flows is the duty cycle applied to the oppositely driven FETs.

Example:
Input: 100V battery
Output: motor at half speed, generating 50V BEMF

Situation 1: duty cycle is 75%, so the motor sees 75V and accelerates - current is draining from battery (buck operation)
Situation 1: duty cycle is 25%, so the motor sees 25V - current flows back into the battery from the motor (boost operation)

The FETs already have the correct configuration in present controllers, but are just not being switched in a way to make regeneration possible. Also needed is a way to read bi-directionnal current flow (like an allegro hall sensor). Note that this is nothing new, as many controllers use this switching scheme already.

 

[Doctorbass]

Many thanks for these great explanation Fechter and Zapat!!

I now understand better how a boost and buck converter work, and that it can be used for regen without any complex voltage booster than a coil a fet and a diode...

That's nice!

Searching about buck and boost converter i found this wich is interesting and thau use these converter with capacitor for acceleration and braking regen!

http://www2.ing.puc.cl/power/paperspdf/dixon/42a.pdf

Doc

 

[ZapPat]

Many thanks for these great explanation Fechter and Zapat!!

I now understand better how a boost and buck converter work, and that it can be used for regen without any complex voltage booster than a coil a fet and a diode...

That's nice!

Searching about buck and boost converter i found this wich is interesting and thau use these converter with capacitor for acceleration and braking regen!

http://www2.ing.puc.cl/power/paperspdf/dixon/42a.pdf

Doc

Good find, Doc!

This is a good example of how a dual-direction (buck-boost) can be used to make energy flow in either direction, no matter what voltage difference there is (but efficiency goes down as the difference gets bigger, specialy in the voltage boosting direction).

Fetcher - Your pulsed shorting of all three phases regen technique works I'm sure, and more to the point we are actually talking about the same thing! You are implementing a simple boost converter using the motor as a coil - exactly the same thing as I have been talking about (I'm not crazy after all I guess!). After all, my first version BLDC motor controller was operating in this way, being able to slow down the motor by pumping it's energy back into the battery bank (yes it works well and is simple!).

The difference resides in the switching sequence of the phases we propose to use to do this. Your method seems a bit brutal to me (I don't mean to bump you or anything here, just my opinion). Instead of treating all three motor phases as thought they are the same, I am very certain that controlling the buck-boost circuit independantly on each phase is a better, more efficient method of recuperating the motor's energy. There is no added hardware complexity with all three phases being controlled seperatly, since the hardware is already there. Mostly all that is needed is a "smarter" controller brain (aka a better micro-controller program).

Building this type of controller has been my goal for a few years now, and now that I have real EV platform to test with (my recently completed ebike), I am certain my next design will be more than good enough to drive it WILD! I'm thinking of making a thread to collect everyone's opinions of what their ideal ebike controller would be... after my holidays though!

 

[fechter]

Yes, my method is extremely brutal and it would be far better if the correct FET was always chosen to short the windings. Synchronous rectification would be great too. Other than the current sensing, all the required hardware is already in most controllers.

It would be best if the regen was limited by the motor current, rather than the battery current. On my Zappy (brushed system), the regen motor current regulated the duty cycle which gave a very uniform braking force over a wide range of speed. The amount of force could be varied by a control.

I guess it would be good to have some kind of overvoltage cutout also, but unless you start out with a full battery at the top of a hill, you'd rarely need to worry about it.

 

[The7]

This is a simple buck circuit using one FET and one diode:
View attachment 1
(Vg here would be the battery, and Vo the motor's Back EMF)

This is a simple boost circuit using one FET and one diode:

(Vg here would be the motor's Back EMF, and Vo the battery)

Notice the differences, but notice also the similarities between the two? Now replace those diodes by FETs (they're synchronous rectifiers now, with an actively controlled FET instead of a passive diode). If you observe well, you see that these two circuits are the same when using synched FETs, just look at it from the opposite end of one circuit to see the other!

The only thing here controlling which way the current flows is the duty cycle applied to the oppositely driven FETs.

Example:
Input: 100V battery
Output: motor at half speed, generating 50V BEMF

Situation 1: duty cycle is 75%, so the motor sees 75V and accelerates - current is draining from battery (buck operation)
Situation 1: duty cycle is 25%, so the motor sees 25V - current flows back into the battery from the motor (boost operation)

Thanks very much for the explanation.
In the your explanation, duty cycle of 75% would mean the buck FET is ON for 75% and OFF for 25% in 1 period. During the boost FET is OFF when the buck FET is ON and vice-vesa.
So you will have acceration for 75% duty and braking for 25% when the motor is running at half speed.
It seems that it is running at 4-quadrant mode!?


Seeming that it could also be switching the buck FET and boost FET in the following method for running and braking.
a) During normal running, only the buck FET is PWM operating (the boost FET is always OFF).
For duty cycle of the buck FET > 50% , you would have acceration when your motor is at half speed.
But for duty cycle < 50%, you will have decceration (but no re-gen braking effect).
b) During re-gen braking, only the boost FET is PWM operating (the buck FET is always OFF). You would have braking all the time (irrespect of the motor speed).
Is this a 2-quadrant mode?

It seems that Fetcher is using the latter method for re-gen braking.

 

[ZapPat]

It seems that it is running at 4-quadrant mode!?

That's right!

Seeming that it could also be switching the buck FET and boost FET in the following method for running and braking.
a) During normal running, only the buck FET is PWM operating (the boost FET is always OFF).
For duty cycle of the buck FET > 50% , you would have acceration when your motor is at half speed.
But for duty cycle < 50%, you will have decceration (but no re-gen braking effect).

Exactly, and this is how most regular ebike controllers seem to work, and also why they don't regen (unless you run your motor fast to make it's voltage higher than the battery's - then it regens automaticly without control). It does not require synchronous FET operation, and can be done with only one actively switched FET (top one), using a flyback diode instead of the other FET. Note that all FETs are also diodes, so can be used passively as a less efficient diode.

b) During re-gen braking, only the boost FET is PWM operating (the buck FET is always OFF). You would have braking all the time (irrespect of the motor speed).
Is this a 2-quadrant mode?
It seems that Fetcher is using the latter method for re-gen braking.

You are now using your non-synchronous FETs in boost mode by changing which FET is actively switched, and the other (top FET) is now just being used as a passive diode. I think this is 2nd quadrant operation, but I will have to refresh my memory on the actual terms used.

Now... what is better about synchronous FET operation, we may wonder, since we could just switch either the top OR bottom FETs while using the opposite one as a passive diode? The answer is that a FET's voltage drop is much inferior to a diode's voltage drop at the currents we are looking at. Example:
FET Rds = 0.004 ohms
Load current = 50 Amps
So, the FET's voltage drop = 0.2 Volts
The diode's voltage drop would be around 1 volt at least, so it is dissipating 5X more energy than the FET at the same current!

Of course, synched FET switching implies good gate drive design, with care taken to be sure that both FETs NEVER conduct at the same time (making huge power spikes across the DC bus). At the same time as wanting to avoid this cross-conduction, you also want to turn one FET on as soon as is possible when the other FET turns off. Happily, we have access to good integrated gate drivers like the IR2011 or similar and also well thought hardware modules available on some micros (like the PIC18F2331) to avoid this.

 

[electraflyit]

I'm not after any regen function only a variable bake.
RC car ESC can vary their braking (some via software give you ultimate control of your braking curve.)
How do they do it? Surely it is something that our controllers should have as standard.
Eddie

 

[methods]

I'm not after any regen function only a variable bake.
RC car ESC can vary their braking (some via software give you ultimate control of your braking curve.)
How do they do it? Surely it is something that our controllers should have as standard.
Eddie

This thread is dead but Eddie is right.
If you have enough hobbies it becomes frustrating to see how little interdisciplinary communication there is. Coming from RC, I was appalled to see people using SLA batteries when I thought that NiMh were for cavemen (and dont even say NiCd). I even feel a little behind the times using LiPo when I know that A123 is such a beautiful cell.

I have ESC's for planes that I have used in RC Cars and the Car guys insist up and down that it wont work and I wont have brakes. . . Then I employ the brakes and tip my hat. Many RC ESC's come with software programs to turn the brakes on and off, set up delay times, intensity, ramp rates, etc.

Another thing is that all the brushless RC ESC's are sensorless, fully Lipo aware (and programmable), completly protect themselves, and weigh only grams.

Here is an ESC that weighs 32 grams but can run 1.2KW
http://www.castlecreations.com/products/phoenix_hv_series.html
50V, 30A, less than 1oz
Fully programmable

Here is its big brother,
50V, 110A, 0.001ohm resistance, 5oz
http://www.castlecreations.com/products/phoenix_hv_series.html

At least Justin is briding the gap with his new controller.

-methods

 

[ZapPat]

I'm not after any regen function only a variable bake.
RC car ESC can vary their braking (some via software give you ultimate control of your braking curve.)
How do they do it? Surely it is something that our controllers should have as standard.
Eddie

This thread is dead but Eddie is right.
If you have enough hobbies it becomes frustrating to see how little interdisciplinary communication there is.
[...]
I have ESC's for planes that I have used in RC Cars and the Car guys insist up and down that it wont work and I wont have brakes. . . Then I employ the brakes and tip my hat. Many RC ESC's come with software programs to turn the brakes on and off, set up delay times, intensity, ramp rates, etc.

Might you be kind enough to share with us the techniques used in these advanced RC controllers to achieve the bracking effect? Is it some kind of simple regen (most likely), or do you know some other method of switching FETs to achieve the braking effect without having to dump energy back into the battery? Please share your great insight with us, methods...

BTW, haven't you seen the "using RC motors on ebike" thread here on ES where a number of people have been using these castle controllers on their ebikes?

Pat

 

[methods]

Although I am an electrical engineer I prefer electronic engineering: I like to work with "known good" gadgets to build larger systems. That in mind, I don't usually get into the details of how a particular component works. I make it my business to know what technology is available as opposed to how that technology is implemented. I have reverse engineered boards in the past but these days everything is dont in microcontrollers so . . . So I wish I could tell you how they are implementing the braking. From what I can see they have a big microcontroller that calls all the shots. Thats about where my investigation ended.

I originally came here with the intention of using a Hobby City brushless motor and Speed Control (RC Stuff) to build a bike. Once I saw the brushless hubs I never looked back. I like simple and clean. The fewer belts, chains, sprockets, mechanical adjustments, etc. the better. I think that the hub motor is one of the most beautiful things I have seen in a long time.

Examples of some of the knowledge gaps I see are:

- Electric Braking: Standard in other hobbies, treated like some sort of Voodoo for electric bikes (though coming around)
- Programmable Controllers: Nearly standard in RC, hard to find here. I just bought a Kelly 120V220A box so we will see how that programs
- RATINGS!!!: In RC the first thing they tell you is the KV of the motor, nominal power, and peak power. In the bike world they -might- tell you the power and -about- how fast it goes.
- Self Protection: I have seen some of this, but it is possible to protect a circuit from Voltage, Current, Temperature, etc. very easy. Should only cost a few dollars more.
- Advertising: Though I hate advertising, it does help one to learn about what is available instead of depending on word of mouth
- Documentation: ARG! Nothing is documented. . . All Crystalyte would have to do is ask this forum to write the documentation for them and we would probably have 30 people working for 3 months to make the best documentation known to man. Instead, we have our own documentation (though it is quite good. Thanks guys!)

So not that anyone asked for a rant, but there it is.
I think we are in the wild west of electric bikes . . . The New Yorkers are on their way and in the next couple of years all this land will get settled.
So in that sense, we are lucky to get to be early adopters of this growing technology.
We just need to work through the growing pains of doing business (nearly) directly with China.

-methods

 

[docnjoj]

Yeah methods but first they have to make Ebikes LEGAL in NY!!!!!
otherDoc

 

[methods]

Oh man, that really sucks.

ebikes are really new and fresh to me. . . I have been drinking in all this information for the last couple of months and it is just amazing. I had no idea that this subculture was here brewing up new ideas. To be honest with you, if they were illegal in CA, I would be spending all my time working on a stealth mode bike.

Sometimes I pull up to people with what I think to be an insane pile of wires and Lipo batteries and they don't even notice. . . People tend to see what they expect to see, which is *not* and electric bike.

Build one anyway!

-methods

 

[dirty_d]

hmm, ok so a normal tripple half bridge is all you need to do regen? how exactly do you need to switch teh fets, you need to short the motor for a bit then open the circuit and let the inductance of the motor push current into the battery, but which fets need to be shorted at what time? it seems to me that you would short the two phases that would be be conducting if the motor were being powered right because the backemf in those two would be at a peakit also seems like shorting them with the top or bottom side fets all the tiem would work, but using both sides would better distrubutethe losses to prevent overheating. on the controller im working on i have control over all 6 fets so i could do that. does this all seem right?

 

[Mathurin]

Way over my head, but on the Velectris Ki controller you get almost no regen at max braking and max regen with softer braking.

 

[PhoenixOSU]

From what I understand, and from my simulations. You just swap control lines between the top and bottom FETs, shut off the FET that is conducting into the coil (while motoring), and PWM the bottom FET to repeatedly short the motor coil. This can be done with logic if you just want to implement an ASIC solution and add regen after the fact.

If while motoring AH is PWMing and BL is on, then to brake you shut off AH and PWM AL, leaving BL on and the other FETs are all off still. This should create a simple single switch boost converter, I'm not sure about the synchronous converter method, but I wish I could figure out the firing sequence for this, I may have to ask zappat.

 

[Doctorbass]

I'm not after any regen function only a variable bake.
RC car ESC can vary their braking (some via software give you ultimate control of your braking curve.)
How do they do it? Surely it is something that our controllers should have as standard.
Eddie

This thread is dead but Eddie is right.
If you have enough hobbies it becomes frustrating to see how little interdisciplinary communication there is.
[...]
I have ESC's for planes that I have used in RC Cars and the Car guys insist up and down that it wont work and I wont have brakes. . . Then I employ the brakes and tip my hat. Many RC ESC's come with software programs to turn the brakes on and off, set up delay times, intensity, ramp rates, etc.

Might you be kind enough to share with us the techniques used in these advanced RC controllers to achieve the bracking effect? Is it some kind of simple regen (most likely), or do you know some other method of switching FETs to achieve the braking effect without having to dump energy back into the battery? Please share your great insight with us, methods...

BTW, haven't you seen the "using RC motors on ebike" thread here on ES where a number of people have been using these castle controllers on their ebikes?

Pat

I have seen these RC motor. They are so powerfull!!! but they make sound!! HF sound

I like the trike that Recumpense built!!.. very high quality contrruction!!!

The castle creation seems sensitive for blowing.. they dont have any heat sink and are packaged in heat shrink!!

But.. before to make negative opinion i should try one!

 

[Arlo1]

wow 355 days later..... LOL I thought Id post here because I was thinking about regen alot and saw Justins numbers across canada ~2.5% which I feel is worth it but it is a super low number I belive in the city would be alot higher. I was thinking simpler things like having your stator switch from delta to wye wound (for thos who are delta during rideing. And I was thinking about having different coils for charging/braking with more windings to increase voltage. I emailed castle creations about a year ago on the topic of regen while asking about there momba monster that was coming and they said non of thier controlers do regen it is just used as a break (think of purposly having your motor coging controled by your breaks thats the easiest way to understand it). I like what I red in this thread and I think I would like to see it progress.