Help Needed for Regenerative Braking Voltage Regulation

[stevet47]

I need some help from people much smarter than myself...so I have come here.

I am not building an electric car, but I am trying to build a regenerative braking system for a small 24V application using Li-Po batteries.

The system has to be very compact, and extremely simple. This is more a proof-of-concept then anything, so it does not have to be dead reliable, or a great solution, it just needs to function for now, and I need to avoid PWM or other other complexities.

I believe I have a schematic that will function as intended, but I am worried about the unstable charge voltage and need a way of efficiently maintaining a safe charge voltage/current.

You can see my circuit below. Each battery shown is 11.1V 3.3ah Lipo.

The circuit doesn't reflect this, but the switches will physically mounted together to activate on a single touch (basically making 1 switch out of the two), with the Drive switch being NO and the Regen switch being NC, therefore the circuit is in default Regen mode, until the switch is pressed, which would then put the circuit in Drive mode.

The batteries will be in series in Drive, providing 22.2V for power, but they will be in parallel for Regen, creating a 11.1V bank. My thought here was that I would be able to capture more energy, as the voltage generated by the motor needs to be greater than the battery voltage in order for the batteries to charge, so by putting the batteries in parallel making them 11.1V, the output voltage of the motor will be greater than the battery voltage for a longer period of time as the motor slows.

My issue, is that I don't want to dump 22V into my 11.1V battery pack, as each cell can only safely charge between 3.8-4.2V, which makes the safe charge voltage for my pack 11.4 - 12.6V.

So what is the easiest way for me to maintain 11.4-12.6V entering my battery pack from the motor?
I was thinking of using a Zener diode, but the current is too high. I found a few sources saying I could use a Zener and NPN transistor together, but I am not entirely sure on that.

Any ideas?

-sorry for the long post!


On the image, ignore the white box, that was just covering up something that didn't need to be there.

[newInRobotics]

Hi there 

You are planning to do something very dangerous. To start with, if You think that PWM is complicated, then You should not undertake this project, as PWM is the easiest part of it.

If You do decide to carry on, be warned that charging Li-Ion is potentially dangerous to life, that is - You'd be much better off using Lead Acid batteries, MUCH SAFER!

If You still want to carry on, first research how Li-Ion's are charged, as it is not simple plug-n-play procedure. It involves maintaining constant current and constant voltage, monitoring over/undercharge, and charge balancing when using battery bank made of cells in series. To achieve all that You need to use microcontroller. If You need a proof, take a look at prices of Li-Ion pack chargers, they are that expensive for a reason. All mentioned above is much easier to achieve when You take power from mains, however in Your case it won't be constant power source making things much more complicated.

If You still want to go with regenerative charging, You will need to use Buck-Converter and Buck-Boost-Converter. Build one of them first to understand what is involved.

If You decide to go for regenerative braking only (without battery charging bit), then it is much easier and can be done simply by unplugging powered motor from battery and shorting motor's terminals. Voltage generated by motor is fed to the opposite terminal, when potential is equal to 0V - motor stops.

[stevet47]

Thank you for the comments. I have read up on charging Li-ion, and what I got out of it, is that it is a much less complicated chemistry to charge than NiCd or NiMH. Lead acid won't work for this application, as it would be too large, and heavy for the needed capacity.

I will look into buck converters and buck boost converters.

The other I had considered is using a ready-made charger, and using the energy out of the motor as the input energy for the charger, however, this would obviously require a charger that uses DC in, and it would either need to work at a wide range on input voltages, or I would have to regulate the input voltage to something like 12vdc.
Do you know of any lipo chargers that opperate with DC-in?

Thanks!

[stevet47]

OK. So I read a bit about buck converters and buck-boost converters.
You said I would need both. Can you explain a little more about how this works, and why this would be a safer way of charging Lipo batteries?

As I understand it in very basic form, the buck converter would take the variable voltage from the motor and stabilize it to 12V to charge the batteries. So what is the purpose of the buck-boost converter after this? All I can find is that the buck converter reverses polarity, and the buck-boose converter will switch it back.

I realize I am over my head in this, but unfortunately I MUST complete this, and I MUST do it within the next couple weeks, so any help would be GREATLY APPRECIATED! And I would very much prefer not to kill my self in the process!

[Soeren]

Hi,

I believe I have a schematic that will function as intended,

A very scary schematic at first sight

I realize that you probably have no training in drawing schematic diagrams or the conventions used, but you made me think you were trying to destroy the batteries, until I had read your post and compared with the schematic a few times.
For instance, the way you drew the switches, they appear to be changeover contacts.

You should always draw switches to reflect the "non-energized" position.
When you have more than one switch, be sure to separate them so that they aren't read as one. Your schematic appears to have the middle + and - poles connected no matter the position of the switch and have the two diodes shorting the motor, while when activated, short out both batteries.

I have attached a drawing of how it should look.

Be sure to get break-before-make switches, or you will short circuit the batteries momentarily while switching.

The circuit doesn't reflect this, but the switches will physically mounted together to activate on a single touch (basically making 1 switch out of the two), with the Drive switch being NO and the Regen switch being NC, therefore the circuit is in default Regen mode, until the switch is pressed, which would then put the circuit in Drive mode.

Just to be absolutely sure... You are talking about 2 dual pole single throw switches always pushed in tandem (with no chance of being on at the same time), right?

My issue, is that I don't want to dump 22V into my 11.1V battery pack, as each cell can only safely charge between 3.8-4.2V, which makes the safe charge voltage for my pack 11.4 - 12.6V.

Don't worry about 22V... Worry about 60..80V or more instead (depending on exact motor etc.).
However, the batteries will limit the voltage, assuming they're properly rated for the motor in the first place. If they're near flat, they will have their highest impedance and will limit the least, but that doesn't matter, as the battery then need a lot of charge and when the battery is full, it will be very low impedance, so will load the voltage a lot more and, face it, a regenerative braking has a very short term voltage above the battery.
Initially, you'd monitor the battery, of course, but the only problem I see is  the lack of equalizing the cells, but since regen won't do more than increase runtime slightly, you have to put it on a charger anyway and that should be one monitoring each cell and equalize as needed.

So what is the easiest way for me to maintain 11.4-12.6V entering my battery pack from the motor?
I was thinking of using a Zener diode, but the current is too high. I found a few sources saying I could use a Zener and NPN transistor together, but I am not entirely sure on that.

Well, you never told us the max. motor current, the capacity of the batteries or similar useful numbers, so I cannot comment on whether zeners can be used or not, but I do have some heavy duty zeners (to around 20..25W), as well as Transil's, Tranzorb's VDR's etc. (to around a couple of KillerWatts), and they come in much much lager types, the largest I have had in hand personally, was the size of a hockey puck and the 2kW types are slightly smaller than the size of a common 3A diode.

I'd think that it would be possible to find something that works for you... If you care to worry, that is.
A Transil like the PFZ12A, just to mention one, has a nominal clamping voltage of 12.0V, (11.4V min and 12.6V max) and can handle 90A in the first 1ms (829A in the first 20µS).


Did you measure the max. back-EMF (and back-EMP) the motor will provide when switched this way?

If the battery wouldn't shunt the overvoltage, you would have to worry just as much about your switches being pitted or downright welded together.


As this is just considered proof of concept, forget about switchers and the like - you only need such to wring the very last drop of charge out of regenerative braking.

[Soeren]

And I would very much prefer not to kill my self in the process!

Wuss

[stevet47]

Thank you, that was very helpful.
And yes, it is quite obvious that I have no training in any of this. I squeaked by in Sparks101 and 102 in college, but I have since forgotten everything I learned.

Thank you very much for demonstrating how to properly draw the schematic. The switching is much clearer now.

Just to be absolutely sure... You are talking about 2 dual pole single throw switches always pushed in tandem (with no chance of being on at the same time), right?

Yes. However, I only used two switched because I wasn't sure how to do it with one.

As for telling you max current out of the motor, how do I determine this? Right now the device is setup to simply turn on and off. The motor terminals are shorted to brake the motor.
I can measure the circuits current when battery is powering the motor, but I am not sure what I need to do in order to measure current out of the motor.
I am not sure what else you need, but you mentioned the battery capacity... There are two 11.1V 3300mAh 20C batteries.

[stevet47]

As for NewInRobotics' suggestion of buck/boost converters.
Something like this would work, no?

http://www.sureelectronics.net/goods.php?id=837
or
http://www.sureelectronics.net/goods.php?id=65




And off topic, Soeren, I just saw that you are from Denmark, I just visited there for the first time this past July. I spent 3 weeks in Copenhagen, beautiful city, unfortunately I didn't get to explore the countryside, so I suppose I need to make a trip back!

[Soeren]

Hi,

Yes. However, I only used two switched because I wasn't sure how to do it with one.

You can find 4 pole switches as well - 4 separate switches all switched at the same time.
A 4-pole relay (or two 2-pole relays) can be used as well, switched with a small single pole switch.

As for telling you max current out of the motor, how do I determine this? Right now the device is setup to simply turn on and off. The motor terminals are shorted to brake the motor.
I can measure the circuits current when battery is powering the motor, but I am not sure what I need to do in order to measure current out of the motor.
I am not sure what else you need, but you mentioned the battery capacity... There are two 11.1V 3300mAh 20C batteries.

OK, the running current will give a pointer as to what the motor might put out, so please take that reading.
Approximate physical size of the motor would be another good clue.
While the batteries may be 20C capable, running them at 20C (66A)would mean a very short discharge time, like 1..2 minutes, so I doubt that you'd use it that hard (except it might have a huge load in the first few moments, until it's up to speed).


Shorting the motor to stop it indicates a fairly small motor, if it survived more than a few shorts - don't try that with a large motor. The best way of "shorting" a motor is through a low value (high power) resistor, which can then dissipate most of the power. With a direct short, the motor windings have to sustain the power and in large motors, with a lot of inertia, this may melt the lacquer isolation and short the motor internally - and the smell of burnt motors is a far cry from roses

[stevet47]

I definitely will not be discharging the batteries at their full 20C rating, I imagine I will hit a max draw of about 3C or 10A.

The only motor specs I have are: 24VDC 13,000rpm @ no load.
Physical size is approximately 2" diameter 3.5" long.

[Soeren]

Hi,

And off topic, Soeren, I just saw that you are from Denmark, I just visited there for the first time this past July. I spent 3 weeks in Copenhagen, beautiful city, unfortunately I didn't get to explore the countryside, so I suppose I need to make a trip back!

The inner city of Copenhagen would look quite a lot better without the constant road repairs and new construction (subway, fiber, heat, you name it) that seems to peak every year during the tourist season, but if you explore it like a Japanese garden and remember to look up, there is a host of treasures that most Danes forget in the daily rush.

If you happen to return some day, just get in touch with me ahead and tell me what kind of things you like to explore and I can surely give you some tips and if I have the time, show you some of the off beat places as well - from the fanciest museums and Michelin restaurants to the spots of the latest gang related first degree murder scene - we have it all

Oh well, back on topic:
If you have access to an oscilloscope and a low value resistor of an adequate power handling, it's quite easy to measure the voltage generated by the motor (and the time of critical voltage if any) and the current can be calculated from this.

[Soeren]

Hi,

I definitely will not be discharging the batteries at their full 20C rating, I imagine I will hit a max draw of about 3C or 10A.

The only motor specs I have are: 24VDC 13,000rpm @ no load.
Physical size is approximately 2" diameter 3.5" long.

Sounds close to what is used in some cordless drills.
It's 7 a.m. here and pitch dark and my GF is getting up in 30 minutes, so I'll let her sleep and not make a lot of noise now, but I can take some readings of a slightly smaller (1.5" diameter, 2.75" long body - less axle, 20,000RPM) motor and post my findings later (I might be ready for a short nap first though).

[newInRobotics]

OK. So I read a bit about buck converters and buck-boost converters.
You said I would need both. Can you explain a little more about how this works, and why this would be a safer way of charging Lipo batteries?

Well, Buck-Boost-Converter can either boost voltage or step it down, so You need only one of them depending on how much power You want to capture. With Buck-Converter You can step voltage down if it's too high for battery, and Boost-Converter steps it up if it's too low; Buck-Boost-Converter will be able to step-up/down voltage depending on situation. It won't do it by itself, uC will have to be used to achieve this (it involves PWM). Safe Li-Ion charging involves strict voltage and current regulation, that's what You use these switching voltage converters for. Soeren offered simpler solution with clamping diodes, it is not as efficient, however it gets job done with less complexity involved.

I was wondering, why not to use big caps to store power generated by motor instead of trying to pump it back to battery? Charging cap is more efficient than charging battery and it is easier as well. The end result will be longer battery life as with every new start of the motor energy will come form cap and not from battery.

[billhowl]

Document that use big caps to store power generated by motor.
http://web.mit.edu/first/kart/everpres.pdf

[stevet47]

Thank you for the suggestion of using capacitors. I had considered this in the very beginning of the project, but I did not think I would be able to store al of the motors energy, so I never looked into it. Although it adds to the physical size of my project, (goal is to keep it compact), this IS just a a working proof-of-concept, so that isn't too big of a deal if using them makes the project easier to complete.
I like the series configuration explained in the provided link. It would allow me to safety charge my batteries with a legit charger out of the system, then simple draw from them, with all regen energy going to the cap. That sounds safer to me. What data would I need in order to determine the required capacitor bank size?π

[newInRobotics]

Series method described is not used to charge battery. Capacitor is used as additional power source to power motor rather than power source to recharge battery.

[stevet47]

Series method described is not used to charge battery. Capacitor is used as additional power source to power motor rather than power source to recharge battery.

Yes, I understand that. I meant that I can use a stand-alone Lipo charger to charge the batteries before they are put into the circuit, then the circuit only draws from the batteries (and cap) and all the regen energy goes only to the caps.

If the experts (you guys) say that going with the capacitor method would be simpler then any safe method of putting the charge into the batteries, than I will peruse this method..... hopefully with some more guidance from you all!

[Soeren]

Hi,

If the experts (you guys) say that going with the capacitor method would be simpler then any safe method of putting the charge into the batteries, than I will peruse this method..... hopefully with some more guidance from you all!

Simplest method is, as I mentioned previously, either nothing at all or a voltage limiter of sorts.
With the motor in question, the net winning is extremely small and doesn't really make up for anything more complicated than a Transil or similar.
(You'd still need rectifier diodes to keep the motor from draining the batteries when the  generated voltage gets lower than the battery voltage of course).

If it was a motor for a full size express train with miles of daily braking, it would be worthwhile sucking every drop of juice out of a regen system, but in your case, you might get a few more meters of running tops.

Any kind of switch mode (SEPIC and Cuk topologies would be more suitable here than Buck/Boost anyway) will give losses that may, in this particular case, be more lossy than a voltage limiter and it would need to be designed to the exact specs (found by testing the motor extensively) to be able to output anything useable at all - lot of work for a marginal net win (if any, compared to a limiter).

Capacitors would need to hold as much of the energy from regeneration as possible and since caps are leaky, this energy needs to be used within a short time.
They need to be able to handle the max. voltage possible and they haven't got the kind of "regulation" that a battery will does.
They need to be able to handle high ripple currents.
The total value of the bank can be calculated as: C = A*s/V
A = Ampere
s = seconds
V = the tolerable voltage drop during s

If you eg. want to be able to get 1A from a capacitor during 2 seconds, with a maximum voltage starting at 25V and still being at least 21V at the end of the 2s period, you'd need a capacity of 1*2/4 = 0.5F = 500,000 µF with a rating of 35V (assuming 25V tops, it needs a margin) and able to handle ripple currents of at least 1A.

Filling the cap, you still need to keep the voltage and current well within their ratings (same formula apply) and the short time you get the regenerative power will not be able to do anything but burp at it softly.

If you want to go with caps, eg. to lighten the load on the battery, when the motor draws a high inrush current starting, you could charge the cap bank with a reasonable low current from the battery, but that's an expensive way of nursing the battery.

I took some measurements on the motor I mentioned and although I had to run it on ~7V (it's a 9v motor), as the scope I used have a low and pretty unforgiving max input voltage of 30V that I wouldn't try exceeding (plus the motor whine is annoying at 20kRPM), but the timing will not be that far off (an inertial load will make the regen somewhat longer).



As can be seen here (click for a larger pic), there's roughly 12µs of 110V to deal with. With your motor running at 22V, the voltage may be 300V to 350mV in a period close to those 12µs (plus what inertia brings), but as long as you don't tell the battery, it'll never know
Even a pulse of 100 times that (i.e. 1.2ms) at 350V won't impress your batteries and will certainly not be dangerous in any way - it's simply too short a time that the impedance of even a flat battery will wart it off like an elephant would a mosquito.

One thing is that the battery won't be hurt from this, another is that the charge you gain from a few brakes here and there wll be negligible and IMO not worth the hassle, unless made from an experimenters perspective on learning.

If you feel you need to add something, a Transil (even a fast 1W zener would do) can be installed in 5 minutes and then you can spend the rest of the week (or how long it was) on something more useful (like practicing Ninja RubberNeckin' Skills^TM, building a table of empty Schotch bottles or whatever your favorite pass-time may be ).

[stevet47]

Hi,

If the experts (you guys) say that going with the capacitor method would be simpler then any safe method of putting the charge into the batteries, than I will peruse this method..... hopefully with some more guidance from you all!

Simplest method is, as I mentioned previously, either nothing at all or a voltage limiter of sorts.
With the motor in question, the net winning is extremely small and doesn't really make up for anything more complicated than a Transil or similar.
(You'd still need rectifier diodes to keep the motor from draining the batteries when the  generated voltage gets lower than the battery voltage of course).

If it was a motor for a full size express train with miles of daily braking, it would be worthwhile sucking every drop of juice out of a regen system, but in your case, you might get a few more meters of running tops.

Any kind of switch mode (SEPIC and Cuk topologies would be more suitable here than Buck/Boost anyway) will give losses that may, in this particular case, be more lossy than a voltage limiter and it would need to be designed to the exact specs (found by testing the motor extensively) to be able to output anything useable at all - lot of work for a marginal net win (if any, compared to a limiter).

Capacitors would need to hold as much of the energy from regeneration as possible and since caps are leaky, this energy needs to be used within a short time.
They need to be able to handle the max. voltage possible and they haven't got the kind of "regulation" that a battery will does.
They need to be able to handle high ripple currents.
The total value of the bank can be calculated as: C = A*s/V
A = Ampere
s = seconds
V = the tolerable voltage drop during s

If you eg. want to be able to get 1A from a capacitor during 2 seconds, with a maximum voltage starting at 25V and still being at least 21V at the end of the 2s period, you'd need a capacity of 1*2/4 = 0.5F = 500,000 µF with a rating of 35V (assuming 25V tops, it needs a margin) and able to handle ripple currents of at least 1A.

Filling the cap, you still need to keep the voltage and current well within their ratings (same formula apply) and the short time you get the regenerative power will not be able to do anything but burp at it softly.

If you want to go with caps, eg. to lighten the load on the battery, when the motor draws a high inrush current starting, you could charge the cap bank with a reasonable low current from the battery, but that's an expensive way of nursing the battery.

I took some measurements on the motor I mentioned and although I had to run it on ~7V (it's a 9v motor), as the scope I used have a low and pretty unforgiving max input voltage of 30V that I wouldn't try exceeding (plus the motor whine is annoying at 20kRPM), but the timing will not be that far off (an inertial load will make the regen somewhat longer).



As can be seen here (click for a larger pic), there's roughly 12µs of 110V to deal with. With your motor running at 22V, the voltage may be 300V to 350mV in a period close to those 12µs (plus what inertia brings), but as long as you don't tell the battery, it'll never know
Even a pulse of 100 times that (i.e. 1.2ms) at 350V won't impress your batteries and will certainly not be dangerous in any way - it's simply too short a time that the impedance of even a flat battery will wart it off like an elephant would a mosquito.

One thing is that the battery won't be hurt from this, another is that the charge you gain from a few brakes here and there wll be negligible and IMO not worth the hassle, unless made from an experimenters perspective on learning.

If you feel you need to add something, a Transil (even a fast 1W zener would do) can be installed in 5 minutes and then you can spend the rest of the week (or how long it was) on something more useful (like practicing Ninja RubberNeckin' Skills^TM, building a table of empty Schotch bottles or whatever your favorite pass-time may be ).

If I am understanding you correctly, you are saying:
1) This is a waste of time... ok, but I've gotta at least try.
2) You do not think I need any type of voltage/current regulation at all, and can use my original schematic? This is surprising to hear, based on the initial comments about how what I was attempting was very dangerous.
3) You think a Transil (that is a clamping diode, correct?), or even a zener would do the job, it I wanted it.

Is my understanding of what you are saying correct?
Safe charging of the Lipo batteries would be in the 11.4-12.6V range at anything under ~12A. So I don't think I have to regulate current, but I thought I would have to regulate voltage to within that range. You don't agree?

I am surprised to hear you say I could use a 1watt zener, I would think at only 1watt I would blow it up, can you explain how you determined 1watt is all I would need?

Also. Thank you so much for all your help on this. Clearly I have no idea what I am doing, and need to start much smaller next time!

[Soeren]

Hi,

Dammit' I just lost a lengthy answer to my router
Well, I got an extra free ride (just trying to squeeze that lemon into lemonade here )

If I am understanding you correctly, you are saying:
1) This is a waste of time... ok, but I've gotta at least try.

You know what Yoda would say about trying, right
It's not a waste of time to learn, if that's the objective, but you have to realize that you won't get much out of it, no matter how intricate a circuit you make.

2) You do not think I need any type of voltage/current regulation at all, and can use my original schematic? This is surprising to hear, based on the initial comments about how what I was attempting was very dangerous.

Are you referring to my initial comments (that was based on your schematic looking like shorting the batteries at first sight), or are you referring to what NewInRobotics said, in which case you'd have to take it up with him - If I thought it would be dangerous, I would have mentioned it
Lithium based batteries can be dangerous if mistreated (overcharge, under-deplenish, physically piercing etc.), but if that's your concern, go for lithium iron batteries (http://en.wikipedia.org/wiki/Lithium_iron_phosphate_battery#Safety)

3) You think a Transil (that is a clamping diode, correct?), or even a zener would do the job, it I wanted it.

Yes, Transil and Tranzorb are trademark names for some types of clamping diodes. There are different ways to make a clamping device, but the main thing is the clamping parameters.

Safe charging of the Lipo batteries would be in the 11.4-12.6V range at anything under ~12A. So I don't think I have to regulate current, but I thought I would have to regulate voltage to within that range. You don't agree?

You are regulating, or rather limiting it by the impedance difference.

However, we have somehow gotten to focus on the some-hundred-volts inductive kick back (must have had my head under the arm, sorry) and this is dealt with in the obligatory diode over the motor terminals (or in case you need the motor to be bidirectional, two back-to-back zeners or a bidirectional Transil).

The 'scope pic shows this short (negative going) peak right before the regenerative voltage (that starts with a short dampened oscillation in the pic).

I'll make another trace, remember to stick a diode on the terminals and get the full regen period in view. Then we have to see what that brings to the table.

And please remember that the motor I use for this is a bit smaller than yours - this is the one I have closest resembling yours.

I am surprised to hear you say I could use a 1watt zener, I would think at only 1watt I would blow it up, can you explain how you determined 1watt is all I would need?

I admit I was thinking of the ~12µs kick back here, but zeners can take a lot - short term.
I have a datasheet in front of me on a 5W device that can handle 200A during an 8.3ms half sine period with a duty cycle of 4 pulses per minute max (i.e. one pulse each 15s).