I'm building a regenerative braking circuit for a bicycle generator hub (AC, .6A max, 30V open circuit, ~3W). The higher power of my two power circuits for this, starts out with a bridge rectifier, then a smoothing cap, then goes into 30V TVS diode, then into a small super capacitor bank in series. I have six 5.5V, 1.5f supercapacitors that I want to be able to charge to the 30V my TVS diode will allow.

The best circuit I've seen for balancing this bank is this one, at instructables:


However, I'm wondering if I couldn't do the same thing with some 5W (or less) zener diodes. The max voltage on a given supercapacitor is 5.5V, so I figure a 5.1V zener will protect that as far as voltage. My current at this step really can't go above 600mA, so I figure my maximum power is 3W, the stated power of the generator anyway. I'd like to avoid the complication of the comparator, MOSFETs, etc. (I'm a software jockey), but I don't know what I don't know about using zeners like this to balance this supercapacitor bank. Can anyone tell me why I don't want to do that?

 

You could probably get away with that... but it depends a lot on your array, the number of caps, and their voltage, etc. The purpose of the opamps are so that when the caps start reaching a certain voltage, they stop charging each individual cap, shunting the current to the resistor instead. The thing you'll be most concerned with is charging not over-charging each capacitor. To keep from over charging your caps you also want to put bleed resistors on your caps so that the voltages stay somewhat balanced between caps - which might defeat the purpose of storing energy in the caps, but I bet you can find the 'sweet spot' though with some trial and error.

I'd think you'd want to hit about 27V or so on your TVS. You're solar array will start putting out less current at about that voltage.

Maybe a 5.1V Zener across each cap will be better. Actually, that's really the best solution now that I think about it... Zeners have a pretty mellow knee and they are also stable with temperature at that voltage.

Something like this will work nicely:
http://www.digikey.com/product-detail/en/1N5338BG/1N5338BGOS-ND/1474078

 

Something like this:

Also, is your SA really 600mA? If it is then you're dealing with a 16W solar array... and each of the 5Vzeners will dissipate about 3W each, worst case. Best to go with a 5W part.

Anyway... this is what I'd try if I was going to simplify that design.

 

Tindel - not solar. That was just a nice example circuit I found, with some similar characteristics. I'm working with an AC bike hub, but it's also 3W, rated at 500mA, but the actual gets slightly higher than that at high speeds. And the open circuit voltage goes to ~24VAC, so ~33VDC after the rectifier. I'm planning to keep a small load on it all the time to keep the voltage below 30VDC when I kick on the braking circuit.

Yes! That's exactly the diode I bought. On-Semi 5.1V 5W. And that is about the circuit I was thinking of. I really need to learn to use Spice. Thanks!

 

AC bike hub? I'm guessing some sort of inductive coupling coil attached to the wheel and the wheel shaft working similar to a bike speedometer.

33V @ 0.5A = 16.5W... much more than 3W... I'm confused.

 

AC bike hub? I'm guessing some sort of inductive coupling coil attached to the wheel and the wheel shaft working similar to a bike speedometer.

33V @ 0.5A = 16.5W... much more than 3W... I'm confused.

33V open circuit. With a load it's more like 6-10V. The generator is rated at 3W, which it produces around 10mph. Max is somewhere around 5W, supposedly. Here's a chart:


Voltage is AC, velocity is km/h. There's a separate chart that shows the amps - quickly rises to 500mA and barely budges from there. Supposedly you can pretty easily blow the standard 3W bulbs if you don't have some kind of voltage regulator in your lights and you get above 20mph or so. But the generator magnetically saturates (or something - there are arguments about the actual effect) above that anyway, so they don't increase wattage forever.

I'm going to put a small load on it all the time to power a nano and display and do a tiny bit of battery charging. Then when I hit either the regular brakes or a drag brake, it will turn on the full power circuit, including those supercaps, another bank of supercaps with a lot more capacity but limited to 5V, and a bunch of battery charging circuits (ni-mh and some li-on).

 

Got it... so it does act a bit like a solar array... at light load, the voltage is high, at heavy load the voltage goes down.... Yeah - I think this should work since your mostly light load. Your breaks might not work too great when you first start your ride... that's concerning... you may need to precharge your caps before you go... Although they'll probably stay charged if you're using a uC with a low power mode - such as the MSP430.

Do you have a mechanical backup at least until you've verified the circuitry works - god forbid you have a circuit failure! Yikes!

 

Do you have a mechanical backup at least until you've verified the circuitry works - god forbid you have a circuit failure! Yikes!

Oh, sure. We'll definitely need them - we're doing loaded touring for a month. 3-5W is nothing for braking. I'll have the usual rim brakes, which are good for regular braking, and a drum brake on the front (the hub is a combination generator / drum brake). The drum brake is a drag brake for downhills. You don't want to do drag braking with rim brakes - you can pretty easily heat the rims enough to blow out the tires. So on the downhills I'll have the electric drag brake, the drum drag brake, *and* the two rim brakes. You have to set the drum brake lightly, use the electric drag as much as it can handle, and pulse the rim brakes - slow down a few miles per hour, release them and speed back up, hit the brakes again, etc.

I'm hoping to just pull enough power out of the hub generator to keep all of our batteries charged for the trip and help with the braking a little.

 

The Zeners look like a simple, straightforward solution. Did you say your supercapacitors are 5.5 volts max? If those are the type that are made to maintain CMOS memory, they might have a high impedance, and might be hard to charge with a quick burst of energy from regenerative braking. I don't know for sure, but it's worth checking.

The first schematic with the shunt regulator resembles something that I was originally going to use for the solar bicycle. However, I disliked that shunt regulators equalize voltage by simply burning up excess energy. This is severe overkill for your application, but the most recent scheme is this

and the writeup is here.

Oh also - did you do a writeup for your bicycle anywhere? It sounds fascinating; I'd love to see it.

 

Roderick - Alright, that bike is awesome. I also love the part that the woman in the electric car next to your bike has a huge grin on her face looking at it (10/25/14 post). It's great when we can make people happy and do good!

Yes, I will do a write-up at some point, but it seems too early for that, and I haven't figured out Eagle or anything enough to do a decent schematic. I can do database diagrams all day, but I can't find half the parts I need in Eagle, and don't have that 'draw-new-parts' thing worked out.

Here's the closest I can get to specs for these supercaps.

They seem to work OK so far, but I've only tested them individually so far.

 

Nice. 10 milliohm ESR, and if I'm reading right from the spec, a discharge rate of 1.5A, implying that you can charge at about that rate, too. Good luck on your energy recovery, sounds like it should work.

 

I am noobi trying to find a simple way to do balance my Caps but I just have 2 in seires 2.7v 100F caps that im charging with basic USB power or 5v 700mA to 1A charging current I am trying to light some LEDs for as long as I can very basic but I am new . I have seen the circuit that kjmclark posted and kinda get it but its seems like alot to me. tindel I like your way seems very easy and simple could I use the same for 2.7v caps? Roderick Young I like what you posted but am have seen that page write up but I just dont get how I would set it up for what I am doing. I have tried passive balancing with resisters but I was not sure what resister to put in parallel with them and just tried a 100ohm did not work and a 11kohm same did not work one cap would still charge hirgher then 2.7v. Im just trying to charge the 2 caps as fast as I can with the given supply and protect them for long term usage if any one can help that would be great.

 

The advantage of the circuit in post#1 is that it has a sharp and clearly defined clamping threshold. The zener diode solution may be simple but performs very poorly in comparison; the charge on the capacitors will start to leak away well before they reach full charge.

 

I am noobi trying to find a simple way to do balance my Caps but I just have 2 in seires 2.7v 100F caps that im charging with basic USB power or 5v 700mA to 1A charging current

The bigger problem is the charge rate, I think. If you limit the charge rate, you could do the simple zener diode trick I ended up using, but if you could be pulling a full amp, that might not work. Or rather, it works out to 5W (5V, 1A), which those zeners were rated for, but it's a little risky using them for their full rating. Also, unless you put some kind current limit on the charging side, you might blow that power supply, since caps will pull lots of current at first.
I'm only a little more than a noob, but it seems like you want a (V=IR -> 5V=1AR -> 5Ω=R) at least 5Ω resistor on the input side to keep from drawing more than 1A from your power supply. If you choose a bigger resistor, and drop your current down to like 800mA (so 5V=.8AR -> 6Ω=R) you can protect your power supply better and drop the current into the range you could use the cheap zener method, but it will take a little longer to charge the capacitors.

Blocco's right about the max charge though. If you want to store as much charge in the capacitors as you can, the zeners aren't great for that, since they start to leak current below their rated voltage. It's not a sharp limit like the original comparator/MOSFET solution. I think that one seems really complicated because we're noobs - to people who work with these things often, that probably looks like a pretty simple circuit. The components aren't that expensive, it's just more work than us novices are used to.

So, fastest / fullest charge would be a current limiting resistor to protect your power supply, and the circuit at the top. Not as fast/not as full, but simpler is to use the zener balancing, as long as you drop the current down to a safer range for a zener (for what I'm doing, I'm not that worried about maximizing my energy storage). The zener concept would go out the window if you were using more current, because it would get too expensive to try and find a zener at that wattage, even if you wanted to.

 

Hi, Stephen. You can ignore the circuit I posted previously. That was just posted for fun - it requires a lot of ground-up design to make it work. Possibly what you want is something like the first circuit kjmclark posted. This was the way I did it:

It's actually a little simpler than it looks. R9 and R10 are just placeholders that aren't used. Two supercaps C5 and C6 are shown, but you only need one. U3 is an inexpensive, rail-to-rail, low voltage op-amp. My circuit uses a Darlington instead of MOSFETs. I had trouble generating enough gate drive voltage to turn on a MOSFET, given the 2.5 volt power supply. Since the Darlington transistor is in linear mode anyway burning power, there is no efficiency to be gained by using a MOSFET. There circuits would be stacked on on top of the next, for as many supercaps as you would put in series. The writeup is here.

The trouble with using only resistors to balance the supercaps is that if you want good balance, the resistors have to be small, and that leads to burning a lot of power, and also the capacitors will discharge themselves if disconnected from the charger, or if you use a charging source that is weak.

 

Wow you guys are AWESOME Thanks Roderick Young , kjmclark and blocco a spirale I love this forum.

It looks like I am go to need to use an active balance for what I am trying to do and I have some questions but just to let you guys know or anyone else who is reading I did try the Diode example and found that it worked but also did not work I mean I tried it with a 2.7v Zener with 5 to 5.4 vlolt 1 amp supply using a CC CV charger from china I got on Aliexress that I have been using to test with on these 2 Caps not knowing to balance them and finally notice that 1 cap was at like 3.5v this may have happened several times with these 2 caps so when I tried using the diode balance I watched the faster charging Cap rise and when it got to close to 2.7v it did stop flutter for a few seconds but then keeped going while the other was still reaching 2v so it did not work BUT is suspect that this Cap was probably damaged do to previous using ( any thoughts ?) because I tried it with 2 more new caps and they worked fine when the faster charging Cap got to the 2.7v range it stooped then even went down but just a .02 volts. I saw a Youtube video with some one who used 3 diode in series on each cap but did not try it (link below) the drain did not last as long as without the diodes on with 4 LED with 100ohm resisters on each so this circuit will not work for me I need LED to last as long as possible.

So for active balancing in your circuit Roderick Young so I understand better I don't need R9 and R10 or are do I use 2 valuse so that the drop between them is my limit voltage for the Op Amp to see? and is R12 a variable resister or how would I connect it as a normal 10k resister? sorry if these questions are dumb I am learning. Here are some otheres I have found I anybody know witch would be the cheapest, simplest but yet good protection or balance please let me know

This has eagle Links here
http://www.insideelectronics.co.uk/ultracapacitor-charge-balancing/
http://www.insideelectronics.co.uk/wp-content/plugins/download-monitor/download.php?id=3
It seems like a much over kill for my 100F but can I tone it dow to make it work for me??

Could anyone help me find the values for all there components???

 

If you only have two capacitors, that circuit you showed with just one op amp in the middle looks good. It will only balance gently, but that may be enough.

In my circuit, it's not necessary to use a 10k potentiometer. It would probably be better to use 1 M in your case, to cut down on power dissipation. Once you tune the potentiometer to get your desired output voltage (in my case, it was 2.5 volts), you can pull the potentiometer out, measure the resistances across the terminals with an ohmmeter, and substitute two fixed resistors. The only reason you would use my circuit, which is more complicated than the single op-amp circuit, is if you have to handle huge surge currents. The application that I had might charge with a large pulse of current, maybe 10 A for a few seconds, and it was important to have a beefy circuit to prevent the capacitors from getting too much voltage.

Also remember that the point of all this is to avoid overvoltage on the capacitors. It's not strictly necessary that you have 2.5 volts on each one if the output is 5 volts. You could have 2.4 volts on one, and 2.6 volts on the other, if the capacitors will tolerate that. The voltage balancer only needs to be that good, not perfect.

 

If you only have two capacitors, that circuit you showed with just one op amp in the middle looks good. It will only balance gently, but that may be enough.

In my circuit, it's not necessary to use a 10k potentiometer. It would probably be better to use 1 M in your case, to cut down on power dissipation. Once you tune the potentiometer to get your desired output voltage (in my case, it was 2.5 volts), you can pull the potentiometer out, measure the resistances across the terminals with an ohmmeter, and substitute two fixed resistors. The only reason you would use my circuit, which is more complicated than the single op-amp circuit, is if you have to handle huge surge currents. The application that I had might charge with a large pulse of current, maybe 10 A for a few seconds, and it was important to have a beefy circuit to prevent the capacitors from getting too much voltage.

Also remember that the point of all this is to avoid overvoltage on the capacitors. It's not strictly necessary that you have 2.5 volts on each one if the output is 5 volts. You could have 2.4 volts on one, and 2.6 volts on the other, if the capacitors will tolerate that. The voltage balancer only needs to be that good, not perfect.

Hi, Stephen. You can ignore the circuit I posted previously. That was just posted for fun - it requires a lot of ground-up design to make it work. Possibly what you want is something like the first circuit kjmclark posted. This was the way I did it:
View attachment 82798
It's actually a little simpler than it looks. R9 and R10 are just placeholders that aren't used. Two supercaps C5 and C6 are shown, but you only need one. U3 is an inexpensive, rail-to-rail, low voltage op-amp. My circuit uses a Darlington instead of MOSFETs. I had trouble generating enough gate drive voltage to turn on a MOSFET, given the 2.5 volt power supply. Since the Darlington transistor is in linear mode anyway burning power, there is no efficiency to be gained by using a MOSFET. There circuits would be stacked on on top of the next, for as many supercaps as you would put in series. The writeup is here.

The trouble with using only resistors to balance the supercaps is that if you want good balance, the resistors have to be small, and that leads to burning a lot of power, and also the capacitors will discharge themselves if disconnected from the charger, or if you use a charging source that is weak.

I am trying your circut I have found it to be the best all around but I have an LM393 dual op amp C2328a transistor I tried it with Vr using 2 10k for divider ref and just connecting the + side op amp to the middle of the caps and found that I needed to connect the output to VCC with 10k R to turn on the transistor I used these models

this was a basic Ideal but I found out its not that simple to set up and in there sch they used something just like what you have here.
But with these I found that my 5v 700ma supply was now only charging at 250ma do you know how I could increase that so the caps would charge faster?

 

I am trying your circut I have found it to be the best all around but I have an LM393 dual op amp C2328a transistor I tried it with Vr using 2 10k for divider ref and just connecting the + side op amp to the middle of the caps and found that I needed to connect the output to VCC with 10k R to turn on the transistor I used these models...

...this was a basic Ideal but I found out its not that simple to set up and in there sch they used something just like what you have here.
But with these I found that my 5v 700ma supply was now only charging at 250ma do you know how I could increase that so the caps would charge faster?

I'm not exactly sure which circuit you're referring to. If you mean the Blueshift one, note that the + and - inputs are reversed for the top and bottom op amps. If you have them connected the same way, you could get both T1 and T2 conducting at the same time, drawing heavy current.

Any of these circuits draw some power during charging, but I would suggest that the voltage limiter type of circuit in post #15 may draw the least, until the capacitors reach their target voltage. I think it's important to keep the big picture in mind - usually, when I charge supercapacitors, my concern is not to exceed the working voltage, not to make sure that every capacitor has exactly the same voltage. The voltage balancer type circuits will draw power while you charge, in an attempt to keep the voltage equal across both capacitors. And the voltage balancer doesn't actually protect from overvoltage, so it's important to make sure that your input supply can never go over twice the capacitor working voltage.