Using supercaps to support/replace battery pack

[vanilla ice]

Soo, I've got these 15v 1/8 farad capacitors collecting dust in the garage. In a past life they were used for a car stereo system. Can I put them to use again? If so, how do I wire them up? I understand the draw is very high, basically like a short, which can damage batteries. Any way around this?

 

[fechter]

The controller usually has a real wimpy version of these built in. They would go across the battery terminals of the controller. You would need to put several in series to have enough voltage rating for the batteries you are using.

There is an extremely high current when the connection is made to the batteries. This won't hurt the batteries, but could blow any fuses in line or kinda burn up the contacts where you make the connection.

The way around this is to precharge the capacitors with a resistor before making the connection. I use a precharging circuit on my scooter and a large relay to make the connection.

I don't know how much improvement it would make. Probably not much. In theory, they should help improve the efficiency of the controller at partial throttle and possibly reduce the peak of any voltage spikes in the system. It would make more of a difference in a very high power system, just like for car stereos.

 

[The7]

I've got these 15v 1/8 farad capacitors

Wow. 1/8F = 125000 uF
A very large capacitor.

 

[vanilla ice]

I think I have like six of these left. Yes they are quite large for caps, but surprisingly light weight. Even in my car stereo days I was frugal (cheap), so I bought 8 electronic surplus 1/8 farad caps and wired them in parallel rather than use one of the commercially available car audio 1 farad caps.

The resistor/relay thing is a bit over my head. Could you go in to more detail on this? Or would I be ok just pre-charging each with a 12v battery charger or something? How fast do capacitors discharge? Will they discharge down to nothing over night when I turn off my bike?

Even if they don't do much I may as well give them a new lease on life. They are so light, and I *AM* at partial throttle most of the time... I have not bumped up the voltage yet, so I can do 2s2p wiring on the caps for now. Mebbe 3s2p when I try for 36!

 

[fechter]

The resistor/relay thing is a bit over my head. Could you go in to more detail on this? Or would I be ok just pre-charging each with a 12v battery charger or something? How fast do capacitors discharge? Will they discharge down to nothing over night when I turn off my bike?

The capacitors will probably discharge to near zero overnight.

Below is a basic precharging setup. With those big caps, it might take 30 seconds or so to charge up. The value of the resistor is not really critical.

It would be interesting to see if you can notice any difference in performance.

 

[vanilla ice]

Ok I think I get what it does. For storage with no battery drain- I'm guessing I would need an additional switch between the resistor and battery?

Appreciate the diagram and help.

 

[fechter]

Yes, for long term storage, you'd want to completely disconnect the battery.
The drain from the controller is mostly leakage current in the capacitors which is very low, so you could leave it connected for like a week with no problem.

 

[bobmcree]

a 1 farad capacitor will discharge 1 amp for 1 second so a 1/8 will do 1/8 of that, and you will need to put several 15v caps in series to get enough voltage, so that will cut the capacitance by the number in series. let's say you put 4 in series to stand off 60v, the least you would want to use on a 48v system, you would end up with 33,000 uf. this is enough to discharge 10 amps for about .003 seconds. you will not see any significant reduction in voltage drop when you hit the throttle, except for the first few milliseconds. the caps just cannot store enough energy to run the motor for any significant length of time.

the caps in the controller are necessary to maintain a stable dc voltage for the fets which switch very quickly, and they are kept charged by a low impedance path to the battery. a precharge resistor will cut down on the inrush of current that occurs when you connect the battery, and will mostly eliminate the spark you see, which will save wear and tear on a power switch if you have one in the battery wire. in the controllers most of us use the caps are kept charged all the time and the switch controls a transistor circuit that provides power to the controller logic circuits.

supercapacitors that have 10 farads or so would provide some real output current, but only for seconds at a time. they are just not a good substitute for a better battery.

in a car stereo the big capacitor helps provide the peak current for transients that require much more than the base line power, so the headlights will not dim on the bass notes, but these peak demands last much less than a second. most people these days use at least a 1 farad 20v cap, and one of those caps is bigger than a beer can. even a sixpack of them would make little difference in performance on a typical ebike.

the big caps on a car stereo power line also serve to protect the battery from the high peak current, which can shorten battery life by eating chunks out of the plates. the cap lets the current come out of the battery more slowly so the chemical reaction can be maintained at a safer level

 

[Dalecv]

This post made me go review some electronic theory.

I always knew to buy capacitors rated at or above the needed voltage but didn't recall that putting capacitors in series to match the voltage would decrease the capacitance.

 

[vanilla ice]

So I guess the answer to the thread title is "Why yes, yes they are useless." As I thought, but wanted to check first. Here they are any how, managed to find four in the garage tonight.

[/url]

 

[xyster]

So I guess the answer to the thread title is "no." As I thought, but wanted to check first. Here they are any how, managed to find four in the garage tonight.

Wow. Be careful not to drink from the wrong one!

 

[Dalecv]

Well the blue can would probably taste like a 9 volt battery for about .03 seconds and then you would speak with an Australian accent and pedal your electric bike at over 120 rpm. :lol:

 

[fechter]

The capacitors built into some controllers are inadequate, so an external one would help reduce heating and losses in the capacitors. But those are sort of overkill for that application. They look cool, though.
Funny, I think I have some that look exactly like that.

 

[The7]

Do you mean these used on my mono block MOSFET amp?

 

[x88x]

So, the Audi thread got me thinking along these lines, but it was getting a bit far OT and I wanted to explore it further so I figured I'd branch it off on its own. This post mostly explains where I was up to now, which I'll sum up below.

Basically, I started looking at supercaps as a means of getting a large chunk of energy into a vehicle in a very short amount of time, then distributing that energy into the battery pack. I ran into two problems. First, the supercap I was looking at (the Maxwell BCAP0310 P270 T10) maxes out at 30A continuous current, though it can take 240A for 1s..so it might be able to be charged in 1S bursts of 240A..but I'll get into that later. The second problem I ran into was that to do a 1:1 dump, you would have to have a capacitor bank the same voltage and capacity of your battery bank...and with the supercaps I was looking at coming out about the same cost as the battery bank...

In the other thread I got started going down one line of thought, and then as I was thinking it over more today I got another one:

Idea 1: Using supercap bank as charger middle-man for battery pack.
The idea here is that since fast-charging (10C+) most if not all batteries tends to shorten their lifetime, have a bank of supercaps that a fast-charger would fill, then the supercaps could dump their charge at a slower rate into a battery pack. Along this train of thought, it occured to me to find out how many 240A pulses it would take to fill the caps I was looking at earlier. The cap specs were as follows:
Maxwell BCAP0310 P270 T10
2.7V
310F
2.2mOhm ESR
>500,000 duty cycles
10 year lifetime
Electrical charge capacity: 6.975Ah
Max continuous current: 30A
Max peak current, 1s: 240A
$19.71 each (decreases if you buy more)

The energy capacity of each cap turns out to be 1129.95J. Now, a 240A 2.7V 1s pulse would be 648J, so if 240A could be maintained as the ESR adjusts, then the cap could be filled in two 1s pulses. The question would then be how much time would be required between these pulses in order to maintain the healthy operation of the cap bank.

This then led me to figure out what would need to be done to build such a bank for a battery bank of a given size. Now, I knew the laws for capacitance scaling, but I had never thought to look into electrical charge capacity scaling with capacitors. As it turns out, it works out the same as it does with batteries, which is nice.
More detailed explanation:
x = # of caps in series
y = # of caps in parallel
z = Ah capacity of bank
C_b = capacitance of each cap
V_b = voltage capacity of each cap

z = (C_b * V_b * y) / 120

I got this by first finding the energy capacity of the cap in Joules (.5*C*V^2), and since 1J = 1Ws, you just have to divide by the energy capacity by the voltage and 60 (to get hours), and you're left with Ah. Then, if all the caps are identical, C scales nicely as (y/x) * C_b. x ends up dropping out since when calculating the entire array you use the total voltage (V_b * x).

Unfortunately, even with almost 7Ah per cap, it would cost $1,330.84 + shipping + hazmat charges to build a bank (27S3P, 5kg) what would be able to do a 1:1 dump into a 72V 20Ah battery pack. For comparison, such a pack built with Turnigy Nano-tech's (2S4P of 10S 5Ah packs, 11.5kg) would be $1,289.92, or for a more similar discharge ability (only 4C with the 30A max), a couple Pings (2S1P of 36V 20Ah packs, 15kg) would be $940.

This led me to the second train of thought:

Idea 2: Using a capacitor bank instead of batteries
At this point I started wondering... If I can get the same charge capacity and at least as good energy density out of a capacitor bank for a price in the same ballpark as a battery pack...why bother with batteries? Obviously if I need more than 4C discharge, right? Well, that led me to look at some of Maxwell's other offerings, and I found this:
BCAP3000 P270 T05
2.7V
3,000F
0.29 mOhm ESR DC
5.2mA Ic (leakage after 72 hours)
10,935J
67.5Ah
The rated Isc (short circuit current) of 4,800A doesn't match with their ESR though..hmmm..maybe 0.5625 mOhm is the ESR at peak capacity?
Rated to 'more than' 1,000,000 charge cycles, with a lifetime of 10 years, with an estimated max 30% drop in capacitance and max 150% rise in ESR.
No max current is listed in the mfg data sheet (PDF), but since they specifically state that these were designed to be used in "hybrid vehicle drive trains, automotive subsystems, and other heavy duty applications", I would guess that it is quite high.
$95 each (orders less than 5)

Not only is this the largest capacity 2.7V supercap that Maxwell makes, it is also the best priced one, at 82-86 cents/kJ (price drops if you buy 5-9 or 10+). For contrast, the 310F supercap I referenced earlier is priced at $1.74-$1.31/kJ depending on quantity.

Now, I'm sure you noticed that this has a huge electrical charge capacity of 67.5Ah. This places it more in line for a car than an ebike or really even motorcycle...and unfortunately the lowest capacity supercaps in this line (2.7V, 1,500F, 33.75Ah) are $75.15, so while it's nice that you can double your capacity for just 25% more, it does rather place the barrier of entry pretty high (ok, so they do have a 650F model, but it's only $3 cheaper per cell than the 1,500F model so in this application I really don't see the point).

So, with that in mind, I threw together some numbers for a 111V 67.5Ah (or closest I could come) pack with these, Turnigy Nano-tech's, and cell_man A123 26650's. Before you ask, no, I picked 100V out of the air and 111V was the easiest to match with the Turnigy packs.

Maxwell:
41S1P
110.7V
67.5Ah
41 @ $89.24 each
$3,658.84
22.55kg

Turnigy 10S 4Ah (the 4Ah packs ended up being ~$160 cheaper and only 459g heavier than the 5Ah packs):
3S17P
111V
68Ah
51 @ $129.62 each
$6,610.62
68.289kg

A123 26650:
34S30P
112.2V
69Ah
1,020 @ $6
$6,240
71.4kg (cells only)

So, from just this analysis, it would seem that the supercaps are a better way to go if you have something this scale...but as with anything, I assume it's not that simple. So, my question to you who have more experience in this (ie, most if not all of you ), is this. Has anyone done anything along either of these lines? Do either of these ideas sound feasible? ...or am I missing something that makes this whole discussion ridiculous?

 

[webfootguy]

Two thoughts.

One, the biggest problem I have seen with Supercaps as battery replacements is that the output voltage drops from the max voltage to zero along flat line. After half of the capacity is used up, the voltage is cut in half (lithium batteries do not behave this way). This makes some kind of DC to DC converter a requirement and thus efficiency losses and some loss of capacity as the input gets too low to generate any power.

Two, there are surplus supercaps available out there. Explore these links:

http://www.goldmine-elec-products.com/prodinfo.asp?number=G17930
http://www.goldmine-elec-products.com/prodinfo.asp?number=G17933
http://www.goldmine-elec-products.com/prodinfo.asp?number=G17934
http://www.goldmine-elec-products.com/prodinfo.asp?number=G17960

Hint: search for boostcap in the search box.

I think you will like the prices. Supply is probably limited but you could buy enough to play with.

 

[Samba]

I got this by first finding the energy capacity of the cap in Joules (.5*C*V^2), and since 1J = 1Ws, you just have to divide by the energy capacity by the voltage and 60 (to get hours), and you're left with Ah.

seconds over 60 = minutes. seconds over 3600 = hours.

 

[x88x]

I got this by first finding the energy capacity of the cap in Joules (.5*C*V^2), and since 1J = 1Ws, you just have to divide by the energy capacity by the voltage and 60 (to get hours), and you're left with Ah.

seconds over 60 = minutes. seconds over 3600 = hours.

Doh! ..whoops...I guess it was the third option then.

EDIT:
Also, thanks webfoot. I hadn't thought to look on any surplus sites. At those prices I think I'll still pick up a couple to play with...and maybe grab another one of their surprise boxes while I'm at it.

 

[liveforphysics]

41x 3000F caps in series makes a 110v 73F cap.

W= 0.5C*(V^2)

This is is 438KJ

Which is 121watt-hours.

So you've got a huge bulky $$$ thing that carries less energy than this single little LiPo pack:

http://www.hobbycity.com/hobbyking/store/uh_viewItem.asp?idProduct=9935

If you picked your favorite criteria to value, the same volume, the same weight, or the same cost, and used that same amount of criteria towards LiPo, you end up with drastically better charge/discharge rates, and land yourself roughly >20-30x the energy storage...

AND, to top it off, the voltage in the cap rapidly reaches a uselessly low voltage level as you begin using the capacity, so good luck getting much of that measly 121watt-hours out of it unless you've got it feeding a wide-range-input DC/DC converter to feed the controller (which makes things rapidly become more absurd).


Ri dropped too low on LiPo for supercaps to have a place in EVs. Unless EEStor finally releases a high-voltage supercap, then the entire EV battery market is eliminated overnight.

 

[grindz145]

Unless EEStor finally releases a high-voltage supercap, then the entire EV battery market is eliminated overnight.

Vaporware money grabbing scheme. My guess is that we'll never see this.

 

[liveforphysics]

Unless EEStor finally releases a high-voltage supercap, then the entire EV battery market is eliminated overnight.

Vaporware money grabbing scheme. My guess is that we'll never see this.

I can still dream, and the physics work (if they do have the claimed insulation values on the particles).

 

[cancerkazoo]

What about a couple of them in combination with LiPo to help hold the "hot off the charger voltage", or let someone regen on a fresh battery sooner, or just soak up the regen faster and lessen stress on the battery?

If you put a couple in series with the batt would that help? or is a little bit or circuitry needed to make it work?

 

[liveforphysics]

What about a couple of them in combination with LiPo to help hold the "hot off the charger voltage", or let someone regen on a fresh battery sooner, or just soak up the regen faster and lessen stress on the battery?

If you put a couple in series with the batt would that help? or is a little bit or circuitry needed to make it work?

It's weird to think of a capacitor being slower than a battery, but say you want to make a 16Ah 100v hot-off-charger pack, and you're using the 4x 4Ah Turnigy NanoTech cells to do it. Each cell is 0.8mOhm, so 4p in 0.2mOhm. 24s x 0.2mOhm = 4.8mOhm total pack Ri.

You make a 100v (hot off charger) string of the 3000F caps at 0.29mOhm each, and you end up with a 37s string, making a 10.73mOhm string. Meaning, this physically huge and many times heavier than the LiPo pack string of cap's paralleled is bringing pack resistance from 4.8mOhm to 3.31mOhm.

So, how much difference in energy does this giant heavy $3,6000 string of caps add to your pack? You go from 1420watt-hours to 1528watt-hours (the caps adding 108w-hrs) IF IF IF you were to discharge the pack down to zero volts... which would be an epic fail. Using the range of the batteries HVC/LVC, you're actually adding about 45 watt-hours to your pack for that $3,600 and massive amount of size and weight added.

So, what about discharge/charge differences? Well, the current rate on the caps is the current delivered if charged fully up to 2.7v, and shorted down to zero. The current that flows then is a large number, however, at zero volts and a thousand amps, you've still go zero watts of useful power. Unlike the misleading datasheets, the reality is, like any component they can only discharge as much useful current/power as the resistance * voltage difference enables them to.
In this example, if drawing 100amps from the charged 100v LiPo pack with no super caps, the voltage drops from 100v to 99.52v. Add in the giant heavy string of supercaps, and your voltage now goes from 99.69v.
Drawing 500amps from the same setup with no supercaps, voltage goes from 100v to 97.6v. Add in the super caps, it goes from 100v to 98.45v.
But don't expect it to keep that fraction of a volt of voltage sag up for long, that 37s string of 3000F caps only has 81F of capacity, so under DC load situations, literally in under 1 second at 500amps the load is entirely on the cells again, and the caps are just there to weight the machine down and use money and volume that could have been spent on battery.


Before modern LiPo (nanotechs and a couple other premium types), super-caps could have had an application in something like a drag racing EV type situation where the pathetic energy storage capacity wasn't a concern. After modern LiPo, it has zero applications left for any aspect of EV use. You can simply charge and discharge faster by using that space/weight/volume to add LiPo than ANY brand of cap available at any price.

Capacitor energy storage for EV's = brain-fail.

 

[x88x]

Yes, yes, I missed a variable, so my calculations were off by a factor of 60..moving on.

If someone were to come out with high voltage caps with similarly low ESR (that weren't stupidly expensive), I still think a charge/discharge middle-man implementation would be useful, both as I suggested in the OP and as is suggested here, because unlike any batteries currently on the market, these supercaps can take over a million charge cycles...more than 10x the highest battery claim I've seen (100,000 for A123 prismatics). The cost would have to be at least 10x lower than that 125V Maxwell for it to be useful at all, but having those absurd current levels available for charging would be nice (I=V/R, for the 125V Maxwell would be a max of almost 7kA). ...not $10,000 nice, but maybe $1,000 if you're talking about a $5,000+ battery pack. It would also be nice for applications where you need an absurd amount of torque for a very short amount of time.

With current tech though, I agree. High-C lithiums would be a better choice, even if they did need to be replaced much more frequently. 125V Maxwell == 1Ah, so a 125V 4Ah bank would cost $41,600 vs $442.33 for 45C Turnigy Nano-techs, would weigh 238kg vs 4.585kg for the Nano-techs, and would take up 85.82L vs 2L for the Nano-techs (ignoring mounting/etc).

 

[liveforphysics]

Even that 10,000usd 63F 125v cap isn't going to make a noticable difference in charge or discharge rates on the battery.

The above described LiPo pack can all ready be charged at 14,200watts continously.

That capacitor pack could only be charged at 7KA while the capacitors are at 0V and connected to an zero resistance 125v source... The extreme charge rate you get is only for the low potental time of the cap charge cycle, which isn't even where it holds useful energy storage, and as it reaches a useful energy storage potential, the rate of charge it can take slows to roughly 1/4th the rate the LiPo pack could be taking it directly.