The battery game

[duivendyk]

Paralleling dissimilar batteries of significantly different capacities is a bad idea,and does not accomplish all that much either,they ought to have at least approximately equal capacities
Monitoring the output voltage of such a parallel combination does not tell you all that much about the current division between them.Looking at the voltage changes (under load !)that is A, B, A+B gives some info.The individual current monitoring using small resistors in series with the batt. tells you a lot more.
The homogenizing scheme of a 12V SLA + 12V NMH (or Nicads) in series and two of these in parallel with the crossover switch to turn them into two separate 24 V SLA/24V NMH batteries has a lot going for it,by sidestepping the crossfeeding& loadsharing problem.
Nicad batteries have a memory effect,if not completely discharged before recharging their capacity tends to decrease.
Low speed hill climbing is the pits for these kind of motors,high torque at low speed means poor eff..But total power available (rider+ motor) will limit motor speed.A lot depends on how steep the incline is.
I have derived a useful expression relating the power P (in Watts) to climb a hill to the weight (W in kg), the incline I (in %) and the speed S (in km/hr).Here goes:
P= 0.027xWxIxS Watt.For instance W= 130kg ,I= 4% and S= 10 km/hr ,P=140 Watt
This does not include the power needed to overcome the rolling&aerodynamic drag,prob. another 25-30W?

 

[Fuzzo]

What's your recharger solution? Dui also suggested that you had some experience mixing battery chemistries. How did that go?

I take it to find bad cells you can firstly measure each output to find the weak one, then just slide the batteries out of that tube to test each one?

Where does any vented gas go?

 

[Fuzzo]

Dui, your series scheme is worth trying, particularly from the safety point of view, although the complications of charging will need to be solved. You already suggested a double-pole-double throw could switch from parallel to series - seems to me like it's worth trying, and goes to one of my key aims - incremental pack building. Am I correct in assuming that I could add as few as 10 1.2 volt batteries in series to each 12V SLA and not risk overdrawing from the little batteries? If so, that's ideal - then I can add more 10-battery packs as I see fit, in parallel with each other. Crossfeed and balance will be less of a problem because they are the same spec and chemistry. They'll still be in series with the SLAs.

Your power equation surprises me. It is linear on speed, so going half the speed should draw half the watts. Which suggests it doesn't matter what speed you are going, the efficiency will be the same - half the watts over twice the time is the same total energy used. Extrapolating a little further, it would actually be most efficient to go slowly, as the drag factors are at a minimum, and the Peukert effect is less.

Have you left something out?

 

[safe]

Nicad batteries have a memory effect,if not completely discharged before recharging their capacity tends to decrease.

Did you know that's a myth? (obviously not)

Apparently NASA sent up a satellite with some NiCads and had them set up in such a way that they discharged and recharged in a pattern that was identical from day to day. It was the regular routine that caused the problem. Ebikes tend to use their energy in unpredictable ways... you can never be sure of exactly how much energy is going to be used and how fast.

I've found with my NiCads that at first they tend to behave rather sluggishly until I start to use them more often. The harder you push them the more they like it... NiCads like hard abuse.

NiCad's only get a memory effect if you do the opposite and use them gently and predictably.

Treat them like a woman... they like a rough ride now and again or they bored...

-----------------------------------

http://www.repairfaq.org/ELE/F_NiCd_Battery.html#NICDBATTERY_014

2.9) Does the memory effect exist?

<Flame shields on> YES

Just as everyone is running around and saying that the memory effect is a myth, here I am, saying that it is true. OK, so, why is this? First of all, the term memory effect is quite unscientific. People tend to attribute any failure of a NiCd to memory.

Let us define memory as the phenomenon where the discharge voltage for a given load is lower than it should be. This can give the appearance of a lowered capacity, while in reality, it is more accurate to term it voltage depression.

Memory is also hard to reproduce, which makes it hard to study. Originally, memory effect was seen in spacecraft batteries subjected to a repeated discharge/charge cycle that was a fixed percentage of total capacity (due to the earth's shadow). After many cycles, when called upon to provide the full capacity, the battery failed to do so. Since we aren't in space, the above is not really relevant...

Let us look at various causes of "memory" or voltage depression.

Memory can be attributed to changes in the negative or cadmium plate. Recall that charging involves converting

Cd(0H) to Cd metal.
2

Ordinarily, and under moderate charging currents, the cadmium that is deposited is microcrystalline (i.e. very small crystals). Now, metallurgical thermodynamics states that grain boundaries (boundaries between the crystals) are high energy regions, and given time, the tendency of metals is for the grains to coalesce and form larger crystals. This is bad for the battery since it makes the cadmium harder to dissolve during high current discharge, and leads to high internal resistance and voltage depression.
The trick to avoiding memory is avoiding forming large crystal cadmium. Very slow charging is bad, as slow growth aids large crystal growth (recall growing rock candy). High temperatures are bad, since the nucleation and growth of crystals is exponentially driven by temperature. The problem is that given time, one will get growth of cadmium crystals, and thus, one needs to reform the material. Partial cycling of the cells means that the material deep with the plate never gets reformed. This leads to a growth of the crystals. By a proper execution of a discharge/charge cycle, one destroys the large crystal cadmium and replace it with a microcrystalline form best for discharge.

This does NOT mean that one needs to cycle one's battery each time it is used. This does more harm than good, and unless it is done on a per cell basis, one risks reversing the cells and that really kills them. Perhaps once in a while, use the pack until it is 90% discharged, or to a cell voltage of 1.0V under light load. Here, about 95% of the cells capacity is used, and for all intensive purposes, is discharged. At this point, recharge it properly, and that's it.

The more common "memory effect" isn't memory at all, but voltage depression caused by overcharging. Positive plate electrochemistry is very complicated, but overcharging changes the crystal structure of the nickelic hydroxide from beta-Nickelic Hydroxide to gamma-Nickelic hydroxide. The electrochemical potential of the gamma form is about 40 to 50 mV less than the beta form. This results in a lower discharge voltage. In a six cell (7.2v) pack, this means a loss of 300 mV. Trick? Don't overcharge. Leaving cells on a trickle charger encourages formation of gamma nickelic hydroxide. Expect the cells to discharge at a lower voltage.

 

[safe]

Extrapolating a little further, it would actually be most efficient to go slowly, as the drag factors are at a minimum, and the Peukert effect is less.

My NiCad SubC's can deliver 10C (that's ten times the rated amp hour capacity) current. They do this with a minimal Peukert's Effect so that means you are much more free in your ebike design as you can drain them rapidly if you so choose. (if you actually try to use 10C all the time the NiCad's get hot)

SLA's have a really bad Peukert's Effect and in general you definitely want to get as close to 1C as possible. So let's do the math:

Four 12 Volt 18 Ah SLA's batteries.

Connect them in series to make a 48 Volt 18 Ah pack.

Now use a controller that has a 20 amp current limit (1C) and you can now have total power going to your motor as:

48 volts * 20 amps = 960 watts

Minus motor losses:

960 watts * (average) 75% = 720 watts

-------------------------------------

So you can get a one horsepower bike this way.

Weight of the batteries:

14 lbs * 4 = 56 lbs

...so you need about 56 lbs of lead to build a basic ebike using SLA's.

-------------------------------------

My NiCad pack stores about the same energy, but weighs about 35 lbs and can deliver a lot of power if you wanted to:

260 amps * 24 volts = 6240 watts (4680 watts output... 6 horsepower)

 

[Fuzzo]

NiCd looked like a good option, but you get a lot less Wh/kg, and Wh/$ than NiMH. I can appreciate if you're loading them up big time, drawing 6kW then yup, you have to have batteries that can handle it. But I'm not doing that, I'm hitting them with 200W. My aim is currently range, not speed.

I'm just extrapolating from the equation Dui presented. I'm sure we both know it's not as simple as that - keeping the motors at efficient RPM doesn't figure in there anywhere.

 

[safe]

NiCads and NiMh end up being the same price.

http://cgi.ebay.com/100-NiCd-Sub-C-...photoQQcmdZViewItemQQ_trksidZp1742.m153.l1262

NiMh cost twice as much for the same size cell, but then deliver twice the energy in total. However, the NiCads win in the end because they last twice as long. But the NiMh win it back because you are getting effectively double the battery for the same weight.

My solderless tubes hold SubC cells of WHATEVER I want to load into them... so if I want to spend more money I can always upgrade. With NiCads and my advancing age (I'm nearly 50 years old) the odds are that the batteries outlive my own useful lifetime. (by 60 years old I won't be able to do much with these road racers anymore)

-----------------------------------

I've written a full scale ebike road course simulation using the GNU language. This is the formula I used in that simulation:

#
# PowerHillClimbing(w)
# --------------------
#
# Global Parameters
# -----------------
#
# slp : slope of hill (percent)
# wt : weight of bike plus rider in kg
#
# Function Parameters
# -------------------
#
# w : rotational speed (rpm)
#
# Returns
# -------
#
# watts
#

PowerHillClimbing(w) = slp * wt * 9.8 * Velocity(w)

...one of these days I'll post that again. There are some bugs and upgrades I want to do on it. (maybe next winter)

"w" is the rotational speed of the motor in the simulation.

The Velocity has to go through a bunch of other calculations in order to connect motor rpm to the rear wheel speed. That's where you're going to run into troubles because you need to consider things like gears and tire size. That 0.027 constant probably only applies to his bike. (it's not a universal constant or anything like that)

Anyway... the only way to really get much information of value is to do the complete simulation of an ebike and run it over an imaginary course and see how it performs. I use it to find ideal gearing for the courses I ride on.

 

[duivendyk]

This equation has nothing to do with the means of propulsion ,it could be a steam engine.It's just basic physics.It takes so many Watts to move a certain weight up a certain incline at a certain speed,disregarding rolling &air resistance and the drive train eff. including the motor. You can rework it if you like HP for power instead of Watts (1.0 HP= 745 W, so the constant becomes 0.027/745=0.000036,or if speed S in mph instead of km/h, multiply constant by 1.61. It becomes 0.0435.
I can give an expression of the probable motor eff. at different speeds based on the eff at max.voltage. (a bit more complicated )

 

[Fuzzo]

Yup, I got that it was a very rough figure of the energy required. You're talking about watts at the axle rather than watts coming out of the battery. The relationship between the two won't be straightforward.

The hard part is working out the incline. It's a winding hilly road with lots of ups and downs. I can give a total ascent over distance to give an average gradient, but that's not going to account for extremes in any way. I could do the same for a ten kilometer ride on the flat with a vertical cliff at the end, but it's pretty clear that the vertical ascent will require more maximum power than a ten kilometer long incline to the same height. Furthermore, Piha is at sea-level, as is the starting point at home, so the average incline there is zero. The ride from Waiatarua to the top of the Piha hill actually drops quite a lot, but that doesn't mean I can coast the whole way - it's a mountain range, with many steep hills along it.

Also, at this point, I'm not really considering changing the power of the motor. Which means that if more power is required at any point than I have available, I simply have to drop my speed, right back down to walking the bike in the extreme cases. I'm fine with this, since actual experience has shown me that the length of those stretches is not that great. It's not hard work walking alongside an electric bike on 1/3 throttle carrying itself and my bag up a hill. If it all gets too much I can always stop and have a rest. Or give up and go home.

What I can't do is put any more power into the battery. So the battery needs to have enough in it for the entire trip, or I'll be walking it most of the way.

 

[Fuzzo]

I guess overvolting is another possibility. I haven't really considered it much, having this gut feeling that it could just totally wreck the motor and/or controller. But it would make the battery game a little easier. I could make 12 volt packs which are parallel with each other, but added in series to the SLA, giving me 36 volts, which I would guess could lift the power from 200 to 300 watts (still legal in NZ). Any thoughts on this possibility are welcome.