Mixing chemistries for most economic performance and range

[swbluto]

That sounds interesting. Still, that of course should be tested on a case by case basis but my "intuition" suggests that the capacity battery would be wasting (resistor value)/(resistor value + load resistance) its energy on the additional resistance. For a normal current draw of 15 amps from a 50 volt battery(load resistance = 50/15 = 3.333) a measly one ohm for current control would waste 1/4.3333 of its energy on the resistor and that sounds pretty wasteful. Also, the booster pack would be continuously drained even more so at lower currents which might necessitate a larger, more expensive booster pack. But, I'm sure there's a "sweet spot" out there - I'm just not going to waste money to experiment with finding it. :wink: Ok, EDIT, you're probably talking about something like an additional .15 ohms or .1 ohms by changing the wiring gauge, which is much less(i.e., .15/3.5 < 5%. At higher currents, though, it'd climb upto 7-15% of the total energy. I'd imagine an average ride would entail depleting around 7% of the battery's capacity on the additional resistance, which sounds less than the 10% I'd expect for a switching circuit. So it could be more efficient.)

After the electronics are designed, it just makes things easy to hook together and go without having to worry about too high discharge/charge currents which can degrade either battery's cycle life. Also, a big bonus, it lets you design the system to work exactly where you want each battery to work, which is much more than you can do when fine-tuning each battery branch's resistance. But, with small ohmic changes on the battery branch(by changing wire gauge, for example), the additional losses would be comparable to the losses of a switching circuit, and the simplicity of that method might be worth it. It still wouldn't have as much "control" and wouldn't ensure safe charge/discharge currents, though.

For simple, "I don't want a lot of improvement" upgrades, it sounds like your method is perfectly fine. I know there's a lot of people that just want to be able to run their 20 amp pings on an ebike at 30+ total amps or so and that should be perfectly fine with the proper testing and fine-tuning. But for those that want to run ratios of 1 to 0 at lower currents(the booster isn't being used at all) and then 1 to 4 amps at high currents with the capacity's battery limit being respected, I really doubt fine-tuning resistances would let you accomplish that without unnecessary expenditure.

Anyways, I was reading http://en.wikipedia.org/wiki/Brushed_DC_Electric_Motor#Speed_control for information on DC motor controllers, and I found this interesting bit:

The effective voltage can be varied by inserting a series resistor or by an electronically controlled switching device made of thyristors, transistors, or, formerly, mercury arc rectifiers [2]. In a circuit known as a chopper, the average voltage applied to the motor is varied by switching the supply voltage very rapidly. As the "on" to "off" ratio is varied to alter the average applied voltage, the speed of the motor varies. The percentage "on" time multiplied by the supply voltage gives the average voltage applied to the motor. Therefore, with a 100 V supply and a 25% "on" time, the average voltage at the motor will be 25 V. During the "off" time, the armature's inductance causes the current to continue through a diode called a "flyback diode", in parallel with the motor. At this point in the cycle, the supply current will be zero, and therefore the average motor current will always be higher than the supply current unless the percentage "on" time is 100%. At 100% "on" time, the supply and motor current are equal. The rapid switching wastes less energy than series resistors. This method is also called pulse-width modulation (PWM) and is often controlled by a microprocessor. An output filter is sometimes installed to smooth the average voltage applied to the motor and reduce motor noise.

I'll accept your claim that "normal" brushed controllers have large input capacitances to smooth out the battery's delivery(Not hard to believe as that'd be hard on the battery), but what I'm not seeing is anything that suggests that the controller naturally outputs a steady current to a brushed motor by its own devices - this steady current is provided by the motor's internal inductance instead of the circuitry of the controller, so I imagine if you were to attach the controller to act like a current limiter on the battery current and parallel it with the booster pack, I think you'd likely be in for some nasty surprises. It seems like "sometimes" you wouldn't be, but that's only a maybe.

 

[Aerowhatt]

That sounds interesting. Still, that of course should be tested on a case by case basis but my "intuition" suggests that the capacity battery would be wasting (resistor value)/(resistor value + load resistance) its energy on the additional resistance. For a normal current draw of 15 amps from a 50 volt battery(load resistance = 50/15 = 3.333) a measly one ohm for current control would waste 1/4.3333 of its energy on the resistor and that sounds pretty wasteful. Also, the booster pack would be continuously drained even more so at lower currents which might necessitate a larger, more expensive booster pack. But, I'm sure there's a "sweet spot" out there - I'm just not going to waste money to experiment with finding it. :wink: Ok, EDIT, you're probably talking about something like an additional .15 ohms or .1 ohms by changing the wiring gauge, which is much less(i.e., .15/3.5 < 5%. At higher currents, though, it'd climb upto 7-15% of the total energy. I'd imagine an average ride would entail depleting around 7% of the battery's capacity on the additional resistance.)

You're still thinking one battery at a time. It's a system with two power sources. It really doesn't effect the booster pack capacity in practice. Unless of course you are doing nothing but flat out top speed run with no stops, or climbing a mountain pass. Most of the applications are for stop and go commuting, so the system works. Really the only time you are wasting any power to speak of is at high current loads. For most cases this is less than 10% of the ride. Ten percent of 7% is a pretty reasonable wasted power number. Considering that whatever the electronics waste will be through the entire range of loads. One hundred percent times whatever the loss is.

I'm not trying to talk you (or anyone else) out of persuing the electronics based solutions. For less well matched systems it becomes a necessity. If EV's take off like they should it could even make you rich. I just want people here to know that they don't have to wait for it. If they choose their battery combination wisely they can be riding long distances with plenty of grunt when needed for minimum weight, bulk, complexity and cost . . . right now.

My motorcycle chassis can't carry enough lead for more than about half the range it has with the hybrid battery. Which means it would be used a lot less replaced by fossil burner on those trips. An all LiFePO4 battery pack with the power density to give the bike the same performance off the line and up hills would have cost at least an additional $2,500.00 over what I have in the bike.

Aerowhatt

 

[swbluto]

That sounds interesting. Still, that of course should be tested on a case by case basis but my "intuition" suggests that the capacity battery would be wasting (resistor value)/(resistor value + load resistance) its energy on the additional resistance. For a normal current draw of 15 amps from a 50 volt battery(load resistance = 50/15 = 3.333) a measly one ohm for current control would waste 1/4.3333 of its energy on the resistor and that sounds pretty wasteful. Also, the booster pack would be continuously drained even more so at lower currents which might necessitate a larger, more expensive booster pack. But, I'm sure there's a "sweet spot" out there - I'm just not going to waste money to experiment with finding it. :wink: Ok, EDIT, you're probably talking about something like an additional .15 ohms or .1 ohms by changing the wiring gauge, which is much less(i.e., .15/3.5 < 5%. At higher currents, though, it'd climb upto 7-15% of the total energy. I'd imagine an average ride would entail depleting around 7% of the battery's capacity on the additional resistance.)

You're still thinking one battery at a time. It's a system with two power sources. It really doesn't effect the booster pack capacity in practice. Unless of course you are doing nothing but flat out top speed run with no stops, or climbing a mountain pass. Most of the applications are for stop and go commuting, so the system works. Really the only time you are wasting any power to speak of is at high current loads. For most cases this is less than 10% of the ride. Ten percent of 7% is a pretty reasonable wasted power number. Considering that whatever the electronics waste will be through the entire range of loads. One hundred percent times whatever the loss is.

Yes, I understand there are two power sources, but I also realize physical and electronic laws, and you're suggesting to increase the resistance on the "capacity pack", not the booster pack, no? This would increase the current coming from the booster pack at a given load. Here's a video for your educational pleasure. http://www.youtube.com/watch?v=ILpsfdkYOoQ

As you can see, at low current draw from the motor, increasing the resistance on the "capacity pack" line(To limit the current of the capacity pack at higher current draws, right?) causes the current to rise significantly on the booster pack. For "Short distances", this might be insignificant, but I'd rather not drain the booster pack if I had the choice. For this video, the ping had an internal resistance of .22 ohms and the a123s had an internal resistance of .14 Ohms.

If there are any values that you suspect would significantly increase the performance in this regard, let me know and I'll be happy to tape it for you. I understand that tools can significantly help the ignorant(It certainly has helped me!).

Anyways, I'll explain my entire set of motives here:

I want exact control on my batteries and I also don't want to shorten the cycle life of my batteries through over-charging currents. There's a reason why it's cautioned that people do not parallel different chemistries...

As far as the circuit goes, I'm glad to report I finally got a prototype circuit working that has the desired performance specs! I'm planning on putting a charging limiter on the "booster pack" using linear regulation, as it's a lot simpler, less costly, less error prone and its comparable in efficiency to other forms of regulation.

If anybody wants to put a current limiter on the "capacity pack" using almost-ready-to-use solutions, I might just suggest ordering a DC-DC converter and place input and output caps on it as needed(Though, finding DC-DC converters seems to be difficult for the powers we work with, as far as my experiences go.). If you want to use a brushed DC controller, I'd suggest using output caps(if needed) and putting an inductor in series to simulate the inductance of the motor. But, I'm sort of wary of that approach as I don't know exactly what the circuitry of common cheap brushed motor controllers are.

 

[ZapPat]

Yes, I understand there are two power sources, but I also realize physical and electronic laws, and you're suggesting to increase the resistance on the "capacity pack", not the booster pack, no?

I think he meant the opposite of this actually, swbluto (I'm however assuming that he is using the LiFePO4 as the power/booster pack and the SLA as the energy/capacity pack... I might be wrong). Personnaly I would not use SLA's as the power pack, since they have terrible peukert losses at high discharge rates. Humm... I think swbluto might be right, and Aero is using his chemistries in the opposite way I was thinking of.

I've tried out the NiMh+Lipo combo, using a 2 cell 460mAh 20C zippy pack (power pack) in parallel to a 6 cell NiMh 4000mAh pack (energy pack). I am running into a similar problem as Aerowhatt describes with his LiFePO4 + SLA combo where one string takes too much of the load, and so I will try experimenting with adding some small resistance in the string that's taking too much load (the LiPo power/booster pack in my case). Like swbluto, I don't like the idea of adding resistance though, since avoiding resistance is why we pay more bucks for these batteries. Even a 0.1ohm resistor at 50A would be dissipating 50A^2 * 0.1ohm = 250W!! Even a 0.01ohm resistor would be dissipating 25W, which makes for a pretty beefy resistor. My problem is that my energy pack (NiMh) has such high resistance that the booster pack (LiPo) is bearing too much of the load, so that I can't even get the NiMh to output at a 1/2C rate while the Lipo is pushing 5C! I'll see what gives with a resistor to try and even things out. Also note that these Nimh's have very bad internal R compared to modern ones I'm sure, which makes this problem worse.

BTW swbluto, a buck regulator working at such a low voltage differential (input/output) with a 10% loss would be a very badly designed affair. You should be able to get something like 96-98% efficiency from such a circuit if well done (fast switching speeds, appropriate FETs, etc).

 

[swbluto]

Thanks ZapPat. I'm not really much of a circuit designer(Not one of much experience anyways) but when I heard the efficiency could reach 98% on Wikipedia for a buck regulator, I was under the impression something "realistic" would be a bit lower, so 90% seemed to be more realistic to me as a "Probable less than ideal case" scenario.

So, ummmmmm, how would you maximize the efficiency? Any particularly important component values to adjust?

As low as frequency as possible? As low of inductance as possible to minimize resistive losses in the inductor(or an inductor with minimized resistance)? It seems those two are competing factors, so it sounds like some optimization would be in order except, I don't believe I know how to calculate realistic power losses for a buck regulator! Ok, more research.

And I think I'll look into the "boost buck" regulator you linked to earlier with the supercapacitor. That seems like it'd be particularly elegant for providing so much control over the direction of current.

 

[Aerowhatt]

Yes, I understand there are two power sources, but I also realize physical and electronic laws, and you're suggesting to increase the resistance on the "capacity pack", not the booster pack, no? This would increase the current coming from the booster pack at a given load. Here's a video for your educational pleasure. http://www.youtube.com/watch?v=ILpsfdkYOoQ

As you can see, at low current draw from the motor, increasing the resistance on the "capacity pack" line(To limit the current of the capacity pack at higher current draws, right?) causes the current to rise significantly on the booster pack. For "Short distances", this might be insignificant, but I'd rather not drain the booster pack if I had the choice. For this video, the ping had an internal resistance of .22 ohms and the a123s had an internal resistance of .14 Ohms.

You are correct that the booster pack provides more current if you increase the resistance on the capacity pack (we must also remember how small the change I recommended was). But that's only one aspect of the real world system performance. Urban Ev's spend significant time coasting and stopped waiting for traffic signals. After a period of discharge you find that the capacity pack is resting at higher voltage and because of this charging the booster pack back up. I'm trying to share real world practical experience here. I started designing, building and using hybrid chemistry packs in 2003. I majored in physics which is what allows me to get close to a good balance of two packs with only qualitative information. In real world use the increased resistence on the capacity pack has little influence on the ability of the booster pack to handle the next big load since the booster is constantly replenished during no load and very low load periods. By way of example, extensive mapping of the performance of the motorcycle battery system shows that in everyday riding on a 75% DOD discharge (around 30 hard riden miles) of the complete hybrid pack the SLA alone is only at almost 50% DOD.

I'm trying to use your terms here to avoid confusion. But if battery longevity and therefore low cost of operation are some of your goals I think the terminoligy should be more representative.

Booster pack = Main pack

Capacity pack = range extender pack (XR pack)

The reason for this is that for these systems the Main pack should be able to handle the full peak load for several minutes all by itself. Preferably ~5 minutes. If you design under this guidline you won't have good battery longevity. This is experience talking!

That said, Congratulations on your control circuit successes! I look forward to reading more about their progress and real world results.

In addition reguarding any confusion expressed in some other posts. On the motorcycle the SLA's are the main pack and the LiFePO4 is the XR pack. It's prohibitively expensive IMO to use anything but high quality Odyssey Sla's to provide amperage peaks of 350 amps (~8.5C) on a regular basis. They have about the lowest pukert numbers of any SLA battery too.

Now, if one were to want to make a true booster pack to suppliment high peak loads for a high capacity main pack capable of handling the average cruising load all by itself. A simple shunt triggered solenoid connection for the booster pack to put it in parallel when, and only when, the current required by the system rises to an adjustable threshold. The rest of the time it would be sitting idle waiting in the highest state of charge possible (since it's only providing current to boost required periods) There would be virtually no losses in this solenoid switch approach and the only concern needing testing and examination would be to make sure that back current into the capacity pack from the booster pack (while breifly connected) would be acceptable if/when the capacity pack is at a very low SOC. My sense is this shouldn't be a problem issue.

Aerowhatt

 

[swbluto]

Now, if one were to want to make a true booster pack to suppliment high peak loads for a high capacity main pack capable of handling the average cruising load all by itself. A simple shunt triggered solenoid connection for the booster pack to put it in parallel when, and only when, the current required by the system rises to an adjustable threshold. The rest of the time it would be sitting idle waiting in the highest state of charge possible (since it's only providing current to boost required periods) There would be virtually no losses in this solenoid switch approach and the only concern needing testing and examination would be to make sure that back current into the capacity pack from the booster pack (while breifly connected) would be acceptable if/when the capacity pack is at a very low SOC. My sense is this shouldn't be a problem issue.

Aerowhatt

That sounds like a very good idea. Ensure that the capacity pack/"XR pack"(I know we humans have our own set of preferred terms. :wink doesn't significantly go over its current limit during high current periods by adding small resistance to its branch as needed and switching the booster pack on when the current is high enough. Since the capacity pack is lower current by definition, the relative power losses by the increased resistance should be pretty manageable for ebike applications, although it might become more serious for things like electric cars or anything else. But maybe not if 2-7% doesn't hurt.

But let's extend functionality without significantly increasing complexity. Let's replace the solenoid switch with an appropriate mosfet that's actuated by the same shunt triggered system, and that also employs current limiting during recharge currents(Only requires a few op-amps.). During "discharge", its resistance could also be comparably minimal(less than .004 ohm resistance I believe which might even be less than a switch's resistance), it can turn off the current as needed, and it can limit the charging current as needed.

But, what about if you're in the scenario where the booster runs dry? Then you'd need to turn off the booster pack's current, meaning all the current demanded would be flowing from the capacity pack/"XR pack"(I'd prefer to call this the "main pack" as it's the one that's supposed to be mainly used.). Unless you have some way of limiting the load current, it sounds like this could be problematic. But, it seems you imply from

"The reason for this is that for these systems the Main pack(booster pack) should be able to handle the full peak load for several minutes all by itself. Preferably ~5 minutes."

That you'd design the booster pack so that'd it never actually deplete to nothing, right? So that'd scenario would never happen. But if it does, you're just SOL? :lol: (Unless you have a way of limiting the load current, but it sounds like that'd be too much complexity, tsk tsk).

Anyways, it sounds like this scenario exposes a flaw in my own embryonic design. If the booster pack were to cut-out and the load was still high, then it'd adjust the current going out of the capacity pack by lowering the output voltage. It might not be a frightening amount in practice, but it sounds like a drastic voltage drop might be "bad" or something. Well, I'm sure the decrease in the discharge current could be "gradually lowered" so that the rate of voltage change isn't problematic with the controller.

Now, if there were a simple way to either solve the "booster pack's" depletion problem or just forget about it in a worry-free way, I'd find the mosfet's dual switch/linear-charge-limiter idea on the booster pack in tandem with adjusting the branch resistance on the capacity pack+testing to be a particularly simple safe implementation of paralleling with minimal losses. And 5 minutes sounds like a good guideline, but 2.2 AH of a123s at 60 amps only provides 2 minutes of current and I want to avoid buying more in the essence of cost minimization. Well, I'll either flirt with the line of danger or it'll be completely no-longer problematic in future projects(Well, actually, it seems like that dynamic would always exist - more "boost time" means more money but the cost can be minimized by getting the lowest cost cells that can provide the performance specs at a safe "peak boost time" guideline. But, even if it turns out a low boost time is problematic, by experimentation, then it's fairly trivial to increase the capacity as needed.).

For now, this is just an exercise of the mind and I'm not actually going to do anything with it RIGHT NOW or even within the next month(I don't have extra batteries just lying around.). But when I get the inevitable itch to make a high powered system(I'm getting tempted with my outrunner-upgraded scooter, lol) or something with huge capacity, I may consider doing something.

To ZapPat: To increase the resistance by .2-.01 ohms would be done by going to a smaller wire size and adjusting the wire's length as needed. It's simple and cost effective without the need for a "beefy" resistor.

 

[swbluto]

Aerowhatt, do you have a link to your build? I'm curious about some of the technical details.

 

[Aerowhatt]

That sounds like a very good idea. Ensure that the capacity pack/"XR pack"(I know we humans have our own set of preferred terms. :wink doesn't significantly go over its current limit during high current periods by adding small resistance to its branch as needed and switching the booster pack on when the current is high enough. Since the capacity pack is lower current by definition, the relative power losses by the increased resistance should be pretty manageable for ebike applications, although it might become more serious for things like electric cars or anything else. But maybe not if 2-7% doesn't hurt.

Yeah, making the main pack the low output pack is the opposite of the systems I have done. I've taken systems with enough power on tap already but insufficient range. Then added range with a different chemistry(bonus is you get more power too). To go the other direction and have a Ping as the main pack (not large enough to handle full peak currents) and a boost pack for filling in the Amperage gap when needed. In that case I would minimize the resistance in the whole harness. If the Boost pack is only used for short peaks and out of the system the rest of the time then it would make more sense to solve any controbution issues of each pack when they are paralleled by raising the voltage of the boost pack (Adding a cell.)

Lets say you have a ping II that is 16ah and that's ~25% more than you need for your range target. OK so at best you can expect 32amps continous out of it. For this battery you should be looking at cruising amps of 20 amps or so, or else have a bigger ping. Now you set up your boost engagement to jump in (and also jump out) at around at about 30 - 35amps system draw. The Ping BMS will protect it by dropping out if you over draw from it. If you build it and find your boost getting tapped out during regular use could do one of three things. You could back off and live with the lower power level for the rest of the trip. You could also use a constant current up converter to charge the boost battery from the ping when the boost isn't engaged. Or you could increase the capacity of the boost pack. In order to design it you would need to do some riding with a pack capable of handling peak currents and map the current with time. Then you could see how many ah of boost current pack would be needed under normal conditions.

I just have to repeat here though that shaving battery capacity really close doesn't get you low battery costs. It does get you lower initial cost but higher operating costs. Usually much higher long term operating costs.

I used to mountain bike here in the Rocky Mountains with a guy who spent thousands to shave a few ounces off of his bike. Overall his aluminum show stopper weighed about 2.5 pounds less than my steel stump jumper ($550 new). So if you account for the rider weight and the weight of the bike you are looking at a less than 1 percent weight savings for about $3,000 more he spent. I never had the heart to share the math with him and frequently kicked his butt on my modest steel bike. I mention this becuase the size and weight of the battery when talking LiFePO4 is fairly insignificant to the whole package (rider motor and bike). If I needed a 60 peak amps I would have ping make me a 24ah unit (adjust the shunt if needed) and be done with it.

I think the hybrid chemistry approach makes far more sense on larger platforms. It is fun to play aroud with such things though!

Aerowhatt

 

[Aerowhatt]

I've tried out the NiMh+Lipo combo, using a 2 cell 460mAh 20C zippy pack (power pack) in parallel to a 6 cell NiMh 4000mAh pack (energy pack). I am running into a similar problem as Aerowhatt describes with his LiFePO4 + SLA combo where one string takes too much of the load, and so I will try experimenting with adding some small resistance in the string that's taking too much load (the LiPo power/booster pack in my case). Like swbluto, I don't like the idea of adding resistance though, since avoiding resistance is why we pay more bucks for these batteries. Even a 0.1ohm resistor at 50A would be dissipating 50A^2 * 0.1ohm = 250W!! Even a 0.01ohm resistor would be dissipating 25W, which makes for a pretty beefy resistor. My problem is that my energy pack (NiMh) has such high resistance that the booster pack (LiPo) is bearing too much of the load, so that I can't even get the NiMh to output at a 1/2C rate while the Lipo is pushing 5C! I'll see what gives with a resistor to try and even things out. Also note that these Nimh's have very bad internal R compared to modern ones I'm sure, which makes this problem worse.

This combination is just too far apart to make work without some serious intervention. Even if you raise the cell count on the NiMh to compensate for the high internal R you would need a step down converter to avoid overvolting the LiPO at the beginning of the discharge. Sometimes it's just too far to reach!

Aerowhatt

 

[Aerowhatt]

Aerowhatt, do you have a link to your build? I'm curious about some of the technical details.

I don't have it documented online. I'll list basics here and I can answer any other details you need. I also have some external photos I could send you. It’s an ugly little motorcycle but a great performer. Just like the plain girlfriend that treats you right. She gets better looking the more you get to know her.

The Base Bike is a TECC Viento it's a 2002 model year . Originally it came as a moped class bike with a top speed of 29mph. It was too big for this speed and cars would slow for me to pull out into traffic. I rod it about 25 miles with the stock gearing. It's powered by a Perm PMG 132 motor. It's German motor, radial gap brushed motor capable of much more than the original setup asked of it. At 36 volts it’s rated at 3600 watts continuos, 7200 watts for 10 minutes.

I initially geared it up to 42mpg (~40%) and then set about increasing the range

Specs are as follows

36 volt system

Curtis controller with 275 amp limit (It has issues clamping quickly so 350 amp spikes happen)

Curb weight 287lbs
Front wheel curb weight 127lbs
Rear wheel curb weight 160lbs
Drive belt - Gates PowerGrip 2 - 1512mm 8mm pitch x 20mm wide

Total weight as ridden 425 lbs
Front wheel weight with rider 175lbs
Rear wheel weight with rider 250lbs

Current battery configuration:
internal - Odyssey PC-1700 SLA (42ah, 10 hr rate)
external - 2 Ping 20ah 36 volt LiFePO4. (40 ah, 3 hr rate)

The two Pings are connected in parallel with each other right after an individual 40amp circuit breaker for each one..Then they are wired from there to the main pack SLA’s with 10 ga cable. The LiFePO4 works really well with the Sla’s. Their resting fully charged voltage closely matches the float charge voltage for the Sla’s. The circuit breakers serve as a disconnect to isolate the 2 pings for a monthly cell balancing charge and as a safety if one pings fails short circuit. In regular use they charge with the Sla’s which brings them to about 97% full. The Ping BMS boards will drop out before the circuit breakers trip. It’s possible to drop the pings out but very difficult to do and won’t happen without an instant full throttle start from a standing stop up a steep hill. The SLA’s have a BMS on them to keep them in balance too. The bike has 1000 watt output (DC side) onboard charging.

The Bike has between 32 and 38 miles range the way I ride it (max with more conservative throttle use should be near 45 miles). The range goal was 35 miles because I need about 22 miles or so between charges on a regular basis and I want the ~$1,400 battery pack to last a good long time. The only way to maximize battery life is to have more range available than one needs.

Aerowhatt

 

[dogman dan]

The only way to maximize battery life is to have more range available than one needs.

Exactly what I keep saying to people who post something, like... My motor draws x per mile therefore to ride my y commute I only need x times y plus one mile. I'll respond, actually you want x times y plus at least 25%. If not 50% Some days it's uphill into a lot of wind to get there. Not to mention range drop when it gets to be colder. You are a rare one Aerowhatt, engineering brains and a bigger than usual dose of commonsense.

 

[ZapPat]

The only way to maximize battery life is to have more range available than one needs.

Exactly what I keep saying to people who post something, like... My motor draws x per mile therefore to ride my y commute I only need x times y plus one mile. I'll respond, actually you want x times y plus at least 25%. If not 50% Some days it's uphill into a lot of wind to get there. Not to mention range drop when it gets to be colder. You are a rare one Aerowhatt, engineering brains and a bigger than usual dose of commonsense.

I was wondering how long it was going to take Dogman to say this!

Good sense for sure... keep the newer chemistries away from being too empty and even too full and you'll keep your battery a looong time.

 

[swbluto]

There's an optimum amount of capacity to minimize long-term costs, for sure, but it's clearly not a value that approaches infinity. If you buy twice as much battery as you need and it's twice as costly, and it ends up having 3/2 the lifetime than a smaller pack, then buying twice the battery is not really long-term cost-optimized. But somewhere between 1 and 2 times you need sounds like a good range. Then there adds on the issue of calendar life and how often people actually use their battery - It appears, with break downs and everything, I only end up using my battery once or twice a week on average. There's no way I would reach the 700 theoretical cycle life(for a stressed battery - a123s show cycle lives a little above 1000 for "stressful conditions", like 20c continuous average) if a given battery deteriorates to unacceptable levels within 5 years no matter its starting capacity, so the bigger battery would then be uneconomically optimal. However, the "security of mind" of never running out has its own price and so that's naturally factored in to how much the extra range is worth.

I think this discussion is deserving of its own thread given the complexities involved.

(In my particular case, I'm only thinking of propelling my scooter with a max range of 5 miles. My ping battery provides upto 12 of those miles, so I'm covered.)

 

[Aerowhatt]

That's a good point, calendar life is a significant consideration. Interestingly though the calender life has to be taken with the fine print. For the older LiPO it was said to loose 20% of capacity in 2 years. Well if you figured some head room on capacity initially that helps some. I'm in a situation now where I have too many EV's to get lot's of cycles on them all (poor me right?). So the solution for me was to start buying different batteries where I can. For SLA's there are some choices. You can find different brands with standby life expectancies of anywhere from 3 to 12 years. LiMn and LiPO can be refridgerated when not in use roughly tripling their calender life. I'm hopeful that LiFePO4 will make this issue, well less of an issue. But we likely won't know that for some time.

So with the right strategy I'm able to fly airplanes with LiPO that I bought in 2001. They were 2,100mah when new and come in at about 1,750 mah now but they still get the job done.

Dogman is very on target with the head winds etc, that can raise power consumption. Last summer I was motoring home on an eGO Cycle. It's about 9 miles and up in elevation ~300 ft. On this particular day it was blowing in a thunderstorm from the southeast. I had brutal gusty head winds all the way home. The eGO cycle has a pretty strong system so you don't notice how hard its working in a big top speed reduction, or anything like that. When I arrived home the battery gauge was a lot lower than usual. So I measured the power to recharge. The consumption on this commute and this bike would normally come in around 51 wh/mi. This day it was a whopping 86 wh/mi.

Aerowhatt

 

[dogman dan]

When people talk about they need to go five miles, I get a lot less likely to scream buy more! Some evidence is trickling in now also, that the ping v2 really is able to do 2c discharge rates in the real world, so some can buy smaller packs now than last spring. But when I see somebody with a 10 mile ride, going, I'll never need more than 8 ah while I ride 30 mph..... Well, I just know that isn't going to work everyday.

My two recreational bikes don't need huge capacity. But my commuter is another story. I bought a 36v 20 ah, thinking I might get 30 mile range. I actually get 20. Then it got cold, there went 2 miles of range. Then seasonal winds turned, and now I have a headwind to get home everyday, AND the dang 15 miles uphill. Pretty soon I was needing every hypermile trick I knew to keep from tripping the pack everyday. Yeah, when I bought I thought I was buying twice the size I needed....... So I try to keep folks from basing an expensive purchase on the lies they read in the ads for their motors.

It is a lot of money, but a lot more to have to buy batteries twice. Since I am a commuter and have to do two cycles a day when I bike to work, the cycle life is crucial to me. I'll burn this ping down in three years for sure. After that, it can power a bike I ride for fun. For a fun bike though, let er rip and if you fry stuff, buy more! Till recently my entertainment budget incuded a hot air balloon. Only $300 an hour for that fun. Sorry if this is hijacking the thread, but cycle life ought to be considered in most battery decisions, unless its entertainmet budget.

 

[dnmun]

lifepo4 is now obviously the way to go. you guys on all these threads should listen to the dogman, that is voice of experience talking. the man who bot the first ping when everyone ridiculed the idea. just to remind us, was it $300 and free shipping? how many people were sending you hopeful messages at the time?

it is really false economy to try to skimp on capacity. we still don't understand what causes the lifepo4 cells to age rapidly in use. or if they really do need the confines of an aluminum cylinder to squeeze against during discharge and charge. but if you have too small a pack, you will regularly be running in the lower voltage range of the cells, which we suspect is what causes them to age because the Ri climbs so rapidly then, along with excessive overcharging beyond 3.65V.

i am of the opinion it is best to charge right up to the 3.65-3.68V top, so that when the pack discharges, it is using juice at the very top of the voltage range, which just seems intuitive to be the best place to use it, from looking at the discharge curves and measuring the internal resistance during the discharge.

Ri changes from low initially to high at the end, so more capacity to start with gives you more power to use before it gets down there, and less sag all along too.

so if you buy 2 packs instead of one, the set will likely last far longer than just buying them sequentially, and who knows what the price will be in 5 years?

and the kicker is that you will always be able to go twice as far, twice as fast while stressing the cells as much as if you had only the one smaller pack. jmho

ps: consider the battery to be a lifetime purchase, doubling capacity is the way to make it happen.

 

[dogman dan]

Gee dnmun,,, I hope you live a bit longer than that! I wasn't exactly the first ping buyer, but when I ordered, the reviews were mixed at best. But the unhappy ones were all buyers of 10 ah packs, and had lots of cut outs, and bms mods to fix it. The v1 pack really did need to be as big as possible. I paid $350 plus $100 shipping at the time. My order was placed in April 2008. It will be very interesting to see what the shelf life of seldom used or lightly used Ping v1's will be. I think 5 years is a reasonable expectation, but the battery may be very usable beyond that, as a shorter range pack for a lower watt bike. The way I'm riding, and what I'm saving vs a car, I will hit the free riding mark on the battey at 3000 miles, just 600 to go! On the bike itself, I may be forever in the hole the way I can smoke motors. :lol: Oh well, if it burns, it's entertainment budget. 8)

 

[ZapPat]

Dnmun and Dogman - I understand your love for the LiFePO4 chemistry, and also your points about being kind on our batteries by using only part of their cacpacity as much as possible. However the point of this thread was to see how we could get the range advantages of a cheap but voltage-sag prone battery and combine this with the power-handling advantages of a smaller but strong battery, and all the while keeping costs low.

For example, take a small 48V/10Ah ping (~250mOhms, ~6kg, 2C = 20A) and combine it with a small but powerfull (and cheap) 5Ah LiPo battery (<50mohms?[TBD], ~1.8kG, 15C = 75A)... and you end up with a very powerfull pack when needed with very little added weight! Plus you actually end up being much kinder to your cheap LiFePO4 cells that don't like seeing high discharge currents, so they will also very likely end up lasting longer too.

Of course in a few years we'll hope to be able to buy high performance LiFePO4 cells pretty cheap, but for now this isn't the case. Hey, even combining cheaper energy-type cells with more expensive power-type cells of the same tech (like LiFePO4) might be a good idea for you guys... if you like the occasionnal burst of power to accelerate or go up a big hill of course.

Cheers!
Pat

 

[torrent99]

For example, take a small 48V/10Ah ping (~250mOhms, ~6kg, 2C = 20A) and combine it with a small but powerfull (and cheap) 5Ah LiPo battery (<50mohms?[TBD], ~1.8kG, 15C = 75A)... and you end up with a very powerfull pack when needed with very little added weight! Plus you actually end up being much kinder to your cheap LiFePO4 cells that don't like seeing high discharge currents, so they will also very likely end up lasting longer too.

Cheers!
Pat

Exactly! That's what I want to mix chemistries for.
* Bulk (long term) = Ping.
* Nitro (semi-disposable) = LiPo ......
By doing this it means you only need to size the Bulk (Ping) for the range you require, not the current you require..... SO LONG AS YOUR CURRENT NEEDS ARE IN SMALLER BURSTS. If your journey is a long drag up a high hill, then you'll need a lot more Nitro (LiPo) or a heck of a lot more Bulk (Ping) or just take the hit on the life expectancy of your Bulk...

Once the recipe & techniques are there for combining and controlling the chemistries, I can see endless fun discussions over what's the best combination in what proportions for particular journeys and riding styles. :lol:

Cheers

Steve

 

[Aerowhatt]

It just sounds like inexperience talking to me. After thousands of EV miles it's versatility that one really ends up wanting/needing. Sizing and tailoring for a specific commute doesn't give versatility. "Can't take the bike today not enough range while pulling the trailer" . . . "Well probably not enough to do the out of the way errand after work, better take the car". The other thing I've learned over the years is that 8 to 9 people out of 10, just don't seem to learn from others experience (the easy way). So they must learn by repeating a completed journey (the hard way). The up side is that everybodies "hard way" is a little different and as a result occasionally a valuable new innovation is stumbled upon.

Have fun guys!

Aerowhatt

 

[swbluto]

Ok, I'm done with the finals, so I can start designing the "actual circuit"(The one I drew up was an idealization to simplify simulation but it's not very economic to have huge capacitors in real life). I haven't crunched the efficiency numbers but it seemed like 90% is reasonable lower bound and only higher numbers like 96% can be achieved using synchrous switching as opposed to a diode, but that complicates the driver circuit.

One thing I'm having trouble with is implementing the drivers for the Mosfet's Gates - it seems this is controlled by the gate-to-source voltage, but the source voltage isn't at ground(I think?) unlike the driving circuitry, so it seems like there must be a special way to generate a voltage signal that's based on the individual mosfet's source. I've heard Op-Amps can be used as "virtual grounds" and that sounds relevant to what I'm trying to solve, but I don't know much about that.

 

[Roy Von Rogers]

How about using SCR's on the output of the batteries to control which battery is used. Use a LVC to stop the discharge on the A123 when it gets to 2.5V, and let it recharge in series to 3.4V from main battery via a HVC control. One could also use a current sense to decide when the extra from A123 was needed, by turning the SCR's on-off.

My understanding is, if the A123's are not overstuffed or over discharged, it shouldnt effect the balance on those M1 cells.

Now mind you I may be way over my head here..lol..since I'm nowhere near as savy as most of you in this field.

Roy

 

[bearing]

One thing I'm having trouble with is implementing the drivers for the Mosfet's Gates - it seems this is controlled by the gate-to-source voltage, but the source voltage isn't at ground(I think?) unlike the driving circuitry, so it seems like there must be a special way to generate a voltage signal that's based on the individual mosfet's source. I've heard Op-Amps can be used as "virtual grounds" and that sounds relevant to what I'm trying to solve, but I don't know much about that.

There are at least two ways to solve this. (That I know of.)

One way is to use a simple isolated SMPS and connect its output ground to the source. The gate driver IC is powered by that voltage. The input to the gate driver will have to be optocoupled from the switch-controller IC.

Another way is to connect the gate driver to the primary of a pulse transformer and connect the secondary of that transformer to the gate. This eliminates the hassle to have an isolated power supply for the gate driver, but won't work with long static ON-periods (I think).

 

[monster]

this seams like a nice application for car batteries (not deep discharge). you can pull hundreds of amps from them for a short time.