Combining A123 M1´s & low C rate lifepo4 questions

John in CR

100 TW
Joined
May 19, 2008
Messages
14,954
Location
Paradise
I want to parallel a 4p string of A123´s with a string of 1-2c 10ah lifepo4 cells. My thought is that the A123´s will protect the low c rate cells due to their much lower IR, so the no name cells would sag more under load and the A123´s would deliver the current. Then when after the few seconds of acceleration are over the low c cells will slowly charge the A123´s because of the small voltage differential.

Assuming that idea holds water how should I connect the packs together, at the ends or fully parallel them? What size gauge wire should I use?

John
 
How many amps is your setup going to be pulling? Continuous, maximum? Voltage?
 
magudaman said:
How many amps is your setup going to be pulling? Continuous, maximum? Voltage?

Voltage 26s, so under 100V fresh off the charger. Current 120-150A 5 second peaks, 50-70A 2 minutes, 30-40A continuous. If works well and the cheapie cells don't heat up, I'll likely explore some higher peak currents. If they do warm up much, then I'll run them at lower power. Note that the pack would rarely get pushed past 60% DOD, and in those cases it would be more conservative riding.
 
Unfortunately, there are a lot of unknowables. IMO, the only way to difinitively answer your questions is to build and characterize it.

I suspect that as long as your peak power * duty cycle is less than the continuous output power rating on the cheap cells, you'll be fine. The most practical way to answer that question is probably to track your average power consumption over several rides and compare that to the continuous power rating on the pack, treating the M1 cells as additional "cheap cell" capacity (i.e. use the lower power rating).


By the way, here's a thread where this topic was discussed previously: http://endless-sphere.com/forums/viewtopic.php?f=14&t=19848&start=0

*edit* Wait... I just noticed you said 30-40A continuous... Assuming you have a total capacity of 19.2Ah (assuming 26650's) and the cheap cells are rated at 2C, then you're probably overstressing the cheap cells.
 
John in CR said:
I want to parallel a 4p string of A123´s with a string of 1-2c 10ah lifepo4 cells. My thought is that the A123´s will protect the low c rate cells due to their much lower IR, so the no name cells would sag more under load and the A123´s would deliver the current. Then when after the few seconds of acceleration are over the low c cells will slowly charge the A123´s because of the small voltage differential.

I thought about this idea also some time ago . I think the problem that makes direct paralleling of same voltage / different Ri cells unappealing is that the recharge of the low Ri cells would be far to slow fromt he small voltage difference one would like to maintain. Or put it another way, during typical use at first the low Ri cells will provide the majority of the current and are likely to discharge almost completely, then after this the higher Ri cells will provide current, but not recharge the low Ri cells. As long as any current is drawn the Ri of the cells will lower the output voltage below any practical recharge voltage.

To make this approach work, one has to build two packs, the higher Ri one with a higher nominal voltage, then connect them together with a charge controller.

Example: combining A123 with Headway.
For 100V we need 30-32 series cells of the A123.

For 100V and a 30A nominal (average) load the Vdrop of series Headways is about 10V. Add a little more voltage for charge control. So 4 additional headways cells are needed.

Each pack gets a BMS.

The headway pack is connected to the other through some charge control mechanism. This could be as simple as a cutoff when the a123 pack is fully charged. People who mode their prius and insight hybrids just use a contactor to cut charge from supplementary pack to primary pack. For an ebike with some component level access it might be doable and more elegant to find a "charge complete" signal on the A123 BMS and feed this one to the Headway BMS so to cut out the Headway BMS when A123 are full. This is a control problem so some care has to be taken to make it work right. Might work with just a cutout at full and a cut in a bit lower, or a linear switching scheme might be better.

In operation, initially current from 0-30A is fed from Headways and goes through both BMS to load.
From 30A and up, 30A is fed from Headways, the rest from A123.
After A123's are discharged a bit, whenever current goes low, surplus current will recharge A123.
 
jag said:
John in CR said:
I want to parallel a 4p string of A123´s with a string of 1-2c 10ah lifepo4 cells. My thought is that the A123´s will protect the low c rate cells due to their much lower IR, so the no name cells would sag more under load and the A123´s would deliver the current. Then when after the few seconds of acceleration are over the low c cells will slowly charge the A123´s because of the small voltage differential.

I thought about this idea also some time ago . I think the problem that makes direct paralleling of same voltage / different Ri cells unappealing is that the recharge of the low Ri cells would be far to slow fromt he small voltage difference one would like to maintain. Or put it another way, during typical use at first the low Ri cells will provide the majority of the current and are likely to discharge almost completely, then after this the higher Ri cells will provide current, but not recharge the low Ri cells. As long as any current is drawn the Ri of the cells will lower the output voltage below any practical recharge voltage.

The slow charging of the RI cells is actually what I'm hoping for, so I don't stress the cheapie cells. I was thinking to maybe only have the cell parallel connections to the A123's, ie no series connection of the cheapies to each other, so all current to the controller is delivered through the A123 series structure. I have no commute, and this would be a blast around errands bike where the rides are in short spurts...a few miles of high performance and then the bike is parked for 20-30min or a couple of hours, not like many of you guys with a 10 or 20 mile ride with a number of hard accelerations mixed in. For the pretty rare long rides I'd either use another bike or go easy on the throttle with this one.

I've got the 4p of A123 just sitting there aging that I'm finally ready to put to use, but it makes for a smaller capacity pack than I'd like. Yes I could make a pack with just them for a short range blaster bike, but I also have these unused 10ah cells gathering dust that I'd never use other than in this manner because they are low power for my needs. They're rectangular cells to they pack nicely.

jag said:
To make this approach work, one has to build two packs, the lower Ri one with a higher nominal voltage, then connect them together with a charge controller.

I have to pick up some charge controllers for some solar stuff anyway, so I really like this idea a lot. I could even combine dissimilar packs, both chemistry and voltage, as well as fix the current drawn from the lower power cells. I love the flexibility of this approach. Kfong came up with a nifty little LVC switch that would be perfect to protect the charging pack from over discharge. I just need to decide on a relatively wide input range and top of charge exact voltage for the solar charge controllers I need for my projects.

That way I run A123's as my pack, and toss any old pack on the bike as my range extender that bulk charges my A123's at 10A both on the fly and while stopped. I'm liking this the more I think of the possibilities.

Thanks,

John
 
John in CR said:
I have no commute, and this would be a blast around errands bike where the rides are in short spurts...a few miles of high performance and then the bike is parked for 20-30min or a couple of hours, not like many of you guys with a 10 or 20 mile ride with a number of hard accelerations mixed in. For the pretty rare long rides I'd either use another bike or go easy on the throttle with this one.

I've got the 4p of A123 just sitting there aging that I'm finally ready to put to use, but it makes for a smaller capacity pack than I'd like. Yes I could make a pack with just them for a short range blaster bike, but I also have these unused 10ah cells gathering dust that I'd never use other than in this manner because they are low power for my needs. They're rectangular cells to they pack nicely.
Charging the A123 from the cheapies during resting might work by just paralleling. It won't be optimal, as charge rate will taper off and become exponentially slow towards the end, and practically it may be that only a small amount of charge transfers. However why not try an experiment? Take a fully charged cheapie, and a discharged 4p A123. Connect together with an Ampere meter. Read amperes at 0min, 5 min, 10min, 20min, 40min, 80min. the unequal spacing of readings is to account for the exponential decline in current, so the total of 6 readings of the Amp meter give us about the same accuracy as if we had read it every 5 min (uniform sampling) for a tedious 30+ total readings. (This is called adaptive sampling, and we will next do "adaptive quadrature". It just saved 30/6 = 5 times of work. If you paid college tuition and didn't learn this demand a refund!).

Now calculate amperehours charged to the A123 as follows:
Amin = 2.5*A0min + 5*A5min + 7.5*A10min + 15*A20min + 30*A40min + 60*80min + 60*160min
Ah = Amin/60
This gives you the number of Ah charged into the A123 after 160min

If you want the Ah charged after 3 h you do:
Amin = 2.5*A0min + 5*A5min + 7.5*A10min + 15*A20min + 30*A40min + 60*80min + 80*160min
Ah3h = Amin/60

2h:
Amin = 2.5*A0min + 5*A5min + 7.5*A10min + 15*A20min + 30*A40min + 60*80min
Ah2h = Amin/60

1h:
Amin = 2.5*A0min + 5*A5min + 7.5*A10min + 15*A20min + 30*A40min
Ah1h = Amin/60

With similar capacity cells and similar chemistry, at best 50% of the charge will transfer, so the charge is equalized between the paralleled cells. I don't know how long that will take though, and the final charge transfer may be lower than 50% due to hysteresis in the system.

John in CR said:
jag said:
To make this approach work, one has to build two packs, the lower Ri one with a higher nominal voltage, then connect them together with a charge controller.

I have to pick up some charge controllers for some solar stuff anyway, so I really like this idea a lot. I could even combine dissimilar packs, both chemistry and voltage, as well as fix the current drawn from the lower power cells. I love the flexibility of this approach. Kfong came up with a nifty little LVC switch that would be perfect to protect the charging pack from over discharge. I just need to decide on a relatively wide input range and top of charge exact voltage for the solar charge controllers I need for my projects.

That way I run A123's as my pack, and toss any old pack on the bike as my range extender that bulk charges my A123's at 10A both on the fly and while stopped. I'm liking this the more I think of the possibilities.

Thanks,

John

First note misprint " the lower Ri one with a higher nominal voltage" should be "the higher Ri one with a higher nominal voltage" (edited and fixed in original post).

You need also to protect any lower voltage battery bank from overcharge. This is normally not a problem with a wall charger as they are CC/CV, and the final rest voltage is the right final charge voltage for the batteries. If you find a solar charge controller with the correct voltage for the A123 pack you can use it for the approach I mentioned above. It will need to handle the max current experienced when connecting the packs together, or (perhaps more practical) implement a current limiting to a reasonable max current.
 
Jag,

I think I'll use "gut feel" or "intuitive" sampling, which will be even better than the adaptive sampling. I may have my answer after just 2 measurements. LOL I think you're right that the "charge" current will be too low to be effective, since the A123 discharge curve is so flat. That approach would probably be much more effective using Lipo... :idea: :idea: :idea: Maybe it's time to buy a 5ah single string of those 45-90c Lipos, and use my Konions to charge them continuously. Maybe the self balancing aspect of 15ah of konions will give me the best of all worlds, incredible peak power and no worries about balance charging. hmmmm

Regarding the charge controller, I would actually use lower voltage packs of the cheapies, making them both easier to manage as well as offer more flexibility for these range extender packs. The charge controller would be one designed for use with solar panels, which are small and efficient, and they can take a wide range of input voltage, and then output a fixed voltage and fixed current that I select. It will be perfect for this implementation, because it doesn't care about a voltage differential. eg My target input range would probably be something like 30-60v with a 10A ouput and voltage cutoff at 92.5v +/- 2%, giving me a nice safe margin for 26s A123s.

With a Kfong LVC switch protecting the cheapie battery from over-discharge, I simply connect any 36-48v cheapie pack to the charge controller, and it will supply a 1c charge to my A123s any time their voltage is below the cutoff. It seems like a great solution to me.

It seems like such a good idea, I wonder what kind of peak currents I can pull through 2p A123s with the factory toolpack tabs, since I may want to do this with 2 bikes instead of one. I have several kwh of low C lifepo4 currently not in service, and lot's of Konions I can use too for that matter.

John
 
Back
Top