DIY Electric Car Forums banner

A123 20Ah pack configuration question

22K views 48 replies 9 participants last post by  EVPowers 
#1 ·
I need some advice from the pros :eek:.

More detail on my project can be found here: http://www.diyelectriccar.com/forums/showthread.php/tennessee-rx8-conversion-68311.html

I've purchased 500 of the A123s (32kW) and want some advice on how I should configure them. My plan is in two phases.
Phase 1: uses one WarP 9 with the transmission. I'm looking into either a Z2k-HV or EHV. I haven't decided yet. HV saves me some upfront costs.
Phase 2: adds a TWarP 9 for a dual motor setup and removes the transmission. I'll also upgrade the drive train at this time.

I was thinking of having a modular design where I would only use about 60-70% of the cells and then have an extension pack that I could hook up to extend the range to +25 miles.
Here's where it starts to get complicated. I don't mind redoing the pack once I get to phase 2, so a completely different arrangement could be configured for each phase if needed.

To start things off, this is my first proposal:
A modules of 70s4p for 224V so that if pack sags, I'll still have >170V for the motor. I could then combine in parallel an additional module of 70s2p or 70s3p for a total pack of 70s6p or 70s7p given I only have 500 of these cells.

I've heard some say, go for the highest voltage your controller allows. I've heard others say, that connecting modules together in parallel isn't a good idea and that it would be better to have the extension pack in series that adds voltage using the same number in parallel.

What are your opinions on how to configure the pack and would it be different for Phase 1 versus phase 2 :confused: ?

Thanks in advance for everyone's input. I'm learning so much about this field everyday. I have a lot of appreciation for the members of this forum.
 
See less See more
#2 ·
Use all cells or nothing. Having a range extending pack really complicates things, especially if they're not the same SOC. If you parallel them for range extension, the pack with lower SOC will cause an arc from the higher SOC pack. If you do things in series, you have to modify all of your controller parameters. If you use a BMS, it complicates things as well. If you use some sort of SOC meter, it can really complicate things.

Why wouldn't you do the large battery pack up front? Why redo it second phase and mix old with new?
 
#3 ·
when you go with the dual motor set up are you putting the motors in series or parallel or doing series/parallel switching?

These A123's don't sag as much as prismatics so having a really high voltage pack doesn't make as big a difference.

My logic could be wrong but I would guess that putting the extra range pack in series with the normal pack would allow you to keep things in better balanced. Putting another string in parallel should be fine in theory, but you wouldn't have the parallel cell connections between the two strings, they would be just tied at the +/- extremes.

Another thing to consider (which is a big task for 500 cells) but I'm in the process of measuring the IR and capacity of every one of my cells (my pack will only be 288 of the 20ah A123's) that way I can group them into perfect parallel sets, in my case 3 parallel that total the same value for every single grouping. That way the series groups should all behave exactly the same, mine will be top balanced, but since they are all the same capacity they should behave like they are bottom balanced as well. So far I have been able to group them to within 0.01Ah.
 
#4 ·
when you go with the dual motor set up are you putting the motors in series or parallel or doing series/parallel switching?

These A123's don't sag as much as prismatics so having a really high voltage pack doesn't make as big a difference.
I've been talking with John Metric from DC Plasma and his calculations show that if you have a high enough pack voltage (like 333V) the series/parallel switching isn't beneficial due to the electromagnetic field emf. So if I string enough together I could avoid all the additional contactors etc.

Another thing to consider (which is a big task for 500 cells) but I'm in the process of measuring the IR and capacity of every one of my cells (my pack will only be 288 of the 20ah A123's) that way I can group them into perfect parallel sets, in my case 3 parallel that total the same value for every single grouping. That way the series groups should all behave exactly the same, mine will be top balanced, but since they are all the same capacity they should behave like they are bottom balanced as well. So far I have been able to group them to within 0.01Ah.
I'm very interested in learning more about this. How long does it take to test the capacity of each cell? What equipment do you use? Perhaps you can guide me on how to do it with a PL6.
 
#6 ·
just curious rwaudio, are you using the discharge or the charge capacity? Or are they the same? I'm doing similar cycling with CALB 130ah cells with my PL6 and I'm seeing about 2 more ah going in than come out on a cycle.

Granted my cells are a lot larger (and take more like 8 hours/cell to cycle) and 2ah on my cells is more like 0.3ah on yours, but was just wondering which number is considered the capacity of the cell.

I'd also be really curious to see if a single cell was cycled three times, how close are the three capacity values. In other words, how repeatable are the results with the PL6? I'm certainly not going to try and find out since it takes me 8 hours to cycle a single cell. :D
 
#7 ·
I'm using the discharge cycle as the capacity (at 2C in my case which is somewhat close to or just above what the car should average), and you are close, I put in about 0.2Ah more than I get out. I did have a higher discrepancy 0.4 to 0.5Ah using the stock banana's and wires at 40A, I have since switched to a better banana and shorter 8awg cables. So a good portion of the loss was in heat, there is still some heat in the cables and connector so I believe actual losses in the cell to be very low. I have done some repetitive testing on a couple of cells and the results were within 0.01Ah (This isn't based on a very big sample though, one cell 3 times, and another cell twice). My testing is in a room with fairly constant temperature (basement with no outside doors that could be opened, and no sources of heat other than the furnace).
 
#12 ·
I didn't read all the details so excuse me if this has been covered but I just want to mention that you don't need multiple PL units to do more batteries. I have a PL 8 and I do 6 cells connected in series, at one time. With balance and charge leads connected, you can get individual IR readings and individual capacities. You can balance charge or not depending on settings. u can finish at top or bottom or anywhere u like . . on the voltage scale.
 
#13 ·
An important point for me would be to know the time it takes. I thought the PL6 could only deliver 40A max. If this is distributed over multiple batteries like 6 since that why it has its name that the amperage would go down and the whole process would take longer.

I haven't used my PL6 yet so I'm not positive about this. My goal here is to measure the capacity while bottom balancing the pack. The idea of 6 batteries at once might be useful. Meaning instead of changing the batteries out every 45 minutes, I could change 6 batteries out every 270min (6*45min). This may be work investigating for working with a large number. I still think that I'll need multiple PL6s in order to complete 500 in a reasonable amount of time.
 
#16 ·
I did try that as well, and didn't get consistent results. During discharge it will stop the lowest cell and only discharge at 1A to the cut off point which is not the same as the 38A setting I was using. Perhaps I didn't have it configured properly to do it this way but that was just my experience.

I would get a true capacity on the first cell to discharge but once the first cell hit the low cut off it messed up the results for the remaining cells. During charging it does work, but if the cells are very far off the whole balancing thing doesn't work very well with cells of this capacity. Shunting a maximum of 1A doesn't do much on a 38A charge if the cells aren't at the same SOC.

I gave up on that and went to a single cell, if you have any tips on how to do that better I would be happy to hear them. For me getting a capacity number under the same conditions every single time is more important than shaving off a bit of time.
 
#19 ·
Well, my batteries arrived and they look great. They all have Made in USA on them and have brand new full size tabs. I'm not sure if I'm using the correct settings, but I managed to charge, discharge and charge again a single cell. I'm using a power supply instead of a 12V battery, so I appear to be limited to 10A up and down. This is terribly slow at about 2 hours each.


rwaudio, when you said you were using a block of A123s in 4s3p for a 12V power supply, how did you get started? Did you balance them individually first or just hook them up and charge them as a pack before balancing? I've measured the initial as-received battery votage of each one and they all appear to be at 3.27-3.28V consistently. Charging the first one up took about 10Ah, so mine appear to be half charged as well. The first cell capacity was 18233mAh. This was from 3.65 to 2.5V. I'm not sure if I'm doing this correctly. so if you have any tips on how to do this more efficiently, I would appreciate it greatly.
 

Attachments

#15 ·
thinking about it now, I may have led u a stray. I'm not sure if I got individual capacities or not. I was more interested in individual IR. . . That may be a problem for you if you are grouping according to capacity. Since mine is a high output application, I was grouping according to IR. I can go back and check or maybe RWAudio knows off top of his head since I think he was interested in grouping by capacity.
 
#34 ·
Very Cool. Thank you for sharing that stuff. Since I will probably not be drilling holes through my tabs I will most likely just get beefier alligator clips or clamps that will help lower the resistance. Additionally I will use some shorter 6or8AWG wire as well for the output leads. This should allow my error to drop below 5% or so hopefully
 
#37 ·
I'm glad to hear everything is working well.

I changed my set up slightly as well, I added a second PL6 and decided to leave the cells discharged just like you. I have found that by skipping the final charging phase it takes much longer for the supply pack to get drained since it basically gets the energy from the cells when they arrive to make up for some of the wasted power (heat, running the charger/fan etc.).

I got through 20 cells in a few batches over the weekend and I'm happy with the process now. The only "problem" I'm finding is that this new box of cells that I opened has a higher average capacity than the first 72 cells that I tested. Meaning that I may have a hard time getting low enough average groups to match the first 72. I haven't assembled those first 72 though, so I may start over and re-average the target capacity and re-group the cells before assembly. So it's really not bad news, but the most recent box of cells are almost all in the range of 19.1 -19.5Ah with a 2C discharge.

The one thing I'm noticing with two cells discharging side by side is two cells of the same capacity (virtually the same anyway) but different IR have much different terminal voltage once they dip below 3.2v so if possible when grouping cells for a street car capacity should be the first consideration but averaging the IR should also be in the back of your mind.

The other thing to note is my 12v pack powering the charger now has well over 100 shallow cycles on it along with some 100% discharges and some high current bursts. I use the pack to move my Porsche in and out of the garage when I need to. This pack (12v 50.75Ah) and the Soliton1 work happily together to drive the car (slowly) I've also used it to jump start my Infiniti when it had a dead battery, and my F150 when it was too cold to start. This is a pack of my "defective" cells, IE low capacity, high IR, and or soft cell.

Basically if these reject cells that I won't be putting in my traction pack work this well I have high hopes for what the "good" cells can do and how they will perform.
 
#38 ·
I got through 20 cells in a few batches over the weekend and I'm happy with the process now. The only "problem" I'm finding is that this new box of cells that I opened has a higher average capacity than the first 72 cells that I tested. Meaning that I may have a hard time getting low enough average groups to match the first 72. I haven't assembled those first 72 though, so I may start over and re-average the target capacity and re-group the cells before assembly. So it's really not bad news, but the most recent box of cells are almost all in the range of 19.1 -19.5Ah with a 2C discharge.

The one thing I'm noticing with two cells discharging side by side is two cells of the same capacity (virtually the same anyway) but different IR have much different terminal voltage once they dip below 3.2v so if possible when grouping cells for a street car capacity should be the first consideration but averaging the IR should also be in the back of your mind.
I had some time to kill last week and thought I would see how matching would work. I created a random set of 20 cell caps and tried to group them as if they were my actual cells. It worked out OK using the manual pick-and-choose method to match the averages. It didn't take as long as I thought either. I'm going to wait until I have all the caps before starting the real matching sudoku game.

Which IR measurment are you using for comparison? Sometimes the very first charge IR is higher like 1.4, then after I discharge, the next charge it's 0.8 on the same cell. This was more in the beginning, because now I'm stopping the cycle on discharge. When I was having supply issues and having to restart the whole cycle multiple times, I noticed that I would get different values. It seems you have to wait several minutes for the averaging to stabilize to give a consistent IR value. Have you seen this in your testing?
 
#40 ·
OK, so here are the pics from my latest setup. I have all 3 PL6s up and running :D. Each have independent power supplies made from a module of 4s5p A123 cells. I'm using a board that has bolts mounted from underneath that go through the output leads so that the cell change-outs can be quick. I crossed the output leads before connecting to the cells so that I could place the cells with their writing on top.

I also made up a jig out of another board where I mounted the punches out of a hole puncher. This provided a very clean and consistent hole for all the cells. After you remove the label, there's always a glue residue left behind. I use lacquer thinner on a rag to wipe this off. Fortunately the labels always stay in one piece when you pull them off.

I'm quite proud of this second generation battery design :cool:. I first used the nylon all-thread concept, but it was a hassle weaving it through the tab holes and I didn't like how the rod would distort with any real amount of force applied. So, instead of a solid rod between all the cells, I used a rope. It's only job is to align the tabs and connectors so it didn't need to be special. I finally used a 5/16" all-thread steel rod in the center between the tabs with steel clamps on the ends in order to provide a good amount of clamping force to hold it all together. I have one of those black plastic wire looms around the all-thread to protect against any unwanted fireworks during installation. The rope consequently becomes a lifting handle for the module as well. Here are my latest pics:

 

Attachments

#41 ·
Cell testing is progressing. I just pulled #258 off the bench. I thought I would update my car pack design ideas to hopefully get some feedback from the group.

Each 4s5p module would have 10 pieces of aluminum flashing cut to approximately 7" by 7" square and placed in between every other cell. So that every cell has one side contacting a piece of flashing. This is to provide about 0.25" fins on each cell edge for wicking away heat from the pack. Below is a top view of the design. The electrical connections on top would be made of 1/4" by 2" angle to provide a solid terminal for each module. To minimize the outer dimensions on the clamp side, I plan to weld one end of the all-thread to one end of the clamp and use a lock washer and lock nut on the other to hold the clamp in place. I will torque this lock nut several times between cycles to ensure that there's no thermal expansion problems.

Then I plan to construct a trunk pack out of 12 of these 4s5p modules in a design that minimizes the connection lengths on top - see attached top view. In addition, I will arrange the modules such that perforated air tubes can deliver air from the floor of the battery box. See the green line depicted below. The air inlets will be routed to a safe location to draw clean fresh air. A fan will be installed in the lid of the box to pull air across all the battery fins.
 

Attachments

#42 ·
Cell testing is progressing. I just pulled #258 off the bench. I thought I would update my car pack design ideas to hopefully get some feedback from the group.

Each 4s5p module would have 10 pieces of aluminum flashing cut to approximately 7" by 7" square and placed in between every other cell. So that every cell has one side contacting a piece of flashing. This is to provide about 0.25" fins on each cell edge for wicking away heat from the pack. Below is a top view of the design. The electrical connections on top would be made of 1/4" by 2" angle to provide a solid terminal for each module. To minimize the outer dimensions on the clamp side, I plan to weld one end of the all-thread to one end of the clamp and use a lock washer and lock nut on the other to hold the clamp in place. I will torque this lock nut several times between cycles to ensure that there's no thermal expansion problems.

Then I plan to construct a trunk pack out of 12 of these 4s5p modules in a design that minimizes the connection lengths on top - see attached top view. In addition, I will arrange the modules such that perforated air tubes can deliver air from the floor of the battery box. See the green line depicted below. The air inlets will be routed to a safe location to draw clean fresh air. A fan will be installed in the lid of the box to pull air across all the battery fins.
I am contemplating the same but do you think you will need the cooling? From bench testing these cells at 2C continuous for 30 minutes I only get a rise of 10°F. And if I do another cycle with little to now cool down, I get a lower IR and a percent or two higher capacity. After 1.5 hours at 2C of charging and discharging, it was only at 98°.
 
#45 ·
A friend of mine is posting data points from his testing of cells from a group buy of 500 cells we did earlier this month.

He recently bought an infrared camera so he can measure the thermal qualities when charging/discharging. Without any cooling, he did a 103 amp discharge and saw a max temp of 105 degrees Fahrenheit.
 
#47 ·
I took another way to assemble the cells.
The attached image is a PCB for a 9s3p module.
You put the tabs of three cells in parallel through one slot, fold them to the soldering pad and fasten them with two screws to the next three cells in series.
It will be a 3.2mm thick PCB, that should be solid enough.
Each terminal has a trace to the BMS connector.

I am still waiting for the produced PCB's. Actually it should take 10 working days. But I am waiting since 11th of June, now.

I will get double the number of PCB's I need, because it was almost the same price.
 

Attachments

This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top