1. Attachments are working again! Check out this thread for more details and to report any other bugs.

Discussions about LiFePO4 cells and Battery University

Discussion in 'Prius PHEV Plug-In Modifications' started by dan2l, Dec 16, 2012.

  1. wb9k

    wb9k 09 Gen II Prius w Hymotion Plug-In Batt

    Joined:
    Oct 27, 2012
    174
    99
    0
    Location:
    Commerce, MI, USA
    Vehicle:
    2009 Prius
    Model:
    N/A
    Interesting....please continue. I asked about the integrity of your cell logs data...let me ask it from a different angle. How do you know the balancers are coming on when they're supposed to and that they're drawing the amount of current they're supposed to? 2A on the Mott boards sounds like total max current for the whole board. This would equal 333mA per cell (Edit: D'oh! make that 666mA per cell). Do you collect any data directly through these control electronics that shows that the controller is seeing what you are seeing in CellLogs?
     
  2. dan2l

    dan2l 2014 Prius v wagon

    Joined:
    Nov 26, 2009
    654
    263
    5
    Location:
    Vancouver Wasington
    Vehicle:
    2014 Prius v wagon
    Model:
    Five
    Hi W,
    I think you are not understanding. The resistor and electronics is buried inside the MottCell Plastic box. the box is sealed.
    DSCF5667.JPG
    There is no way to know the draining current. The only way I know that they are being balanced is when I see the voltages getting closer together as shown in the charts.

    Note that the opened box was made before the resistors were added so you do not see resistors but you do see the blue Cylindrical cells and what looks like RTV to pot them in place.

    Thanks,
    Dan
     
    [email protected] likes this.
  3. wb9k

    wb9k 09 Gen II Prius w Hymotion Plug-In Batt

    Joined:
    Oct 27, 2012
    174
    99
    0
    Location:
    Commerce, MI, USA
    Vehicle:
    2009 Prius
    Model:
    N/A
    Ok, at least that is now clear to me. It's really unfortunate there is no way to service or even verify operation of the balancing electronics. This could be a real source of trouble, but it may also be really difficult to pinpoint something without performing what would presumably be destructive teardown of the whole assembly. Your cell replacement options are limited too. Do you have a schematic of the electronics? You appear to have instances of adjacent cells (6&7) diverging, which can be a symptom of erroneous Voltage readings in control electronics.

    Looking forward to the next installation....
     
    [email protected] likes this.
  4. dan2l

    dan2l 2014 Prius v wagon

    Joined:
    Nov 26, 2009
    654
    263
    5
    Location:
    Vancouver Wasington
    Vehicle:
    2014 Prius v wagon
    Model:
    Five
    Hi W,
    Let me resond to a couple of your comments.
    No, Enginer did not give information on any of their designs.

    I hope I did not add to confusion with that chart. It was very early in the process and so it was before the pack really got balanced. I have seen eroneous voltage reading where adjacent cells diverged before but that is not what is happening here. In this case it was just a balance issue. 7 is low and 6 is high. They each show as a spike at the end of discharge or the end of charge. This is normal when one cell is low or high.
     
  5. dan2l

    dan2l 2014 Prius v wagon

    Joined:
    Nov 26, 2009
    654
    263
    5
    Location:
    Vancouver Wasington
    Vehicle:
    2014 Prius v wagon
    Model:
    Five
    So on with the story....
    After about 1.5 years of good operation I started to get a cell in the RealForce 2kwhr pack going low earlier and earlier. I started paying more attention to my data but the issue did not show in that data. I first thought it was a balance problem so I added some charge to that cell. Over the next couple of weeks that cell got worse.

    I finally pulled that 2kwhr pack out and then with the 2 packs separated I could see that the bad cell was also going way high at the end of charge. I estimate that I was getting maybe 15ah out if that cell and that was limiting the functional capacity of the whole system. This cell was #16. I am assuming that the long sensor wires that connected the two packs together had enough resistance that the high and low cell voltages became hidden from the CellLog. I expect this is because sugnificant current flow in those wires caused a voltage drop in the wire and that confonded the readings.

    Anyway I then replaced cell #16 and balanced the 2kwhr pack on the bench. I then put it back in parallel with the 4kwhr MottCell pack but I did not tie the 2 sets of sensor wires together. So now I have The Mottcells on the Enginer BMS with high voltage and low voltage cutoffs working. Here is yesterdays data for the Mottcells... 2z19Mottcell.jpg
    My plan was to have the Realforce pack so well balanced that it would never go high at the end of charge and I think I have acomplished that. However #11 goes low and still limits the systems functional capacity. Here is the same data for the Realforce pack.
    2z19RealForce.jpg

    Hopefully now you can see and understand better my entire system.

    Thanks,
    Dan
     
  6. wb9k

    wb9k 09 Gen II Prius w Hymotion Plug-In Batt

    Joined:
    Oct 27, 2012
    174
    99
    0
    Location:
    Commerce, MI, USA
    Vehicle:
    2009 Prius
    Model:
    N/A
    Dan,

    Are the first four frames of data here Mott cells only, on four consecutive days? If so, I would say there is either a problem with the balancers or some fairly severe cell defects, or both. The balance is getting worse and worse for four days in a row--if the balance scheme is working, it should be getting better--sort of like what it happening in the final frame, but much more slowly. By the first day with the BMS (second frame), the imbalance is severe enough that I would want to manually pull down those top three cells at least. If these were A123 cells, I would say you are definitely doing some serious Li plating with all the time above 3.8, 3.9, and even 4 Volts. You do not want to expose any cells to that--over time they will become your weaklings unless you act quickly to stop the abuse. However, these are Mott Cells and I have no idea what the recommended Voltage range for those is. You may be fine. Still, the way balance appears to worsen on a daily basis under normal use tells me either the balance scheme needs refinement, the hardware is not working properly, some cells are defective, or some combination of those things. It sounds like you go from 15 A right to the float charge. I would be more comfortable seeing other current levels in between, taking hold when any cell hits 3.8 or maybe less--so that would be a refinement in the controls strategy which would help limit plating. At the same time, I have to question if you've got some balancers that are always stuck on or off, OR you've got several cells with serious self-discharge problems or impedance growth. You'd have to run several tests to get a handle on what parameters have changed over time. If you have daily data, or enought periodically collected data you might be able to see what the trends are. If cells that started off spending a lot of time at the high side of the balance spectrum are the same cells that are now fading the fastest, this would be a strong indicator for Li plating. Going too low doesn't appear to be a problem for you at all (but again, I'm thinking in A123 numbers, don't know what Mott or RealForce want to see). I would want to be sure whatever the permissible Voltage numbers are, that they were the same for both Mott and RealForce since they are presumably being charged by the same system using the same rules.

    I'll stop there for now and see how I'm doing with staying in the right ballpark....
     
  7. lopezjm2001

    lopezjm2001 Senior Member

    Joined:
    Apr 14, 2009
    1,146
    407
    5
    Location:
    Sydney Australia
    Vehicle:
    Other Electric Vehicle
    Model:
    N/A
    I am dubious about how well these CellLogs keep their accuracy over time and also taking the range of ambient temperature that they would normally be subjected to into account. May be worth checking each cell volts CellLog value against a 3.000 volts calibration cell or source at the highest ambient temperature expected during their use.

    You might find them a bit inaccurate when they are working at an ambient temperature of 40 Deg.Celcius which may happen if the hot air given off by your charger increases the ambeint specially by the end of charge which is when they need to be at their most accurate.
     
  8. wb9k

    wb9k 09 Gen II Prius w Hymotion Plug-In Batt

    Joined:
    Oct 27, 2012
    174
    99
    0
    Location:
    Commerce, MI, USA
    Vehicle:
    2009 Prius
    Model:
    N/A
    What would be the mechanism causing such a loss of accuracy? This would surprise me, unless the design of the electronics is extraordinarily bad.
     
  9. dan2l

    dan2l 2014 Prius v wagon

    Joined:
    Nov 26, 2009
    654
    263
    5
    Location:
    Vancouver Wasington
    Vehicle:
    2014 Prius v wagon
    Model:
    Five
    Hello Lopez,
    I am confident that the CellLogs are acurate. They repeat day to day and they repeat when I change CellLogs. I have used about 8 of them and never seen one go out of calibration. Also I have ocationally double checked the cell with a good quality DVM.

    I do not see the temp extreems that you have. Seldom below freezing and seldom over 100F. Have you ever had a CellLog go out of calibration?

    Thanks,
    Dan
     
    [email protected] likes this.
  10. dan2l

    dan2l 2014 Prius v wagon

    Joined:
    Nov 26, 2009
    654
    263
    5
    Location:
    Vancouver Wasington
    Vehicle:
    2014 Prius v wagon
    Model:
    Five
    Hello W,
    Yes, the BMS Balancing left a lot to be desired. Yes, I am sure that I had some plating in the Mottcells.

    Enginer has never had a good balancing BMS. They started with 2 DB8's. Most every system had cell failures because the user did not hear the beeping. Their was no automatic shut down at that time so the converter eventually pulled a cell low and caused damage. Later we also found that the DB8s failed from getting too hot.

    Then Enginer shipped without a BMS for a while. The converters were set to turn off at a total voltage of 46v so if the pack was in balance, that was OK. They did ship CellLogs for monitoring but did not turn off the system. I connected the CellLog allarm to a very very loud siren. This allerted me to turn off the charger or stop the discharge. Even if I was not inside the car. It was very, very loud.

    Then Enginer developed the BMS16D which is what you see in the data. So I had a pack that was balanced because it was new cells from the factory. Then the BMS pulled trickle power from some cells and not others. Then during charge we see that if a single cell or 3 cells were highest they went up to 4v. Jack said this was OK for 10 minutes or so. Then the balancing slowly equallized the cells at the top.

    I later found that I had reduced capacity at discharge rates above 1C but did much better when I added the second pack because that dropped the discharge rate below 1C. That led me to believe that I had degraded the cells.

    Realize that this data was from the summer of 2010. There was no one on any forums giving good information on LiFePO4 cells at that time. We were getting opinoins from various people and learning as we went. The best information I had at the time was that short periods at 4v were OK as long as I did not let the cells get warm.

    Thanks,
    Dan
     
    [email protected] likes this.
  11. wb9k

    wb9k 09 Gen II Prius w Hymotion Plug-In Batt

    Joined:
    Oct 27, 2012
    174
    99
    0
    Location:
    Commerce, MI, USA
    Vehicle:
    2009 Prius
    Model:
    N/A
    The more I hear about Enginer, the less I like the system. Seems like a lot of fumbling around in the dark. It's fine to develop a system that way, but to sell a system like that seems crazy to me. Anyway....I'm not sure if I've been of any help to you here or not. Has any of this conversation been useful for you? Any further questions or ideas you want to bounce off me? I'm enjoying the discussion...not sure about anyone else. :)
     
  12. dan2l

    dan2l 2014 Prius v wagon

    Joined:
    Nov 26, 2009
    654
    263
    5
    Location:
    Vancouver Wasington
    Vehicle:
    2014 Prius v wagon
    Model:
    Five
    Hello W,
    Yes, This has been helpful to me. You are the battery expert that we all needed 3 years ago. Much of what you discribe, I have experianced and guessed at over the three years. It is not important to me that you have confermed much if it from independant work.



    Thanks,
    Dan
     
  13. wb9k

    wb9k 09 Gen II Prius w Hymotion Plug-In Batt

    Joined:
    Oct 27, 2012
    174
    99
    0
    Location:
    Commerce, MI, USA
    Vehicle:
    2009 Prius
    Model:
    N/A
    Dan,

    I'm glad to help advance the cause and I appreciate the dialogue. 3 years ago I would have been right there trying this stuff with you if I could have afforded it.... About 5 years ago I was playing with WVO instead--a PITA, but do-able on a shoestring if you've got a couple simple things going for you. Anyway, I've only been at A123 for just under 2 years now, and while I've been dabbling in battery and solar power for many years (my WVO pumps at the house all ran on solar-charged LA batteries), I've had to dive MUCH deeper since arriving here. It's been a baptism in electrons--this is still a frontier industry and we are ALL learning, mostly the hard way. You've been at this longer than me, but I'm sure with a fraction of the resources--and my resources are nothing to write home about. Point is, I'm sure you could show me a thing or two because you've lived with a system for a long time now and have witnessed certain trajectories.

    More to come....
     
    [email protected] likes this.
  14. lopezjm2001

    lopezjm2001 Senior Member

    Joined:
    Apr 14, 2009
    1,146
    407
    5
    Location:
    Sydney Australia
    Vehicle:
    Other Electric Vehicle
    Model:
    N/A
    Hi Dan,

    I used the word "dubious" because I have never checked for accuracy as I have never used them for voltage protection on my PHEV kits so in my case there is not much point in checking them for accuracy. Maybe I am wrong but I think that they were never intended for PHEV kits even though they seem to be doing a good job for you and others. I only use them as a quick visual check at the end of a discharge. I just do not have the time and will to do all the data logging collection that you have done and I cannot say with any confidence that the Li plating (or what ever is causing your cells to lose capacity) of your RFE cells was caused by your Pacific EV BMS or whether the RFE batteries just do not last the distance as the RFE 20-ah cell Spec. sheets claim.
    Anyhow this thread has been interesting so far.
     
  15. dan2l

    dan2l 2014 Prius v wagon

    Joined:
    Nov 26, 2009
    654
    263
    5
    Location:
    Vancouver Wasington
    Vehicle:
    2014 Prius v wagon
    Model:
    Five
    Hello Everyone,
    I drove 35 miles yesterday and estimate that 12 mile of that was from electrons from the Enginer system. I am guessing that means that I used a little more than 3kwhr out of my 6kwhr packs. The charge was 4 hours at 15a and then another hour ramping down the current. So that might be about 3.5kwhr put back in during charging.

    Here is the data from the MottCells 2Z20Mottcell.jpg
    and below is the data from the Realforce pack. Note that I changed the linetype for cell 11 so it would stand out a little more 2Z20RealForce.jpg
    The CellLog on the Realforce pack beeps when any cell goes down to 2.5v. That happened at 9:57 on the end of discharge chart. I turned off the system for about 3 minutes and the cell recovered about 0.6v to 3.1v. Then I would turn the pack on again and then off when I heard the beep. I assume this is OK because the beep is way higher than the 1.5 v level. You can see that at the end, the cell was recovering only to about 2.7v. Everything I have read indicated that 2.7v at rest was OK for LiFePO4 cells.

    Then the charge again looks like it always has been, with 11 giving a bump up to 3.5v and holding there at during the chargers constant voltage mode.

    I have some very small cells. About 3ah each. I am planning to charge one up and put it in parallel with cell 11. I am thinking that this will improve cell 11 enough to be able to see what the next weakest cell might be.

    Thanks,
    Dan
     
  16. wb9k

    wb9k 09 Gen II Prius w Hymotion Plug-In Batt

    Joined:
    Oct 27, 2012
    174
    99
    0
    Location:
    Commerce, MI, USA
    Vehicle:
    2009 Prius
    Model:
    N/A
    The bit of extra capacity trick should work, but of course there are limits to how bad the problem can get before it makes more sense to just replace the whole cell.

    Can you repost the "all data" charts for both packs with the weak cells (8 and 11) highlighted like what you did with the close-ups for the RealForce pack? I may be able to suggest some strategies for improving your performance and your understanding of what has gone wrong with these cells.
     
  17. dan2l

    dan2l 2014 Prius v wagon

    Joined:
    Nov 26, 2009
    654
    263
    5
    Location:
    Vancouver Wasington
    Vehicle:
    2014 Prius v wagon
    Model:
    Five
    I don't think I can make anything clearer in an image that I can post. If you want more detail I can Email you the raw data.

    Thanks,
    Dan
     
  18. dan2l

    dan2l 2014 Prius v wagon

    Joined:
    Nov 26, 2009
    654
    263
    5
    Location:
    Vancouver Wasington
    Vehicle:
    2014 Prius v wagon
    Model:
    Five
    Hello Everyone,
    I put 2ea 3.3ah cells in parallel with my week cell 11. The realforce pack is sealed in a metal box, so this added capacity is on the 22awg sensor wires. The power flow is done slowly so the current is low. However the voltage drop on the wires is enough to see in the voltage readings of the adjacent cells. So when current is flowing out of the added cells, 11 reads a little higher than the actual voltage at the Realforce cell, and cells 10 and 12 read a little lower. As soon as the current flow is stopped the reading become acurate again. So while I and using the system, at the end of discharge 10 and 12 go low and give the alarm. as soon as I turn off the pack I see 2.8v on 13 and 3.0v on the others. so 13 is the next lowest cell, but 12 is actually giving the alarm.

    Also I did two runs of about 30 miles each with a 95% charge between them. Both runs gave about 1/2kwhr more power. This is more than I would expect from the 6.6ah capacity that I added to cell 11. So I am thinking that the lower discharge rate on cell 11 also allowed it to use a little more of its cappacity that was not available before.

    I am thinking that if I could get real readings on 10 and 12, then I could even safely get more capacity from the pack. However I am afraid of making an assumtion that all is OK to lowere voltages so I will leave it where it is for now.

    Thanks,
    Dan
     
    [email protected] likes this.
  19. wb9k

    wb9k 09 Gen II Prius w Hymotion Plug-In Batt

    Joined:
    Oct 27, 2012
    174
    99
    0
    Location:
    Commerce, MI, USA
    Vehicle:
    2009 Prius
    Model:
    N/A
    I strongly recommend finding a low impedance path to connect to cell group (CG) 11 in the Realforce pack. Current flowing in your sense lines causes all kinds of problems with monitoring, as you are seeing here. You'll definitely want to disconnect those extra cells during charging of the pack and charge them separately. Left connected, you'll get the same issues during charge, and mismanagement by the control system will be the result. In the longer term this can result in serious electrical abuse of the cells. Next to that, the problems you've been having might not look so bad anymore. There should be virtually zero current in your sense lines. Putting current flow (power) there is trouble, period. It's possible you are also defeating the design intent of the control electronics in other ways too.



    This is a sure sign of impedance growth in your Realforce cells. As impedance grows, higher C-rate capacity falls, but at lower C-rates the total capacity will show less change.




    Do Mott and Realforce provide any specs for the batteries they are selling?
     
    [email protected] likes this.
  20. wb9k

    wb9k 09 Gen II Prius w Hymotion Plug-In Batt

    Joined:
    Oct 27, 2012
    174
    99
    0
    Location:
    Commerce, MI, USA
    Vehicle:
    2009 Prius
    Model:
    N/A

    Actually, you may be within the margin of error depending on how you arrived at your 1/2 kWh number. 6.6Ah * 3.0 V=19.8Wh. Cell 11 having this additional capacity enables the rest of the cells in the pack to contribute just as much (assuming nobody else bottoms out first): 19.8Wh * 16 cell groups = 316 Wh, or about 1/3 kWh. Does that seem close enough to explain your result?
     
    [email protected] likes this.