They will definitely sag less than 30Q.
1 Like
Great Test!
I have been thinking about something similar for weeks now. Our problem when choosing cells is the lack of representative data. We usually only have capacity at 0.2A and then continuous 5A, 10A, 20A etc. tests. None of them really represent our use case.
My idea is to come up with some kind of standardized esk8 cell test so we can compare cells better. I think we should have at least 3 tests like: City riding, Medium load and aggressive racing.
This profile would be very aggressive riding/racing in my case. My rides are never shorter than 1 hour and this test is done in less than 15 minutes.
If you think about it. On a normal ride in a city you have free rolling, breaking and even waiting phases and you probably donāt accelerate like a mad man. Just holding up with city traffic speeds does not take more than 5A per cell on my 12s5p.
When choosing cells there is no universal right answer and I donāt think you can generalise that 30Q Ā“s are worse than P26aĀ“s. For racing or when you have a very small pack P26aĀ“s are probably better. But with bigger batteryās or less load the advantage goes back to the 30Q.
The hard part now is coming up with comparable tests that you can repeat for every cell. We could have 3 standard curves that only go up to 5A (something every cell can handle) and then have standard curve *2 and *3 so we have the same profile just with different intensity. Then you can look at every cell and your desired load and know how many you need in parallel to get the desired curve.
If that doesnāt make any sense I can try to explain it better 
I donāt have the equipment to performe those tests. I could build something crappy or we can try and come up with something mooch can test with his professional stuff.
9 Likes
When comparing cell performance though there is almost no difference between the results you get when discharging continuously and when using pulsed discharge profile. The best performing cell when discharged continuously will almost always be the best performing one when pulsed. There is only a couple percent difference or so in performance between cells when comparing continuous/pulsed results.
This is because pulsing a li-ion cell for anything longer than a second or two is essentially the same as running it continuously until itās empty. Nothing new happens inside the cell after that first second or two. Youāre just continuing to move ions from one side of the cell to the other as the discharge continues.
Pulsing a cell is just a series of shorter continuous runs.
Continuously discharging a cell until empty is just one long pulse.
As far as the battery is concerned they are just different lengths of the same thing and therefore the comparative performance of the cells you are testing doesnāt change much.
The small changes between continuous and pulsed performance that do occur are related to internal resistance vs temperature differences between the cells being tested. Some cells change their internal resistance more than other when warm. This can lead to small differences between the relative performance of cells when pulsed and when run continuously, perhaps a couple percent. This is unnoticeable when riding IMO.
Bottom lineā¦thereās very little difference between the relative performance of cells when pulsed versus being run continuously.
This can be done but starts turning into a significant amount of testing. We need to define our goals.
Do we want to know which cells are the better performers at different discharge current/power levels? Or do we want to be able to estimate how much riding time we get for the cells used by the esk8 community?
Comparing performance can be done using continuous discharge testing at different current/power levels or by using a representative pulsed discharge profile. The results between these two will be essentially the same.
Estimating run time needs much more specific pulsed discharges that take a lot more time to run. Hereās where things can quickly expand into days and days of testing, trying to replicate different riding styles.
Which type of testing is more important?
I know just about everyone will say that both sets of results will be great to have but that is a lot of testing. Since pack size is typically limited by the enclosure being used, and thereby fixing the running time to a certain length, does all the work needed to estimate run time make the best use of our limited test time?
Would just knowing which cell would perform better at the estimated current/power levels be good enough? Especially since run time estimation is affected by the cutoff voltage, pack construction quality (voltage drops), and degree of regen being used?
How do we best account for those variables without adding loads of more testing? If we have to tell riders to add or subtract rough percentages for different usage scenarios then how much better are the estimations than what we do now?
Iām not condemning run time estimation testing. Iām just letting my brain run free and writing down things that can be discussed. 
I think some kind of standardized, community-defined testing for esk8 use could be great though.
16 Likes
You have a gift for explaining things in a way even a bowl cut like me can understand. Really appreciate your approach.
9 Likes
Thank you thatās good to know. I thought Running them continuously would make a bigger difference for performance. So I can look at the continuous graphs to see the performance at the specific Current. Still the temperature would probably be a lot different.
My goal however was more about the total energy. My personal approach when choosing cells is I want the most energy possible with enough performance now and then. I very seldom come close to the maximum performance since that just throws me off anyway. My ride is only like 5% over 10A and 1% over 15A. I still want to have the power when I feel suicidal but its really not as important as the energy I need all the time.
If I ride for one hour my average cell current has to be less than 3A.
Yes thatās a problem. Everyone probably has different criteria and it could escalate very quickly. Its hard to come up with a load test that could that could be compared to other cells. You would need at least 3 tests per cell if not more that last around one hour. I donāt know how easy and automated those testes are. I could throw together a microcontroller and electronic loads that could do some in parallel but I also donāt have a lot of different cells and it would not be to accurate.
Im thinking about maybe calculating the runtime based of other tests. Could you combine the different load times together and like 50%3A 40%5A 10&15A and get the energy this way? So basically use existing tests and piece them together?
I would just focus on one cell. Battery construction would be equally better or worse no matter what cell you use.
4 Likes
Performance for comparing different cells, yes. Performance for estimating riding time, definitely not.
Yes, the temperature difference between a continuous discharge and a pulsed profile can be huge.
Three one-hour tests takes at least 13 hours of test time. You have the charging time and time needed for the cells to come back to room temperature.
The tests can be automated if using the right equipment but wouldnāt be a huge issue to do manually. It is one very long day or two days of testing for each cell for just those three tests.
You canāt test just one cell of each type either. Since all the cells we buy, except Molicels, are excess inventory or unauthorized purchases it is much too easy for any particular cell to have been damaged in storage or be over a couple of years old. Different batches of cells are thrown together and then sold to vendors. At least two cells of each type needs to be tested, preferably four. This allows you to not be caught by a particularly bad or good cell but also lets you quantify the cell-to-cell consistency, a big issue with the cells made by the smaller China factories.
This means testing one cell type would take up to a week to do properly. Capacity and internal resistance testing adds another day.
This is why I do the continuous current testing.
It takes āonlyā 2-3 days for the capacity, IR, and performance testing of 2-4 cells.
Iām not sure I understand.
The energy delivered at different current levels for every one of the cell types I test is included in the test report. You can use these numbers to estimate run time.
One cell? Sorry, what are you referring to?
Any cell type would need at least two cells tested.
5 Likes