Lipo Battery Standards, open discussion

[johnrobholmes]

This thread is not for debate of any particular brand, only to propose guidelines that may help standardize the ratings of lipo batteries. If nothing else, it will allow comparison between batteries on a level playing field. If you have hard discharge data on a particular cell, fell free to post it up with your interpretation on how it may fit in with said proposed guidelines.


I will start. This is what I think could be a good basis for battery ratings.



Continuous discharge rate with average voltage no lower than 3.3v/cell
Continuous discharge rate with no lower than 90% of 1C discharge capacity
Starting and ending temperature noted, along with any cooling and setup details. Temperature could be overlayed on the discharge graph.
Cutoff voltage should be 3.0v.

Burst ratings are much more difficult to pin down, but duty cycle may be a useful parameter to use with this.

10% duty cycle for burst rating (maybe 5%?), average pack voltage of 3.7v for instance.
Sustained duty cycle at 50%, average voltage at 3.7.
Pack should never drop below 3.05v or 3.1v under load.


To compare with enerland, they rate capacity aggressively. Their 2100LP cells generally hold closer to 2050 at 1C discharge, but at the continuous rated discharge at 18C yields right at 3.4v/cell and more than 90% of rated capacity. They do not show temperature or pack cooling during testing. They are up front that there is 70% capacity retention at 16C discharge after 50 cycles.


Please add your thoughts, objections, additions to this.

[johnrobholmes]

No thoughts or suggestions? I figure that one or two of you might have some insight. These can't be accepted standards if there is no feedback! I suppose I could just force them on you though

[BrianG]

This sounds good to me! It would definitely help substantiate some manufacturers claims and weed out the packs with inflated ratings.

For continuous C rating:

Discharge graphs at 1C, 5C, 10C, 15C, 20C, etc should be done while measuring voltages and pack temperatures taken at 30 second intervals for each C rate.

The voltage drop at a particular current would allow you to graph internal resistance since it's non-linear and is a function of C rate.

The measured temps should then dictate the max discharge C rating the pack is good for. And anything over 120*F or 30*F above ambient (whichever comes first) should be considered non-useable.

Not sure what the minimum allowed voltage should be during the test for the associated C rate to be considered valid. 3.5v would be nice, but is that too optimistic? I think 3.4v would be acceptable.

For burst C rating:

This is a little trickier since I doubt anyone is going to agree what exactly constitutes a burst. Heli burst is probably different than land vehicle burst. And racing has a different burst profile than bashing. And what is the period (frequency) and the "resting" C value between bursts? We know current does not go to 0A in between these bursts, so that has to be taken into account. So, I would suggest using a worst-case test.

Discharge graphs of 15C, 20C, 25C, 30C should be take while measuring voltage and temps. A burst should consist of 2.5 second 10C "rest" period and 0.5 second bursts (~17% duty cycle). Anything over 130*F (or 40*F above ambient) should be considered unusable. That should cover basically anything.

Again. not sure what the minimum allowed voltage should be during the test for the associated C rate to be considered valid. I'm thinking 3.2v would be acceptable.

Results

Using the results from these tests/graphs, manufacturers could then rate batteries using something like:

This pack is rated for 20C continous with no less than 3.4v (for 90% of the cycle) @ no higher than 120*F temperature.
This pack is rated for 30C burst with no less than 3.2v (for 90% of the cycle) @ no higher than 130*F temperature.

Sounds similar to how amplifiers are rated "rated 100 watts RMS X 2 channels into 4ohms from 20Hz-20kHz with no more than 0.05% THD"...

Originally Posted by BrianG

Results

Using the results from these tests/graphs, manufacturers could then rate batteries using something like:

This pack is rated for 20C continous with no less than 3.4v (for 90% of the cycle) @ no higher than 120*F temperature.
This pack is rated for 30C burst with no less than 3.2v (for 90% of the cycle) @ no higher than 130*F temperature.

As a calibrator (someone who measures things, like voltages, for a living to test for accuracy and consistency) I really like the sound of that. Precise, consistent, and most importantly, repeatable.

[johnrobholmes]

I would be a bit more liberal with temperatures, around 140*f is my max temp for lipo. Of course as temp goes up R goes down. That is really what is missing in most discharge recommendations, and it may be why many cells do not seem to live up to specs. Testing a cell at 80*f and 120*f makes a HUGE difference in voltage under load. It really would be as easy as stating a few temps and the correlating bursts that the pack can handle without voltage dropout.

I like your idea for burst rating. I have come across 10 sec burst, 60 sec rest from a few factories, and that seems to be the same numbers I get when I talk to importers. The important point is that the method of determining burst is rarely disclosed by the retailer. I know of none that do right now.


Pack cooling will also play a larger roll. If airflow is used to cool a pack then the C ratings may be inflated since the temperature may not hit the cutoff.

[BrianG]

yeah, the temps are a little on the conservative side, but you also don't want to give people the idea that it is OK to run them at 140*F because you KNOW people will push it (just like everything else).

As far as the rating, maybe change to:

This pack is rated for 20C continous with no less than 3.4v (for 90% of the cycle) @ 120*F (+/- 5*F).
This pack is rated for 30C burst with no less than 3.2v (for 90% of the cycle) @ 130*F (+/- 5*F).

And then a little note stating the absolute max temperature and the fact that performance will suffer with cooler temps. Also, maybe a little blurb to summarize the burst test: Burst = repetitive 30C rate for 0.5s with 10C rate for 2.5 seconds.

[johnrobholmes]

If the cont. ratings are 3.4v average it will exclude many many brands from their ratings. Even some of the enerland cells don't hold that.

This is where the 90% rule can come into play. If you can get 90% capacity at 3.3v under cont. load you are good. If you can get 90% capacity at 3.4v under cont. load you are better.

Almost too much to consider.

[BrianG]

Ok, then, how about this:

This pack is rated for 20C continous discharge with no less than 3.3v (for 90% of the cycle) @ 120*F (+/- 5*F).
This pack is rated for 30C burst discharge with no less than 3.1v (for 90% of the cycle) @ 130*F (+/- 5*F).

The maximum temperature allowed for safe lipo use is 140*F. Performance will be reduced when running cell temperatures other than the ones listed. Burst rating is defined as the repetitive 30C rate for 0.5s with 10C rate for 2.5 seconds.

[johnrobholmes]

yeah, thats reasonable. The burst voltage@load will certainly be very relative to pack temp too. Man I really want to get some test equipment for burst now. I need to make some phone calls.

[aqwut]

Originally Posted by johnrobholmes

yeah, thats reasonable. The burst voltage@load will certainly be very relative to pack temp too. Man I really want to get some test equipment for burst now. I need to make some phone calls.

Get them... I love looking at graphs...

[BrianG]

Originally Posted by johnrobholmes

yeah, thats reasonable. The burst voltage@load will certainly be very relative to pack temp too. Man I really want to get some test equipment for burst now. I need to make some phone calls.

No offense, but I doubt you would be considered an "impartial third party" for some people.

Ideally, it should be someone with the right test equipment that the industry would recognize. That way, resellers would be willing to send sample cells to rate - heck they might even WANT to do this! Then, various resellers can post the graphs/specs of the cells on their site (maybe even a datasheet?) and put little rating stickers on the packs containing these cells.

Even though the testing isn't perfect due to the various variables at play, at least it would be a VERY big step in the right direction.

[johnrobholmes]

The only problem is that the people willing to spend the cash and time will generally never be "impartial third party". I really don't care that I won't be considered impartial. I just like having complete data when I make a decision on something and to me that is worth it. If anybody could be considered impartial with the knowledge to test this stuff, it would be you. Do you have the time and resources though?

[BrianG]

I could make the time, but don't have the resources.

If I was to do it, I'd probably get a bunch of dummy loads and put them in combinations of parallel configs to get the approximate load needed. Then, hook the cell up to an Eagletree (or something similar) and measure voltages and temps in 10 second intervals for the continuous test, and 10-50 milli-second intervals for the burst test (to capture the 0.5s pulse readings). This is assuming the ET device can be modded to handle higher currents without insertion losses (only being used for voltage). Also, I assume the raw data can be saved? Need those for custom calculations and graphs.

That said, what reseller is going to go on the word of just some dude doing these tests in their basement? I was thinking more of an already established third party testing company.

Originally Posted by BrianG

I could make the time, but don't have the resources.

If I was to do it, I'd probably get a bunch of dummy loads and put them in combinations of parallel configs to get the approximate load needed. Then, hook the cell up to an Eagletree (or something similar) and measure voltages and temps in 10 second intervals for the continuous test, and 10-50 milli-second intervals for the burst test (to capture the 0.5s pulse readings). This is assuming the ET device can be modded to handle higher currents without insertion losses (only being used for voltage). Also, I assume the raw data can be saved? Need those for custom calculations and graphs.

That said, what reseller is going to go on the word of just some dude doing these tests in their basement? I was thinking more of an already established third party testing company.

Just an afterthought. Look into a calibration company to do the testing that you are looking for. They are certified and inspected by NIST on a regular basis, have extensive, and expensive test equipment, and tons of dummy loads, as well as be able to provide you the raw data and what their findings are.

Not a lot of people know about calibration so it is overlooked. But this may be the way to go. Like I said, just an afterthought.

[hoober]

A number of testers on rcgroups have really slacked off producing the graphs. There were at least two that were very good and had capacity to test at high current and power.

Why have the battery graph vault slowed down? Maybe it can't keep up with the changes and various vendors. I definately checked the graphs and like the format of the CBA tester. At least there's one constant (the current) a pull for each current is done and overlayed nicely.

I also like the way kunlang.. is sorting his currents all on one graph too. It is a lot of info on one graph.

Would you like to see single cells or pre-made packs?