Proposal for simple lipo performance figures of merit - RC Groups
Oct 29, 2009, 05:36 PM
Registered User
Idea

# Proposal for simple lipo performance figures of merit

A topic of much interest to many is "which lipo is best". Comparing lipos to each other can be a complex business. Many of us value the time and effort that folks put in to perform tests and post results, frequently in the form of discharge curves.

Discharge curves provide good insight into pack performance but can become challenging to read when many test runs are plotted together. Not everyone can easily read the many-coloured traces and not everyone can easily interpret what they see.

I have an idea for a way to condense the results from a discharge curve into three simple normalised numbers, each in the range 0 to 100. A perfect battery would score 100/100/100.

Using this approach it becomes easy to compare scores between packs one-on-one or many-to-many in either a simple table where each row represents a particular test run on a single pack or in a simple scatter plot where each data point represents a particular test run on a single pack and up-to-the-right is best. Such a table could be sorted by the voltage column or the capacity column depending upon whether higher power or higher duration was more important for a given user or application.

Normalised figures-of-merit derived from a discharge curve
• Average voltage/cell, excluding the first and last 10% of values, as a percentage of the nominal cell voltage (3.7V for lipo).
• Capacity (mAh) as a percentage of the nominal capacity
• Final temperature rise as (100 - percentage of (140F - startTempF)) or (100 - percentage of (60C - startTempC))

The first parameter gives an indication of the power under load (indicative of IR).
The second measures usable capacity under load.
The third provides insight into the likely lifetime of the pack under load (temperatures above 140F seriously degrade lipo cycle life).

A perfect battery would score 100/100/100 indicating 3.7V average voltage under load, 100% nominal capacity delivered and zero temperature rise.

Using these figures-of-merit it would be very easy to populate a table that makes it easy to compare packs under the same discharge conditions.

Further, you could create an easy to understand plot with a single point for each pack on a set of axes where x-axis = delivered_capacity_% and y-axis = loaded_voltage_% and it would be clear which packs were most up_and_to_the_right and hence the best under those test conditions.

A results table might look like this:
Code:
```----------  ---------- ---------  --------  ------  -----  -------------------  ------------------------------
Brand       Model      Rated mAh  C-rating  Weight  Cycle  Discharge            Score
(g)   count  current (A)  C-rate  Voltage  Capacity  Temperature
----------  ---------- ---------  --------  ------  -----  -------------------  ------------------------------

----------  ---------- ---------  --------  ------  -----  -------------------  ------------------------------```
In my opinion these three figures of merit say a lot about a pack's performance in a way that is concise, easy to understand and easy to compare. I am very interested to hear from other members about these ideas.

Last edited by kgfly; Oct 29, 2009 at 06:22 PM.
 Oct 29, 2009, 05:47 PM Southern Pride Way back https://www.rcgroups.com/forums/showthread.php?t=489564 more recently https://www.rcgroups.com/forums/show....php?t=1055837 most recently here Steve Neu proposed 10c only testing. On anothee Site a couple had ideas of simple to understand graphs but never did anything with the data. Good luck. Charles
 Oct 29, 2009, 05:51 PM Registered User Steve's thread... https://www.rcgroups.com/forums/show....php?t=1104509 Mark
 Oct 29, 2009, 06:02 PM Intensely Calm I think the first step to prove or see the value of this proposal will be to crunch some data into this format to see how it looks, it certainly appears sounds in principal. The hardest part is getting the average voltage (excluding the first and last 10%), but if eyeballing it on a graph is good enough, then it would be fairly simple. Also, method of measuring temp needs to be consistent and accurate, many temp probes (such as the one on the CBA) have considerable lag at high rates of discharge and don't accurately measure the final temp. - David
 Oct 29, 2009, 06:10 PM Registered User Thanks, I followed Steve's thread for a while although have not read every post. It seemed to circle mostly around the merit of "C" ratings and suggestions for a standardised test profile. The end results are still discharge curves, although some did suggest results based on a cluster of figures. I did not see (although again, I have not read every post) a suggestion like this one for a normalised set of figures-of-merit that can provide simple comparison between packs. Of course for sensible comparison the scores would have to be published along with the test information as shown in the example table above. One option might be to add a fourth number which is the load current as a percentage of the rated constant load. That would ensure presenting the score always included that critical aspect of the test conditions. Mark/Charles, you are both very experienced with testing batteries and I suspect with the frustration of trying to explain the results. What do you think of my suggestion ?
 Oct 29, 2009, 06:19 PM Registered User David - I agree and when I get some time I will try to prepare some examples. Although not ideal, I think that the average voltage could indeed be estimated by eyeballing a graph. One simple heuristic could be to simply pick the voltage at the 50% discharged point since most lipos are fairly linear in the middle 80% of discharge. Sometimes cell heating at higher C-rates can cause a rise in voltage during the discharge which could make estimating the average a little harder, but I still feel that a reasonable estimate could be made. If there is enough interest I could potentially create a tool for processing CBA results files to generate the average voltage according to the proposed rule. It is true that comparisons between results gathered by different folk with different equipment or under varying conditions will not be precisely comparable, but that is true regardless of how the results are presented. In many cases we see discharge curves from the same individuals who are sharing their results across multiple cycles or multiple packs. Whatever their temperature measurement technique, it is usually constant for all the tests they run, so the comparisons within their results are perfectly valid, even if the absolute temperature values may differ from those that might be obtained by someone else performing the same test. There will always be debates about whether one person's results can be compared to another's. Using these figures-of-merit does not address that problem, butI think it would make it easier to compare either way.
 Oct 29, 2009, 06:27 PM Southern Pride IMO no matter how you show the data most will still want you to make a recomendation and then want you to expalin why 12 different ways. David went to a great deal of trouble to create hig references for chargers, Lipolys,Gyros etc. . If you point someone to say his chargers spread sheet they still have no idea how to make a selection. I like CBA II graphs. IMO they are not that dificult to read and understand if ones really wants to learn. Yes those with a color blindness issue are at a great disadvantage but what percentage of members is that? To me the biggest problem is reaching a consenses how batteries really need to be tested. I have never really cared for constant current testing as I do not fly at CC in fact it is imposible to do so. If I test at 12 to 15 C CC which IMO is more demanding that most flying many say where are the max. C rated discharges. If I do burst cycles at say 7 to 18C with a 12.5C average then few seem to get what they are showing. Charles
Oct 29, 2009, 06:29 PM
Registered User
Quote:
 Originally Posted by kgfly Mark/Charles, you are both very experienced with testing batteries and I suspect with the frustration of trying to explain the results. What do you think of my suggestion ?
I think it's a great suggestion but one that may ultimately be a significant challenge to implement. Those who have the ability to understand and interpret graphs and other performance data base their purchase decisions on such. For others, they look at vendor C rating, pack price, and little else. Some are in between.

My thoughts regarding Steve's proposal here:
https://www.rcgroups.com/forums/show...74&postcount=6

One other very significant consideration for me is pack longevity. I've purchased some lipolys in the past that were amazing for the first ~30 cycles / 6 months and then went straight to crap. Pack longevity is actually a larger consideration than voltage under load for me personally. I'll take a slightly lower performing pack if it'll last twice as long. And I'm willing to pay more for it.

That said, easy to interpret normalized testing would be beneficial. Might be an interesting exercise to take some existing performance data and repackage it as you suggest just to see where the chips fall...

Mark
 Oct 29, 2009, 06:55 PM Registered User Mark/Charles, Thank you for your comments. Kenneth