PDA

View Full Version : Lipo discharge and V = SOC?



podolefsky
29 November 2011, 2058
[I posted this on Endless Sphere as well...just to get ideas from multiple sources...]

I've believed for a while that voltage is not a good predictor of SOC. I know this is generally true with lifepo4, but wasn't sure about lipo. Some folks on ES were saying you could use voltage to determine SOC if you had a look up table. I'm not an expert on lipo, so I'm posting this to see if my thinking is right (or if what missing, because I'm probably missing a lot...)

So I took a Turnigy 20C, 2S, 5.0 Ah pack and did two discharges. First was ~0.6C, the other at 6C. At certain points during the discharge, I removed the load and let the voltage rise for 1 minute and recorded the value. All this is shown in the graph below.

Here's what I see. Discharging at 0.6C, removing the load makes the voltage pop up, but it pops up about the same amount each time (red squares). So looks like this should let you use voltage for SOC (if you have a 0.6C curve to compare).

Discharging at 6C, you get the same behavior but the voltage jump is a lot larger (blue squares). What's interesting here is that the blue squares don't match up with the red squares. If you had the 0.6C curve, but were running at 6C, you would predict a SOC that is about 0.7 Ah too low. That is, you actually have more Ah left than you would predict. (The dashed lines show where the blue lines up with red for voltage, and then the Ah you would predict).

Now, if you had the 6C curve, you would predict the correct SOC. But in the real world, you never go along at constant C-rate. I don't actually know what happens in that case - I need to do something like pull 6C, then switch to 0.6C or 2C, then go back to 6C and see what happens to voltage.

If someone's already done this, I'd love to see it so I don't have to spend more time testing.

http://www.colorado.edu/physics/EducationIssues/podolefsky/EV_project/lipo_discharge.png

lugnut
29 November 2011, 2143
I've believed for a while that voltage is not a good predictor of SOC.......

I think you demonstrated one difficulty with the method in that it is load dependent. Other factors which will present problems are that the slope and shape of the discharge curve depends on specific chemistry and cell construction, age of cell, temperature and likely several other things. Not all LiPo are the same. By the time you take in account all these factors and contend with the accuracy of the very small voltage difference measurements for SOC, it is easier to Coulomb count, IMO.

You can become familiar enough with a specific battery in a specific application to judge its SOC from voltage within a certain margin of error. But I find this takes many cycles where you have observed the discharge and charge quantities and related that to voltage. I would not assume that such an experience would be valid on a different brand or type of LiPo.

podolefsky
29 November 2011, 2145
Yup - all that is what I think.

I'm just arguing with people on ES (because I hate myself I think)

thanks lugnut.

Nuts & Volts
29 November 2011, 2148
IMHO I believe resting voltage of LiPo cells is the best determination of SOC. Problem is most of use dont have time during a ride to let the voltage settle out. I believe that your theory that you would have 0.7Ah is incorrect. The voltage doesnt perk up as much because the more Ah has been converted to heat rather than useable electricity (other losses exist that I can't really explain. Thus at 6C you get about 0.7Ah less or retain about 86% capacity. This seems like more than I would expect. Can you do a 3C or something higher than 6C to get a little bit more data??

PS - I was trolling on that Thread on ES, that guy wasn't making any sense and wasn't listening to anything.

podolefsky
29 November 2011, 2201
A lightbulb just went off...at 6C I lose about 0.7 Ah. Shift the blue to the right and it all works out. :D

Thank you N&V.

DaveAK
29 November 2011, 2204
A lightbulb just went off...at 6C I lose about 0.7 Ah. Shift the blue to the right and it all works out. :D

Thank you N&V.
What wattage bulb? That's going to effect your data to some extent.

lugnut
29 November 2011, 2214
I believe that your theory that you would have 0.7Ah is incorrect. The voltage doesnt perk up as much because the more Ah has been converted to heat rather than useable electricity (other losses exist that I can't really explain. Thus at 6C you get about 0.7Ah less or retain about 86% capacity.

The 0.7 Ah is also the difference where the 6C discharge curve knees over at 4.3 instead of 5.0 Ah. The cells are not 100% discharged here. They would only take 4.3 Ah to charge fully. The 0.7 Ah is still in the cells but unavailable at that time. If you waited a while and discharged again you could get it. The Coulombic efficiency of Li cells is very nearly 100%. In other words you only need to put in the Ah you take out. The loss (heat) in the cell comes from the I^2*R and does not subtract from charge (Ah).

Nuts & Volts
29 November 2011, 2220
Yeps

I also agree with lugnut about different batteries and chemistry needing different options. Like for my EIG cells I have some specs that tell me resting voltage at 100%, 90%, 80%...etc. Turnigy cells arent going to be the same. Also much easier to coulomb count and then use a LVC to determine an "empty" spot at higher discharge rates.

podolefsky
29 November 2011, 2304
The 0.7 Ah is also the difference where the 6C discharge curve knees over at 4.3 instead of 5.0 Ah. The cells are not 100% discharged here. They would only take 4.3 Ah to charge fully. The 0.7 Ah is still in the cells but unavailable at that time. If you waited a while and discharged again you could get it. The Coulombic efficiency of Li cells is very nearly 100%. In other words you only need to put in the Ah you take out. The loss (heat) in the cell comes from the I^2*R and does not subtract from charge (Ah).

Yup - just noticed that about shifting by 0.7Ah. Interesting about being able to still get those 0.7Ah. I can see that, since the apparent loss in Ah is due to concentration polarization, which is temporary, not an actual loss in energy.

frodus
29 November 2011, 2356
Does the voltage of the cell keep rising when you let it sit more? does it eventually settle at a certain voltage?

Don't listen to the ES guys, most of them are way off....

ARC EV Racing
30 November 2011, 0038
are you not using the spyglass meter with your set up? That shows soc and is as accurate as you need it to be, the curtis controller does a very good job of measuring it.

teddillard
30 November 2011, 0329
Can you link that thread on ES please?
(edit found it:http://endless-sphere.com/forums/viewtopic.php?f=10&t=34247 Sorry, was looking in the battery section)

Dunno about this particular thread, but as far as "most of them are way off", I don't think that's particularly fair, considering what they've done over there with lipo (basically my entire strategy for lipo came from there via RC). Seems when I started this madness they were the source for Headway information, too... maybe you were just talking about this particular thread, frodus?

(edit edit: On the contrary, I found that little discussion to be pretty interesting and informative... simple miscommunication to start, that's all. 'Course that never happens here. :D For example, I do believe he answered your question, frodus: "Also, a cell can take a verrry long time to rise back up in voltage after a hard load. i've seen a pack go from 3.3/cell average to about 3.45v/cell average after sitting for ~30 minutes after i beat the daylights out of it at 9C constant and it got all warm/hot.")

podolefsky
30 November 2011, 0507
Does the voltage of the cell keep rising when you let it sit more? does it eventually settle at a certain voltage?

For the 0.6C curve, I actually let it sit overnight part way through. In the morning it was at the same voltage. Seemed to peak out after 1-2 minutes and didn't rise more than 0.01V after that.



Don't listen to the ES guys, most of them are way off....

I know...that's why I did the test myself. I thought the data I collected showed that voltage / SOC relationship depended on C-rate, but what lugnut and N&V said makes sense.

Still, I'm gonna keep testing this little pack in all sorts of ways. I'm really curious about getting those 0.7 Ah back at lower discharge rate. I always thought the Peukert effect was a loss in capacity and energy...but sounds like lugnut is saying that it's only an apparent loss in capacity at high discharge, that the charge is still there. That sort of makes sense, since, as I understand it, Peukert is a diffusion limited effect, not an energy limited effect. If you reduce the C-rate, maybe the voltage will pop back up and you can get the Ah back.

But the cell does heat up more, so there must be energy loss...you might have Ah left, but they're not useful if you don't have Wh. Guess I'll just have to run some more tests :)

podolefsky
30 November 2011, 0523
Can you link that thread on ES please?
(edit found it:http://endless-sphere.com/forums/viewtopic.php?f=10&t=34247 Sorry, was looking in the battery section)

Dunno about this particular thread, but as far as "most of them are way off", I don't think that's particularly fair, considering what they've done over there with lipo (basically my entire strategy for lipo came from there via RC). Seems when I started this madness they were the source for Headway information, too... maybe you were just talking about this particular thread, frodus?


There are definitely some folks that know their lipo...what with all them e-bike nuts over there.

But there are also some schmucks. I think there are just a lot more people on ES, so more people talking out of their ass stand out (not like over here...we do perfectly well with just a few people talking out of their ass :) )

If you can stomach it, this thread (http://endless-sphere.com/forums/viewtopic.php?f=10&t=34110)is kind of amusing (it's what got me thinking about this V = SOC thing...turns out I learned something :D )

podolefsky
30 November 2011, 0525
are you not using the spyglass meter with your set up? That shows soc and is as accurate as you need it to be, the curtis controller does a very good job of measuring it.

this was purely academic - I was just testing this little lipo cell because of a discussion on ES about voltage as a measure of SOC

teddillard
30 November 2011, 0658
this thread (http://endless-sphere.com/forums/viewtopic.php?f=10&t=34110)is kind of amusing

Hysterical! Thanks for sharing, made my morning... I think they need a Celine Dion Time Out!

ZoomSmith
30 November 2011, 0744
LOL.

My only question: who's the Village Idiot on this forum?

teddillard
30 November 2011, 0816
LOL.

My only question: who's the Village Idiot on this forum?

STOP looking at me, bitch. :cool:

frodus
30 November 2011, 0921
Dunno about this particular thread, but as far as "most of them are way off", I don't think that's particularly fair, considering what they've done over there with lipo (basically my entire strategy for lipo came from there via RC). Seems when I started this madness they were the source for Headway information, too... maybe you were just talking about this particular thread, frodus?

(edit edit: On the contrary, I found that little discussion to be pretty interesting and informative... simple miscommunication to start, that's all. 'Course that never happens here. :D For example, I do believe he answered your question, frodus: "Also, a cell can take a verrry long time to rise back up in voltage after a hard load. i've seen a pack go from 3.3/cell average to about 3.45v/cell average after sitting for ~30 minutes after i beat the daylights out of it at 9C constant and it got all warm/hot.")
While unfair, I've found it to be true for many cases. I've got a lot of posts and most were spent discussing simple topics that people just couldnt wrap their heads around. Lots of misinformation being thrown about by people that think they are experts because they read it online. I think there's only a handful of people there that actually know a fair amount when it comes to batteries. Most doesn't mean all though.

Thanks for pointing out his answer. He didn't answer that in this thread and I read mostly neptronix' stuff.


Sent from my SGH-I897 using Tapatalk

teddillard
30 November 2011, 0947
That was neptronix comment, just to be clear.

Yeah, I think Noah is on the money. It's a bigger forum so there's a lot more noise, but the fact remains there's a crap-ton of really valuable information there that shouldn't be just offhand discounted, for anyone reading this who hasn't spent much time there.

Hell, I think it was you that sent me over there re/Headways in the first place. :D

lugnut
30 November 2011, 0954
are you not using the spyglass meter with your set up? That shows soc and is as accurate as you need it to be, the curtis controller does a very good job of measuring it.

I looked at that briefly. It looked as though it was fashioned for lead acid. Were you able to set it up for Lithium and get it to work well? Is it possible for you to post the parameter changes to do that?

podolefsky
30 November 2011, 1028
Hysterical! Thanks for sharing, made my morning... I think they need a Celine Dion Time Out!

he changed it, but i especially liked the part where he said the Etek was made by Black and Decker...

oh, and i gave them a celine dion time out...but ES doesn't let you embed video (or it doesn't work), so it's buried in a link.

teddillard
30 November 2011, 1038
Oh. good. BTW, yeah you can:
"to embed a video: you put the id# in the you tube brackets the forum provides, just the blue #s you see in your code line
http: //http:// www(dot)youtube.com/watch?v=0rOYaf5A7UM &noredirect=1"

podolefsky
30 November 2011, 1043
ah! thanks Ted...i guess it helps to read the directions

celine dion thread kill video now embedded...you can't miss it