Optimal Charge Voltage - (LiCo)

Many scoff at charging Lithium Cobalt (LiCo) cells at less than 4.20V.
(LiCo includes most all 18650 and LiPo cells)

There seems to be a consensus of knowledgeable opinions that every 1/10thV reduction in charge voltage doubles the number of usable cycles.
So, I intend to overwhelm the naysayers with sources.

"Pictures are worth a thousand words" so I'll start there ...

Interesting stuff !
Can you reference your sources so we can get more background.

still based on opinions and not the facts as presented in that lecture by the guy who did the research using the precision charger/dischargers to evaluate the parasitic reactions at full charge. all of this is old news, representing batteries kept at full charge, not batteries that were charged only directly before use.

dnmun said:

still based on opinions and not the facts as presented in that lecture by the guy who did the research using the precision charger/dischargers to evaluate the parasitic reactions at full charge. all of this is old news, representing batteries kept at full charge, not batteries that were charged only directly before use.

Click to expand...

Quite to the contrary!
Most of these tests involved continuous cycling.
Charging immediately before discharging.

This is why the Chevy Volt used only something like 55% of it's 16kWh pack on the first gen LG chem cells. They could endure many 10's of thousands of shallow 55% DOD cycles. The new gen cells get charged a little higher and discharged a little deeper, and still out-cycle the previous generation.

More modern designs of high purity electrolytes filled with exotic additives can be made to survive cycling in the higher voltage ranges. The smartphone and tablet cell market has some 4.35-4.4V cells in mass market devices, and they can cycle fine. It takes some $$$ electrolyte to make it possible.

The biggest advantage for the ebikers running hobby-grade cells or B-stock 18650's is how much more friendly lower charge voltages happen to be with respect to the electrolyte decomposing from impurities and things. Some hobbycells will puff sitting at 4.35V in a warm place in just 1 day. Most EV grade cells can sit at 4.35V in a hot room all year long and not be puffed (though they would be micro-gassing at a way higher rate than they would be at a lower voltage and temp).

A-lot of old and incomplete data here.

Here is how you do a real test:
http://lygte-info.dk/info/BatteryChargeVoltageCurrent%20UK.html

Note that 60-minute resting voltage, which is itself a function of charge voltage and termination current, is determinate of capacity, and it's rather linear for Li-Co.
Not simply charge voltage.

And Li-Co is old hat by now.
NCA is the game in town these days.

Lithium Titanate it seems...... Li4Ti5O12 ( LTO ) Li-titanate.

http://batteryuniversity.com/learn/article/types_of_lithium_ion

My Dillenger pack uses Lithium Manganese Oxide........ LiMn2O4 ( LMO ) Li-manganese.

After 6 months with 16 FULL charge cycles and 1,023 Km I'm happy.

I'v only ridden it to LVC twice and again, never ridden within hours of a FULL charge to give it time to 'settle'.

One of those charts shows why I stopped charging to 4.1v. It just sacrificed too much capacity. 10% is one thing, but closer to 15% shortened my range too much. But the main reason I like a charge less than 4.2v, is I'm charging my HK packs the risky way. No bms, with a bulk charger.

Worth noting though, on that chart it shows 300 cycles that have more capacity than a 4.1v charge ever has. For lots of us, 300 cycles is years.

So I am charging to about 4.15v, getting a decent capacity, and still have .5v of slack before any cells that got inbalanced get to 4.2v. I get two solid years of use from these cheap ass hobby packs, and I'm fine with that. I still have useable ones 4 years old, but they are very low capacity now, like 60%. I just carry them when I need extreme range once or twice a year. In between, I store them at 3.8v or so.

If I were to charge to 4.05v I would need to carry 20% more battery. I'm not keen on that. It would also extend their life to 5 years if cycled 5 days a week. I don't see the point in that, the calendar life is under 5 years anyway. It is trying too hard. You can't extend the cycle life beyond the best before date. Most users doing this won't fully cycle 5 days a week, they may never fully cycle at all. Further extending the cycle life past the best before date. They could have a projected cycle life over 3000 cycles, but still never use a third of them as they go out of date.


I actually target 160 full cycles 4.2-3.0 as Turnigy state. It never happens each time though as my pack is 10% bigger than my longest journey requires(8 miles). It's well over twice the size required for my average run. These facts extend that 160 cycles considerably. I should of seen 1600 miles to 80% but at 2000 miles I see no end in sight. This is from 5Ah costing $80. This sort of mileage represents two years use of this pack. As it is what it is, I see no point trying to get any longer from one. It's best years are over. I'm about to retire it on to a cheap bike for a mate. 4 cents per mile.

The story would be different if the bike got used differently. The important thing is deciding when your pack should die, rather than doing everything you can to extend it. There is little point having 2000 cycle life, if it won't be used 1000 times before it's old. That would be carrying too much weight and money about for no reason.

If you spec for winter use, in summer your generally taking it easier. A pack that lasts a year gets an easy time in summer. Within 3 years, It would be hard to put 2000 cycles on a pack before it retired. Maximum nannying is only for commuters that size the pack close to their requirements. Or people that charge many times a day, though such use is usually irregular, thus hard to calculate.

I think the moral of this story was not to find the holy grail of charging voltage, instead try to put people in a perspective so they can maximize the use of their packs by choosing the appropriate charge voltage cut off for each cycle depending on intended ride. If today I need to ride no more than 10km on a pack that can go 30km, it has no sense to charge it over 4V. If I intend to ride the full 30km twice in a year, and there is a slight headwind, charging to 4,25V is more than justified.

frendly1uk wrote

Or people that charge' many times a day, though such use is usually irregular, thus hard to calculate.

Click to expand...

That would be me as I used my bike to and from my house to the winery and then back to my house for almost an entire year. I was using 4.15 as my charging voltage had a 12s 15ah turnigy 20c lipo pack and in that year saw very little internal resistance change. I also used this pack on some 30 mile plus journeys usually opportunity charging at 600w and still saw a good cycle life on the pack coming down the pike. That is until someone stole my bicycle and my packs. This would have been the first lipo pack that actually would have given me some life cycles.

"Going out of date"? (Aging)?

I wonder about Lithium cell aging ...
For example:
I just built packs using NOS (New Old Stock) laptop LiPo cells dated 2008.
They supplied 95% of new rated capacity using a metered discharge (iMax B8 - Genuine)

DrkAngel said:

To determine optimal voltage range I am capacity mapping samples of these cells
3s2p 2008 Rev 4 Lipo ... for present build and
3s2p 2010 Rev 5 Lipo ... for future builds
(4320mAh OEM rated)

Step 1
Charged all cells precisely to 4.18V
Then metered discharged and equalized to 3.52V
Rev 4 (2008) metered 4128mAh
Rev 5 (2010) metered 4100mAh + (accidentally discharged below 3.52V then recharged - so approximated)
4128 / 4320 = 95.5% of oem rated
(Might supply 100% of rated with 4.20V - 3.50V discharge!)
2008 = ~6 years old cells ... Makes me question "age deterioration"! ... ???

Both 6 packs discharged evenly - all 3 banks within 1/100th V at 3.52V!

Step 2
Rigged 2 identical mAh meters to MeanWell S-150-12 and
will charge 2 - 3s2p in 6/100th V increments (2/100th V per cell)
(35 - mAh readings per 6 pack)

Step 3
Will graph data and present

Click to expand...

I previously built a 25.9V 25.92 pack with used 2004 - 2008 cells of this type and ran it daily for 3+ years and 7000++ miles before retiring the still usable cells to 12V power supplies for AC inverter and LED lighting.

"#1 Laptop LiPo Build 2004-08 used cells
2011 - 24.0Ah (25.9 to 29.4V)
2012 - 20.8Ah (25.9 to 29.4V)
2013 - 15.8Ah (25.9 to 28.7V) 6500+ miles
2014 - 12Ah+ (25.9V to 29.2) 7000+ miles"

See - Homemade Battery Packs
I would hate to believe that these LiPo cells are somewhat unique in there lack of aging! ... ?

RC LiPo, being designed as volatile rather than stable and reliable, would likely "age" much more quickly?

NCA has two advantages over LiCo worth mentioning in regards to a topic many are ignoring.

1.) More energy is stored at the lower voltages. This is good because if you charge a NCA cell to 4.10, it's like 93% capacity compared to 88% for LiCo.
2.) NCA lasts longer when stored at high voltage. Put an NCA cell and a LiCo cell in storage at 4.17V, and the NCA cell will last three times longer.

Comparing apples to apples:

Charge a LiCo to 90%
Charge a NCA cell to 90%...

The NCA cell will probably last six times longer in storage.

This doesn't even touch on cycles. Many people just want to be able to leave their bike with a decent charge when not riding.

Hombré said:

Lithium Titanate it seems...... Li4Ti5O12 ( LTO ) Li-titanate.

http://batteryuniversity.com/learn/article/types_of_lithium_ion

My Dillenger pack uses Lithium Manganese Oxide........ LiMn2O4 ( LMO ) Li-manganese.

After 6 months with 16 FULL charge cycles and 1,023 Km I'm happy.

I'v only ridden it to LVC twice and again, never ridden within hours of a FULL charge to give it time to 'settle'.

Click to expand...

It's better to ride it right away after a full charge. Storage at full charge is what degrades the cell.
Model S owners will time their "full" range charge so that the charge completes exactly when they plan to leave. (and they only go to 4.15V with NCA cells! No wonder there hasn't been any significant reported degradation)

Li-Mn and Li-Mn hybrid types are particularly bad offenders in this regard (they are worse then both NCA and LiCo)

okashira said:

NCA has two advantages over LiCo worth mentioning in regards to a topic many are ignoring.

1.) More energy is stored at the lower voltages. This is good because if you charge a NCA cell to 4.10, it's like 93% capacity compared to 88% for LiCo.
2.) NCA lasts longer when stored at high voltage. Put an NCA cell and a LiCo cell in storage at 4.17V, and the NCA cell will last three times longer.

Comparing apples to apples:

Charge a LiCo to 90%
Charge a NCA cell to 90%...

The NCA cell will probably last six times longer in storage.

This doesn't even touch on cycles. Many people just want to be able to leave their bike with a decent charge when not riding.

Click to expand...

While I agree NCA is the cat's pajamas, you are attributing too much to the cathode material chemistry. The cathode chemistry itself is neat, and everyone loves to talk about it, but as far as making a cell safe or durable in cycling and calendar life etc, the cathode chemistry is one of the lesser significant factors.

The stuff that makes cells durable and long lasting and safe comes down largely to the surface geometry of the surfaces wetted by the electrolyte, and the electrolyte itself, and even the type and design of the separator.

For example, there are sketchy dangerous short lived LiFePO4 cells, and there are some extremely long-life safe LiCo cells.

While NCA is an awesome cathode material, it is pretty reactive and if you built a cell up with NCA cathode materials and didn't design an optimized particle surface shape and used low-quality hobby-grade electrolyte, and gave it a no-safety features hobby-cell separator, it would make a pretty shitty short-lived potentially dangerous cell.

If you take the NCA cathode material, and then also build a quality cell around it, only then do you get a long lasting durable safe cell. It's not something that auto-magically occurs because the cathode slurry contains NCA.

This is why it's pretty bogus to attempt to draw conclusions about a cells performance or safety or calendar life based around whatever it's cathode material may happen to contain alone.

Yeah, yeah, yeah, NCA is great. Now if you could find some 20C rated cells for the same price as 20C Turnigy at 31 cents per wh it would be great. But as it is now, you can't even get 3C NCA at anywhere close to that. My 10ah 24s 20C pack cost $275 and is still going strong 30 months and ~10K miles later.

wesnewell said:

Yeah, yeah, yeah, NCA is great. Now if you could find some 20C rated cells for the same price as 20C Turnigy at 31 cents per wh it would be great. But as it is now, you can't even get 3C NCA at anywhere close to that. My 10ah 24s 20C pack cost $275 and is still going strong 30 months and ~10K miles later.

Click to expand...

Hmm, cost is not everything to everyone.
There are other properties to consider, not least being size and weight, and "real" actual "C" rate...not what is printed on the sticker !
Your "20C" pack ( real life more like 5-10C ?) weighs 6.4kg ,.. in cells alone
A similar 10 Ahr 18650 pack ( 4P, 24S of 25R cells) would weigh 4.3 kg, and cost $390 in cells ( EVVA prices )
Samsung rate that cell for 20A continuous (8C) and 100A pulse !...so it can match the Lipo without a problem.
So the 18650's are 2kg lighter, more compact ( 2/3 the volume..) and only $100 more !..
With all the other benefits of 18650 technology !

Also much safer and long lasting.

wesnewell said:

Yeah, yeah, yeah, NCA is great. Now if you could find some 20C rated cells for the same price as 20C Turnigy at 31 cents per wh it would be great. But as it is now, you can't even get 3C NCA at anywhere close to that. My 10ah 24s 20C pack cost $275 and is still going strong 30 months and ~10K miles later.

Click to expand...

Yes those lipos are heard to beat for power/capacity and cost.

I would be interested to hear more about your usage... charge/discharge voltage, ambient temperature and have you tested their capacity recently?
Good to hear they made it 2 years.

You could build an NCA pack that is 150% more capacity for the same weight, and be quite close in cost per storage. And with your controller you are only using up to 4C anyway, there are NCA"s that can handle that just fine

From a few months back.
http://endless-sphere.com/forums/viewtopic.php?f=14&t=61446&p=918065&hilit=+2+years+#p918065

.
Was researching a cell type and found a manufacturer who tested 4.20V vs 4.10V cycling.



Through near 700 cycles, 4.20v charged capacity degradation was >200% that of the 4.10V charged.
Contrary to assertions of some, "time setting at full charged voltage" seems not a major factor as each cycle only spent 10 minutes at charged voltage.

Capacity at various charged voltage
4.20V = 2568mAh
4.10V = 2356mAh
91.74% or 8.26% capacity loss charging to only 4.10V
.

.
.... 4.20V vs 4.10V
New 2568 - 2356 - 8.26%
300 2416 - 2299 - 4.85%
500 2318 - 2259 - 2,55%
800 2158 - 2148 - .47%
1000 2103 - 2113 + .53%
1200 2095 - 2061 - 1.63%
1500 2056 - 2035 - 1.03%

4.10V charged after 1000 cycles actually has better capacity than the 4.20V charged.
However, the 4.20V charged seemed to stabilize near the 80% mark and deterioration slowed remarkably!
Lost
9.7% after 1st 500 cycles
8.4% after 2nd 500 cycles
1.8% after 3rd 500 cycles >> inexplicable?
These are from .5C charges and 1C discharges

my personal scooter spent most of its time at 4.05V in summer and 4.1 in winter float charging.
after 50k miles i lost 6% capacity.

in general its best to look at peeps here that did the actual testing or companies like tesla that are dumping hundreds of millions in battery tech and their whole business is dependant on batteries.
their first gen batteries topped out at 4.05V and their "range mode" is 4.12V. (with slight overshooting to 4.15V on some newer generation packs)
that means that they can do more then 10k cycles (aka: more then half a million miles) with less then 20% capacity loss.
the latter model 3 with the "new hotness" 21700 cells is less conservative due to their tweaking of the temperature control systems and chemistry and means they can get away with 4.1V regular range and the range mode is actually 4.2V. but you do get a warning that exxessive use of range mode increases battery wear and it gets logged. their bms is acutally seriously impressive shit.

i would not use data from car companies like GM that use weird chemistries and are still decades behind on tesla.

DrkAngel said:

.
Was researching a cell type and found a manufacturer who tested 4.20V vs 4.10V cycling.

Click to expand...

So lets make some unwanted conclusion from uncle Pajda.

1. All of the graphs in the first post of this topic (source is Battery University?) are minimally obsolete and was minimally obsolete even when they were released. By my opinon they were created by someone who never tested cycle life for a single cell. And this is exactly the example that a good intent can cause a lot of damage.

2. Data from this last post are in accordance with the current state of knowledge of 18650 cell production at least 5 years back. So again data in the first post are sucked from someone finger.

3. Data from this last post clearly shows that for modern cell it is not problem to be operated at high DoD with 4.2V charging voltage. Usable absolute capacity at 4.2-3V SoC window is still higher up to 1500 cycles than cycled at 4.1-3V. (The better result around 1000 cycle for lower SoC is caused by measurement error - most probably by lower room temperature in this particular region). No one really proof that 4.2V resting voltage have significant impact (5% or more capacity loss) on calendar age of modern NMC cell. Is highly probable that modern NMC cell operated at 4.2-3V will hold for first 4-5 years higher absolute usable capacity than cell operated at 4.1-3V.

4. I do not understand the reason why present graph in % of wear level without presenting the graph with absoulute capacity drop? It is highly misleading and it only confuses the average reader that at the first sight there is a siginficant benefit to do that, which is obviously not true.

flippy said:

in general its best to look at peeps here that did the actual testing or companies like tesla that are dumping hundreds of millions in battery tech and their whole business is dependant on batteries.
their first gen batteries topped out at 4.05V and their "range mode" is 4.12V. (with slight overshooting to 4.15V on some newer generation packs)
that means that they can do more then 10k cycles (aka: more then half a million miles) with less then 20% capacity loss.
the latter model 3 with the "new hotness" 21700 cells is less conservative due to their tweaking of the temperature control systems and chemistry and means they can get away with 4.1V regular range and the range mode is actually 4.2V. but you do get a warning that exxessive use of range mode increases battery wear and it gets logged. their bms is acutally seriously impressive shit.

i would not use data from car companies like GM that use weird chemistries and are still decades behind on tesla.

Click to expand...

Flippy I told you many times that your information about Tesla batteries are not true. All Teslas from the first Model S could be charged to the cell 100% SoC which is 4.2V per cell or 403,6V per battery pack. There are available many videos showing that in range mode or 100% battery the display is showing the tappering voltage of 404V. Also they can be discharged almost to zero, lower than 3V per cell before bricking safety. Actually they lowered the charging voltage to ca 4.15V for the oldest 85 batteries just few months ago due to the problems with accidentl fire during charging sessions of this old packs.

Tesla 18650 and 21700 cells definitelly can not do 10k cycles, at least not at more than 50% DoD. I perform A lot tests of both TMS and TM3 cells and there is nothing interesting about their performance in comparison especially with modern NMC 18650/21700 production.