Battery(61Wh) discharge curve

Framework Laptop 13
1

I used ectool to measure the open-circuit voltage vs remaining capacity, the battery us pretty new with only 37 cycles and the last full charge is 3971mAh(101.4%). Charge or discharge to a certain percentage and stay there for 10 minutes, measure the voltage, change the charge limit, repeat.

%, remaining capacity, open-circuit voltage
91%, 3618mAh, 17312mV
80%, 3180mAh, 16742mV
70%, 2782mAh, 16310mV
60%, 2387mAh, 15876mV
50%, 1988mAh, 15518mV
40%, 1589mAh, 15283mV
30%, 1194mAh, 15134mV
20%, 795mAh, 14959mV
30%, 1194mAh, 15209mV
40%, 1589mAh, 15396mV
50%, 1986mAh, 15640mV
60%, 2384mAh, 16011mV
70%, 2783mAh, 16460mV
80%, 3178mAh, 16838mV
91%, 3618mAh, 17331mV
100%, 3972mAh, 17784mV
Note: there’s no 0~20% as the EC does not allow chargelimit that low, and without this it’s a bit hard to measure open-circuit voltage. I chose 91% is because 3572/3915=91%, if the open circuit voltage is at 17.3V (limited charging voltage is slightly higher as the voltage will drop after charging completion due to internal resistance), then the 61W batterpack is like an “unlocked CPU” compared to 55Wh if the dischrage curve is identical from 0 to 3562mAh, but I’m not pretty sure.

image
image1041×585 38.5 KB

2

Can you elaborate what you mean by that? The more I read it the less I understand it.

Anyway though, careful with doing voltage to state of charge cause that is only ideally a constant thing and the state of charge mostly depends on the coulomb counting on the bms not the voltage and can drift somewhat when it has not seen 100 or 0% for a while.

Afaik the charge limit is state of charge based not voltage based so it gets even messier there with that testing method.

1 Like
3

Sorry, English is not my first language. I know that the charge limit is SoC based not voltage. The 91% is the ratio of old vs new battery pack (zoom in the photos of the links). What I was trying to say is that if you have a 55Wh and a 61Wh battery pack and you set the charge limit % in a way that they both stop charging at the same mAh(like 3000mAh for example) they might have the same open-circuit voltage. If that’s really the case, it’s similar to “unlocked CPU vs locked CPU” because if you set the limit of the unlocked CPU the same as the locked CPU you can get identical performance and power consumption. if you limit the charge percentage to 72% on the 61Wh and 80% on the 55Wh you get the same mAh, if the voltage is also the same, the batteries might have the same discharge curve and discharge cut-off voltage, in this case, the degradation will also be the same and the only difference is max unplugged runtime when absolutely necessary(set chargelimit to 100%) and the 55Wh might be more economical(?)

4

The whole point about the 61Wh battery is the tweaked chemistry that allows them to charge to higher voltages so that isn’t really a good comparison. An unloked cpu is letting the same thing go higher, this is letting a different thing go higher, like a not unlocked cpu a generation later XD. If your ran the 55Wh chemistry to 61Wh voltages it probably won’t blow up but it will have much lower cylce life.

5

Sorry I didn’t explain it clearly. To increase a battery’s capacity one could either increase the “width” of the battery i.e higher capacity or more cells in parallel, or “height” i.e increase the charging voltage, or both. Given the specification of the battery packs the 55Wh → 61Wh upgrade is likely to be the latter.

The “unlocked CPU” analogy means if you set the unlocked CPU to the locked counterpart they’ll get the same performance but if you hack the locked CPU to increase the power to unlocked it might explode so they are not really identical

6

It is according to the blog post about it. Most of the capacity gain is in higher charge voltage at about the same cycle life thx to tweaked chemistry.

That’s just different bins, it’s really not a great comparison. It’s more like a 13900 vs a 14900, almost the same but tweaked slightly.

1 Like
7

Which raises a question. If you charge both of them to 4.2V, or more specifically adjust the chargelimit % so they ended up at the same voltage, do they degrade the same? If you charge both of them to 3000mAh, do they degrade the same? If you charge the 55Wh to 80% and the 61Wh to 72% and if both the remaining capacity and the open-circuit voltage are almost identical, do they degrade the same? or the 61Wh degrades a bit slower?
In comparison, if you charge both of them to 80% do they degrade the same? or the 61Wh degrades a bit faster?

8

Almost certainly not.

Since they share very similar chemistries and both are rated to lose 20% after 1000 cycles it it would indicate limiting each of them to 80% should have a similar impact to cycle life in both.

It is very hard to actually know though. Where you stop charging a discharging os only one piece in the battery degradation puzzle, there is also temperature and charge/discharge rate (which to be fair are mostly also temperature). If you cook the battery in a hot laptop it may loose capacity without being cycled at all.

9

I’m asking this question because I found that older laptops with a lower limited charge voltage(4.1V to 4.2V/cell) degrade slower than modern laptop batteries, HP Omen is notoriously bad for that (4.4V/cell).

10

It could be a bad battery because it is 4.4V or it could be a bad battery that just happened to be 4.4V the 2 may not be related. Also gaming laptops are also known to absolutely cook the battery which doesn’t help.

Ultimately we’ll see how the 61Wh shakes out and the good thing is it’s easily replacable and available first party XD

11

There’s one thing in common on HP Omens and Framework Laptops, is that the battery charging does not have any “hysteresis loop”. Most quality laptop resume charging at 5% below the charge limit i.e. only restart charging below 95% when the charge limit is set to 100%, hence the hysteresis loop. This problem makes swollen batteries a commonplace in HP Omen, until HP pushed a BIOS update to allow the users to set charge limit to 80%. I also witnessed during the measurement of the charge curve, is that the FL13 pushed the battery really hard as I spent half an hour charging from 98% to 100%. Since Framework is a small business compared to HP and most of Framework Laptop users are DIYers and know how and why to set a lower charge limit. I’m wondering about whether the lack of hysteresis loop is a potential risk of faster degradation, if the BIOS charge limit is untouched (defaults to 100%)

12

I’m not actually convinced turning a couple short charge cycles into one long one is actually helpful from a battery wear perspective, appart from shifting the average state of charge of the cycles down a smidge. Though charge “mileage” at the top end (and low end) of soc are known to produce a lot more wear than the same “mileage” more to the middle.

Allowing a lower charge limit is probably a massively bigger factor than the hysteresis thing, a few small charges and discharges are about equivalent to one bigger one wear wise, but doing the same dance at 80% vs at 100 is going to produce a lot less wear.

That’s probably more the bms figuring out what 100% even means.

The charge limit being set to 100% is probably the much bigger fish than the hysteresis loop, doing 5x 1Wh out and 1Wh in isn’t significantly worse than doing 5Wh out 5Wh in but doing either of those at 100% is going to cause more wear than at 80% which in turn does more than at 60%. Though all of that info comes from research with much older lion battery chemistry, with the newer one the impact of higher soc charging may not be as big anymore.

But hey they claim 20% capacity loss after 1000 cycles at full capacity so I am curious how it actually turns out.

13

Hi.

I don’t have the ectool, but will work on that. I monitor the battery wear multiple times a day and graph the lowest value.

The battery is the 55W: 26 month old and has just done it’s 101 cycle, however the wear is at least 12% not 2% as I would have hoped.

My standing BCL is 78%

Every month or so, or if I notice sudden increase in ‘wear’ I do a full discharge and 100% for a day to try and reset the BMS.

In the first 22 months this worked fine and wear reduced over that time.

Then in December it almost double after my forced BMS reset to to 100 and again a week ago it doubled from around 6% to 12%…

If this info is a bit off and unhelpful I will delete, otherwise I will provide graphs and data.

14

I am aware of you tracking and am very curious where it goes, especially the 200 cycle mark at the current rate. I am pretty sure degradation is non linear but we’ll see eventually.

1 Like
15

Too little information.
What’s the temperature during charging discharging? A higher temperature means much lower cycle life.
What’s the discharge cut-off voltage? A lower cut-off voltage means deeper discharge and higher degradation.
What’s the charge cut-off current? During constant-voltage phase, the charging current goes lower as the battery % increases, when the current is below a certain value the current is cut and the charging is finished, continue charging nonstop will overcharge the battery even the limited charging voltage is never exceeded. Most classic 18650/21700 battery manufacture specify a 0.03C cut-off current.
What’s the charging C-rate? A higher rate means lower cycle life, batteries rated 1C charging often charge at 0.2C during cycle life testing, in this case if you always charge it at 1C it’s cycle life will be significantly lower than it’s specification.
What’s the discharging C-rate? same as above, batteries rated 2C or even 5C might still discharge at 0.5C during cycle life testing.
Here’s an example, note that the battery cell is capable of doing 2C discharge but cycle life test is done at 1.0C
No specification about calendar life, battery degrades over time, the higher the % the worse the degradation, the best storage % is 30~40%. Although 30%~40% is better than store at 100%, the battery will still degrade. To get the max cycle life one would charge/discharge non-stop to minimize calender life loss, which is the opposite of actual usage.

Practically, battery almost never get to the rated cycle life on the datasheet.

  1. Prolonged storage. Longer time is spent per 100 cycle in actual use regardless of %, compared to test condition. (worse)
  2. Elevated temperature. No battery cooling, heat is conducted via the metal casing, while the CPU has dedicated cooling but still conduct heat to the surrounding parts, meaining the battery temperature sits between CPU and ambient. To make matters worse. There’s no temperature reading on the AMD7840U w/61Wh battery (worse)

And the CPU runs straight to the wall (100C) before underclocking. And there’s no official way to enable S3 and disable S0ix sleep. In modern standby the CPU won’t underclock if M$ is abusing your laptop’s computational power till it’s last leg(I witnessed much higher power comsumption on modren standby compared to idling on desktop or browsing websites or documents). Either use Linux(modren standby consumes 2.5 to 25 times of wattage compared to s2idle), or use some sort of hacks such as UMAF to emable S3, otherwise when sleeping(more like pretending to be asleep) in enclosed space such as inside you backpack, the CPU will constant temperature operation while conducting heat to other components such as the battery. Since there’s no indication of battery temperature the battery will continue heating up to 100C before death instead of shutting down the system upon overheating. If there’s no official fix, an onboard fire aviation accident caused by modern standby is going to happen sooner or later.

  1. Higher charge/discharge rate. The battery discharges at 3915mA(1C) max and during charging and if the power supply is sufficient, 3915mA to 40% then 2740mA(0.7C) when above 40%. If the cycle test is done on 0.2C or 0.5C, the actual cycle life will be lower. (worse)

  2. Lowered limited charging voltage. If the user use 80% limit at 99% of the time, the cycle life will be higher compared to testing. (better)

Conclusion: since there are more factors that lowers the actual cycle life compared to those extend it. The battery will more likely to degrade faster

About the HP Omen battery degradation, thing.

  1. It does not undervolt, the limited charging voltage is 4.4V/cell. Many older computer runs on 18650 batteries that has a limited charging voltage of 4.2V/cell but the laptop manufactures program the BMS to limit the charging voltage to 4.1V/cell.
  2. It does not have a hysteresis loop, mentioned earlier.
  3. It did not have user-configurable charge %, before a later BIOS update.

I found that most laptops with swappable battery pack with cylinder cells, the battery pack degrades much slower compared to modern laptops or smartphone where the cells are in pouch format, even if most of the former don’t have charge limit adjustment. which means that 1 & 2 can be as effective as 3.

1 Like
16

1 is definitely as effective as 3 as it is pretty much the same thing but different (imo setting the max cell voltage would probably be better than a percentage limit but also a lot more confusing for a regular consumer). There is also a bit of selection bias as consumery stuff (which tend to have shorter warranties and fewer incentives for the manufacturer to increase durability) switched to non swappable and then pouch cells a lot earlier. The last holdouts with sappable and cylindrical cells were busyness grade stuff which has more incentives to not run the batteries into the ground and less to just have “bigger number”. Similar to how consumer ssds tend to have as little spare area as necessary and enterprise ones a lot more giving them more endurance.

2 probably had a much lesser impact if at all.

17

My previous laptop, a Panasonic Let’s Note, has batter battery longevity

As shown 
--- TLP 1.6.1 --------------------------------------------

+++ Battery Care
Plugin: generic
Supported features: none available

+++ Battery Status: BAT1
/sys/class/power_supply/BAT1/manufacturer                   = Panasonic
/sys/class/power_supply/BAT1/model_name                     = CF-VZSU1C
/sys/class/power_supply/BAT1/cycle_count                    =    405
/sys/class/power_supply/BAT1/energy_full_design             =  42480 [mWh]
/sys/class/power_supply/BAT1/energy_full                    =  43120 [mWh]
/sys/class/power_supply/BAT1/energy_now                     =  36170 [mWh]
/sys/class/power_supply/BAT1/power_now                      =  14600 [mW]
/sys/class/power_supply/BAT1/status                         = Charging

/sys/class/power_supply/BAT1/charge_control_start_threshold = (not available) 
/sys/class/power_supply/BAT1/charge_control_end_threshold   = (not available) 

Charge                                                      =   83.9 [%]
Capacity                                                    =  101.5 [%]

5 years, 405 cycles, 92% capacity. Things to note:

  1. The chemistry is NCA, not LCO. The former has better longevity.
  2. The cells are cylindrical which don’t swell, unlike pouch “lithium polymer” cells. Better longetivity
  3. Hot-swappable battery pack, which means better cooling due to separate case away from the laptop’s heat source.
  4. No undervolting, the cells’ limited charging voltage is 4.2V, and the pack’s limited charging voltage is 8.4V(4.2V/cell).
  5. No charging limit setting or any kind of battery care. The battery will charge to 100% when plugged in.
1 Like
18

Such reports are just that they are calculations not facts.

Yes that the same as the Framework spec. I’ll just have to wait a few more years to how this one fares.