That graph gives you the capacity loss when drained from 4.2V to 4.1V per cell. Which does not tell you what the capacity would be when charged to 4.1V and drained from there.

 

I used a practical drain load of 2A on my 42V 5Ahr battery as that was the current used when on a typical ride, assist level 2 at 14mph or so.

I noticed that point in your experiment but you have not yet repeated the same experiment with Sturmey 's diode idea.

I noticed that point in your experiment but you have not yet repeated the same experiment with Sturmey 's diode idea.

 

Not yet, waiting for a 1A charger to arrive.

 

I want to see if the very extended (150 minute) CV period with the diode used is replicated when the charge current is reduced to a more sensible 0.2C.

So, the capacity check...

 

I charged the pack to 41V. I left it to stand for a week, then did a discharge test. I found the heating element I had lying around from a heat gun, with the intention of discharging at around 1.5 to 2A, depending on how the resistance of the element changes as it heats up, compared to it's measured resistance at room temperature.

 

I set up the battery with a new Watt meter. Open circuit, I was getting 41V although it is probably about 40.9 as the Watt meter has an an error of about 0.1V compared to the calibrated DVM.

 

Initial current was around 1.6A and I took readings at intervals over a 10 hour period.

 

 

After 10 hours, I had to go to bed (it was 1am by this time). The pack was down to 33V or so, but it wasn't exactly falling off a cliff at this point and I still think more could have been taken out of it. So from 41V to 33.38, I got 544 Wh.

 

Here is the raw data:

 

Battery Capacity Test


























Time (Sec)
Voltage (V)
Current (A)
Watts (W)
Energy (Wh)

0

41

0

0

0

1

40.9

1.6

65.4

0

100

40.8

1.59

64.8

2

220

40.74

1.59

64.7

4

300

40.71

1.58

64.3

5

1080

40.42

1.57

63.4

19

1920

40.16

1.56

62.6

34

2700

39.92

1.55

61.8

47

3600

39.66

1.55

61

63

4680

39.36

1.54

60.6

81

8520

38.36

1.5

57.5

144

10260

37.91

1.49

56.4

172

11820

37.54

1.47

55.1

195

14580

36.88

1.45

53.4

237

16860

36.48

1.43

52.1

270

19680

36.16

1.42

51.3

311

21720

36

1.41

50.7

333

22560

35.91

1.41

50.6

352

25200

35.68

1.4

49.9

388

28440

35.33

1.38

48.7

434

29280

35.21

1.38

48.5

444

29820

35.12

1.38

48.4

452

30600

35

1.38

48.2

462

31200

34.88

1.37

47.7

470

32520

34.55

1.37

47.3

488

33300

34.29

1.39

47.6

497

33840

34.09

1.39

47.3

505

34440

34.08

1.39

47.3

513

34980

33.98

1.38

46.8

520

36060

33.73

1.37

46.2

534

36120

33.71

1.37

46.1

535

36240

33.66

1.37

46.1

536

36360

33.63

1.37

46

538

36480

33.57

1.37

45.9

539

36600

33.52

1.36

45.5

541

36720

33.45

1.36

45.4

542

36840

33.38

1.36

45.3

544

 

 

I was a bit lazy and didn't label the axes, x axis is in seconds, y axis is volts. current is also plotted, but is lost a bit on this scale. You can take the data and re-plot if interested.

 

So, it's got me scratching my head a bit. These are supposed to be HG2 cells at 3 Ah each, giving a total of 18Ah (648 Wh). I have always been sceptical about this as they were bought from Hong Kong on eBay for £2 each and I never felt that their capacity was that large and the pack seemed to perform similarly to my 15Ah packs. I assumed they were LG but 2.5 Ah versions which had been re-sleeved.

 

Anyway, if the pack is at 40.9 V, you can expect to lose approx 10% there (compared to a fully charged pack to 42V), so say 64Wh. I would guess there is still probably another 50 Wh to be had if the pack is driven right down to 25 V, which would give a figure just over 648 Wh. Which would mean zero loss in capacity for cells that have done hundreds of cycles and are several years old. However, if you look at Lygate's review of LG HG2 cells:

 

https://lygte-info.dk/review/batteries2012/LG%2018650%20HG2%203000mAh%20(Brown)%20UK.html

 

 

At 1 to 2 A discharge rate the capacity achieved is less that 2.8 Ah/cell, so 605 Wh for a 6p pack. Hence these cells of mine seem to be giving better results than brand new cells. This makes me wonder if they are actually LG MH1 cells that have been re-sleeved as HG2 cells.

 

https://lygte-info.dk/review/batteries2012/LG%2018650%20MH1%203200mAh%20(Cyan)%20UK.html

 

 

These are nominally 3.2Ah, but were found to give just over 3 Ah when discharged between 1 and 2 A.

 

So there you go, discuss...

Edited August 17, 20232 yr by WheezyRider

So, the capacity check...

 

I charged the pack to 41V. I left it to stand for a week, then did a discharge test. I found the heating element I had lying around from a heat gun, with the intention of discharging at around 1.5 to 2A, depending on how the resistance of the element changes as it heats up, compared to it's measured resistance at room temperature.

 

I set up the battery with a new Watt meter. Open circuit, I was getting 41V although it is probably about 40.9 as the Watt meter has an an error of about 0.1V compared to the calibrated DVM.

 

Initial current was around 1.6A and I took readings at intervals over a 10 hour period.

 

[ATTACH type=full" alt="53417]53417[/ATTACH]

 

After 10 hours, I had to go to bed (it was 1am by this time). The pack was down to 33V or so, but it wasn't exactly falling off a cliff at this point and I still think more could have been taken out of it. So from 41V to 33.38, I got 544 Wh.

 

Here is the raw data:

 

Battery Capacity Test


























Time (Sec)
Voltage (V)
Current (A)
Watts (W)
Energy (Wh)

0

41

0

0

0

1

40.9

1.6

65.4

0

100

40.8

1.59

64.8

2

220

40.74

1.59

64.7

4

300

40.71

1.58

64.3

5

1080

40.42

1.57

63.4

19

1920

40.16

1.56

62.6

34

2700

39.92

1.55

61.8

47

3600

39.66

1.55

61

63

4680

39.36

1.54

60.6

81

8520

38.36

1.5

57.5

144

10260

37.91

1.49

56.4

172

11820

37.54

1.47

55.1

195

14580

36.88

1.45

53.4

237

16860

36.48

1.43

52.1

270

19680

36.16

1.42

51.3

311

21720

36

1.41

50.7

333

22560

35.91

1.41

50.6

352

25200

35.68

1.4

49.9

388

28440

35.33

1.38

48.7

434

29280

35.21

1.38

48.5

444

29820

35.12

1.38

48.4

452

30600

35

1.38

48.2

462

31200

34.88

1.37

47.7

470

32520

34.55

1.37

47.3

488

33300

34.29

1.39

47.6

497

33840

34.09

1.39

47.3

505

34440

34.08

1.39

47.3

513

34980

33.98

1.38

46.8

520

36060

33.73

1.37

46.2

534

36120

33.71

1.37

46.1

535

36240

33.66

1.37

46.1

536

36360

33.63

1.37

46

538

36480

33.57

1.37

45.9

539

36600

33.52

1.36

45.5

541

36720

33.45

1.36

45.4

542

36840

33.38

1.36

45.3

544

 

[ATTACH type=full" alt="53418]53418[/ATTACH]

 

So, it's got me scratching my head a bit. These are supposed to be HG2 cells at 3 Ah each, giving a total of 18Ah (648 Wh). I have always been sceptical about this as they were bought from Hong Kong on eBay for £2 each and I never felt that their capacity was that large and the pack seemed to perform similarly to my 15Ah packs. I assumed they were LG but 2.5 Ah versions which had been re-sleeved.

 

Anyway, if the pack is at 40.9 V, you can expect to lose approx 10% there (compared to a fully charged pack to 42V), so say 64Wh. I would guess there is still probably another 50 Wh to be had if the pack is driven right down to 25 V, which would give a figure just over 648 Wh. Which would mean zero loss in capacity for cells that have done hundreds of cycles and are several years old. However, if you look at Lygate's review of LG HG2 cells:

 

https://lygte-info.dk/review/batteries2012/LG%2018650%20HG2%203000mAh%20(Brown)%20UK.html

 

[ATTACH type=full" alt="53419]53419[/ATTACH]

 

At 1 to 2 A discharge rate the capacity achieved is less that 2.8 Ah/cell, so 605 Wh for a 6p pack. Hence these cells of mine seem to be giving better results than brand new cells. This makes me wonder if they are actually LG MH1 cells that have been re-sleeved as HG2 cells.

 

https://lygte-info.dk/review/batteries2012/LG%2018650%20MH1%203200mAh%20(Cyan)%20UK.html

 

[ATTACH type=full" alt="53420]53420[/ATTACH]

 

These are nominally 3.2Ah, but were found to give just over 3 Ah when discharged between 1 and 2 A.

 

So there you go, discuss...

Your assumption is incorrect. It's impossible to have 10% of the charge between 40.9v and 42v. I alreay showed you that it's about 3%. Any 18560 discharge chart will show the same. That means that your battery lost 7% not zero %. It's impossible to lose zero % after hundreds of cycles. Not even LiFePO4 can do that.

Interestingly, Samsung quote an exact figure in my case with 21700-33j cells. The Samsung data below per cell shows a capacity of 12 Wh at 4.2v and 10.83Wh at 4.1v. Perhaps the exact figure is dependent on the actual cell.

Here is the actual discharge curve of that cell. Let's be generous and say it crosses the 4.1v line at 90 mah. The total capacity is 3270, so 90/3270 = 2.75%, which is close to what I demonstrated with my battery tester.

https://www.thunderheartreviews.com/2019/02/Samsung-33j-21700-Li-ion-battery-for-Tesla.html

Anyway, if the pack is at 40.9 V, you can expect to lose approx 10% there (compared to a fully charged pack to 42V), so say 64Wh.

 

And that is the real World way of doing the capacity comparision;

 

Stop charging at 41V, measure the capacity.

Wait a bit

Stop charging at 42V, measure the capacity.

Compare the capacities.

And that is the real World way of doing the capacity comparision;

 

Stop charging at 41V, measure the capacity.

Wait a bit

Stop charging at 42V, measure the capacity.

Compare the capacities.

 

Exactly. You have to let the battery rest to get a true picture. When discharging and you hit 41 V, if you stop and let the battery rest it will go back above 41 V after resting. In my discharge experiment, the next day the voltage of the pack had rebounded from 33.38 to 34.2 V.

 

In experiments where I have charged from 41 to 41.9 V I have been able to put in about 7 to 8% of the pack's rated capacity. Most of that is between 41 and 41.5 V. Once you get to 41.8 V, balancing starts and then it is hard to judge how much is going into the battery and how much is being bled off by the BMS. Even with fairly rapid charging (~4.7 A for 15 Ah pack) I've found charging efficiency to be around 95%.

 

If the pack is at 40.9 V, ie less than 41V, you lose another % or two of capacity. So overall approx 10%.

Edited August 17, 20232 yr by WheezyRider

Your assumption is incorrect. It's impossible to have 10% of the charge between 40.9v and 42v. I alreay showed you that it's about 3%. Any 18560 discharge chart will show the same. That means that your battery lost 7% not zero %. It's impossible to lose zero % after hundreds of cycles. Not even LiFePO4 can do that.

 

As I said above, you have to let the battery rest in order to find out how much capacity is between 4.1 V/cell and 4.2 V/cell. I have found it to be typically 7 to 8% and this is in line with what is widely reported online.

 

There are a lot of issues that have to be considered before it can be claimed how much capacity has been lost for this pack.

 

1) It is not known if these are genuine LG HG2 cells. They have the correct type of wrapping, but they were cheap and were not obtained from a reputable supplier. My guess was that they were perhaps cheaper LG cells re-wrapped in HG2 sleeves, but they could even be some Chinese cells re-wrapped as HG2 cells. Hence we have no concrete idea what the manufacturer's stated initial capacity should be.

 

2) This experiment was mainly focussed on testing cell balance rather than cycle life. To get an idea of what has been lost, a capacity test should have been done at the start, but I didn't do this.

 

3) This is a single experiment with a single Watt meter. Even if working perfectly, it will have an error of a % or two and this has not been considered.

 

4) Say you are right, and it is 3% between 40.9V and 42V. For a 648 Wh pack that would be about 19 Wh. Add that to what I got out in my discharge experiment (544), you get 553 Wh. There is still capacity left in the pack, the voltage is not falling off a cliff, so I think there is easily another 40 to 50 Wh available if charged right down to 2.5 V/cell. This would give about 600 Wh.

 

Although LG state these their HG2 cell's capacity to be 3 Ah/cell, Lygate tests even at modest discharge rates have found them to be only approx 2.8Ah/cell:

 

https://lygte-info.dk/review/batteries2012/LG%2018650%20HG2%203000mAh%20(Brown)%20UK.html

 

Which would be about 605 Wh for a 10s 6p pack. This is very close.

 

So, maybe you are right about the capacity between 41 and 42 V, they are HG2 cells, but they have lost about zero capacity...or they are higher capacity but cheaper cells re-sleeved and sold as HG2s, but because we don't know for sure what they are and I didn't do a capacity test from new, we can't draw a firm conclusion over capacity loss.

 

Although we can't draw definitive conclusions, I still think it is interesting in that after several years of charging to 41V, and hundreds of cycles, the pack is still in balance and capacity is still excellent, with an estimated total capacity of around 3Ah/cell.

Edited August 17, 20232 yr by WheezyRider

As I said above, you have to let the battery rest in order to find out how much capacity is between 4.1 V/cell and 4.2 V/cell. I have found it to be typically 7 to 8% and this is in line with what is widely reported online.

 

There are a lot of issues that have to be considered before it can be claimed how much capacity has been lost.

Can you confirm that with your method of reducing the charging voltage (was it NealH who suggested using a diode inline?), the effective reduction in full charge voltage is from 41.7V to 41V? That would giive about 0.7V difference. Saneagle did not give much details how he charged the pack to 41V so it's difficult to interpret his 3% reduction.

Can you confirm that with your method of reducing the charging voltage (was it NealH who suggested using a diode inline?), the effective reduction in full charge voltage is from 41.7V to 41V? That would giive about 0.7V difference. Saneagle did not give much details how he charged the pack to 41V so it's difficult to interpret his 3% reduction.

 

No, I don't use diodes inline to control voltage. I either adjust the voltage adjustment pot inside the charger, or as in most cases with chargers recently, change the resistor on the voltage reference IC to give the output I want. I have discussed this in my other posts, most recently:

 

https://www.pedelecs.co.uk/forum/threads/5a-36v-charger-evaluation-and-modding.45767/#post-687863

 

I have given more details here:

 

https://www.pedelecs.co.uk/forum/threads/tuning-cheap-42v-charger.43837/

 

https://www.pedelecs.co.uk/forum/threads/mikes-ebikes-3a-charger.43833/

 

It is quite easy to do once you have identified the 431 voltage reference IC for controlling the voltage. Some chargers have two 431 ICs, one for monitoring voltage and another for setting the current. So if you see two, you need to check which one does what. Start with very small differences in resistance and monitor the output.

 

Only do this if you know what you are doing and have the appropriate expertise. Use a calibrated voltmeter to ensure you get the right result.

It is quite easy to do once you have identified the 431 voltage reference IC for controlling the voltage. Some chargers have two 431 ICs, one for monitoring voltage and another for setting the current. So if you see two, you need to check which one does what. Start with very small differences in resistance and monitor the output.

Thank you for publishing your results.

I can order the chargers with 41V output when I am convinced that they are a better choice for those who want longer life and lower risk for their batteries.

Sometimes I put a diode inline with the charger circuit on a battery, but this is only to stop a voltage being present at the charger port when the charger is not plugged in. The junction voltage drop of a diode depends on temperature and current level, so it is not an ideal way of accurately controlling voltage level.

I can order the chargers with 41V output when I am convinced that they are a better choice for those who want longer life and lower risk for their batteries.

 

And which chargers are those ?

 

I realise you would not sell them seperatly.

Thank you for publishing your results.

I can order the chargers with 41V output when I am convinced that they are a better choice for those who want longer life and lower risk for their batteries.

 

It would be nice to be able to buy 41V chargers without having to modify them. What would be great is a switchable charger, so most of the time you can charge to 41V and then every now and then you could do a charge at say 41.8V to give extra capacity, or to balance. I would not recommend going as high as 42V, keep it to the minimum needed to get the BMS to initiate balancing. Ideally BMS manufacturers should start balance at a slightly lower voltage, eg, 41.6V.

 

But, you need very good quality control to ensure voltage output is what it says it is. Most manufacturers these days don't care and supply chargers at 42V plus/minus quite a bit. Some are too low to initiate balance, some are too high and damage cells.

Grin have been doing it for years:

 

https://ebikes.ca/product-info/grin-products/cycle-satiator.html

 

Ego use packs they call "56 V". Inside they are 14s, so should be giving 58.8. So it looks like they are only charging to 4 V/cell.

 

https://egopowerplus.co.uk/products/batteries-chargers/ba5600t-10-amp-hour-battery

 

Edited August 17, 20232 yr by WheezyRider

I think it's bad of Ego to advertise the Ah of their packs based on the cell manufacturers Ah rating, when that rating is for when the battery is charged to 4.2 V /cell and not 4 V/cell as seems to be the case in Ego devices. So an Ego 10Ah pack could not give 10Ah in reality if they really have used Samsung 2.5 Ah cells in 4p config, charged only to 4 V/cell.

 

[mention=6303]Woosh[/mention], you might want to bear this in mind when advertising pack capacity if you do charge to lower voltages.

At present, I assume that changing to 41V would cause a reduction in capacity around 7% and up to 10% so it's not very desirable in view that the life expectancy of the 42V charging regime is well beyond the recommended 5 years of use.

And which chargers are those ?

 

I only use SANS charger. I have 2A, 3A and 4A SANS chargers.

I only use SANS charger. I have 2A, 3A and 4A SANS chargers.

 

Do you have a calibrated DVM to measure the output of the chargers? If so, what sort of range of voltages do you get?

I have an ordinary non-calibrated DVM.

I have an ordinary non-calibrated DVM.

 

I recommend you get a decent DVM that is calibrated. Even a 1% error (often good by Chinese DVM standards) in the 200 V range can make a big difference when measuring 42 V, that's a difference of 0.42 V, so you could be way down at 41.58 V or way up at 42.42 V, the difference between not balancing, or overcooking your cells and you'd be none the wiser.

I've been looking at DVMs on RS components. Even the calibrated DVMs only seem to have a best DC accuracy of 0.5%. The desktop models seem better at 0.015%

Even with a non calibrated DVM, you can still learn something from it, ie what is the charger to charger output variability like? You might not know the exact output, but you can make relative measurements from one charger to the next and see if there is much variability.

I've been looking at DVMs on RS components. Even the calibrated DVMs only seem to have a best DC accuracy of 0.5%. The desktop models seem better at 0.015%

 

Looks like you need to spend around £500 to get anything decent. I'm lucky, I managed to get a couple of Fluke desktop machines that were being thrown out. May be possible to get something on eBay, but it's advisable to re-check the calibration on second hand kit.