Paralleling dissimilar cells (hybrid), interesting results

[dogman dan]

But you will still be leaving some capacity on the table, since you can't discharge one to as low voltage as the other. Maybe you plan on a shorter stopping point for both anyway, which aint a bad idea.

Other than the stopping point, no reason not to build your pack. But since it's going to be a pretty big pack, build it in two sections. Each section one type. Then parallel the two. That gives you some flexibility later, to build another section out of the same cells as the best ones.

 

[Overclocker]

But you will still be leaving some capacity on the table, since you can't discharge one to as low voltage as the other. Maybe you plan on a shorter stopping point for both anyway, which aint a bad idea.

Other than the stopping point, no reason not to build your pack. But since it's going to be a pretty big pack, build it in two sections. Each section one type. Then parallel the two. That gives you some flexibility later, to build another section out of the same cells as the best ones.

i just did discharge tests (to 2.7v) on the iCharger:

26F (solo at 3.0A): 2456mah
ATL(solo at 2.5A): 2208mah
26F + ATL (in parallel at 5.5A): 4540mah*

i call that a success 8) minimal inter-charging well below 0.5C. should i require more range i could go to this comparator: http://lygte-info.dk/review/batteries20 ... arator.php and select something with a similar curve


*4540 is less than (2456+2208). the icharger probably terminated a bit early coz it doesn't use the balance port. it senses end voltage over the main leads and 5.5A induces a bigger voltage drop

that's great thing about it. you get to extract all the juice out 8)

but yes i'll be packing them separately, to be charged separately also, for better health monitoring

 

[Overclocker]

works well full-scale!

 

[billvon]

but that's when doing CONTINUOUS discharge which isn't representative of typical ebike usage. suppose at half-capacity the pack is allowed to rest, the ATL would charge the sony and their voltages would equalize. this gives the advantage back to the sony and would take the lion's share again

Agreed. You want to minimize that sort of cross-charging behavior because every amp-hour that gets transferred from one cell to another incurs loss; the (significant) voltage loss and the (slight) loss because of coulombic inefficiency. Some systems I've seen intended for multiple batteries use active diodes to prevent this sort of cross-charging behavior - but that doesn't work for regen.

 

[docw009]

YEs, this is an old thread, but the topic doesn't get looked in detail at very much.

A year ago, I combined three 2200 mah 40 milliohm Samsung 22P's with two 3100 mah 30 milliohm LG MH1 cells. This was one bank in a 10S-5P battery with all the other banks consisting of 22P's. Expected a nominal 10Ah out of the pack. These were used cells, and I had gone thru the labor of measuring all 50 cells. By the way, not worth the bother of using old cells/

I ran it on my bike, and was surprised when that combo bank hit LVC at 3v after delivering 6Ah, when the other nine banks were at 50% SOC. It was like the MH1's did all the work, and I guess they did. This thread helps me understand it better..

 

[eMark]

I ran it on my bike, and was surprised when that combo bank hit LVC at 3v after delivering 6Ah, when the other nine banks were at 50% SOC. It was like those MH1's did all the work, and I guess they did. This thread helps me understand it better.

Not sure i'm following that "the MH1's did all the work" (only 2-MH1 cells).

I took it to mean that the two MH1 cells were quickly depleted in capacity moreso than the three 22P cells in that 5p bank/group. That's assuming all 48 22P's were of similar/equal capacity ... including the three 22P cells in the 5p bank with the two MH1 cells.

How do you conclude that those two MH1 cells "did all the work"? To me it sounds like they were so weak/worn out to begin with that their supposed(?) 6200mAh? (2 cells) was quickly depleted being nowhere close to 6200mAh. Thus reducing that 5p bank/group to 3V; while the other nine 5p banks/groups were at 50% (3.65V).

What am i missing when you conclude that those two MH1 cells were the better cells ("did all the work")? Wouldn't it be more likely that those two MH1 cells were the worse performers (doing the least amount of work)?

How sure are you that each of the two MH1 cells were anywhere even close to 3100mAh?

 

[lnanek]

Guess he's saying a 10s2p battery of the newer, lower internal resistance cells would have given him 6Ah anyway (2 x 3.1Ah). So it was pointless to make a hybrid battery with an additional 3p of old, used, higher internal resistance cells in combination since the hybrid battery died at 6Ah anyway.

Judging from the results in the thread, the lower IR cells drained more first, then whenever he was coasting, charged the higher IR cells which hadn't lost as much voltage, which is a waste of capacity.

 

[eMark]

This was one bank in a 10S-5P battery with all the other banks consisting of 22P's.

Sounds like he only had 48 22P cells instead of 50. If that was the scenario then one would have to conclude that neither of the two MH1 cells had anywhere close to 3100mAh capacity for that "one bank" (3-22P and 2-MH1). Maybe, the 3-22P cells in that one 5p bank/group were also sub-par ?

Wish he would explain what he is really saying about those two MH1's "did all the work" ...

I ran it on my bike, and was surprised when that combo bank hit LVC at 3v after delivering 6Ah, when the other nine banks were at 50% SOC. It was like the MH1's did all the work, and I guess they did.

How does he figure the two MH1 cells "did all the work" when it's apparent neither of them had 3100mAh of usable capacity for that one bank to be at 3.00V; while the other nine banks/groups were at 50% SOC (3.65V).

 

[docw009]

Sure. I ran the pack on my ebike and based on my past knowledge of its consumption, using wattmeters, etc, my mileage when the battery shut off corresponded to 6AH. Every cell in that battery had been measured for both capacity and IR. All the 22P's over 2000 mah. The two MH1's at 3100 mah, The simple model gives 10ah for banks 1-9, and 12.2AH for the last bank. Got a real life value of 6AH,

I replaced the LG cells, and the battery with all 22Ps delivered 9AH on my ebike which is 1.8AH per cell, about 90% of what my capacity tester, a Liitokala 500S predicted. I also put the LG cells into another battery, all MH1's, and that battery says the LG's average 2.8 Ah per cell. also about 90% lower than the battery tester.

So I exaggerated when I said the LG's did all the work, That last bank of cells somehow managed to burn 5AH and not send it to my motor. I'll take the explanation above that the strong cells went low, and the weaker ones recharged them.

 

[eMark]

The following is more for someone building their FIRST ebike 18650 battery using salvaged cells that may be too "dissimilar" in capacity and internal resistance. ALL the cells first need to be thoroughly tested so as to eliminate cells of less capacity or higher internal resistance (outliers/duds). At least 10% of the salvaged cells may be outliers because of lessor capacity, higher IR or higher than normal self-discharge (more of a problem with high energy dense salvaged cells). That said would appreciate any feedback from docw009 and others that a knowledgeable first-time DIY builder can grasp ... "dissimilar" versus what qualifies as "similar" for a general purpose 10s5p DIY ebike battery build using salvaged cells ?

This thread about "dissimilar" cells (DIY build) is worth discussing further. For example at what point are salvaged cells considered "similar" enuf (not "dissimilar") in both capacity and IR for fabricating a worthwhile DIY ebike 18650 battery build ? We can agree that the example by docw009 is definitely with "dissimilar" cells and definitely a No-No.

Whether to use "dissimilar" Li-ion 18650 salvaged cells for your 1st DIY build is always a gamble with the odds against the DIY builder. First-off most all agree that paralleling "dissimilar" salvaged cells, especially no name brand cells, is a NO-NO. By definition "dissimilar" means it's too much of a gamble that in the end wasn't worth the effort even if the cells were FREE.

HOWEVER, as long as the salvaged cells have same chemistry (e.g. NMC), similar capacity and internal resistance (w/data sheet) it's worth the effort (depending on your application). If the cells (unused, but outdated) aren't a familiar brand with no datasheet don't waste your money no matter how cheap the lot of cells.

You need the right equipment to thoroughly test all of the "similar" salvaged cells so as to eliminate the outliers/duds. Doing so will then make it possible to arrange the cells so each parallel group is as similar as possible in both capacity and internal resistance. The higher the current demand (using high energy dense cells) the more important it is for each parallel group to have similar capacity and similar low internal resistance.

Sure. I ran the pack on my ebike and based on my past knowledge of its consumption, using wattmeters, etc, my mileage when the battery shut off corresponded to 6AH. Every cell in that battery had been measured for both capacity and IR. All the 22P's over 2000 mah. The two MH1's at 3100 mah, The simple model gives 10ah for banks 1-9, and 12.2AH for the last bank. Got a real life value of 6AH,

Summary, IMO: The BMS LVC set at 3.0V was triggered by the #10 5p bank (2-MH1, 3-22P), while the other nine parallel banks were at 50% SOC. The one factor never mentioned was your average amperage drain as that 10s5p battery neared and reached the BMS LVC of 3.0V (#10 5p bank). If the average ending amperage drain was at a 7Ah drain rate versus say a 14Ah drain rate then wouldn't the "real life value" have been noticeably higher (e.g. 8.0Ah vs 6.0Ah guesstimate). In other words you wouldn't have hit #10p 3.0V BMS LVC as soon if the bldc motor was only pulling a 7Ah rate instead of a 14Ah rate.

So I exaggerated when I said the LG's did all the work, That last bank of cells somehow managed to burn 5AH and not send it to my motor. I'll take the explanation above that the strong cells went low, and the weaker ones recharged them.

Summary, IMO: Indirectly the #10 5p bank of cells (2-MH1, 3-22P) was actually supplying current to the bldc motor. The reason the two MH1 cells drained faster was due to its lower IR and greater current demand (e.g. 14amp drain) by the bldc motor (probably a 20amp Controller). Thus #10 5p bank first reaching the BMS LVC of 3.0V; while banks #1-#9 5p where still higher (50% SOC=3.65V). Even if the LVC of the Controller was set at 32.0V (3.2V x 10p = 32.0V) the battery was still at 35.85V (3.65V x 1p-9p = 32.85V + 3.0V (#10p) = 35.85V.

One factor that i don't believe has ever been discussed in this ES forum is the difference in resistance between 22P cell and MH1 cell when under a load (e.g. 14Ah) compared to when at rest. Is it possible that the IR difference between MH1 cells and 22P cells is even more noticeable when suppling a high demand for currrent (at rest vs 14Ah drain). For example when tested individually at rest the IR difference between the two is only a difference of 10 milliohms (40mO vs 30mO). However, when drained at a high rate (14Ah) the difference in IR may be at least 30mO instead of just 10mO (IR at rest).

Questions:
1. What was the bldc motors' current demand on your battery leading to only 3.0V on #10 5p bank triggering the BMS LVC ?
2. Is there more than a 10mO difference in IR of a 18650 Li-ion 5p bank when tested at rest compared to when tested being drained at 14Ah? Thus the IR of 5p 22P vs 5p 2-MH1, 3-22P is more signifigant under a load then at rest ?
3. What is the IR of a 5p 22P bank at rest (_______) vs a 5p 2-MH1, 3-22P bank at rest (_______) ?
4. Do you know if the IR difference of a 5p 22P bank vs a 5p (2-MH1, 3-22P) bank under a 14Ah drain rate is more than the same 5p IR difference at rest? My guess is YES ... thus why the #10 5p bank with lower IR (e.g. 14Ah rate) was drained even faster. When #10 5p bank hit, 3.0V triggering the BMS LVC the two MH1 cells may have been as low as 2.5V, and the three 22P cells at 3.45V (5V + 10.2V - 15.2 / 5 = 3.04V). However, after a 2 hr rest with bounce back voltage the MH1 cells were closer to say 3.0V and the 22P cells closer to say 3.60V (just a guesstimate).

The greater the current drain (14Ah vs 7Ah) the sooner the BMS LVC of 3.0V is triggered. Likewise, the greater is the bounce back voltage at 14Ah drain rate than 7Ah drain rate.