Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
High and low drain cells mixed
#81
(09-10-2019, 12:06 PM)Wolf Wrote:
(09-09-2019, 04:18 PM)OffGridInTheCity Wrote:  which goes back to my original question.  Use the MOLIs or set them aside and by extra to have 100% Panasonic in my packs..

As you are using 100% Panasonic NCR18650As right now and it is working great why change and or chance a good thing. If you have the "means" by all means I would go with 100% same cell technology. It works the best and all cells are in harmony.
Now to the charts.
Looking at the discharge cycles with the 3 NCR18650As at ~3.599V the MOLI takes a hard hit giving up ~739 mA. 37% of the pack load.
Temperature does not seem to be a problem as the MOLY only gains maybe ~1°C compared to the others.

So even if you run your packs in the 4.1 V to 3.4 V range the MOLIs cycle is much heavier thus I believe will reduce the effectiveness of the pack over time. By how much time is anyone's guess.

Wolf

Thank you so much for this work/info - fantastic!   I'm going to go with 100% Panasonics and re purpose the MOLIs to other projects.
Reply
#82
Results of the 3 min on 3min off discharge cycle
of theses cells.

104 MOLI-ICR18650M  61.1  4.1844  2551
7200 NCR18650A       34.3  4.1844  2563
101 MOLI-ICR18650M  63.8  4.1844  2660
99   MOLI-ICR18650M  58.1  4.1844  2706

Discharge 3min Rest 3min to 3V 3Ω resistor


Lots of stuff happening after ~3.65V

Wolf
OffGridInTheCity likes this post
If 18 X 650 = 2200+mAh then we have power! 
May all your Cells have an IR of 75mΩ or less Smile
Last count as of 8/7/2019
Total Number of Cells Recorded and processed                 6149
Total Cells required for PowIRwall                                   2856
Total Cells ≥2200mAh, ≥80%, ≥35mΩ, ≤75mΩ, ≥4.12V   2760
For Info Google Drive
Not your average Wolf       
            Cool
Reply
#83
Series of charts to follow.... all of the charts are covering the 3 min off 3 min off data so start with


The chart is specifically the same time point in each 3 minute discharge cycle, of which there are 186 on/off intervals in the test data.

Adding in the chart for all 4 cells together (cell2 is the lone cell) - makes me think to view this a bit differently for the second half then it all mixes up


In the temperature alone it is quite interesting that the lower grade of Cell 1 even with the lower IR than cell 3 is actually having a harder time during discharge, which is more inline with the cell relative capacity (Cell 1 is lower than Cell 3) while Cell4 has a lower IR and higher capacity. Even though Cell2 is nearly half the IR the relative temperature is quite interesting pointing to a portion of the internal heat just from the chemical reaction of discharge and not IR.
If you can't quantify how much they cost, it's a deal, I'll buy 5 of them for 3 lumps of rocking horse ......
Reply
#84



The lone cell initially is being recharged by the other three until they are out of energy and then the situation reverses where the lone cell is then having to re-charge the other 3 cells to cope with the discharge current. To make it clear 300mA (near on 1W) is still going out of the lone cell to recharge the other three cells. Scale that up form 4p to 40P and you have 10W of rebalancing going on.. enough for a light just for one parallel pack.
If you can't quantify how much they cost, it's a deal, I'll buy 5 of them for 3 lumps of rocking horse ......
Reply
#85


This one is a bit more complex to understand.

This one is showing a series made up from the difference one time point in each 3 minute recovery interval (off time) showing how much the cell current changes within the 3 minute off period. i.e it may start at 50mA and then end up as 10mA towards the end of the 3 minute off time and therefor show as 40mA. Plot these results onto a scatter (i.e. plotted against cell voltage at the time) and you then start to see more clearly when the cell is happy and when it is providing a lot of rebalancing, which is not ideal as your re-charging one cell by discharging another.

Down to 3.6V things are reasonably happy, 3.5V borderline, lower than 3.5V and a lot of rebalancing and then 3.2V cliff edge of exhaustion.

This one is a bit more interesting as to the effect of a cell under load and the difference between the two types of cell in the test.



The chart is showing the resting voltage as the cell voltage at the end of the 3 minute off time vs the voltage recovery (voltage under load vs voltage at the end of the 3 minutes) divided by the cell Amps. This sort of gives a rough proxy as to the expected voltage drop a cell may see when put under load for any given cell voltage and where a pack or cell runs out of steam because it does not recover as well. Sort of like the dregs of the barrel (that awful beer when they have nearly run out...).

The main difference in the chart is due to the different currents  as the voltages are effectively tied to within 50mV (shunt drop is taken out of the calculations).

The area between the two green verticals is roughly how the pack as a whole is coping between roughly 3.65V and 4.1V which seems to be relatively consistent before a lagrer cell imbalance issue takes over

Conclusion at this point, mixed cell builds need to have the voltage range restricted to above 3.6V in order to prevent excessive cell to cell realancing. For the 4 cell pack in this test I would restrict the operating voltage range between 3.65V (maybe 3.60V) and 4.00V (little to gain, much to loose above 4V).

Now to add in the continuous test data for the same cells...
If you can't quantify how much they cost, it's a deal, I'll buy 5 of them for 3 lumps of rocking horse ......
Reply
#86
Another chart (cell 1)...


The wording on the chart explains most of it.

Below 3.4V to 3.5V the cells are out of consistent energy and becoming more like supercaps with a rapid decrease in voltage with discharge currents.

Considering in the 4 cell pack (under test) below 3.6V Cell1 is starting to get nearly 100mA rebalancing charge, scale this up to a larger parallel pack and add up the size of the energy flow. Also considering this is close to 0.1C the higher currents at the lower discharge point may start to blow 2A wire quite easily if your pack is designed with even 0.5C.

Awesome data Wolf.

Just realised with this pack setup (which would be worse with a lower portion of cells, i.e. 1 cell in 10), cell fusing could be an issue even at 3.6V as the current differences increase dramatically even over just a 40mV drop below 3.64V.





Compare this to how much energy you may actually still have in a pack at 3.6V with a different chemistry.


Or like this




Starting to make me think the real main issue with mixing a low quantity of high discharge cells into a pack is that at a low charge state the cells could easily blow the fuses even at a 3.6V, which may still have a voltage indicating only 50% DoD depending on pack mix.

Ok, another chart. This time the aggregate discharge energy of the "4 cell pack" showing how really different the two sets of cells are in relation to the chemistry.



While the 3 cells bail and are exhausted at 3.6V the lone cell is quite happy to continue to a much lower voltage visible quite easily as there is no real drop off in rate (linear) while the three cells have an obvious knee point.

Last one before I get banned for chart spamming.



Converted the discharge to % DoD to more clearly show the difference for each of the cells and the discharge energy as a whole.

In the chart the balancing issue starts before you get to a 60% DoD on the pack and then the real fuse popping issues (for this pack mix) would start at just over 70% D0D so that 30% contingency is low load only or not even accesible at all depending on cell chemistry....... hmmmmmm.
If you can't quantify how much they cost, it's a deal, I'll buy 5 of them for 3 lumps of rocking horse ......
Reply
#87
@completelycharged: just keep up the chartspamming. it makes it a lot easier to unterstand the data from wolf! you 2 do great work for the community!

@worf: i have read that the current sensors are from the same batch, but i think it would be quite interressting if you repeat a testcycle with the same cells in a different arrangement in the cell holders. right now we have excellent datapoints that are consistant, but they are not "verified". (if it is possible to say that in english that way?)


it was interesting to see the discrepancy between the cells under 3.6V. The same happens on high current cells (6s1p 1,3Ah and advertiesed 120C) that we use on our racedrones. from 4.2V down to 3,6V the cells are normaly inbetween 20mV but under 3,6V it goes rapidly up to over 100mV


keep up the excellent work!
Reply
#88
(09-12-2019, 08:48 AM)PAF Wrote: @worf: i have read that the current sensors are from the same batch, but i think it would be quite interressting if you repeat a testcycle with the same cells in a different arrangement in the cell holders. right now we have excellent datapoints that are consistant, but they are not "verified". (if it is possible to say that in english that way?)
PAF,

Yes I was thinking of doing the same thing. One thing I have done is change the position of the odd cell from position 2 to  position 3 occasionally.
Also just with the batteries inserted and them being in a static state (not charging or discharging) the base readings are quite even. Obviously there is some noise in the µA range but that is to be expected on a measurement that small.


OK on to the final test results for OffGridInTheCity with this combination of cells.
I will be moving on to different cells and chemistries now that OffGridInTheCity's
test are done.
7194 NCR18650A        34.2  4.1845  2896
118 MOLI-ICR18650M 56.6  4.1844  2709
7195 NCR18650A        34.9  4.1844  2871
7207 NCR18650A        36.2  4.1844  2840
3 min discharge 3 min rest on these cells to 3V with 3Ω resistor.

At ~2AM there was some glitching not sure what happened there I would blame it on a Micro$oft update but this is running on Linux Tongue
Not quite such a dramatic ending as the test with 3 MOLIs and 1 Panasonic although still interesting.
I will also incorporate a 'Room temp" reading in °C very soon.
I already have a °F home sensor for that room which seems to sync up with the ambient on the board.

Wolf
If 18 X 650 = 2200+mAh then we have power! 
May all your Cells have an IR of 75mΩ or less Smile
Last count as of 8/7/2019
Total Number of Cells Recorded and processed                 6149
Total Cells required for PowIRwall                                   2856
Total Cells ≥2200mAh, ≥80%, ≥35mΩ, ≤75mΩ, ≥4.12V   2760
For Info Google Drive
Not your average Wolf       
            Cool
Reply
#89
Wolf, you have an issue with the values of t1 in the graph data lining up, minor issue, no real problem..

Chart spamming starting up....


Cells keeping relatively cool during discharge, less than 1C warming.

 
The different chemistry reactions start to show up for the cells here, switching between which cell still has higer voltage electrons to give/push.

 
This one is showing the voltage drop from the end of the rebalancing 3 minutes compared to the end of the 3 minute discharge period, which should show how much voltage the pack (difference between cells if predominantly the shunt, impacted by current).

This gives a pointer as to why much above 4V is not that great... voltage drops quite quickly with loading. Then the lower end at 3.3V starts to show an empty cell state of a rapid increment in the voltage drop.

The "blip" at 3.8V ignore as this is just an issue with the data, it is the point of imagination, just imagine it does not exist, lol.


This starts to really show the real difference as to what is happening in the pack as Cell 2 runs out of juice around 3.6V and is then being re-charged by the rest of the pack in order to provide the initial surges after a rest. Mixing these cells and running the pack below 3.6V would just end up with the one cell being discharged and re-charged by the pack, offering no real benefit. Next up DoD.

 

Taking the average voltage during discharge and the average current to work out the energy per 3 minute interval and then stacking this up. Within the data there is around 3-4% of data points missing so the values should in thoery be very slightly higher, however no real impact on the shape.

The 3.6V point becomes a little more apparent, however not really that clear unless you look at the DoD as a percentage...
If you can't quantify how much they cost, it's a deal, I'll buy 5 of them for 3 lumps of rocking horse ......
Reply
#90
 

Yes, all 3 cells are covering the same line, while Cell2 strides ahead deliverying charge and then fades like a marathon runner who forgot to have breakfast. This is showing the 3.6V difference and in this pack an 80% DoD limit would be beneficial looking at this one chart.

The series here is calculated as a nominal 100% DoD is the discharge energy delivered by the time the cells get to near 3.1V, even though some cells could have delivered more at a lower voltage. The view here is the relative energy delivered by the cells in the pack.

Spam over, brilliant data Wolf.
Wolf likes this post
If you can't quantify how much they cost, it's a deal, I'll buy 5 of them for 3 lumps of rocking horse ......
Reply


Who read this thread?
72 User(s) read this thread:
EnergyOasis (09-10-2019, 12:29 AM), stok (09-12-2019, 12:30 PM), piglet (Yesterday, 11:44 AM), smileypete (09-03-2019, 10:07 AM), mackoffgrid (09-13-2019, 03:56 AM), Emiel (09-13-2019, 10:09 AM), foamovercast (08-30-2019, 01:49 PM), drbacke (09-11-2019, 08:49 PM), juanjo57 (09-13-2019, 02:19 PM), Doin it (Yesterday, 11:54 PM), 100kwh-hunter (09-03-2019, 01:28 PM), winny (09-13-2019, 08:51 AM), intra (09-10-2019, 12:26 AM), OffGridInTheCity (09-13-2019, 04:01 PM), kevinjaye (09-11-2019, 09:07 PM), ajw22 (09-03-2019, 10:10 PM), madmike (08-27-2019, 09:48 AM), metalorn (08-29-2019, 12:31 PM), 7h0r (09-01-2019, 05:15 AM), Whiteraven1386 (09-09-2019, 07:10 AM), Eduardo Kang (09-12-2019, 05:59 PM), donnib (09-05-2019, 01:18 PM), James-w (09-12-2019, 08:29 PM), hschlech (09-08-2019, 07:03 AM), chad (09-06-2019, 02:27 AM), diib (08-23-2019, 04:03 AM), MBard (08-27-2019, 06:30 AM), Cobra_Phil (09-02-2019, 03:07 AM), berthfield (08-22-2019, 03:17 AM), dragonfly (08-25-2019, 05:36 AM), LPG (09-06-2019, 01:40 AM), drnickriviera (Today, 12:27 PM), 3xtr3m3 (09-03-2019, 01:20 PM), Otakar (08-25-2019, 08:23 AM), Jonisingt35r (09-02-2019, 11:11 PM), stevelectric (08-29-2019, 12:00 PM), Generic (09-11-2019, 04:11 AM), Oz18650 (Yesterday, 09:10 PM), ChrisD5710 (09-12-2019, 04:41 PM), Grumplestiltskin (08-29-2019, 03:34 AM), Jon (08-26-2019, 06:41 PM), SWEwall (09-11-2019, 07:44 PM), Geek (09-13-2019, 12:15 AM), Hanssing (09-12-2019, 11:39 PM), jesusangel (09-11-2019, 06:13 PM), watts-on (09-06-2019, 03:50 PM), jdeadman (09-13-2019, 12:53 PM), Franky Beuselinck (09-09-2019, 01:59 PM), PAF (09-13-2019, 06:13 AM), A + A (09-11-2019, 09:40 AM), Korishan (09-12-2019, 12:21 PM), mike (09-03-2019, 10:32 PM), Sean (09-03-2019, 12:16 PM), djuro (09-04-2019, 05:50 AM), jm1 (09-09-2019, 12:50 AM), Freedom3737 (08-30-2019, 03:29 PM), Wolf (09-13-2019, 07:07 PM), OhmGrown (08-23-2019, 07:15 AM), clupton (09-01-2019, 03:36 PM), jaeger-audiodesign (09-11-2019, 02:17 PM), emuland-metroman (09-06-2019, 05:09 PM), chuckp (09-13-2019, 12:37 PM), Jim Jr. (08-31-2019, 02:33 PM), Riplash (09-11-2019, 03:01 AM), Bubba (09-13-2019, 10:49 PM), avm131 (08-23-2019, 05:52 PM), Zvonko000 (09-13-2019, 07:04 PM), Ibiza (09-13-2019, 05:34 PM), completelycharged (09-13-2019, 02:47 PM), Crimp Daddy (09-11-2019, 01:13 AM), Chablis_m (08-24-2019, 01:15 AM), daromer (09-11-2019, 06:24 PM)

Forum Jump:


Users browsing this thread: 1 Guest(s)