So we have these two massive 4/0 AWG cables that connect to the battery pack's B+ and B- terminals. Does it matter where we connect them on the pack ?
Nah, "A wire or 2 is still a wire" and "Ground is ground wherever it's found". Right ?
Plus, there are countless battery-pack-build videos and schematics out there showing packs built with this 'same side' main wiring sequence:
 |
| 4S3P example Note: not showing the parallel interconnects for each 1S pool |
Yes, absolutely, it will work. Is it optimal though ? Not really...
At same SOC (state of charge) the cells on the left will always experience more current than the cells on the right. Meaning that they will discharge, or will be charged, a tad more aggressively than the other cells. That will result into higher temperatures, more electro-chemical stress, etc. So they'll age faster.
Why ? Because of the interconnect resistance (Rint) between the parallel strings:
Each interconnect is typically in the 0.1 to 0.2mΩ range. That is tiny, but since they are in the same ballpark as our LF280 LiFePo4 cells' ~0.17mΩ internal resistance, at high current they'll have a non negligeable impact.
For instance, if 150A flow in or out of the pack, that's 50A per series string. So, assuming Rint = 0.15mΩ:
- A string: 0mV Vdrop, same voltage as Vbatt (VB+ to VB-)
- B string: 50A x 2 (cumulative currents for B+C) x 2 (upper & lower Rint) x 0.00015 = 30mV Vdrop
- C string: B's Vdrop + 50A x 2 x 0.00015 = 45mV
- Conclusion: Max drop Delta between strings = 45 - 0 = 45mV
So, string A will experience a higher voltage than B, therefore a higher current than the simplified 50A assumption above. The next post will show that this current imbalance can be significant, at least until the differing SOC depletion between strings finally leads to a 50A equilibrium.
Meaning that the A string will work harder than B. And A & B will work harder than C. Consequently, A cells will degrade a bit faster than B cells, which in turn will degrade a smidge faster than C cells.
By a lot ? Good news: not really, at least if there are more cells in series than in parallel. In that case the impact on lifetime probably falls in the Obsessive Compulsive Disorder category...
Over ~3000 cycles it will likely only mean a few more cycles of untapped life left in B & C cells once A hits the XYZ% degradation threshold. I.e. an unrealized couple more weeks over 10 years of use, or something (lots of caveats here, glossing over...).
Note: XYZ typically equals 80% for EV applications, but can be lower for solar / building use.
But this effect gets worse with more cells in parallel, or with higher charge & discharge currents. And since it does not cost anything to connect the main wiring to this or that location, we might as well pick an optimal connection point.
Ok, but where is it ?
Is this 'midpoint' connection strategy best ?
Or this 'opposite corners' strategy ?
In the 1st case, A & C will experience a 50 x 2 x 0.00015 = 15mV drop versus B (0mV). So, the Max Delta between the hardest working and the laziest strings is 15mV.
In the 2nd case, A & C will experience: 50 x 0.00015 + 100 x 0.00015 = 22.5mV drop versus Vbatt. B will experience 100 x 2 x 0.00015 = 30mV. Max Delta = 7.5mV.
And the respective Power losses are ( 0.015 + 0.015 + 0 ) x 50 = 1.5W and ( 0.0225 + 0.0225 + 0.030 ) x 50 = 3.75W. Quite a difference, but dwarfed by the total power draw of 150A x 4 x 3.2Vnominal = ~1920W.
Clearly, for 4S3P these last 2 wiring strategies are much better than the 'same side' strategy. 15mV or 7.5mV of 4S string voltage imbalance is more optimal on the long run than 45mV. And an opposite corners wiring strategy might even be ~50% better than a midpoint one from that standpoint.
Disclaimer zone... 🚫
- So far we have ignored the parallel interconnects within each 1S pool. By providing alternate current paths from VB+ to VB- those will impact the current imbalances
- Also, a 150A charge / discharge current was assumed. Obviously, the imbalances will be better or
worse depending on each application's max current (250A for this van)
Anyway, conclusion: for now opposite corners seems to be the best wiring strategy for the 4S3P 'Summer' pack. To be confirmed, though, once 1S interconnects are added. But simulations will be needed for that. More on that in the next posts.
Now, moving on to the 4S5P 'Winter' pack, it should therefore also be wired in an opposite corners fashion to be optimal. Right ?
Wrong ! Here the V drops are:
- A & E strings: ( 1 + 2 + 3 + 4) x I x Rint = 10 x I x Rint
- B & D strings: ( 4 + 2 + 3 + 4 ) x I x Rint = 13 x I x Rint
- C string: ( 4 + 3 + 3 + 4 ) x I x Rint = 14 x I x Rint
- Max Delta = 14 - 10 = 4 x I x Rint
It is indeed way better than the 20 x I x Rint Max Delta from a same side 4S5P wiring strategy. But this one is even better:
But wait ✋ In a 4S5P pack with only 1 BMS, the parallel cells within each 1S pool are also all connected together. So that the BMS can monitor each 1S pool and do balancing, or take action if a pool gets close to an overvoltage or undervoltage condition. Do these additional connections change the conclusion ? (spoiler: yes ! read on...)
Well, playing like we just did with drawings and arrows, ignoring the pool and series interconnects, and neglecting the cells' internal resistance, is too much simplification to tackle that question.
And reasoning based on minute voltage deltas is not ideal when dealing with Lithium Ion cells, where current and SOC are the actual key indicators for performance, imbalance, and life expectancy.
So, time to get the heavy artillery out and do some simulations 💪 In the next episodes...
Spoiler: several of the conclusions above are going to change, and new insights will emerge.
Comments
Post a Comment