LG Chem Batteries from Battery Hookup. 5.94 KWh

[t3kn0f1l3]

Taking off the buss bar covers is not very difficult. There are a lot of clips, but each one is easy to release. If you have a high power soldering iron, I was able to solder my balance wires to the tab where each buss bar is soldered to the PC board that goes to the balance connectors. I wanted to use the balance connectors for a few reasons, the main one being, there are already fuses on the PC board for each balance lead, but I was not able to source the connectors. Since I am connected right to the buss bars, I did use inline ATC fuse holders with 7.5 amp fuses incase a balance lead shorts, it will just pop a fuse. I drilled a hole and cut a slot for each balance wire in the buss bar covers so I could put them all back on with the wires poking out.

[I had to delete the rest of the original b/c of word limits. My reply follows.]

Hi GXMnow,

Thanks for taking the time to write up all these v. helpful details - one thing - I think I found the connector as I was searching last night. From what I can tell, the module's boards have this, which pairs to the part we need (I think?): Molex Part 347910080, Mini50 Unsealed Receptacle, Single Row, Non-Bridged, 8 Circuits, Polarization Option A, Black. Two details on this - first there are (3) different polarization options for this part, and I think we need Polarization option A, but I am going to order and test the part before I know - I will post to confirm either way. 2nd detail: If I am right about this part, then the above is just the casing - you have to also order the crimp tins (8 pieces/connector), Molex Part 5600230448‎ - N.B. this particular one is the 22AWG tin; there is a different tin for the 24AWG connector. Now I have to figure out what wire and what crimper to get - not to mention trying to determine the pin-out - I am stuck in the middle of trying to figure out the 3p 8s module, as I attempt to explain here.

Your configuration in 14s seems reasonable to me. A few have also suggested this to me in the thread I linked above, but encouraging to get confirmation from you that it works well in the field. Separating the cells in the 3p 8s module using a dremmel sounds like a good, clean option - liquid electrical tape, gap over 3/16ths inch, some small holes in the bus bar cover, minimal additional resistance through your brackets, balancing leads soldering to the bus-bar/PCB connection - thanks for all these (and more) details - v., v. helpful! One q. here: you mention that the max. charge in this part of your module is 16.8V, so 4.2V/cell group. Is that just the nominal limit to illustrate the low risk of arcing, or are you actually charging to this maximum? If not, what max. (and min.) have you set? I ask because I think I would like to charge cell groups maybe to 4.0V or 4.1V, and probably limit the low to somewhere around 3.0V or 3.1V for longevity of battery. Do you have any thoughts on this? I'm a bit concerned by the Chevy recall that 'solves' the battery fire problem by limiting charge to 90%, depending on the year of manufacture . . . ?

[EDIT: I just saw that you answered my q. about voltages in an earlier post in this thread: "They have been charging to 4.1 volts per cell, and then discharging to 3.6 volts per cell for several month now with no issues at all."]

As for inverter/chargers, I am planning a dc-coupled system (don't even know if I'm going to use the grid-tie option, although I want to have it there in the inverter for a later design update) I had completely overlooked the Radian's "spring loaded wire terminals for a 30 amp 120/240 connection." Thanks for the tip - I'll take a closer look now. Glad to hear the Schneider XW is working for you, minus the software problems. I'll be comparing the Radian and Schneider more carefully after your comments, so will keep in mind the advice you got from them as I plan (yes, it is inadequate and they should allow full functionality with non-Schneider charge controllers. The daily work-around you're using sounds inconvenient for you to say the least. As you say, the solution shouldn't be hard for their software people. I'd want to see at least some willingness to provide a solution esp. when straightforward and low-cost to them.) As for the skybox, I was initially attracted, but for my own DC-coupled purposes, I believe the XW would be better - also one of their tech people told me that the transformerless design makes it problematic for surges like any electric motor start. I've got a bunch of surging loads - sump pump, refrigerator, pellet stove, etc., but I'm sure it is a good option for other applications.


I'll use your description as a ref. going forward, so will likely have more questions for you in future. Thanks again for taking the time to provide all this v. helpful info.!

 

[GXMnow]

It's always good to hear my typing in appreciated.

If you do end up with connectors that fit, I will keep that in mind if I ever come up with another set. I would not bother re-wiring the set I have now. Here is a decent pic of the terminal side before I installed the battery into the cabinet. The main negative terminals are the right end of the two 10S sections. The main positive outputs are the far right terminal of the 8S section, and the left of the two clamp terminals that I made. The Black and yellow cables are the jump from #10 cells to the #11cells.



My two clamped on connectors are on the two halves of the buss bar that I cut. So yes, the potential across the cut buss bar is 4 cells, and the maximum would be 16.8 at 4.2 per cell. I only brought them that high once when I was balancing the pack. As you found in my post, the normal full charge is stopping at 4.1 volts, actually 57.2 volts total at the Schneider XW, 4.086 volts per cell. Measuring at the cells with my Fluke meter, it does differ by about 6 millivolts between charging and discharging due to the wire and BMS resistances. The Chevy recall did spook me a little bit, so I have been following any news. My brother and his wife both own a Bolt, yes, they have 2 in their garage. They got the notices and have both set to only charge to 90%. From the un-official reports I have found, it seems the bms/charge control system in the Bolt may be the problem. The voltage divider resistors are a bit off spec, so what it thinks was 4.2 volts, may actually be 4.3 or so. So when set for a full charge, it was actually overcharging a little, then ad it bumping the batteries around on our not so smooth roads and a cell somewhere in one of the bricks becomes over stressed and the separator fails and the cell shorts and get very hot and bursts. The heat from one failing cell cause the neighboring cells to go into thermal runaway. Take away bouncing the cells down the road, and never getting past 4.1 volts, and I think we are good to go. My 2 BMS temp sensors are on top of the center of the 8S module now, and wire tied to my clamped on positive output terminal. If either of them get over 65C the BMS will shut down the system. I may also add a smoke alarm inside the cabinet. It is on wheels, if anything goes wrong, I just yank the Anderson connector and roll it out of my garage.

Here is a pic with it all mounted up in a 19 inch rack cabinet.

The red knob is the main shut off switch. The first 3 breakers feed a 12 volt 300 watt buck converter, a 24 volt 400 watt buck converter, and a direct 48 volt battery output on a terminal strip. The 4th breaker is planned to feed a small 120V inverter in the battery cabinet. That would allow me to have some power for a light and my laptop if I needed to work on the XW during a power failure. All of those breakers are off as I am only using the high current battery connection to the XW-Pro in normal use.

The battery is now wired through the 350 amp Anderson connector you see out the top. The blue heat sink is mounted on the aluminum plate, right in front of where the FET's of the BMS are bolted on the back side. The highest current I have run yet is 80 amps, and it barely got warm. The sides and front door are perforated for air flow, and the top has openings for 2 fans, but I have not seen any need to the fans as it runs stone cold.

 

[t3kn0f1l3]

That is a really nice cabinet - and plenty of good details worked into it! I like the red shut-off switch and the cleanly-mounted breakers alot.

I am just wondering about the Anderson connector coming up out the top - is this just half of the connector and is the battery currently disconnected from the half that feeds your inverter? - I'm not familiar with the connector and am wondering why it looks like it terminates there.

My cabinet will be going into a shed, so part of my design is to insulate, heat, and have a temp. cut-off in case the heat (probably just an incandescent light bulb or two) fails. I am afraid, though, that it will have none of the class of your fancy cabinet. I like your idea of a smoke alarm. I was wondering about somehow installing an automatically-deploying fire extinguisher. Recently, I saw this product by Elide Fire, which looks like a possibility for me - maybe mounted inside my cabinet at the top, but I'm a ways off from the cabinet stage right now.

I've been turning this over since yesterday, and have decided that I am going to basically copy your cell configuration. The 3p 14s set up makes so much sense with these modules. A few q.'s for you here: 1st, could you tell me what size your black/yellow and red jumper cables are - You mentioned in an earlier post 2/0 (00), but just want to confirm. I am trying to determine cable sizing now and have come across several calculators, but would love to know what calculations you used to arrive at both your jumper cable and main battery cable gauges. Also, I am going to use your suggestion to cut the middle bus bar of the 3p 8s module in the middle - just where I've got a red wire hanging down in my pic. below. In attaching your two clamp terminals in the middle of your 8p module, did you have any problems with the physical stability of the clamp terminals? When I cut the copper bus in half, it looks like I will be losing the stability provided by the three black rivets and am wondering if I will have to add stability to support the new clamp terminals that I install there.

Thanks again for your help in sharing your work!

 

[GXMnow]

The cabinet I am using is a NavePoint I found on Amazon. It is only rated for 80 pounds when wall mounted, so I am a bit over loaded. I used a pair of Mid Atlantic brand rack rails to hold the batteries, they ae about double the thickness of the stock rails that came in the cabinet. I am using the supplied rails to hold the BMS, fuses, shut off switch, and the output Anderson connector. I mounted the Harbor Freight casters on the bottom, braced to the corners to help support the weight. The cabinet was right about $200, a Mid Atlantic brand cabinet like that is over $500.

I used to make all kinds of custom rack panels, so I had a few left over panels and the tools. So it was not a big deal for me to drill the panels to mount all the parts. The main thing to making them look good is measure twice and cut once. The breakers were a bit of a pain as I had to drill the large center hole for the lever and then get the screw holes dead on. I got it close enough on the first try, but I did have to open the holes a tick bigger for it all to bolt up.

Good catch on the Anderson connector. In that picture, the cables are still going straight to the shut off switch. The connector sticking up is bolted in but does not have the cables going to it yet. The cables from the inverter did not have the connector on the end yet. It is connected now. The connector body sticking up does not have the cables in it yet. Here is what it looks like now.

The 2/0 cable actually looks small going into a 350 amp connector.

Each brick of Bolt batteries is 3P of 60 Amp Hour cells for 180 AH total. My setup places 2 of them in parallel for a total of 360 AH. My BMS is rated for 200 amps. Each single string is connected with #2 awg Windy Nation cable from Amazon again. The two strings connect to a marine MRBF fused buss bar. Each string is going to it's own 125 amp fuse. The main output from the buss bar then goes to the disconnect switch and then the 200 amp Class "T" fuse right before the Anderson connector. I went with the largest Anderson I found on Amazon, rated for 350 amps. The cables after the two battery strings join is all 2/0 Windy Nation cable. I used the NEC ampacity charts to make sure the cables were safe. My runs are short, so voltage drop here was no concern, and they are high temp "welding" cable jacket rated at 100 degrees C. Windy Nation rates the #2 at 205 amps, and the 2/0 cable is rated at 325 amps. I used all marine grade tin plated crimp connectors and bought a hydraulic crimper.

I thought about the same thing when cutting the buss bar. I looked at those 3 rivets, and came up with a plan. I cut it at an angle, so I only lost the middle rivet. The top rivet is still holing one side of the buss bar, and the bottom rivet is holding the other side of the bar. I still did a single straight cut, but it is tilted a fair amount so I have a solid 1/4 inch past the rivet to the cut line. I thought I had pictures of the cut. This is the best I could find, right before I did the cut.

That shows my clamp very nicely. All of the current flows from the copper plate to the copper plated buss bar. The aluminum is only a clamp holding the copper parts together. On final assembly I did coat the aluminum parts with NoAlOx. The cut in the buss bar goes from the left of the lower rivet to the right of the upper rivet. The buss bar still felt pretty solid, but the cables are also strain reliefed pretty well so there is little stress on the buss bar.

Most of my balance wires are soldered right were the factory buss bars solder to the PC board, where it says "CV4" in that picture. But on this connection, since I am cutting the bar, I did have to add one balance wire to my clamp. That went on that extra 6-32 screw in the picture.

With a single BMS, I did get creative with the balance leads to the two separate strings of cells. I wanted to be sure it was safe no matter what bad thing could happen. I bought 30 ATC inline fuse holders. Each lead from both strings goes through a fuse, and the matching pairs then tie together to the balance lead to the BMS. The small wire resistance is enough to act like a voltage divider if the two cells are not at the exact same voltage. A few millivolts is no big deal. The BMS sees the average between the two strings. The wire and fuses between the two strings tends to self balance the strings to each other. I got the two strings balanced very close before I connected them, so when I did tie them together, I only saw a few milliamps bleeding between the two strings. I tested each junction before plugging the fuses in. The cables from the two negative ends join at a large buss bar before going to the BMS. The two positive cables join at the marine fuse holder buss bar. All of the intermediate cell connections are only tied together through these small 5 amp fuses. If a bank goes out of balance, the worst case is a blown fuse. If one of the main 125 amp fuses blows, then a lot of the balancer fuses might blow, but that is not a big deal, and it should never happen unless there is another serious problem. Hope for the best, but design for the worst. In normal operation, these fuses only see the occasional 2 amps of balance current from the BMS, and that is split to 2 fuses, about 1 amp each.

I just noticed in your picture, the positive is on the left, mine has it on the right. We certainly need to be sure we check which way the modules are before making any connections.

 

[t3kn0f1l3]

The cabinet I am using is a NavePoint I found on Amazon. It is only rated for 80 pounds when wall mounted, so I am a bit over loaded. I used a pair of Mid Atlantic brand rack rails to hold the batteries, they ae about double the thickness of the stock rails that came in the cabinet. I am using the supplied rails to hold the BMS, fuses, shut off switch, and the output Anderson connector.

I see re: the Anderson connector now. That is another idea I might steal from you just to be able to disconnect quickly if necessary. I was planning to use an old file cabinet I bought used last year, weld up a rack from angle iron to go inside the file cabinet enclosure, then cast a fire-resistant 2" concrete shell around that, followed by polystyrene insulation for the winter cold. At least that's my present plan. Needless to say, it will have neither the elegance nor the portability of yours.

Nice idea on the angled cut of the 8s3p module's middle bus-bar. My debt to you gets bigger every time you post! I understand that you used the aluminum parts only for structural clamping and coated them with NoAlOx, too. That's a great detail. I've been wondering also about screws in these terminal clamping applications - do we have to worry about possible corrosion with them, too?

On a similar note, what type of copper plate did you use for your clamped-on terminals?

I've just started looking into the wiring of the BMS's and your description of how to wire up one BMS for both modules is tempting-if I can be safe and spend less on the BMS portion, I will. I just started reading through this Victron manual, which also outlines one possibility for using one single BMS (? -they call it 'BMV') by connecting to the mid-point of the battery bank (pp. 21-22).

I take your point about checking the polarity of ea. battery carefully. Someone else (I think Marty) on this thread also warned me about this and I am grateful for these cautions.

Thanks again for being willing to share the details of your build!

 

[GXMnow]

The copper I used was some scrap plate I had left lying around. So I don't know the exact alloy, but it is a hard copper, similar to water pipe. It is 1/16 inch thick 0.0625 or about 1.6 mm. I kept it wide to spread out the load across the whole buss bar height and reduce the resistance inthe copper as well. My voltage drop is lower than the factor end buss bars to the studs. The one change I think I am going to make it doubling up the black/yellow #2 cables from the 10S to the 4S sections. It is certainly safe and working just fine, but the BMS is actually sensitive enough to see about a 2-3 millivolt difference from charging to discharging compared to the factory buss bars between cells. This change is not at all about ampacity, it is just trying to match the voltage drop between cells for consistent measurements.

Speaking of the BMS, the Victron BMV and the Schneider Battery Monitor do not measure all of the cells. They only measure 2 or 3 points. For a lithium pack, you really do need a BMS that monitors the voltage of every cell. In a perfect world, all of the cells would stay balanced, and mine are doing incredibly well, but it does not take much for one cell to get off balance and cause a problem. Every lithium battery failure (without physical damage) that I have heard about first hand has been the result of a cell being either over charged or discharged. I had several years of experience with RC Car LiPo packs, and 3 e-bike packs that really drove it home, how important it is to make sure no cell get out of the safe range. For the first 6 months, my cells are holding rock solid within 3 millivolts and the balancer basically never intervenes, but it is there if it needs to. But anything can happen. One idea I did have for the Schneider Batt Mon if I got one was to make my own circuit that would monitor all of the cells, and output a scaled version of the lowest and highest cells to the two mid point sensing leads. That way, the Schneider app could report the worst case imbalance. I figured that could be an interesting Arduino project. The Mega2560 board does have 16 analog input and a pair of PWM outputs could produce the scaled mid point voltages. I even found instrumentation amp chips that can measure each cell voltage with up to 250 volts of offset, and reference the reading from the chips ground pin. Analog devices AD629 or TI INA149 But the chips are not cheap. And needing 14 of them adds up. Here is the data sheet for the INA149

https://www.ti.com/lit/ds/symlink/ina149.pdf?ts=1616195004134&ref_url=https%253A%252F%252Fwww.ti.com%252Fproduct%252FINA149

Page 18 shows it being used to measure cell voltages in a series string. But it is also a small surface mount chip. I thought I had seen it in a dip package.

Of course, I ended up getting the JK BMS with active balancing, and it does all I need. The only pain is I have to go into my garage to check on it. The Blue Tooth only has a 30 foot range. Here is the one I am using, but this is a different seller.

7.5US $ 25% OFF|1a/2a Active Balance Battery Protection Board Smart Bms 7s ~ 24s 100a 150a 300a Can Rs485 Gps App Lifepo4 Li-ion Lto Jk 16s 20s - Battery Accessories & Charger Accessories - AliExpress

Smarter Shopping, Better Living! Aliexpress.com

www.aliexpress.com

Once the system is all dialed in, I should not need to even bother looking in on it. Maybe a once a month if I don't get any errors on the Schneider Insight page. But for now, I need to start the charger each day. I hope to get that fixed soon. And if I do get the 6 DC panels connected, I will be in great shape. That adds about 40% more power to my system.

 

[t3kn0f1l3]

OK, thanks for the detail on the copper bus bar. Not being an EE, I'm guessing the flatter the bus bar, the more surface area, so the better you're taking advantage of the 'skin effect'? The Victron manual, Wiring Unlimited, that I've been reviewing says in the section on sizing the bus bar (p. 26) to use the cross-sectional surface area, but it seems that b/c of the skin effect, the outer surface area of the bus bar is important, too? I'm just starting to figure out bus bars.

I see what you are saying re: the connection b/t the 10S to the 4S sections. I was wondering about that connection only in terms of keeping the two different black and yellow cables sized equally - which it looks like you have done. Interesting to know also that there can be a greater resistance in each cable than in the factory bus bars. I'll take another look and see how difficult it would be to physically separate the 8s3p module into two equal parts - maybe locating each 4s3p part of that module directly adjacent to the 10s3p part and connecting with a thick bar would solve that?
I'm interested to learn, though, whether your doubling up the black and yellow cables solves it because that's likely an easier fix.

In terms of your description of the chip and an Arduino program, I have to say 'uncle' - I'm at max processor speed trying to climb up the other learning curves and don't have enough of a background to be able to follow you there. I'll probably circle back, though, and work on some of what you are describing later on.

In terms of the manufactured BMS's, I took a look at the link you posted and re-read your description of your BMS wiring. I also read through this .pdf called Strings, Parallel Cells, and Parallel Strings (att.) that an Orion distributor sent to me. No idea what BMS I'm going to get yet, but this document cautions strongly against parallel strings for a number of reasons. I've begun to think that rather than trying to parallel my 3 14s3p strings, I am better off parallelling all the cells, so I end up with 14s9p. One advantage this seems to bring is that I could use 1 BMS with the entire battery - they said I'd need 3 BMS's otherwise.

If we leave aside the significant extra labour required to configure (3) 14s3p modules as one 14s9p battery, do you see significant upsides/downsides of going this route rather than the BMS wiring you've adopted? Am I crazy to even think of configuring my 3 modules as 14s9p? Honestly, I will spend the money on 3 separate BMS's (not Orion, though ) if that is optimal, but the cautions against this in the Orion document have me rethinking my configuration. Any of your input appreciated.

 

[GXMnow]

On the buss bars, the cross section area really determines the resistance and most of the current capacity. The surface area can help with cooling when they are pushed real hard. The factory bars are half the thickness of the terminals I made. There is no skin effect at DC, on as AC frequencies get high.

The slight difference in resistance form the black/yellow cables vs the factory buss bars is not really a problem. The only reason I can tell is that my BMS is very sensitive and giving me 1 mv resolution. The difference is less than 5 or 6 mv at 100 amps. So the difference is next to nothing. But for a while, I did have the BMS set to balance down to 4 mv. So it would trigger from time to time, and I noticed it was always the same 2 cells. Charging at 40 amps, the voltage is high by 0.003 volts and when discharging at 40 amps it is low by 0.003 volts. Just enough to trip the balancer to try and fix the difference. But since then I have set the balancer to only adjust if the cell voltage is off by 0.006 volts (6 mv) and now it is just fine, but it still highlights those 2 cells as the highest while charging and the lowest whole discharging. By doubling those cables, it appears the resistance between cells 10 and 11 will be brought down to match the buss bars between all of the other cells. But this will have zero effect on the operation.

I am certainly a bit of a techie. I have built many electronic devices for all kinds of little projects. But I am in my 50's now, and the desire to make a PC board from scratch is pretty much gone. Making my own cell level monitor seemed like a good idea, but it is again, just not needed with a good BMS. This is a discussion we had on other threads on this forum. The BMS is a safety device. A well setup system should never shut down. The device charging the battery bank should stop at the correct voltage, and the load should shut down before the battery is brought too low. The BMS just looks at all of the cells and can shut off the system if any single cell charges to high or gets pulled down too low. Most BMS units also have some form of balancing. Most of them will just use a small resistor to pull some current from the cells that are running higher than the rest. As long as the cells are balanced to begin with, and are reasonably well matched, even the small 30 milliamps should be enough to keep them in line. Larger cells, might need a little more current, and some BMS's can pull 200 or 250 ma for balancing. Most will only pull balance current when there is charging current, so they will try to "top balance" that pack. The balance current slows the charging of the high cells so the low cells can catch up. The BMS I have does true active balancing. It finds the cell with the highest voltage, and it can pull up to 2 amps ( 2,000 ma) of current from that cell, and pump it into a super capacitor. It then pushes the energy from the super capacitor into the lowest cell at the same 2 amps. This is good for a few miss matched cells, but since it can only work on one cell at a time, it is not much faster on a badly mismatched pack than a 250 ma resistive balancer, but I do like that it does not waste power into heat, it just moves it from cell to cell.

My parallel strings setup is not ideal. It is a half way between a 6P pack and two separate strings. My BMS was $200 and I was being cheap and didn't want to buy 2 of them. 2 separate BMS units has the one big advantage that if something goes wrong, you have a second pack that could keep you running at half power. Depending on a single BMS has already bit me once, but the grid was up, so it was not too big of a deal. One of the BMS lead wires was not crimped properly. This was the harness from China that came with the BMS. It cause the cells on both sides of the failed connection to read no volts, so it shut my system down. I was very worried that my $200 BMS board had just quit. But it was just the bad connection. Once I figured that out, it all came back up and has been working for a month since with no issues. When I do expand my battery bank, which I will do, the next string of cells will have it's own BMS. I might get a second JK like I have and split my bank into 2 separate systems, but it is working fine, I just have no backup at this point. I did the big NO NO and ran for 2 full days without a BMS, but I set the charge voltage lower and checked the cells about every 6 hours. I knew they were holding balance fine, and I was here to watch it. But I still stand by what I have said many times in the forum, "Always use a BMS!" I only did that as a test, and we had a rain storm come in and I wanted to have battery available in case we lost power.

How much current are you going to run? 3 strings of cells... I think three separate 100 amp BMS units would be a good setup. Too bad, even the 150 amp rated JK like I have are still $150 each, so 3 of them does add up. But how much did we spend on batteries? So in hindsight, I think I would still do 2 for my battery bank, 14S3P x 2 I spent close to $200 again on some more conduit fittings and such to tie in my new air compressor and welder outlets. I know the way I set it up is safe. If any cell goes out of balance for any reason, it might pop a fuse and the BMS will shut it all down. What I have done is a hybrid between the 2P setup and the 2 separate string in the Orion pdf your attached. To make it a true 2P (or 6P in our case) we would need to use a jumper cable or buss bar capable of taking the full current of the pack. That would be 15 of those black/yellow cables jumping each buss bar pair together. But 99.999% of the time, they will not pull any current at all. So I did a bit of math and figured the fuse holder wires which are #16 awg about 2 foot long would be able to easily handle 5 amps if the two cells groups did go a little out of balance from each other. The two ends of the strings are tied together with the #2 awg cables and 125 amp fuses. All of the high current cables are cut to within 1/16th inch of each other. And I measured with a DC current clamp meter to verify the 2 strings are carrying within 2% of each other at all times. So the cell connections should all stay balanced just fine, and if there is a little drift or mismatch, the strings can pull up to 5 amps through my balance lead fuse holders to bring the two strings into balance with each other. So the cell 1 of both strings have to stay in balance, and the cell 2's, and cell 3's etc. But if the cell 2's are 5 mv low and the cell 7's are 4 mv high, then the BMS has to deal with that with 2 amps of balance current. If something does cause the 2 strings to go out of balance from each other by enough that it will pull over 5 amps through the balance lead, then the fuse will pop. The BMS should then see a cell out of balance and if it goes too far off, it will shut down the system. I actually put an amp meter in place of the fuse in a few random locations and ran some high charge and discharge currents, and I never saw more than a few milliamps going between the 2 strings. These Chevy Bolt cells are very well matched. To give my balancer a torture test, I connected my 12 volt 700 watt inverter to the 4S group. Ran about 200 watts for a while which caused a pretty bad imbalance. The 4S section was 60 mv lower per cell than the 10S sections. The JK balancer saw this and worked the active balancing overnight pulling power from all of the 10S cells, and pushing it into the 4S cells. The next morning, the balancing had stopped, and all of the cells were within 4 mv again.

So I am confident that this solution on these cells, works just fine. If they were surplus or grade B cells, they might not behave so well, but with these, I am not worried at all. But there are some BMS units I could have bought for the cost of those 30 fuses and holders.

The system just completed the absorption charge, it had dropped down to just 6 amps of charge current. The lowest cell group is at 4.064 volts, and the highest cell group is at 4.067 volts. The battery pack is at 21C or 70F and the outside air temp right now is 65F. The charge current dropped to zero (no float) as I was watching the BMS, the absorb cycle only lasted 15 minutes. The XW-Pro says it put 7.6 KWH into the battery on this charge cycle. The BMS shows 142 amp hours went in. At an average voltage of 54 volts, that is about 7.668 KWH's so very close. That is putting almost 40% back into the battery, but I am only going up to about 88%. And I had only run it down to about 50%. SO the numbers all work. Well within 5% error. The sun is shining strong now, so it will be a bit before the solar dips low enough for the battery to start running the house again. Since the battery is now full, I am running all my loads on solar and pushing back 1,700 watts out to the grid.

 

[t3kn0f1l3]

Thank you for clarifying that skin effect is not a phenomenon in DC current. I understand re: the thermal (cooling) benefit of having a thinner plate, though, given the conductor's equal cross-sectional area.

As for the very minimal ~3mv (.003V) difference caused by the black/yellow connection b/t the 10S to the 4S sections, I understand that this is minimal resistance, so not a serious concern. I'm still interested to know your measurements on this detail after you have hooked up the second set of cables and have had the chance to test.

If cells fail only due to under/over-charge, and if the charge controller and inverter/charger cut off as they are supposed to, then I see that your conclusion follows: "A well setup system should never shut down." I am following you here and in the supporting point that these LG Chem cells are in good shape, everything is well-matched, etc. Your description of active balancing is really clear. I also like that the energy is transferred in this set-up rather than simply wasted (passive balancing). I'm not sure what it would add up to over time, but the concept is attractive.

One q. for you here about your JK BMS: I am now forgetting the source, but I read that in some cases when a BMS is set for a max charge voltage lower than the cell's max voltage and there is top-balancing, the BMS might not engage the balancing feature, which only kicks in at the cell's max voltage. Given your description of various instances of balancing, I am assuming that you do not have this problem with the JK BMS. Did you have to apply a particular setting beyond the maximum voltage setting to get the JK to balance at your 4.1V max charge, or was this unnecessary?
Good to know that your JK has worked well aside from the loose connection that you fixed.

In terms of current - I plan to use as much as I can produce by solar. My household use is ~33KwH/day after having reduced and changed out a number of appliances, etc. over the past few years. My plan for this summer is to put in ~3Kw of panels, then in a year or so another ~3Kw. (As I think about it, it might be best to just get the Schneider XW Pro 6800 W model so I don't have to get another inverter in a year or so for the second panel array . . . ). In this case, I would not be at the level where I could satisfy all my loads + export to the grid, as you, but eventually want to run 6000W of power at 48V. Using I=P/V (just showing my math, so it can be corrected, if wrong), I'm getting 125A. If this math is correct, I should be fine with a single 150A BMS, as you have? - or if I have (1) BMS / module (x3) modules, do I divide the 125 by 3 = 41.67 A / module?, so 100A is fine (as you suggested)? Please bear with my newbie-level calculation questions here.

When you say, "To make it a true 2P (or 6P in our case) we would need to use a jumper cable or buss bar capable of taking the full current of the pack. That would be 15 of those black/yellow cables jumping each buss bar pair together. But 99.999% of the time, they will not pull any current at all," can you please tell me what the v. unlikely case is that would require the 15 black/yellow jump cables to be sized for the full current of the pack? (this is so I better understand what I would be doing)

If I go this route, I am thinking of maybe using 15 copper buss bars to connect the 3p-3p-3p groups of cells. I just did some surgery on one of my 8s3p packs to see what kind of physical orientation I can get for the (3x) 14s3p packs to see if a bus bar set-up might work. Here's a pic that shows my debt to your diagonal cut idea.

Thanks again for all your answers and clear descriptions!

 

[GXMnow]

I found my left over #2 wire, so I might make up the extra pair of cables to double the cell 10 to 11 jumpers soon, but now I need to do paying work for a bit. If/When I get those added, I will try to document any differences in operation. My Fluke meter can measure the voltage drop with 0.1 mv resolution, so it should show up any difference at 25 amps.

The JK BMS balancing has a few adjustments to tailor it to your battery pack. It does allow you to set a minimum voltage that a cell must be above for it to turn on balance current. I have this set lower than I ever run the pack, so even at my current 51 volt minimum, it will still balance if it sees the highest to lowest cell difference exceed 6 mv. If I was running LFP cells, I might set this to only balance when the cells are above 3.2 volts as you don't want to pull them if they diverge at low state of charge. Just bring them together when at high state of charge. But our NMC cells have a more linear state of charge to voltage curve, so each cell should be at the same voltage at each point along the charge curve. With the active energy transfer balancing, I think it would also work lower on LFP cells, but with the passive resistor balancing, it certainly should be turned off when the cells are lower, it would just waste power trying to line them up anywhere but full. The whole (Top Balancing) idea is that even if the cells are not well matched in capacity, having them all reach full charge together is going to get the most out of the pack. For example,. let's take a 4S pack with three 120 AH cells, and one that really only holds 100 AH. You have them all perfectly top balanced at full charge. As the system pulls power, the 3 120 AH cells all stay in balance, but the one weak cell drops a bit faster. After you pull out 50% from the good cells, they are all at 60 AH remaining and 50% state of charge, but that 4th cell is at just 40 AH remaining and 40% state of charge. If you had the JK active balance running, it would be pulling power from the 3 good cells and trying it's best to stuff it in the low cell. This sounds like a good idea, but it actually causes a problem when the system charges again. Over a 5 hour span, the balancer manages to push 5 AH into the low cell, bringing it up to 45% SoC. And at the same time, it pulled the other 3 down, a bit less than 2 AH each, to 58 AH remaining. If you kept discharging, this would help extend the run time as the weak cell is getting help, but if we charge back up from here, the weak cell can only accept another 55 AH before it is full. That could cause the BMS to shut down charging at that point. The balancer would be working in reverse, trying to slow that cell and put the extra power back into the higher capacity cells, but unless the charge rate is low, it won't keep up. So what happens is the one weak cell causes charging to stop when it is 100% full SoC but the other three cells are now just at 113 AH remaining, or 94% SoC. The balancer, UN balanced the system. If we don't balance at lower SoC, and only at the top, the cells will still diverge as they discharge, but with the weak cell being down to 40%, while the good cells are at 50%, they will all still take 60AH again to reach full charged all at the same point and be back "top balanced". So the balancer only working at your chosen full charge voltage works the best. I let mine balance much lower because the capacities have proven to be extremely well matched and this problem is not happening.

If you are going to DC couple the solar into the battery bank, then the XW-Pro is an excellent rock solid unit. My AC coupling from Enphase microinverters is working well, but not as seamless as I had hoped. It is really a dumb programming choice by Schneider. There is no way in their software to trigger it into charging the batteries at a specific time or even when solar is good etc. when using AC coupling only. Every morning, I have to click "Force Bulk Charge" and then it works. But if I am not here to do it, it leave my battery down at 50% charge, and won't charge or feed power to the house the next day. But if the grid fails, it jumps to action, powers the house, gets the solar back working, and will charge if the solar makes more than i am using. Go figure?

Your numbers are about right. I sized my system to be able to run the full 6,800 watts, 140 amps if needed. When you parallel 2 battery banks, you won't quite get double the current. No matter how well you matched everything, one bank will take a little more current than the other. I got my two strings very close, but I still see about a 2 amp difference with a clamp meter when running about 80 amps. That is very close, and should not be expected in most situations. In my normal day to day use, I only run 30 amps, with surges to 50 amps, so a single string can run the system. I fused each 3P string at 125 amps, and the whole bank at 200 amps. With 3 strings I would probably do 100 amps per string and still 200 amps total. Use a bigger main fuse, and you need bigger cables to be safe and legal. My JK BMS claims to be 200 amps continuous, and up to 300 amps surge. I like to use the China built 50% rating. 100 amps all day and surges to 150 should be fine. But I may still just put in a second BMS. The $200 each is really not that bad when you add it all up. And if one does fail, the other stays working. My one bad balance wire connection shut my whole system down.

The main difference between my 2 strings, of 3P and a single 6P system is how the current is shared. If the 2 strings are perfectly balance and matched, it won't make any difference. But lets say one 3P group is a little weaker, in one string, and a different cell group is a little weaker in another string. The total pack voltages are the same, and all looks fine. Having 2 separate BMS's, they would both be trying to balance those weak cells up, one is pushing power to cell group 4, while the other is pushing to cell group 7. (just picked random cell numbers). This all works just fine. But I am only using a single BMS. The different voltages at the 4th and 7th cell cause a problem. Even though the 5th and 6th cells are balanced just fine, the voltages on both of the cells are slid a bit up in one string, and a bit down in the other string. My tiny #16 balance wires are seeing this voltage difference and carrying a fair bit of current between the two strings. The balancer is forced to push energy into the 4th and 7th cell groups of both strings, and the stronger string is also pushing power to the weaker string at those cells. If the mismatch between the two strings is too much, the balance leads might carry a lot of current. This is why I had to fuse them at 5 amps to protect the wire. A true 6P setup would be able to carry the full pack current across the cells and force all 6 to always be at the same voltage. The 3P groups have the whole large tab all welded together. Ideally, you would want the other 3P to be tied with the same large buss bar to share the load. If you look at a Tesla battery, they parallel a huge number of cells. If some cells are 10% more or 10% less capacity, it does not matter, as long as the whole group in each parallel array adds up to the same total capacity. The same thing happens with our 3P groups. And that can spread to a 6P or 9P group as well. If they are all tied with a heavy enough bar, all that matters is the total capacity. But when I do it with 2 strings, we need the two strings to match or there could be high currents flowing between the two strings. If I saw this as a problem, I would have gotten the second BMS right away. But as it turns out, the Bolt packs are very well matched, at least my 3 bricks are.

Just think about all those buss bars and the current on them. Try to insulate as much as you can. I dropped a meter test lead and it tapped a buss bar and burned the end off the test lead. A row of open buss bars could be a disaster. These batteries can put out crazy amounts of energy. I am trying to come up with reasonable covers for my exposed studs.

 

[Mart Hale]

GXMnow, I love how you compression fit the copper to split the packs..... great ideas there.​

 

[t3kn0f1l3]

I agree, Mart. GXMnow's shared a whole series of ideas I am really benefitting from, the compression fit copper busses included.

I also really appreciate the caution about my multiple vertical bus-bar design. Does either of you think that using heat shrink to insulate copper bus bars is sufficient protection? I suppose it would make a difference exactly what heat shrink I propose, so if that is true, I'll come back and post some links to part#'s. I have been thinking about whether I should use insulated bus bars or alternatively go with something like 2/0 cable to connect the 3p-3p-3p cell groups. I like the idea of the bus bars because it seems to me that there are fewer possible failure points - crimps, terminals, etc.

Any suggestions/advice on this appreciated. Either way, though, I'll likely use the compression fit copper connection to each 3p cell group tab.

 

[GXMnow]

I spent a little time today to do an experiment. While my system was charging at 25 amps, I went and measured the voltage drops across a bunch of the connections in my battery cabinet. All of my connections are still rock solid. The voltage drop across any cable was less than 4 mv end to end. Even down to 0.1 mv, I did not see any drop from connector to buss bar. The cables jumping between cell 10 and cell 11 were dropping 4.4 mv. The error on the BMS shows cells 10 and 11 about 2 mv off. So just to see what it would do, I paralleled a #8 awg wire with the existing #2 awg cable. This wire cut the voltage drop to just 3 mv. That cut the error at the BMS in half. This proved my point that the cable drop is the cause of my small BMS cell voltage error. Next time I shut the system down, I will replace these small #8 wires with a second #2 cable, and the error should be completely gone.

 

[Mart Hale]

Many of my problems just went away when I moved from a #4 cable to a #2 cable with power supply. but my BMS does not extend across batteries, each battery has it's own BMS

 

[t3kn0f1l3]

I spent a little time today to do an experiment. While my system was charging at 25 amps, I went and measured the voltage drops across a bunch of the connections in my battery cabinet. All of my connections are still rock solid. The voltage drop across any cable was less than 4 mv end to end. Even down to 0.1 mv, I did not see any drop from connector to buss bar. The cables jumping between cell 10 and cell 11 were dropping 4.4 mv. The error on the BMS shows cells 10 and 11 about 2 mv off. So just to see what it would do, I paralleled a #8 awg wire with the existing #2 awg cable. This wire cut the voltage drop to just 3 mv. That cut the error at the BMS in half. This proved my point that the cable drop is the cause of my small BMS cell voltage error. Next time I shut the system down, I will replace these small #8 wires with a second #2 cable, and the error should be completely gone.

Thanks for the update on this. I am using your finding in my build - I ordered some copper sheet, same size as factory bus bar, and once it arrives will use it to make the same connection on the newly-configured module at "this electrical connection is still unfinished" in the pic.

 

[t3kn0f1l3]

I'm trying to figure out what is needed for our Bolt modules in terms of charging or balancing before connecting the parts of the modules together into a battery bank. I've been looking at the tutorial on top balancing here and watched the recommended Will Prowse video to get a sense of the principles I would need to follow, but that is for LiFePo4 chemistry and also is intended for those starting with individual cells rather than with modules, so I'm not sure how closely I should follow the protocol (other than knowing not to charge up to only 3.65V, etc.). Any tips/guidance on how to charge/discharge/balance the parts of the modules before connecting, first into (3x) 14s3p config. and then into a nicely-blanced 14s9p config. much appreciated.

 

[GXMnow]

The NMC chemistry Bolt cells have a much steeper voltage curve than LFP cells. This actually makes balancing them a bit easier. But having them already connected in series groups makes it a little harder and much slower to complete. The first step is to measure all of the cells and make a chart to keep track of it all. If you have any luck, many of the pre connected cells should be pretty close to balanced already.

Let's say you have 4 that are in series that all measure within 10 mv. You can work on that group as one higher voltage section. With NMC you just need to get the terminal voltages all equal, within 5 to 10 mv should be fine. And it does not matter where, as long as you are above 3.5 volts per cell, and under 4.1 volts per cell. So if you have 5 cells at 3.8, 4 cells higher, and 5 cells lower, you can charge the low ones, and discharge the high ones. If 3 high ones in series are balanced to each other, just connect a 12 car headlight and watch the voltage. You should get a good idea of how fast it is moving. If it drops 0.3 volts in an hour, and you need to drop another 0.1 volts, check it in less than 5 minutes. It is pretty linear. So it should take 20 minutes at that rate to go 1/3 of the voltage it moved in an hour. It will probably be slower with just one headlight. My 6P pack takes over 20 amps to move 1 volt per hour on the full 14S string. That is just .07 volts per hour per cell. When you get close, watch it close, and maybe drop to one cell at a time to bring them down to match the middle cells. the cells that are low need to be charged up. I used my 5 amp RC car charger, on up to 4 series cells at a time. it was very slow, but it worked. My 4 series cells were within 0.007 volts, and stayed that well balanced as they charged up slow at the 5 amps. When they got close to my middle cells, i lowered the current to just 1 amp and brought them to within 0.007 of the middle cells. At that point, all 28 of my 3P cell groups were within 0.007 from lowest to highest. and they stayed there after resting a full day. I connected up the series connection and tied the negative ends of the strings together. i then checked between the buss bars of the two strings to make sure each inter cell connection was at the same voltage. And the full pack voltages matched within 0.01 volts. I was then able to connect the two strings with a 16 gauge wire about 3 feet long and it pulled very little current between the strings. I connected each cell group together with fused 16 gauge wire and then to my BMS since I am using just one. With your separate BMS units, that won't be needed. My bolts cells, even with one group having a very different build date are all holding excellent balance.

 

[t3kn0f1l3]

This is really helpful, GMX. Thank you. I'll take a few days and follow your protocol and will probably be back with a q. or two. Among other things, I was wondering was how to draw current, since I don't have any type of 48v inverter yet. Your suggestion re: the 12V headlight works for me - maybe run two or more in series, depending on the # of cells I need to draw from. Thanks again.

 

[t3kn0f1l3]

View attachment 41548
That shows my clamp very nicely. All of the current flows from the copper plate to the copper plated buss bar. The aluminum is only a clamp holding the copper parts together. On final assembly I did coat the aluminum parts with NoAlOx. The cut in the buss bar goes from the left of the lower rivet to the right of the upper rivet. The buss bar still felt pretty solid, but the cables are also strain reliefed pretty well so there is little stress on the buss bar.

Hi GMXnow,

I'm studying your clamp design more closely today - my copper sheeting has arrived - and I am wondering something: is the reason why you elected to make a clamp that does not require holes in the original factory welded tabs so as not to reduce the conductivity at that juncture? Or, was it simply to bring your connection to the cable lug further out to make it more accessible? Or was there maybe some other reason? I am asking because I am tempted to use a design that requires holes in the original factory welded tabs (I believe like Mart). Just wanted to see what your thinking was on this point before I take the plunge. Thanks!

 

[GXMnow]

I had a few reasons for my clamp design. My biggest concern was I didn't want to disturb the factory weld. I expect that it is fully able to conduct the power from the 3 cells to that buss bar as is. But if I drilled through those welds, I was not sure how much of the connection I would mess up. The way I cut the clamp pieces, the clamp force is all concentrated in the middle of the tab. The force is about 1/2 of the force at the screws. I used stainless steel 6-32 screws. Along with lock washers and washers, it can produce a very high clamp pressure. The pressure is along a line the full height of the existing buss bar. I sanded the copper clean and put some dielectric grease on the contact area to keep out moisture and stop corrosion. I then coated the contact area of the aluminum clamp pieces with NoAlox so it shouldn't react against the copper. As I tightened the clamp, it was obviously deforming the battery tab side of the connection. I didn't want to smash the weld too much, but it certainly squeezed the air space between the welded area. Stainless 6-32 screws can be torqued to about 10 inch pounds. That produces over 300 pounds of clamp force per screw. I was going to use Grade 5 allen cap screws. That would increase the clamp load to over 500 pounds per screw, but that didn't seem necessary as I was already crashing the tab at this load. And I also figured that 2 screws was just fine with my 1/8 inch thick clamp plates spreading the load along the tab. I have checked the torque a few times, and they have not loosened at all. And the voltage drop is still less than the drop at the factory positive output terminal. I am not too surprised that my terminal have less voltage drop. The stock end buss bar out to the connection post is not very thick, less than an inch wide, and over an inch long. My copper plate is thicker, wider, and shorter.