Spot-Welding a123 Prisimatic tabs

[nechaus]

very good, i think i will use this method when i get my a123's

I have seen a guy mig weld around a coke can, pretty remarkable

[oatnet]

Been seek for a few days, bad for getting projects done but but gives me a chance to get around to things like trimming the pics for this post on how I addressed terminating the cells.

While spot welding is a great way to join the tabs on a123 prisimatics, it was not effective for connecting power leads to the pack. Instead I put a pair of copper straps on either side of the a123 tab, soldered wires to them, and screwed the straps together.

I cut the copper straps to length, and drilled (2) holes to solder the power wires, and (3) holes for the screws. The ones on the bottom have been tinned to avoid oxidation.

00_IMG_1504.jpg (51.2 KiB) Viewed 1481 times

To create a mechanical join, I put one wire on each side of the strap, and feed half of its diameter through the hole. The half that remains behind, gets soldered to the half of the other wire coming through the hole. With solder through the wires and through the holes, it is pretty solid.

This gives me the low resistance of 8ga throughput, with the the thin profile and flexibility of (4) 14ga wires. The rule of thumb is that a pair of wires is the equivalent to a wire 3ga lower. Each strap has (2) 14ga wires soldered to it, which are equivalent to 11ga. I put a strap on either side of each battery tab, so the (2) 11ga are equivalent to 8ga.

05_IMG_1505.jpg (36.08 KiB) Viewed 1481 times

Looks like I only took pics at this stage of the first one I did, which was my crappiest of some crappy jobs, but you get the idea.

10_IMG_1506.jpg (38.74 KiB) Viewed 1481 times

Here is a completed tab. Wierd how the tinning looks so ripply in the pics.

15_IMG_1955.JPG (53.42 KiB) Viewed 1481 times

Left, a stack of finished straps.

Center, a strap as it will be used on a single a123 cell tab. The pair of red wires gets cut in the middle, to give 2 wires on each strap.
Right, a pair of straps bolted together, ready for an a123 cell tab.

20_IMG_1955.JPG (61.09 KiB) Viewed 1481 times

8 sets of power takeoffs built up for (4) packs. On the lower right, you can see a disassembled power takeoff.

25_IMG_1956.JPG (67.53 KiB) Viewed 1481 times

The 'sandwich' is stacked up in this picture:

(3) stainless Bolts
nickel strip
(future, folded-over a123 cell tab-end)
tinned copper strap
(future, main a123 cell tab)tinned copper strap
Nyloc nut

The nickel strip is from my spot welding stash, it isn't very thick, but it helps keep the folded a123 tab-end in place. The holes in the straps get threaded by the bolts on first insertion, which takes a little effort. The nyloc nuts allow for a lot of torque, and the abundant adhesive from shrinkwrap helps keep them from loosening.

30_IMG_1956.JPG (82.64 KiB) Viewed 1481 times

Clamp the straps around the a123 tab's base; leave clearance such that the tab does not short out against the body. A misappropriated wood auguer easily perforates the cell tab for the bolt. I tried to do both the main tab and the fold at the same time, but too many parts were in motion for it to work out.

35_IMG_1967.JPG (53.45 KiB) Viewed 1481 times

with all three bolts firmly in place, I fold the tab halfway over the straps.

40_IMG_1969.JPG (62.05 KiB) Viewed 1481 times

Back out the two outside bolts as much as possible, while still holding the sandwich in place. Remove the center bolt, fold the cell tab over the rest of the way, and auger it through center hole. Because this techninique dictated the direction of the fold, the nickel strip ended up on the bolt side as pictured.

45_IMG_1970.JPG (59.91 KiB) Viewed 1481 times

Once the center bolt and nickel strip are on, the other two are easy.

50_IMG_1971.JPG (51.01 KiB) Viewed 1481 times

A finished join.

55_IMG_1972.JPG (49.97 KiB) Viewed 1481 times

On the left, a lug has been added for the balance wire. Note the nickel strip under the nuts, holding the folded cell tab against the tinned copper strap. Say that three times fast for a gold star.

60_IMG_1974.JPG (48.23 KiB) Viewed 1481 times

2 pieces of heavy-adhesive heatshrink will seal the whole thing in amber.

65_IMG_1975.JPG (55.27 KiB) Viewed 1481 times

A wide piece fits over the whole sandwich. It is tapered to make a completely encapuslate the right side, but fit under the power wires on the left.

70_IMG_1978.JPG (43.37 KiB) Viewed 1481 times

Another view of the wide heatshrink before shrunk.

75_IMG_1996.JPG (46.56 KiB) Viewed 1481 times

Shrunk, ready for the next piece.

80_IMG_1999.JPG (34.42 KiB) Viewed 1481 times

The second piece slides down the red power wires on the right, encapsulating the copper straps all the way up to the cell tab. Cut this piece at an angle, so the point covers the seam of the first piece of heatshrink. The balance wire emerges to the left.

85_IMG_2004.JPG (49.61 KiB) Viewed 1481 times

Another, darker view, but you can see how the point in the second piece of heatshrink covers the seam in the first piece of heatshrink.

90_IMG_2001.JPG (27.3 KiB) Viewed 1481 times

[docnjoj]

Thanks oatnet. I was thinking about something similar using copper strips but now I don't have to think as hard! This seems like a pretty foolproof way of connecting pouch a123 cells. What sizes are the nuts and bolts?
otherDoc

[oatnet]

Thanks oatnet. I was thinking about something similar using copper strips but now I don't have to think as hard! This seems like a pretty foolproof way of connecting pouch a123 cells. What sizes are the nuts and bolts?
otherDoc

I think the bolts were 6/32 x 3/8", I realized I needed more length and went to a larger size so I don't remember clearly. I'm sure a lot of folks have thought of doing something similar, I hope this gives them ideas on what they would like to do (or not do! ). Mostly I hope I've shown that these cells are pretty easy to work with.

-JD

[dnmun]

your stuff is inspirational. you do such good and thoughtful layout and then you have the good sense to do it right the first time.

you are right, this is how 'engineers' learn to do stuff. the right way.

better than learning from mistakes, like the current blowout on the total platform in the north sea. more bad engineering there.

[oatnet]

your stuff is inspirational. you do such good and thoughtful layout and then you have the good sense to do it right the first time.

Thanks dnmun, that is nice of you to say!

-JD

[oatnet]

I bought (110) cells from Victpower, thanks to Doc Bass for pointing that deal out to me. I tested them all for volltage, most were around 3.27v, 2 were 3.24v. I noticed (1) with a misaligned tab, it jumped out at me, if there are others I didn't notice them on this quick check.

I also have a few cells from the cell_man era, in the 16ah and 20ah flavors, so I thought I would compare and contrast. I noticed right away that the foil used on the new USA cells feels thicker and more solid, more puncture and flex resistant. Both 20ah cells are exactly the same dimensions, and tabs line up in the same spot and are the same dimensions. The new USA cells weigh 6.7% more, which I would attribute to the thicker foil.

I think it would be really hard to counterfit these cells, and have the same dimensions, really similar weight, and have identical tab placement and construction. They look clean and new and unused.

(8) boxes of cells!

boxed_IMG_2964.jpg (45.86 KiB) Viewed 1281 times

On the left, a new USA 20ah cell from Victpower. Center, a cell_man 16ah cell. Right, cell_man 20ah cell.

compared_IMG_2986.jpg (39.21 KiB) Viewed 1281 times

One side of 20ah victpower vs cell_man

head to head 1 IMG_2987.jpg (63.23 KiB) Viewed 1281 times

the other side of 20ah victpower vs cell_man

head to head 1 IMG_2988.jpg (60.52 KiB) Viewed 1281 times

comparing the tops

top_to_top IMG_2989.jpg (45.67 KiB) Viewed 1281 times

The edges

edge_IMG_2992.jpg (47.57 KiB) Viewed 1281 times

The USA 20ah cell weighs 494g, 6.% more than the original unlabeled cell_man 20ah cell.

USA_20_494g_IMG_2993.jpg (52.37 KiB) Viewed 1281 times

The cell_man cell weighs 466g

UNKNWN_20_466g_IMG_2994.jpg (37.97 KiB) Viewed 1281 times

The 16ah cell weighs 399g.

KOR_16_399g_IMG_2995.jpg (37.72 KiB) Viewed 1281 times

[docnjoj]

Thanks oatnet for that timely information. My order of 14 is on it's way fromVictpower. They seem like a class operation so far.
otherDoc

[oatnet]

Someone asked for this a while ago, I got around to cutting open a dead 16ah a123 prisimatic, that had previously been used for my spot-welding testing. There was a chemical smell, but no juice or powder flying around. Like any prisimatic, the layers of subcells are stacked on each other like a book. Enjoy!

-JD

0 new cell IMG_2236.JPG (238.27 KiB) Viewed 1133 times

1 unfolded IMG_2237.JPG (235.39 KiB) Viewed 1133 times

3 tab IMG_2244.JPG (205.84 KiB) Viewed 1133 times

5 tab IMG_2245.JPG (192.88 KiB) Viewed 1133 times

7 tab side IMG_2246.JPG (98.75 KiB) Viewed 1133 times

8 tab top IMG_2248.JPG (117.19 KiB) Viewed 1133 times

[oatnet]

In this thread dnmun speculated on spot-welding cotter pins to join a123 packs. Today I tried it out, and I like it. It does not shift around like the two steel plates used to so it is easier to setup. I don't have to adjust the two pairs of vice grips for each weld, just clamp it on one side with vice grips and zap zap zap down the row, it is much faster. This is a simple, effective method for welding cells together in high-quality join, without heat-damaging them.

I used stainless steel 1/8" x 2" cotter pins, .65 each from my local Ace hardware. The rest of the technique is the same stuff I have been posting here for two years.

Position the pin. The tough part is getting the round end of the pin over the tabs, it is quite tight there.

IMG_4290.jpg (138.96 KiB) Viewed 1092 times

First weld completed, spot welder still in place, and the vice grips where they stayed for the whole operation.

IMG_4291.jpg (133.22 KiB) Viewed 1092 times

A completed weld - note how the entire edge of the tabs are melted, welded together.

IMG_4294.jpg (63.55 KiB) Viewed 1092 times

The 18ga balance wire doesn't carry much load, so it just rests the middle of the roll. I make it a little long, and bend the end up into the next fold, so it doesn't pull out.

IMG_4295.jpg (141.43 KiB) Viewed 1092 times

like this, on the next fold.

IMG_4296.jpg (72.55 KiB) Viewed 1092 times

Last fold complete, and the whole mess crimped with electrician's pliers. The round cotter pin keeps it from crimping as flat as it did with my folding method, but between the weld, and both sides of both tabs being in full contact with each other, it just isn't an issue.

IMG_4297.jpg (84.92 KiB) Viewed 1092 times

A layer of adhesive heat shrink to seal it in amber

IMG_4299.jpg (97.35 KiB) Viewed 1092 times

Heat shrink folded over at the top

IMG_4300.jpg (70.07 KiB) Viewed 1092 times

Sliide on one of the plastic report clips I use (bought more at staples today!)

IMG_4301.jpg (48.74 KiB) Viewed 1092 times

A layer of fiberglass-reinforced tape to keep the plastic clip from stretching out.

IMG_4302.jpg (100.15 KiB) Viewed 1092 times

The whole assembly from the side.

IMG_4303.jpg (133.85 KiB) Viewed 1092 times

[dnmun]

report clips, who woulda thunk it?

rolling it over and then putting on the plastic clip allows the next set of tabs 2 cells over to be exposed so you can work on them in series without the spot welder shorting to them during the assembly. just assemble it in series.

your sense wire is as big as most use for the main conductor.

this split pin approach was a great idea that started with evb on the other thread, i just think out loud, but this is so simple now, reliable connections, easy to go back and replace defective cells.

now we just need to find something to use for those end plates to bind the pouches together in compression like the A123 packs use. with the smooth radius on the outside to distribute the force uniformly across the pouch.

[ohzee]

yea I am seriously considering going out and buying one of these spot welders now.

Looks like a lot better then compression and even safer as it's all covered up.

thanks for sharing everyone very good info here.

[oatnet]

Figured I'd lump this post into my a123 thread, as the 7s module thread has devolved into tangents.

I've been buying a bunch of the 28s3p a123 modules from victpower. The latest one came without the "Active Cell Balancers" that all the other ones had - which is OK, as they are described as having no BMS, so all the others were bonus. However, the pack arrived pretty well discharged, and without the BMS's one cell was down in the 1.8v territory, and another was a little bit lower than the rest, but the other 26 were all in balance.

I've been wondering why these packs were discarded, and pondering whether an underperforming cell was flagged by the BMS. Maybe I didn't see it on my other packs, because the BMS was discharging the other cells to boost the voltage on the bad one. The two low cells in this pack makes that seem likely - and if so, that is bad news for the 7s packs, 1 bad in 7 is worse than 2 in 28.

I decided to break out my old CBA-II and see how bad the cell is - and do I need to solder on a booster cell to keep it from draining flat? I wanted to compare the discharge curve of the bad cell to one of the good cells. Bad news is that my CBA-II appears to have issues; before I start a test, it reads .08v higher than my multimeter. Once the 26.5a test is running, it reads about 0.48v _lower_ than my multimeter. During the test, it apparently runs hot enough to melt the cover off the fuse (see bonus pictures below). Rebooting my PC and replacing the fuse did not help, so my test results be taken with a grain of salt. So instead of running a WH test - which would be skewed by the incorrect voltage - I ran an AH test, and dropped the cutoff to 2.2v instead of my preferred 2.7v. This worked out pretty well, at the 2.2v cutoff, it was just above 2.7v on my multimeter. The discharge is at 26.5a - less than .5c - but that is all my CBA can handle.

The red line is a 'good' cell, and the black line is the 'bad' cell. The bad cell discharges at a lower voltage, demonstrating higher internal resistance. Interestingly, both cells discharged a full 60ah right as they started going off the 'cliff', so that gives me a lot of confidence. Interestingly, the bad cell gave more AH - but I have seen that on c-rating comparison charts, where a high-c discharge resulting in lower voltage makes up the WH by yielding more amp-hours.

This made me re-think my perspective on LiFePO4 balancing a bit. I have been comparing cells from a WH capacity standpoint for a long time. However, all cells will discharge the same number of amps, so maybe I should think of capacity in amp-hours, not WH. True, a bad cell with higher internal resistance will discharge those amps at a lower voltage, and consequently deliver less watt-hours, but if it has equivalent AH, it won't go flat before the good cell. In fact, as we have seen in the discharge comparison below, it might outlast the good cell! Of course, the bad cell means a higher self-discharge rate requiring more frequent balancing, waste heat in the pack, and less watts in the motor, but it doesn't go off a cliff early.

At any rate, I'll take this to mean that although I need to monitor the degraded cells, I can take us much as a full 60ah from this pack without really worrying about it

-JD

NOTE: ACTUAL VOLTAGE DURING TEST IS 0.45v - 0.50v HIGHER THAN SHOWN IN GRAPH

cell 11 -2.png (31.42 KiB) Viewed 761 times

Lesson 1 - detecting a bad fuse.

IMG_4748.jpg (123.36 KiB) Viewed 761 times

Lesson 2 - this fuse is still good! Not that I would use it, mind you...

IMG_4749.jpg (140.81 KiB) Viewed 761 times