Common pack design mistakes, how to avoid?

Thats an awesome suggestion, cheap, and readily available. If the part of the copper series bus-bars that connects to the nickel paralleling strips is tinned with solder, then the solder can be taped off and the entire copper bar sealed in this corrosion sealant. I'm sure a non-conductive sealant could also be used, but the end-user who is building a custom shape of battery pack might not connect the series bus bars in the same place as another one.

In the graphic below, I now feel I understand what doctor bass was saying, and I agree with him. I have modified the drawing (originally drawn by doc bass, and located in the first post) to have the series connections in red.The black paralleling lines are still spot-welded nickel strips. I am uninterested in pack design principles for low current packs. This is because I can't imagine a DIY pack-builder going to all of this trouble just to make a low-current pack, when those are the most mass-produced, and are very cheaply available without the trouble of home construction.

Black lines are nickel strips, red lines are thick copper

BatteryPackDesign1.png
 
When it comes to contemplating the size and shape of copper series bus-bars, what is the exact distance (in mm) from the center of one 18650 cell to the next one, when using the plastic cell-separators?
 
spinningmagnets said:
When it comes to electrical conductivity, aluminum and zinc are both similar or better than nickel, and also much cheaper than nickel. I am persuaded that using 0.20mm thick nickel strips to form the paralleled sub-groups is reasonable, but for the thicker high-current series bus-bars, it seems to me that copper is easy and cheap to use there...
I've made a few ampacity tables for different metal strips in order to try to understand what is "reasonable" for weld metal strips.
Here's my original tables in an excel file : View attachment Metal Strips Ampacity & Resistances by Matador.xlsx

The thing is, "Ampacity" is NOT a "Cast in Stone " value that is only proportionnal to the gauge (aka cross-sectional area) of the metal used as the conductor. Ampacity not only depends on the gauge (ex : AWG) and the type of metal, but it also depends on the lenght of conductor, the increase in temperature and the magnitude of the voltage drop that you're willing to tolerate. I mean, Ampacity can be really high-rated if you don't mind your conductor rising at 105°C (221°F) ( :x loosing precious watts) in an open-air environnement, and if you don't care about loosing 10 volts in voltage drop just in one meter of conductor. Also, ampacity depends on the length of the conductor used... The shorter, the better.

Standard ampacity tables seldom specify these parameters when they give ampacity values. See for example : http://www.powerstream.com/Wire_Size.htm.

From these charts, and looking at what Ohms/km (mOhms/m) values correspond to what ampacity values for different gauges of copper, you can extrapolate ampacity values for different metals of known resistivities : Hence I made this table extrapolating data from PowerStream.com (http://www.powerstream.com/Wire_Size.htm) and from wikipedia (resistivity of different metals : https://en.wikipedia.org/wiki/Electrical_resistivity_and_conductivity ).
Ampacity (Powestream extrapolation).jpg
Just as an indicator of what ampacity values PowerStream.com expects from copper wire, they rate 10AWG at 55 amps (I personally think that would get hot). So if you trust PowerStream.com's ratings, use this first table.

So for you answer... Is 0.20mm thick Nickel (let's assume 10mm width) enough for each parallel cell ? Well, it depends on the cell you choose to use and what amperage you intend to pull out of it. 0.20 x 10 mm should be okay for 7.9 amps. So if you intend to pull max 7.9 amps out of each cell (as in 31.9 amps in a 4P pack), you should be okay, but the strips might get a little warm or even a little bit hot at these sustained currents...

On the other hand, if you plan to use 20A or 30A high-current rated 18650 cells (ex : Sony IMR 18650 VTC4 2100 mAh - 30A; Samsung 18650 30Q 3000mAh - 20A, LG 18650 HG2 3000mAh - 20A, Sony IMR 18650 VTC5 2600 mAh - 20A) then those Nickel Strips will be a serious bottleneck to what these cell could be capable of delivering with proper wiring. At 20 amps per cell or 30 amps per cell, these 0.20 mm thick nickel strips will heat up and fry (remember this : https://endless-sphere.com/forums/viewtopic.php?f=14&t=83505&start=50#p1234386). Even if there'is a copper bussbar collecting current from individual parrallel nickel strips. Even 10A cells (ex : Panasonic NCR18650PF 2900 mAh - 10A or SANYO NCR18650GA 3300mAh - 10A) in my opinion would suffer a bit from bottlenecking if welded with these nickel strips rated for 7.9A. It's a bit like uprgading a car with a new big turbocompressor, but without upgrading the exhaust and intake pipe system ... The restriction from the way too small pipes will make the car underperform, counteracting the potential added benefit of the new turbocompressor ...

Personnally, I don't think the PowerStream.com Ampacity chart is suited to my own battery build. I mean, I wouldn't want my conductors to go over 50°Celcius (to prevent lithium cells degradation from heat). Bare in mind that these conductors would also be enclosed (not in open-air) and I would then be using about 10 meters total of conductor strips... So with theses parameters in mind, the ampacity values has to be seriously derated (see this website for a automated calculator : http://www.electriciancalculators.com/calculators/vd_calculatoradv.htm and this website for Neher-McGrath theory on calculation of ampacity : http://www.electrician2.com/articles/ampacity.htm)
So extrapolating again (ampacity for max 50°C), I made another ampacity chart that is more stringent and more suited to my needs. Only 2 volts of drop when max amps approach ampacity. Here it is :
View attachment 1
Just as an indicator of what ampacity values I personally expect from copper wire (based on calculation for max V drop of 2 volts for 10 meter, and max temp 0f 50°C), I rate 10AWG at 30 amps. So if you trust that rating more than PowerStream's ratings, use this second table .

Just to give you an idea, in your house, your mains are required to be at least 14AWG for 15 amps rated circuits.
Well according to the PowerStream charts, 18AWG would be enough for 16 amps. Hence, I think they overrate.
With my own rating standards (from calculations made for 50°C, 2 V drop per 10 meters) ,14AWG would be rated for 17 amps... I think that is more comparable to household wiring expectations.
 
Wow, what a great job on the chart.

Feel bad knowing pulling 200 amps out of my Koinion pack is bottlenecked at the nickel strips.

Need to figure out this copper welding thing for my next pack of VTC5 or 6s.

Tom
 
litespeed said:
Wow, what a great job on the chart.

Feel bad knowing pulling 200 amps out of my Koinion pack is bottlenecked at the nickel strips.

Need to figure out this copper welding thing for my next pack of VTC5 or 6s.

Tom

Thanks. I thought the exercise would make us more consciouss of what to expect from metal strip conductors.
We really need to work out a way to handle copper either for spot-welding or bolt/screw solutions...

If only 18650 manufacturers could make 18650 with already attached coppertabs, we could just then drill the tabs and directly bolt the cells 18650's to copper bussbars...
A bit like what we see with som cylindrical industrial fuses that have thick copper tabs already on them :

100442-eco-eon-200-one-time-200a-250v-dual-element-class-h-fuse-lot-of-3.jpgfuses_classrk5_200.jpg

We need this : https://www.youtube.com/watch?v=JPhTOHxF3hM#t=30

Avoid that : https://www.youtube.com/watch?v=hvcDIDyrdJw
 
Matador said:
If only 18650 manufacturers could make 18650 with already attached coppertabs
.. or had screw terminals like the Headway cells but made of brass or copper.
 
LewTwo said:
Matador said:
If only 18650 manufacturers could make 18650 with already attached coppertabs
.. or had screw terminals like the Headway cells but made of brass or copper.

That'd be totally awsome !

According to this video, you can see (at 6:45) that the metal strips to attach the Headway cells are actually Copper (Probably Nickel-, Zinc- or Tin- plated Copper) :

https://www.youtube.com/watch?v=erS3b22FgmU

I wonder if the headeway contacts/screws terminal are also plated copper as well...
I'm getting temped on the headway cells (expensive!), but I wonder If they're rated to push big amp loads.
 
Matador said:
.....I would then be using about 10 meters total of conductor strips...

Interesting. Take a hypothetical pack of 14Sx16P. Below I created a circuit diagram of the sub-group of the sketch I made previously. The current from these four cell Ctot is then dumped into a cable. Even in this simi-ideal situation, the current from the outer cells is not equal to the inner cells due to the extra resistance. To get to 16P, we need four of these sub-groups wired in parallel. Each sub-group uses about 14in of strip. Total strip in sub-pack is 4*14in = 56in. To get 14S, we wire 14 sub-packs in series. Total strip is 14*56in = 784in (20m). But is that really what's important or is the effective length of strip from the outer cell to the cable? I haven't worked out the math but I believe it is the latter. (Still in thought here but out of time. will edit later)

30873464221_bb17664b9f_c.jpg


30154147384_2ba88cfd66_z.jpg
 
mkp007 said:
Matador said:
.....I would then be using about 10 meters total of conductor strips...
But is that really what's important or is the effective length of strip from the outer cell to the cable? I haven't worked out the math but I believe it is the latter. (Still in thought here but out of time. will edit later)

You're right... I volontarly oversimplified things by using 10 meters (32.8084 fts) in my calculations. On the other extreme of oversimplifications, if I only look at the 13 series connections of my current battery pack (copper links being 3 cm length ) lenght of conductor could be oversimplified to 3 x 13 = 39 cm or 0.39 meter... My old experimental build : https://endless-sphere.com/forums/viewtopic.php?f=14&t=57810&start=150#p1213147
If I put 2 feet instead of 32 feet lenght of conductor, temperature rise of conductor is the same. I've check using the calculator from here : http://www.electriciancalculators.com/calculators/vd_calculatoradv.htm
The thing that changes from smaller or longer lenght of conductor is the voltage drop (smaller when using shorter lenght). But temperature rise (aka heat lost per meter of conductor) stays the same.

In this calculater I tried 2 series of parameters :

SERIE 1 : Copper, 10AWG, 48V DC, 20°C ambiant, 32 feet one-way length, load 29.8066084 amps.
SERIE 2 : Copper, 10AWG, 48V DC, 20°C ambiant, 2 feet one-way length, load 29.8066084 amps..

In both case, the resulting conductor temperature will be 50°C, which is acceptable (that's at max load as defined by ampacity).
The thing that changes is the voltage drop.... Being 2.2 volts if I use 32 feet (serie 1) versus 0.1 volt if I use 2 feet (serie 2).

Try it youself here : http://www.electriciancalculators.com/calculators/vd_calculatoradv.htm
 
Thinking outside the box ....
I seem to remember a ES member built a battery by inserting 18650 cells inserted into pieces of PVC pipe and capping it at both ends ... kind of like the old "D" cell flashlights that were so popular among various law enforcement agencies. The advantage being there is no welding involved and the battery is easily disassembled. The disadvantage being ballance wires or rather the lack of them.FlashLight(799).jpgWhat I have in my perverted mind is clear lexan tubes with brass caps threaded into each end. The caps would also have 5mmx0.8mm female threads for bus bars/terminations. Between each cell would be a rubber O ring and a brass spring washer. The O rings would insure no contact between the negative casings. The spring washers would insure constant contact regardless of bending or expansion of the lexan tubes.Detail.jpg
 
Perhaps a useful design to consider is two cells spot-welded with a nickel strip, and some accommodation in the center of the strip for series connections. Maybe a hole for bolting (like Headway cells) or a raised and folded-over tab for soldering (like the fuse mount shown above).

There are examples in the stickey archive of thick copper wire soldered directly to the cell ends. The issue is that...even when the best techniques are used in the process, a few cells are sometimes damaged by the heat. Not only that, finding and removing the bad cell(s), is a time-consuming and annoying task.

Spot-welding has proven to be consistent. If you spot-weld a 100-cell pack, you can be fairly certain that all 100 cells are working properly with no damage from the spot-welding process. However, spot-welding requires a big investment up front. Almost every pack-building ebiker already has a soldering iron (or at least should have one). I am fascinated by the spot-welded paralleled sub-packs from a few years ago by supower111, and now available from Micah Toll in his "Maker Battery" kits. He spot-welds the paralleled sub-packs, and the home pack "assemblers" solder the series connections.

I am convinced that copper bars are the appropriate choice for these series connections, and the only issue I have seen so far is that raw copper bars experience corrosion over time, but now we know there are several ways to coat the copper flat bars. Now I am hearing that it may be possible to quickly zero-in on the specs needed for a DIY spot-welder to spot-weld copper onto the cell ends.

Copper strips are cheaper than nickel. The common nickel strips became famous in professional packs from established companies. However, they are almost all what we would consider low amp packs. Thicker nickel strips are not the answer. Nickel is shiney and corrosion-resistant, but if all we have to do is spray a coating over the series copper bars, I say we do it.
 
Matador said:
... but I wonder If they're rated to push big amp loads.
Headway cells have a slightly higher energy density than the larger prismatic cells but offer much higher discharge rates at around 10C (100A from one 10Ah cell)
Reference: http://www.evworks.com.au/headway-38120-10ah-cell-lifepo4-cell-3.2v-10ah

I think that you will find the disadvantage is energy density. Headway Cells use LiFePO4 chemistry.
 
I have recently been impressed by the PF cell, when used at 10A peaks. Imagine someone spot-welds six cells in-line, we solder thick copper flat bars in between every 2-cell group to make the series connection. Three series-bars connections from one 6-cell paralleled sub-group to the next. Copper is cheap, so we make them thicker than necessary. The distance from the center of one cell to the next is 20-ish mm, so each cells' 10A-worth of peak current only has to travel 10mm to reach a series connection.

Only 0.35 milliOhms according to this chart (0.20mm thick nickel strips flowing current a distance of one centimeter): https://endless-sphere.com/forums/viewtopic.php?f=14&t=84412&start=50#p1238471

10A X six paralleled cells is 60A peaks, which is what I am aiming for in an initial prototype. Of course the 30Q cell costs a little more, but at a rated 15A per cell, 6P would be 90A?
 
spinningmagnets said:
I am convinced that copper bars are the appropriate choice for these series connections, and the only issue I have seen so far is that raw copper bars experience corrosion over time, but now we know there are several ways to coat the copper flat bars. Now I am hearing that it may be possible to quickly zero-in on the specs needed for a DIY spot-welder to spot-weld copper onto the cell ends.

Copper strips are cheaper than nickel. The common nickel strips became famous in professional packs from established companies. However, they are almost all what we would consider low amp packs. Thicker nickel strips are not the answer. Nickel is shiney and corrosion-resistant, but if all we have to do is spray a coating over the series copper bars, I say we do it.

I don't know if it's really possible to "spray" a coating of pure metal,
but I have a feeling that electroplating copper at a low ripple-free DC voltage is the way to go. It's easily doable (there are many tutorials on youtube)
I've even had good results in surface Nickel-electroplating Copper with an artisal setup, so I'm pretty sure one could get perfect result with minimum investment in nickel electroplating.
See my pics : https://endless-sphere.com/forums/viewtopic.php?f=14&t=60364&start=25#p1235640

I've had good result plating copper with Nickel (using homemade nickel acetate... ususally profesionnals use nickel sulfate instead).
I havent succeded in the opposite though, that is, I havent been able to plate nickel with copper with good results.
 
Plenty of shops cheaply do zinc-plating, it is what covers galvanized nails for outdoor use. The amount of zinc can be quite thin, it only has to make the copper series flat bars corrosion resistant without adding too much electrical resistance. If zinc conductivity is near the same as nickel, these could only perform better than pure nickel strips, and would also be cheaper.
 
Matador said:
spinningmagnets said:
I've had good result plating copper with Nickel (using homemade nickel acetate... ususally profesionnals use nickel sulfate instead).
I havent succeded in the opposite though, that is, I havent been able to plate nickel with copper with good results.
If memory serves the best automotive chrome plating was Copper, followed by nickel, followed by Chromium. The chrome plating is actually VERY reactive in air. So reactive that it forms a thin HARD non-porous chrome oxide layer which protects it from further oxidation.
 
spinningmagnets said:
Plenty of shops cheaply do zinc-plating, it is what covers galvanized nails for outdoor use. The amount of zinc can be quite thin, it only has to make the copper series flat bars corrosion resistant without adding too much electrical resistance. If zinc conductivity is near the same as nickel, these could only perform better than pure nickel strips, and would also be cheaper.
I do not think you want to use zinc as a conductor. Zinc is very reactive and zinc oxide is very soft and porous. In 'galvanizing' iron alloys it is used as a 'sacrificial' anode ... i.e. it corrodes rather than the iron alloy it is protecting. I would seriously question how pushing a current through a zinc conductor would affect its corrosive properties as well. Zinc oxide is also a 'semi-conductor'. There may unintended consequences.
 
spinningmagnets said:
Plenty of shops cheaply do zinc-plating, it is what covers galvanized nails for outdoor use. The amount of zinc can be quite thin, it only has to make the copper series flat bars corrosion resistant without adding too much electrical resistance. If zinc conductivity is near the same as nickel, these could only perform better than pure nickel strips, and would also be cheaper.

It's true that the plating would add just a minimal amount of resistance to the copper core of the bussbar, but it's still way better than using pure nickel strips.

From this reference http://www.iewc.ca/~/media/Files/PDFs/Technical%20Guide/Suggested-Ampacities.pdf (see end of page one), multiply the ampacity of the bare copper bussbar by a 0.87 derating factor and you obtain the ampacity of nickel plated-copper bussbar intsead of bare copper bussbar. Still excellent conductivity ! And tremedously reduced oxidation formation.

I can only think that it would be even better with Zinc-plated copper, since Zinc is just a small tad more conductive than Nickel.
 
LewTwo said:
spinningmagnets said:
Plenty of shops cheaply do zinc-plating, it is what covers galvanized nails for outdoor use. The amount of zinc can be quite thin, it only has to make the copper series flat bars corrosion resistant without adding too much electrical resistance. If zinc conductivity is near the same as nickel, these could only perform better than pure nickel strips, and would also be cheaper.
I do not think you want to use zinc as a conductor. Zinc is very reactive and zinc oxide is very soft and porous. In 'galvanizing' iron alloys it is used as a 'sacrificial' anode ... i.e. it corrodes rather than the iron alloy it is protecting. I would seriously question how pushing a current through a zinc conductor would affect its corrosive properties as well. Zinc oxide is also a 'semi-conductor'. There may unintended consequences.

Well then, Nickel-plated copper seems like the way to go I guess :mrgreen:



And from that 0.87 derating factor and the very slight increase in resistance that results from nickel plating, this could place nickel plated copper tab in the middle between nickel strips and pure copper strips in terms of difficulty of welding...

I'm sur it's possible to spotweld nickel-plated copper strips : https://www.youtube.com/watch?v=1v84tOF9y4g
In this video, they do 10 mils (0.01", which is 0.254 mm thickness). They seem to by using 13 mm wide strips... So that'd be 3.302 mm2 area.... So if it was bare naked copper that would be around 38 amps ampacity (powerstream rating) or 19 amps ampacity according to my second table...
Apply the 0.87 derating factor accounting for the nickel-plating.... We're talking 33amps/cell (Powerstream rating) or 16.5 amps (according to my second table).

Sounds good
 
Edit (we were typing at the same time) Those are exactly what I imagine. What are the dimensions of those? Whats the minimum thickness for 8mm wide copper flatbars to make the series connections? 30mm long...

If a 6P pack puts out 60A, then each bus bar (three between each paralleled sub-pack) needs to carry 20A. Of course thicker is better, I just want to know the minimum to make the first experiment closer to the final form.

8-ga wire is equal to 8.4mm squared area of cross section, 3.4mm in diameter using round cross-section of twisted strands. So...8-ga is roughly equal to 8mm X 1mm thick bar? meaning 8mm X 2mm would be overkill, and is what I will shop for...
 
spinningmagnets said:
Edit (we were typing at the same time) Those are exactly what I imagine. What are the dimensions of those? Whats the minimum thickness for 8mm wide copper flatbars to make the series connections? 30mm long...

If a 6P pack puts out 60A, then each bus bar (three between each paralleled sub-pack) needs to carry 20A. Of course thicker is better, I just want to know the minimum to make the first experiment closer to the final form.

8-ga wire is equal to 8.4mm squared area of cross section, 3.4mm in diameter using round cross-section of twisted strands. So...8-ga is roughly equal to 8mm X 1mm thick bar? meaning 8mm X 2mm would be overkill, and is what I will shop for...

For 6P at 60 amps... Two cells per bussbar (20A per bussbar), Yes 8mmx1mm thick copper bars are bigtime overkill ! 8 mm2 copper area is close to 8AWG (8.36 mm2), so good for 73 amps according to powerstream.com's ampacity table.

EDIT : Bare minimum for 20 amps (bare copper) would be 1.5mm2 according to Powerstream.com or 3.5 mm2 according to my second ampacity table.
For nickel-plated copper (derating 0.87 factor) bare minimum would be 1.7 mm2 according to Powerstream.com or 4 mm2 according to my second ampacity table.

So for you, if your bars are 8mm wide. Minimum required thickness for nickel-plated copper would be :
Thickness = 1.7 mm2 / 8 mm = 0.2125 mm (or 8.4 mils) if you trust Powerstream.com's ampacity table rating
Or Thickness = 4 mm2 / 8 mm = 0.5 mm (or 19.7 mils) if you'd rather trust my second ampacity table

My copper bussbar pieces were made from flattened 1/4 inch outer diameter copper pipes !! They are 3 cm long, 1 cm wide and 1.76 mm thick (see here for details https://endless-sphere.com/forums/viewtopic.php?f=14&t=57810&start=150#p1213424). My bars have a cross-sectionnal area of 17.6 mm2 (equivalent to 5AWG-16.76 mm2) so good for a bit more than 118 amps according to powerstream.com's website. Way overkill LOL

Now that I've nickel plated them : 0.87 x 118 amps = 103 amps (still good :mrgreen: )
 
I know you are using data from reference tables, but i am struggling to understand why a microscopic plating thickness of Nickle on the surface of a copper conductor would cause its Ampacity to be reduced by 13%...??
For sure , its Conductivity would not be affected.
Can anyone explain ?
 
Hillhater said:
I know you are using data from reference tables, but i am struggling to understand why a microscopic plating thickness of Nickle on the surface of a copper conductor would cause its Ampacity to be reduced by 13%...??
For sure , its Conductivity would not be affected.
Can anyone explain ?

Good point, and honestly, I don't know the answer. I just took the derating factor of 0.87 from the reference I gave before as cash.
Maybe Nickel has less termal dissipation ability than copper. Maybe this is what reduces the ampacity rating...

I'd also like to know if someone knows... This guy could probably help us LOL, he's a electrician who likes to make crazy experiments (just for jokes) : https://www.youtube.com/watch?v=ut5DXxK1dvk
 
Back
Top