soldering vs spot welding debate

aethyr

10 W
Joined
Mar 20, 2017
Messages
76
I'd like to open this discussion because with my experiments and observations, I'm now convinced that soldering is ok. I'm willing to be proven wrong, but take my following observation:

The heat under a weld is so hot, it melts solder. I can take heavily tinned wire, place a nickel strip on top of the tinned wire, apply 1 or 2 weld pulses on the nickel strip and it will melt the solder in the wire underneath. I use the strip because otherwise the heat is so quick and violent that it can make the solder blob explode! I want to emphasize this again - the heat of a spot weld is hot enough to instantly melt a substantial blob of solder. That means the total energy of a spot weld can be equal to soldering, and possibly greater, again depending on the wattage, temperature and duration of the applied iron.

That energy of the spot weld (at least equivalent to the energy of soldering), by the laws of physics we hold dear, has to dissipate into the cell. It doesn't just disappear into thin air. It goes into the cell, the gaskets, protection rings, chemistry, everything. And the notion that because a spot weld is quick (on the order of tens or hundreds of milliseconds), that somehow its less damaging is itself a ridiculous notion. In fact, the faster the energy is applied, the more wattage has occurred. For example, if I apply 100 joules of energy in 1ms, that is far more wattage than if I applied that 100 joules in 1s - 100,000 watts vs 100 watts. It is equivalent to taking a magnifying glass to concentrate sunlight. That concentrated sunlight is far more damaging, because it has more watts, but it still has the same joules of energy as regular unconcentrated sunlight of the same area of the magnifying glass. So if I shine magnified sunlight light into 1g of water for say, 1 minute, the temperature of the water will be equivalent to applying regular sunlight that has the same area as the magnifying glass, for the same amount of time. This is one way to measure joules - take a measured amount of liquid that of a known specific heat, apply energy and then measure the temperature increase.

This same concept applies to melting a given mass of metal, and then temperature of the resulting liquified metal. While knowing the exact joules is harder to extract by melting metal, if two applications of energy on similar masses of metal both melt those masses of metal, then their energy is going to be similar to each other. There will be differences, of course, depending on the temperature of the liquid metal, but the majority of the energy was spent in melting.

Now, I think the assumption has been that the heat energy of a spot weld is less joules than soldering, so perhaps the wattage of spot welding and soldering is the same. If a spot weld occurs 1/100 of the time as a soldering iron, but if it has 1/100th of the energy of the iron, then they will have same wattage. If this were actually the case, then I think the argument can be made that spot welding is less damaging. However we know they have similar energy because they both melt a similar mass of solder. That means that regardless of the exact measurement of joules, the wattage inflicted on the battery is far greater than the wattage applied from soldering.

Now, we can actually measure the heat energy of the spot weld and compare it to soldering. One joule is equivalent to increasing the temp of 1g of water 0.24C. I can spot weld nickel strips in a specific volume of water and measure the temperature increase to measure the joules generated by the spot weld. I then repeat, using a soldering iron placed straight into the water, measuring the temp increase, and calculating the joules. At some point the soldering iron will generate more joules by simply holding it in the water too long.

The question now becomes how many seconds of soldering heat applied, at what temperature, and wattage will it become more damaging than spot welding. But at some point, before that critical time point, soldering is at worst, equal to spot welding, and at best, better.
 
There's not a debate. Spot welding is imperfect, soldering is worse. Ultrasonic welding is really good. All methods can build a useable pack.

I soldered my first 18650 pack because I had no spot welder. It turned out ok, but I'm sure there is damage done from soldering to the case.
 
There is no debate and few will be foolish enough to engage in something so non controversial. Have you ever seen a soldered commercial pack...
 
tomjasz said:
There is no debate and few will be foolish enough to engage in something so non controversial. Have you ever seen a soldered commercial pack...

Spot welding is faster and cheaper than soldering. If a single layer of nickel strips are sufficient given the pack design and usage constraints, then welding is the obvious choice. Welding is the choice for commercial packs for economic reasons. I've also seen plenty of crappy commercial packs.

Can you provide a counterargument other than saying there is no counter argument? Do you refute that spot welding produces enough energy to melt solder?
 
Hmmm...

Try spot welding a tab onto a cell and immediately touch it with your finger. Do the same with a soldered connection.

There's like 10x more energy pumped into a soldered joint. Spot welds heat up only a tiny area so it dissipates quickly into the surrounding metal. There are plastic separators and insulators inside the cells that can melt if too hot, not to mention the active material and electrolyte that doesn't like heat either.

I have soldered a rather large pack together and it is still working to spec, so even though soldering is way more heat, a good quality cell usually survives it. Optimal soldering technique to minimize heating helps a lot too.
 
flat tire said:
There's not a debate. Spot welding is imperfect, soldering is worse. Ultrasonic welding is really good. All methods can build a useable pack.

I soldered my first 18650 pack because I had no spot welder. It turned out ok, but I'm sure there is damage done from soldering to the case.

I agree the soldering is less precise than spot welding - but I'm not convinced its always more harmful to the cell, yet. Once I have a chance to complete a test and measure the heat energy of a spot weld and compare that to soldering, I'll have a better idea. But one clear fact remains for me - I can instantly melt blobs of solder with the spot welder. So riddle me this, if I use the spot welder to solder on the battery, is that bad? :shock:

My skepticism comes from the fact the nobody seems to have done these measurements, and empirically determined that soldering is indeed dangerous. Instead, its all guesswork. Yes, the datasheet for the cell says don't solder. It also says don't leave it in the car. Of course you can safely leave an 18650 in the car. But what's easier to say: don't leave it in an Arizona summer's car where the temperature can reach 180 in the car, or simply say, don't leave it in the car?
 
Anyone with half a brain who's done both can tell you it's not even CLOSE. Soldering puts WAY, WAY more heat into the battery.

But which way causes least damage is a discussion with mostly academic ramifications. For practical purposes, go ahead and solder your battery. A bigger problem than damaging the cell will be stiff wires and cold joints, because it's difficult to solder to the case in particular without REALLY letting the iron dwell there.

For some quantified data, my soldered-to Samsung 25R cells have about 2100-2200Mah capacity at 1C discharge 2 years after I made my battery. It's a ~2500Mah cell new. The pack only had like 30 cycles and a few hundred miles, and was stored for long periods of time at unknown SOC.

That's not really that bad, so I don't think the soldering was too damaging. I kept it to 5 seconds of contact or less though.
 
I think this is still worth discussing. Its easy to do a bad solder job, but with some discussion, maybe we can assemble some tips to make soldered packs a better method than it currently is most often used.

By that I mean...I remember back when I was struggling to get-by financially...very rough times. At that time, I might have been able to solder together a pack, but there was no way I could afford a cheap spot-welder, much less a good spot-welder (I now have an arduino pocket-welder, and also the parts to make a 6-pence spot-welder, but haven't had the time to experiment with them yet I also want a MOT).

There is a guy on youtube who took thousands of laptop cells he acquired cheaply (low current), and assembled them into a DIY Tesla wall power-backup.

https://www.youtube.com/channel/UC0pBauLp63yzf6sVdEOIUbA

I found him when I was looking for info and examples on fuse-wire. He also embraced the construction of having a thick copper series bar, and thin parallel connections, so using fuse-wire for parallel was intuitive.

The parallel connections carry the charging current, which can usually be around 3A-5A..the series connections collectively carry the full pack current divided by the number of series bars between P-sub-pack. Using doctor bass's method of one series bar in-between every two cells, a 6P string would have three series bars between every P-string. If you are using the 25R cell for high performance, the factory rates them at 20A, so 6P would be 120A for the pack (just throwing out round number to establish design principle). Three series bars for 120A is 40A each, so...the P-connections might see 5A during charging and even less when it gets to the balancing charge at the end, and the series connection is 40A each.

We have all seen the "ladder" style bus-plates made of nickel. There is one series connection per cell, and...the series and parallel connections are the same thickness and width. Nickel has a 20/100 conductivity rating compared to copper, which is poor. Why not copper? copper is soft so it adds no structural rigidity, plus it corrodes in humid and salty air (of course, if you use plastic cell-holders, pack rigidity is taken care of without needing stiff busses to help). Why not nickel-plated copper which has a much better 80/100 rating for conductivity over pure copper? It's more expensive, and pure nickel works OK for average performance.

Bottom line, in my humble opinion (IMHO)...nickel is used because it is corrosion-resistant, and it spot-welds very easily. I have seen videos of battery factories that have robots spot-welding dozens of batteries in the same amount of time it would take to one person to solder one battery pack, using the best methods available.

A fat soldering Iron tip means that it will not "cool off" during a 3 second soldering operation, and a high-watt soldering iron means that it will recover its highest possible tip temperature more rapidly. I have a 100W soldering iron with a fat tip. Once I used it to solder wires to connectors, I would never want to go back to a weaker soldering iron.

There is also a new idea that went un-noticed by most. An ES member mentioned that he put a skim-coat of solder on copper parallel strips, and it made the spot-welding onto the cell-ends easy. That is because copper is highly conductive (making spot-welding difficult), but...solder has some resistance.

That is my only issue with solder. It is actually not a good conductor, and it adds resistance to any connection. It DOES add a solid connection that seals out corrosion from reactive air (oxygen, humidity, and saltiness near oceans). Like nickel-plated copper, a thin skim-coat of solder might not be too bad. Pure copper with a skim-coat of solder "might" be better than pure nickel. Or at least not worse, with the added benefit of a serious cost-savings over buying a spot-welder. My 100W "fat tip" soldering iron is over ten years old, can most spot-welder owners say that?

When I researched flat copper ribbon (as opposed to common round cross-section wire), I found solar-PV panel connection "flat wire" that was made of copper, and coated in solder. The skim-coat of solder made it easy for field-techs to make connections without needing three hands, and the solder (tin / lead mix) is fairly corrosion-resistant when left outdoors in the weather (unlike the copper ribbon core).

I suspect that if I was in a situation where soldering onto the cell-ends was my only option, I would want to know what is the best method...even if I preferred spot-welding, if the situation allowed it.

I'm guessing:

Rough-up and clean the wire and cell-ends.

Use flux (I don't know what it is, or how it works, but people smarter than me insist it is a requirement for best results)

Pre-tin the wire (or flat ribbon) with solder

Pre-tin the cell-end with a thin skim-coat, and do this rapidly so the heat does not penetrate into the cell.

let everything cool off, then...

Press the wire/bus against the cell-end, and apply "just enough" heat for the solder on both parts to bond.

side note: If you also embrace the doctor bass method of having one series connection in-between each pair of P-group cells...connect the series bars to the parallel ribbon (whether nickel or copper), BEFORE you make the final attachment to the cell-ends.

HBPowerwall...individual cell fusing
https://youtu.be/xOVEnbbhbRY?t=924

HBPowerwall...soldering vs spot-welding (soldering can be OK)
https://www.youtube.com/watch?v=isa3CXD8-8c

Ebikeschool.com...Can you solder 18650 batteries? (he thinks, not a good idea)
https://www.youtube.com/watch?v=0sj5DvG-qPI
 
If someone wants to do it, it's easy enough to instrument a test setup for determining energy input into the cell with each method.

Take two identical single cells, and two identical strips of interconnect material, and set them up so that a good FLIR camera can clearly see the cells, preferably multiple cameras from various angles simultaneously.

Perform a spot weld, recording the FLIR data from before the weld to after cooldown back to pre-weld thermal conditions.

Perform a solder connection, doing the same recording.

If the energy input is the same, both FLIR data sets will show the same temperatures reached and the same cooldown time.

More likely they will show one of the temperatures as higher and taking longer to cool down.

Math performed on the FLIR data and cooldown time will tell you the actual energy input in each case.


If no FLIR is available, individual temperature sensors could be used.

A completely discharged cell could be cut open so the sensors can be placed inside but leave the endcap area intact so the energy still flows around the same way it would without the instrumentation.
 
Separately, one problem with soldering is that unless one has a machine to do it with, each joint is different than the others, heated less or more, different amounts of pressure, different amounts of solder, etc.

The spotwelding machines at least can help automate how much energy is used to create each weld, and some of them can also automate the amount of pressure.


So consistency is another difference between soldering vs spot welding.
 
I agree that is the biggest benefit...spot-welding with a machine that can adjust the timing and also energy per-pulse can be very consistent from one weld to the next.

Soldering between one builder and another has sooo many variables from one cell to the next...time of contact, heat of soldering iron, thickness of solder on contact point, quality of flux and solder, pressure that is applied, etc
 
What about combination of both:
1. dirt cheap
2. No 18650 heat damage
3. Tesla style fuses (0.1mm ni plated steel)
4. high current
1Hm9VU0.jpg
 
amberwolf said:
If someone wants to do it, it's easy enough to instrument a test setup for determining energy input into the cell with each method.

Take two identical single cells, and two identical strips of interconnect material, and set them up so that a good FLIR camera can clearly see the cells, preferably multiple cameras from various angles simultaneously.

Perform a spot weld, recording the FLIR data from before the weld to after cooldown back to pre-weld thermal conditions.

Perform a solder connection, doing the same recording.

If the energy input is the same, both FLIR data sets will show the same temperatures reached and the same cooldown time.

More likely they will show one of the temperatures as higher and taking longer to cool down.

Math performed on the FLIR data and cooldown time will tell you the actual energy input in each case.


If no FLIR is available, individual temperature sensors could be used.

A completely discharged cell could be cut open so the sensors can be placed inside but leave the endcap area intact so the energy still flows around the same way it would without the instrumentation.

I think you can measure the number of joules of heat without a FLIR and if the temperature increase of the cell is too small to accurately measure, you can also do the same to a quantity of water and measuring the temperature increase. So place a nickel strip in water, spot weld it and measure the temperature increase of that water. Then do the same with a soldering iron.
 
I would like to report additional findings. I did 2 experiments:

1) welding, soldering directly onto a steel washer
2) soldering onto a nickel strip that has been spot-welded onto the steel washer. Note, I'm soldering directly above the washer, so the nickel is sandwiched between the iron and the washer.

Methodology:
* washer was used as heat store
* infrared thermometer to measure heat increase from a baseline reading

Spot welding:
I spot welded directly onto the washer, then measured temperature increase from baseline.

A single tap of weld: 6F increase
double tap (1 sec delay between taps): 13F
triple tap (1 sec delay between taps): 20F

Soldering:
direct on washer, 3 sec of heat: 22F
on nickel strip, 3 sec of heat: 7F increase!

Directly soldering for 3 secs onto the washer is equivalent to about 3 successive welds, with my soldering settings and spot welder.

However, what is interesting is that soldering onto a nickel strip that sits on top of the washer results in a dramatic decrease in the temperature of the washer, such that a 3 sec solder time is roughly equal to a single weld. So by simply adding a nickel strip as a buffer, it appears that soldering can be equivalent to welding...
 
The fastest solder method i tried was the "kepler pack" using Goot Wick, solder wick. It made a workable pack and with some oractice it is doable. But spot welding with the Jakov and Aduino welders is so damn quick, easy, and cheap.

https://endless-sphere.com/forums/viewtopic.php?f=14&t=65719&hilit=Solder

file.php
 
A few relavent comments.
A) the +ve end cap of most 18650 cells is a separate component, removed from direct contact with the cell internal contents, so any heat effect is minimal. Heat effect concerns are mainly confined to the -ve base of the can.
B) a good soldered joint needs the can base to be "tinned" before the final solder joint is made. That requires a separate application of the iron, such that the final solderd joint is a two heat application process, not just one dab of the iron !
C). If solder is considered a poor conductor, why is it still used so extensively in all electrical/electronic components, and why are more electronic components not welded (lazer, ultrasonic, laser, etc) to their circuit boards ?
 
Cost & time in manufacturing processes--wavesoldering and / or paste-application / baking of PCB mounted components is very fast, many times faster than even a well-optimized robotic welding operation would be.

Interconnects with some types of ribbon cables *are* welded in some cases; it makes a thinner connection so you may find this in phones, tablets, etc.


Many of the applications where soldering is still used are those in which either a good mechanical connection is already present, negating the issues with it's tendency to crack under bending loads, or where resistance of a connection isn't important enough to require anything better than that (cost, again).
 
I wonder if my technique with tabs (makita uses 0.15 mm nickel plated-COPPER tabs) is okay.
Soldering the tab ears to the bussbar. I think the metal mass of the welded tabs on cell, the metal mass of the bussbar and the fact that weld is done at a distance from the battery post will help a lot in dissipating heat so as to not damage the cells.

Here's my try in multiple pictures : https://endless-sphere.com/forums/viewtopic.php?t=84791&p=1274120#p1274105
08.jpg

I used a 100W iron with flat tip
40%leaded soldering wire
Sony VTC4 cell
13.4 square-millimiter cross-sectional area pure copper bussbars (85% IACS).
 
Also, consider this

[youtube]YrqsAGr41uU[/youtube]

and this awesome engeneering hack to "spotweld" copper (check out 9:20) and also (check out 16:20) :
[youtube]ZgZztGSYeKY[/youtube]

Matador
 
I've seen the first vid where he cuts open the cell after soldering. A couple of observations:

* he puts the soldering iron on the terminals for a LONG time, far longer than I do, and frankly, far longer than he needed to.

* He said that the cell was quite warm after his soldering. My cell (with nickel strip on top, soldered on top of that) don't get warm to the touch at all, which means the temp increase of the cell was under 10 deg F.

* at some point, too much contact time of the iron against the terminal WILL pump too much heat into the cell and cause damage, which is what it seemed to do at the (-) terminal chemistry.

I have a messed up 30Q cell and a dremel, I'm going to try to replicate his work later. What I really want to do is cut open 2 cells - one with direct soldering and another with semi-direct (nickel strip on top) and compare the results. I just don't want to waste $5 sacrificing a cell :D
 
aethyr said:
I've seen the first vid where he cuts open the cell after soldering. A couple of observations:

* he puts the soldering iron on the terminals for a LONG time, far longer than I do, and frankly, far longer than he needed to.

* He said that the cell was quite warm after his soldering. My cell (with nickel strip on top, soldered on top of that) don't get warm to the touch at all, which means the temp increase of the cell was under 10 deg F.

* at some point, too much contact time of the iron against the terminal WILL pump too much heat into the cell and cause damage, which is what it seemed to do at the (-) terminal chemistry.

I have a messed up 30Q cell and a dremel, I'm going to try to replicate his work later. What I really want to do is cut open 2 cells - one with direct soldering and another with semi-direct (nickel strip on top) and compare the results. I just don't want to waste $5 sacrificing a cell :D


How about using a copperpipe cutter like this
Gut the cell content
Put your finger inside the emptied 18650 casing while soldering to see if it really gets hot ? (Or taping a temperature probe inside)
 
Matador said:
aethyr said:
I've seen the first vid where he cuts open the cell after soldering. A couple of observations:

* he puts the soldering iron on the terminals for a LONG time, far longer than I do, and frankly, far longer than he needed to.

* He said that the cell was quite warm after his soldering. My cell (with nickel strip on top, soldered on top of that) don't get warm to the touch at all, which means the temp increase of the cell was under 10 deg F.

* at some point, too much contact time of the iron against the terminal WILL pump too much heat into the cell and cause damage, which is what it seemed to do at the (-) terminal chemistry.

I have a messed up 30Q cell and a dremel, I'm going to try to replicate his work later. What I really want to do is cut open 2 cells - one with direct soldering and another with semi-direct (nickel strip on top) and compare the results. I just don't want to waste $5 sacrificing a cell :D


How about using a copperpipe cutter like this
Gut the cell content
Put your finger inside the emptied 18650 casing while soldering to see if it really gets hot ? (Or taping a temperature probe inside)
I do have a pipe cutter, maybe that will be easier/cleaner than the dremel.

Don't you think that using a metal washer that has the same mass/volume of the 18650 casing would provide similar results? The only issue is ensuring that the washer has similar specific heat characteristic of the casing. The casing material is just steel, right?

I really do want to see how the heat, if any, transfers from the casing into the chemistry as well. But I suppose if the casing itself doesn't get hot, then there's no way any heat could transfer into the chemistry.

What is really interesting to me is why adding a nickel strip as a buffer on top of the terminal seems to dramatically decrease the temperature of the terminal. Both steel and nickel have similar thermal capacity, but maybe the strip, with greater surface area exposed to air, dissipates the heat into the air, rather than transmitting it into the steel volume beneath it...
 
aethyr said:
Matador said:
aethyr said:
I've seen the first vid where he cuts open the cell after soldering. A couple of observations:

* he puts the soldering iron on the terminals for a LONG time, far longer than I do, and frankly, far longer than he needed to.

* He said that the cell was quite warm after his soldering. My cell (with nickel strip on top, soldered on top of that) don't get warm to the touch at all, which means the temp increase of the cell was under 10 deg F.

* at some point, too much contact time of the iron against the terminal WILL pump too much heat into the cell and cause damage, which is what it seemed to do at the (-) terminal chemistry.

I have a messed up 30Q cell and a dremel, I'm going to try to replicate his work later. What I really want to do is cut open 2 cells - one with direct soldering and another with semi-direct (nickel strip on top) and compare the results. I just don't want to waste $5 sacrificing a cell :D


How about using a copperpipe cutter like this
Gut the cell content
Put your finger inside the emptied 18650 casing while soldering to see if it really gets hot ? (Or taping a temperature probe inside)
I do have a pipe cutter, maybe that will be easier/cleaner than the dremel.

Don't you think that using a metal washer that has the same mass/volume of the 18650 casing would provide similar results? The only issue is ensuring that the washer has similar specific heat characteristic of the casing. The casing material is just steel, right?

I really do want to see how the heat, if any, transfers from the casing into the chemistry as well. But I suppose if the casing itself doesn't get hot, then there's no way any heat could transfer into the chemistry.

What is really interesting to me is why adding a nickel strip as a buffer on top of the terminal seems to dramatically decrease the temperature of the terminal. Both steel and nickel have similar thermal capacity, but maybe the strip, with greater surface area exposed to air, dissipates the heat into the air, rather than transmitting it into the steel volume beneath it...

Not sure about which metal the casing is...
I really think the only way to know for sure aout temperature effect is through doing the tests in real life. Measure temp in the inside of the cell, on the inner surface of the battery post to be soldered. It wont get hotter that this spot. Do the same test soldering with or without the nickel tab to dissipate heat. Also comparing with an electrical spotweld could help judging if there's a real measurable difference.

It seem to be logical that spotwelding will produce heat for less amounts of time. I've just never seen anybody quantitatively measure how much the difference is. Is it really that much significant ?
I mean if a soldering technique get's the inside of the cell post to 55°C while the spotweld gets it to 32°C (fictif exemple to illustrate my point), we are still in the manufacturer tolerance range for temperatures meaning no damage will be done in either case. Who knows....
But if you do it with a small iron tip like the guy in this vid, staying on it for very long.... than most probably, the inside temperature will get above 60°C which is considered too much...
Only way to know for sure is to test.
 
The guy in the vid.... I calculated the total time that he stayed ther with it's iron :

Negative : 25s + 30 s = 55 s (max recorded 166°C at on the external surface of post and liquified solder with the iron ON it)...

Positive : 14s + 3 s + 8 s = 26 sec (max recorded 133°C at on the external surface of post and liquified solder with the iron ON it)...

Center of cell casing 43°C
Cell casing near the negative 46°C.

So let's say he applied 25Watts to the negative post for 55s, that's 1365 W.s of energy...
My 100W iron would have to stay there for 14 seconds....
Of course this is very much crude ballpark maths....

With my 100W soldering iron, I can do it in less than 5 seconds if I prep well (sanding post and bussbar, applying flux on two surfaces, solder on iron tip first).
 
Holy hell :shock:

Almost 1 minute on the negative side! No wonder the chemistry got messed up. Yeah, I only need about 3 seconds to melt a blob on the terminal, then later after it has completely cooled, another 1-2 seconds to melt a pre-tinned wire onto the blob. And this is with a 60w temp controlled hakko iron.
 
Back
Top