Build your own CD battery tab welder for about $100.00+-

[The Mighty Volt]

Ha ha Texas I am an SCR noob right now. I am like an ape that has discovered flint. I will make the leap to titanium and plutonium in due course, right now I am trying to determine the limits of elastic bands and wondering at the magic of nickel tabs!!!!

 

[The Mighty Volt]

Ok Gentlemen, the promised photos and video's

Thats a 0.375" MTO Nickel Tab, belted onto a Stainless-Steel LiFePo4 cell, with the weld probes being sat into some dimples I made with a soldering iron tip. By placing the tab on soft wood, and then pressing down with the iron, the dimples are made quite easily.

Findings/Observations: Bigger sparks created when welding nickel tabs {as opposed to the nickel roll} to the stainless steel frames. Dimples help enormously. Weld is excellent quality, the tab pulls off only where there is no weld. Note the jagged endings on the top right of the photo, where the tab shears at the weak points, i.e not at the welds. There is a weld visible which is not holding any tab- this is from a previous experiment. Voltage for this was around 13.8v, with Capacitance rated at 3.0F

 

[The Mighty Volt]

Video:

10 A123 on a strip being suspended after only a few tab welds. The welds hold the combined weight of the pack with ease.

Specific Observations: Lack of a spark when welding A123 to Nickel Roll.

General Observations: The Dewalt tabs, welded by resistance, peel off the positive terminal with far greater ease than the negative terminal. Inevitably, weld puncturs are found in the bigger, positive terminal, with weld studs being found in the negative terminal. This finding is largely replicated when using the SCR based welder with non-Dewalt tabs.

http://www.flickr.com/photos/43650746@N07/4846972518/

 

[The Mighty Volt]

This is my attempt at dimpling a strip of nickel.

Any comments would be welcome. It seems to work quite well, but that may be purely coincidental.

 

[The Mighty Volt]

The same dimpled strip is now placed on a row of A123. As you can see, the strip is on the negative terminal and is ready for welding.

I make many dimples quickly as I do not know where the dimples are or are not going to line up.

If you place a weld-probe on the tab, right over the vent hole, expect a large spark and probable blow-out.

Question: If you get a "blowout", basically a small hole in the tab, how is it best remedied? Lay some fresh tab down over it and try again, or rip it all off and start again?

I think the above was a non-dimpled affair. And it also worked quite well.

 

[texaspyro]

Lookin' good...

I would probably try for shallower, less pointy dimples. Maybe the flat end of a drill bit on a harder wood for less penetration.

Also (I don't know how pointy your probes are) try a flat probe tip. A lot of people seem to want needle sharp probes. These give tiny welds. You want wider weld spots that carry more current. The ones in your first photo look to be pretty good.

Unless it's pretty big, I would probably not mess with patching a blow out. Or maybe a quick drop of solder with a high-watt iron.

 

[texaspyro]

After playing with using my welder to anodize aluminum, I added a more general purpose power supply mode.

You set the desired voltage from the touch screen. You can use the foot switch for controlling the output, or you can do it from the touch screen. You can also set a time to keep the output on for (with millisecond resolution). The output fets get turned off and the charge fets get turned on when the capacitor voltage drops below the normal regulation threshold. If your load does not like the chopped output voltage, you can add a filter cap.

You need to be real careful about shorting the probes with the output on... remember this thing can provide 20,000 amp pulses and is not current limited. I haven't tested it under heavy loads to see how it behaves. I've only run it at a few amps. My charger transformer is around 400 watts. I suspect the power supply mode is good for 10-15 amps. The main limiting factor would be I*I*R heating in the charge fets.

 

[The Mighty Volt]

Wow Texas you are up there in the high-end specs. Oddly enough, after my earlier fruitless efforts with the SCR, I felt like packing in and buying the plans for a FET welder. Now with the SCR working so good, that is simply a distraction, albeit a luxurious and high-tech one.

One thing I wondered about was whether I should solder the balance wires to the terminals before or after the tabs are welded on.

How does a balance-wire differentiate between one cell and the next in a parallel row??

 

[texaspyro]

You only use one balance wire per parallel string of cells. All of the paralleled cells act as one big cell. They are all at the same voltage when paralleled. BTW, when paralleling cells first make sure they are all at the same voltage. Too much difference and a hugh amount of current can flow... just like a short circuit. A 10P24S pack would have 25 balance wires (one more than the number of cells)

Solder the balance wires to the tabs, not the cells. Solder it near the center of the string. I would put it on the tab at the gap between two cells so that you are not heating the cells. The tabs tend to wick heat away. You will probably need a fairly heavy duty iron so that you can quickly heat the material. Too small of an iron causes you to keep the hot iron on the connection for a long time. This gives time for the heat to spread to the cells. Ideally, you should be able to do the connection in under a second.

 

[The Mighty Volt]

Cheers Texas!

Yes I noticed last night a lot of what you are talking about. I tried to charge a 10p1s strip {resting voltage of 3.3v} up to 3.6 using a 20v 5A switching supply. Wow. Talk about slow. I could have left it there all night just to make up the extra 0.3 volts across the string.

Also....how many strips of tab are you putting down? Just the one?

I notice that the Dewalt tabs are a noticeable bit thicker than the MTO stuff, so I was wondering if it might not be in my best interests to double-up, and put down more than one layer. It could be overkill, but it wouldn't hurt either.

Thanks.

 

[texaspyro]

Lets see... 10P * 2.8 AH -> 28 AH (call it 30). To fully charge from 0 would take 6 hours at 5 amps. How are you charging 3.65V cells with a 20V power supply? Sounds like an explosion waiting to happen (Can I watch? ).

Dewalt/A123 uses 0.01" thick tabs. Yours are half that. I don't know what your current requirements are. Or what the current handling capacity of 5 mil nickel is (would depend upon the length). I calculated (by extrapolating from some 1/8" wide, 3 mil that I have) that the resistance of 5 mil, 3/8" nickel is around .03 ohms per foot.

I also measured the resistance of an A123 tab from the center of the cell to the end of the tab (around 0.75"). It was 0.00075 ohms... call it 1 milliohm per inch. That tab was around 0.35" wide, .010" thick.

Uh oh... this ain't lookin' good. I hope I'm wrong. I bet you hope even more... Put 100 amps through .03 ohms, P = I * I * R. You have 300 watts of power loss/heating in the nickel. Also .03 ohms * 100 amps gives a voltage drop of 3V. Leaves 0.6 V from a fully charged cell for your motor. Granted, not all the current flows through the whole strip. Feeding the current out from the center of the strip cuts losses around half. Still, a LOT of energy is going out as heat. Back to the drawing board...

 

[The Mighty Volt]

Well I am using a switching supply. I am not putting 20v into the batteries as such, right now it is drawing only 3.8v or so @5 amps. As the batteries charge, the power unit draws slightly more voltage. By the time the batteries are at 3.6v, the supply will be drawing 4.0v or so. Something like that. It is not a simple case that am belting 20v into each cell, that would be fun.

The unit is capable of a peak 20v, but it is not drawing that.

My controller {Crystalyte and Lyen editions} are in the 45-50A range.

That said, of course, 10p A123 is capable of far more than that.

I think I should double up on the tab!!!

 

[texaspyro]

What I might do is to solder nickel tab material to copper sheet. Space the tabs to match the battery spacing. Weld the nickel tabs to the batteries. This would minimize the amount of high resistance nickel in the path.

Another thing to do is cut the nickel sheet so it is wide enough to reach the adjacent row of parallel cells. Alternate the polarity of the rows of paralleled cells. That way a single piece of nickel parallels all the cells in a row and makes the series connection to the next row. This spreads the current out so all of it does not try to flow through one little section of tab material.

You would make the high-power wire connections to the pack across the lengths of the tab material (on the end rows of cells) that are not connected to a parallel row of cells. Strip the wire half the width of the row. Spread half the strands of wire in each direction so it makes a T shape. Solder the T to the tab material so that the wire is in the center of the row. I would solder the wire to the tab material before attaching the tabs to the end rows of cells.

 

[texaspyro]

I just found another use for the welder... coil gun driver. Not only can this puppy burn your house down, it can protect it from alien invaders... 720 watt seconds will toss a mean chunk 'o iron.

Now, will it crush cans...

 

[The Mighty Volt]

What I might do is to solder nickel tab material to copper sheet. Space the tabs to match the battery spacing. Weld the nickel tabs to the batteries. This would minimize the amount of high resistance nickel in the path.

Another thing to do is cut the nickel sheet so it is wide enough to reach the adjacent row of parallel cells. Alternate the polarity of the rows of paralleled cells. That way a single piece of nickel parallels all the cells in a row and makes the series connection to the next row. This spreads the current out so all of it does not try to flow through one little section of tab material.

You would make the high-power wire connections to the pack across the lengths of the tab material (on the end rows of cells) that are not connected to a parallel row of cells. Strip the wire half the width of the row. Spread half the strands of wire in each direction so it makes a T shape. Solder the T to the tab material so that the wire is in the center of the row. I would solder the wire to the tab material before attaching the tabs to the end rows of cells.

Hey great suggestions. Thanks.

One thing I think I might have to do is weld up 12 batteries in series and start to conduct some practical tests, at 20Amps for openers, just to see what happens, and make some practical observations.

I have a hub and a 36v 20Amp controller that should allow me to do this. It should give me a good idea where I will stand at higher voltages and higher amps.

 

[The Mighty Volt]

I have it figured out. I think.

A bi-metal strip.

Copper strip, of a similar thickness, is cut to the same length and width as the strip of nickel.

It is then punctured/drilled in line with the terminals.

Then lie the copper strip flush and flat, and underneath the nickel strip, and dimple the nickel into the holes in the copper. The dimples are made into the nickel so that the two strips now sit into each other, like lego, the protrusion from the nickel dimple sitting into the hole in the copper.

Put the strip back on the batteries and begin welding.

In theory, the welder should "blast" the nickel, aided by the dimpling, through the congruous holes in the copper, and pin the nickel to the battery in the usual fashion, without too much of the weld energy being dissipated into the copper.

The copper strip, which is the lower strip, is then held physically in place against the battery terminals by the simple fact that it is pinned there by the nickel, which is itself welded in.

I will try this tomorrow if I can find any copper strip.

 

[texaspyro]

I'm not sure of the long term reliability of this... Not sure of how well the copper will be held in place. The copper will oxidize over time. Also, it will probably shunt a lot of the weld current. If you were going to use copper, solder nickel tabs along the length of the copper, weld the tabs to the battery.

 

[texaspyro]

I cleaned up my power supply mode code and did some more testing. I've had the current up to around 20 amps (12V, 1/2 ohm load) which is really more than what my charger transformer is rated at.

As the load current increases, the output "ripple" increases since it is periodically chopping the output voltage to keep the caps charged to a fairly constant voltage (my welder does not let one charge the caps and turn on the output fets at the same time). The higher the current demand, the longer the output voltage gets choppped while the caps are being refreshed. 12V at 1/2 ohm should be 24 amps. I get an average 20 amp output... the difference is due to the time the outputs are off while the caps are being recharged.

Also, the higher the voltage setting, the longer it takes to refresh the cap. And somewhat unexpectedly, the lower the output voltage is set to, the hotter the charge fets get. They are on for shorter times, but the current flowing through them is quite a bit higher (I = (Vcharger - Vcap) / Rcharger). Heat loss is proportional to current squared and time on.

All in all, it makes a good anodizing and plating supply. Not very good for things that want a pure, regulated DC voltage. If ever I need that, I can tap off the capacitor bank directly. I just cant turn the voltage on and off in that mode. And the consequences of a shorted output is smoked charger fets.

 

[texaspyro]

Well, I added another operating mode to my welder... battery/cell analyzer. In this mode you can connect a battery across (or in place of) the capacitors. Then using either the built in 1 ohm discharge resistor or an external resistor placed across the probes, the cell is drained by the resistor. The cell voltage is monitored and recorded during the discharge and also sent out over the serial port to an external computer for analysis. Cell discharge stops at a preset time/voltage/mAh/mWh. Whatever happens first triggers the shurdown.

If the capacitors are left installed (it's a pain to disconnect them, especially if the unit is mounted in a box) the unit can measure the cell voltage first by placing the cell across the probes. The caps are then charged to that voltage. You then connect the cell across the caps. This prevents a large inrush current as the empty caps represent essentially a dead short.

The internal 1 ohm discharge resistor is good for single cell testing. With an appropriate external discharge resistor, the unit can handle huge discharge currents. At 100 amps, the 18 IRFP2907 mosfets are dissipating under 2 watts since their combined resistance is under 200 micro ohms!

Although it can't discharge the cells as a constant current, the discharge current is within a fairly limited range since the cell might be drained from 3.65 to 2.5V (at 1 ohm that is 3.65 to 2.5 amps). The design of the welder limits the battery voltage to under 30V (and 20V-24V with most car audio caps left installed).

 

[The Mighty Volt]

I am going to have to get me one of those things!!!!!

 

[texaspyro]

One problem with testing cells with the welder caps still installed is that you can't measure the cell internal resistance. The caps make the battery appear to have an iR of around 250 micro ohms.

I can disconnect the caps fairly easily by taking out the three capacitor negative terminal bolts and sticking a piece of paper between the cap terminals and the PC board. My welder is not mounted in a box so this is not a real problem, just a pesky inconvenience.

One small problem with running the welder without the caps installed is that it wants to charge them when you power on. Without the caps installed, the voltage regulation code goes a bit batty trying to maintain a constant voltage on a non-existent capacitor bank. I do have a diagnostic mode so that if you power it up or reset it while the footswitch is pressed, it goes directly to a diagnostic menu without enabling the charger. The CBA mode is accessed from the diagnostic menu.

 

[texaspyro]

I tested a couple of A123 cells on the welder in CBA mode. I disconnected the caps and ran it with the internal 1 ohm discharge resistor. I charged up an A123 cell last week and ran it down to LVC on the welder. It showed 2313 mah. Not too bad for a 2300 mah cell. The temp sensor on the 1 ohm/60 watt discharge resistor showed a peak of 190 degrees F without any forced air cooling.

One issue that it does have is the voltage drop in the leads to the cell connection. I was using a couple of clip leads with 80 millohms resistance... gives a 0.2V drop. Makes automatically calculating the cell internal resistance a bit iffy. It can be compensated for in software or I could kludge in a 4-wire Kelvin connection to the cell.

 

[The Mighty Volt]

I thought they were 2600mAh rated, no?

 

[texaspyro]

I thought they were 2600mAh rated, no?

Nope... definitely 2300 maH. I have the official A123 spec sheet in my grubby little hands as I am typing this with my nose.

 

[texaspyro]

Here are a few screen shots of the welder in CBA mode discharging a brand spankin' new A123 cell at 4C. This was done with a 0.250 ohm, 100 watt resistor across the welder probes. The load resistor was switched by the firing FETs.

I am compensating for the 80 milliohm resistance of the cheap clip leads to the cell in software. The proper way to do that is to make a 4 wire connection to the cell... that would require breaking out the soldering iron... to much work for now.

At this discharge rate it yielded 2080 maH. The screen images were from a laptop screen that echos what is on the LCD screen (closeups of the LCD screen have too much reflection of the camera).


View attachment 1




The cell has been disconnected here. The "Count" field is some debug info (how many A/D samples were averaged over 5 seconds) INT and EXT select the internal (1 ohm) or external discharge resistor.