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

[The Mighty Volt]

I got me some of the fattest, softest, multi-strand copper you have ever seen.

Probably not... I had a friend that made some jumper cables for his truck out of military surplus starter cable for tanks. It was over an inch in diameter, weighed a ton, was more flexible than wet spaghetti. The conductors looked like Litz wire. The insulation looked like some sort of silicone based material. It felt... unnatural...

And again you have to be careful about what you are actually getting. Unless you can actually measure the resistance, you never know what you have. The slimy tricks that a lot of the car audio industry and shady Chinese suppliers have been pulling has made me VERY skeptical about such things... " Don't trust... Verify!"

The 4 gauge Scosche EFX Flux cable that I am using right now is good stuff. It has the proper resistance. You can coil it into a two inch diameter coil. Part of the flexibility seems to come from the fact that the wires are not gripped by the insulation. You can feel the insulation sliding over the wire bundle.

Scosche also makes 0 gauge cable that I may try. That stuff is over 50% fatter than the 4 ga. It would have around 1/3 the resistance of the 4 ga cable at the expense of twice the weight and considerably less flexibility. It would cut the total external resistance of the welder to around a milliohm and boost the available weld current back up to around 10,000 amps.

The Excelene welder cables that I have tested are also very good. Not as flexible as the Scosche stuff, but is real copper. I've only seen it in black. I like my cables color coded for polarity.

I know this copper I have is good because...and get this....nobody actually paid for it or bought it. It was pulled out of a skip. I believe it was 200 Amp rated jump lead copper.

I have taken Jeremy Harris advice and I have secured a few 100 Ohm resistors. First thing I will do with the RockFord Caps is measure their true capacitance.

If they are not up to scratch, good old Mr Ebay gets asked to do a return.

I think they will be though. These 0.5F numbers were actually much more expensive than the Caps which were supposedly 4 times their rating.

Here is the copper....perhaps not the beastly stuff you have but I cannot see it being overly resistant.

 

[amberwolf]

I don't know where it is, but I have some "nice" looking wire I pulled out of some no-name server UPS systems, and some jumper cables from a freecycle box, which looks like some sort of copper-color-dyed aluminum in the first case, and copper-color-dyed steel in the latter; I can actually stick a magnet on the jumper cable wire! (the latter also used steel for the battery clamps; cheap steel I can bend by hand).

I guess with copper prices being what they have been, both of those metals plus the dye process and any other complications to manufacturing from the different-than-usual materials could still be cheaper than using copper.

 

[The Mighty Volt]

I don't know where it is, but I have some "nice" looking wire I pulled out of some no-name server UPS systems, and some jumper cables from a freecycle box, which looks like some sort of copper-color-dyed aluminum in the first case, and copper-color-dyed steel in the latter; I can actually stick a magnet on the jumper cable wire! (the latter also used steel for the battery clamps; cheap steel I can bend by hand).

I guess with copper prices being what they have been, both of those metals plus the dye process and any other complications to manufacturing from the different-than-usual materials could still be cheaper than using copper.

The effort some people will go to makes the mind boggle. I mean...dyeing steel. :x

 

[texaspyro]

A friend of mine owns an electronics parts store. He had been buying (Chinese) lead acid batteries from the same supplier for years. Last year he got in a box of batteries that were quite noticeably lighter than what he had been getting. He weighed a new battery and an old one. The new one weighed over a third less than the old ones.

When he asked his supplier, they said "yeah, we have four cargo containers of bogus batteries going back to China." The price of lead had gone up, and the Chinese battery manufacturer had put smaller battery guts in the bigger packages.

Then there were those reels of magnetic 100% pure oxygen free copper speaker wire... The Chinese gotta have some AWESOME physicists to make magnetic copper and low density lead. We are falling so far behind in the field of bogo-physics that we may never catch up...

 

[The Mighty Volt]

A friend of mine owns an electronics parts store. He had been buying (Chinese) lead acid batteries from the same supplier for years. Last year he got in a box of batteries that were quite noticeably lighter than what he had been getting. He weighed a new battery and an old one. The new one weighed over a third less than the old ones.

When he asked his supplier, they said "yeah, we have four cargo containers of bogus batteries going back to China." The price of lead had gone up, and the Chinese battery manufacturer had put smaller battery guts in the bigger packages.

Then there were those reels of magnetic 100% pure oxygen free copper speaker wire... The Chinese gotta have some AWESOME physicists to make magnetic copper and low density lead. We are falling so far behind in the field of bogo-physics that we may never catch up...

@ "bogo-physics".

Some of these dudes think Physics is a form of carbonated drink.

 

[amberwolf]

F I Z Z I X. Drink of SuperStars. I keep tellin' ya, it'd have awesome profit potential. :lol:

 

[texaspyro]

Well, I have now tested a half dozen of the Monster Cable capacitors and am rather impressed. These units have no model number on the package or the capacitor. The instruction sheet calls them "Intellicap". The cap cans have a silver background with an image that looks like a circuit board. Another Monster cap with a white background performed similarly.

All measured over 1 farad with an ESR under 1 milliohms. Four of them had ESR's around 0.7 milliohms. Rated at 20 volts... working or surge unknown.

 

[texaspyro]

One thing to be aware of in CD welders is the inductive kickback you get from the welder leads when the pulse shuts off. I see up to a 5V pulse around 20 microseconds long getting onto the capacitor (added to it's voltage). This pulse can cause you to exceed the capacitor voltage rating (albeit for a very short time).

This is not a problem if the cap voltage is 5V below it's surge rating when the pulse shuts off... which it will be with an SCR welder. It is most likely to be a potential problem if you are using a FET welder to fire short, low energy pulses with the cap charged above 4-5V less than it's surge rating.

 

[texaspyro]

The circuit boards for my FET welder came in today... freakin' awesome. The internal resistance is about half what I was expecting... and that is without adding additional bus bars (which is trivial to do since all the high power busses are straight lines).

With three of the Monster caps in parallel and 0 gauge wire, the resistance would be right at 1 milliohm... 20,000 amps!

I calculate and plot the power going into the weld... pushin' half a megawatt peak power!

 

[The Mighty Volt]

Congrats. Make sure to post a vid when it is up and running !

 

[texaspyro]

It's up and running... will be a while before I can shoot some pics.

I added a bunch of new fault detection/protection circuitry to the PCB version. There were a couple of minor tweaks I had to do to that section ('doh). Basically there is no single point point of failure that can lead to a destructive fault... nothing can go wrong... wo grong... go wgong...

The new layout and FETs are much better than the hand wired prototype. I'm not seeing any of the inductive spikes, etc on the capacitor. It's spookily clean.

Did you ever get your caps?

 

[The Mighty Volt]

Hi Texas, congrats on your new setup.

Can anyone here tell me the specific difference {and the importance of same} between {a} Inverted DC Welding and {b] AC Pulse Welding?

Thanks.

 

[texaspyro]

Check out http://www.tjsnow.com/supplies/miyachi/DC_invert.htm for a quick review.

Basically AC welders deliver two pulses per cycle of the AC line current. Very little control of the weld current. You can get lots of cooling/heating changes as the weld current goes through its sine wave shape.

Inverter welders pulse at kilohertz rates. No time for heat changes during the weld pulse, plus much finer weld pulse control. You can also do voltage/current feedback during the weld pulse. Makes for very precisely controlled and repeatable welds.

 

[texaspyro]

In the PCB version of my FET welder, I added a bunch of fault monitoring and protection stuff. One thing that I added was support for temperature sensor chips.

I built up a board with three Rockford 1F caps on it and put a temp sensor on the 1 ohm/25 watt resistor that is used to drain the caps (either when changing the charge voltage or when shutting the machine down).

I got a little surprise when I shut down the machine with the caps charged to 18V... the temp sensor tripped (was set to 120F) as the resistor got rather warm dissipating the 600 watt-seconds of energy in the cap. Looks like some heat sinking is in order...

 

[texaspyro]

Here is a pic of my FET welder. The three Rockford RFC1 caps are visible behind the board. Total capacitance is 2.97 farads. ESR is around 0.3 milliohms. They can hold up to 600 watt-seconds of juicy juice.

The PCB board is 3 oz copper on each side. You can easily add external bus bars, but the PCB busses have less than 0.3 milliohms of resistance cap to cap... not much to be gained there. The caps bolt directly to the PCB... minimizes interconnect losses.

There are 18 IRFP2907 FETs in the power switch. Net ON resistance is around 0.17 milliohms. They do not get even slightly above room temp... no heatsinks or fans needed. I bought 100 FETs and matched them in 5 groups of 18 for gate threshold voltage. Matching for gate threshold voltage assures that all the paralled FETs turn on at the same time and minimizes stress on the FETs. All the FETs had identical full-on resistances of 3.100 +/- 0.100 milliohms, so not much to be gained by matching for full-on resistance.

The charger FETs are on heatsinks visible behind the LCD. They are two pairs of two paralled IRF4905's p-channel fets connected in series. The series connection is backup insurance against a shorted FET causing the caps to overcharge if your charger supply exceeds the capacitor rating.

Without the heatsinks, they sweat big time... you can't hold your finger on them. The caps basically look like a short circuit during a charge cycle and will suck all the current that your charger supply can provide. With the heatinks, you cannot feel any temperature rise at all.

Max charger voltage is 30V. The unit can run off of 9-30V DC (including car batteries). My charger is a 30V center tapped, 400 watt transformer/400 amp dual schottky diode/60,000 uF cap. It peaks out at around 19.5 VDC... no way to overvolt a 20V cap.

On the backside of the board is the 1 ohm discharge resistor... it is going to need some help getting rid of the heat if you drain those caps into it a few times in a row. It only gets used when you shutdown the unit or reduce the cap charge voltage.

The image on the 160x80 LCD touchscreen is the weld current of a dual pulse weld into a 1 milliohm current shunt. It was set to 20 Joules in the first pulse and 80 in the second. Caps charged to 12V. Even at these low settings/high load resistance, this thing can dump 100 joules in a couple of milliseconds.

There are two serial ports on the microcontroller (a MegaDonkey from mega-donkey.com (n.b. I designed it)). One port echos the LCD screen to a laptop. Anything you can do on the LCD touchscreen, you can do on the laptop. The other port is a text based interface for remote control of the welder by your own progams. You can download new firmware via either port. The CPU is an ATMEL ATMEGA-2561. There is around 7500 lines of C code in the welder app. It is backed up by another 20,000+ lines of MegaDonkey library code)

 

[rkosiorek]

really a nice neat board. is the LCD just a standard serial or parallel I/O type?

any idea if you want to sell either bare boards or kits? how much? how much for a pre-programmed MCU?

i have some cornell-dublier 1F caps that should fit that board nicely.

but that is a really KEWL!!! setup. GOOD JOB! now entering seriously envious mode......

rick

 

[The Mighty Volt]

Check out http://www.tjsnow.com/supplies/miyachi/DC_invert.htm for a quick review.

Basically AC welders deliver two pulses per cycle of the AC line current. Very little control of the weld current. You can get lots of cooling/heating changes as the weld current goes through its sine wave shape.

Inverter welders pulse at kilohertz rates. No time for heat changes during the weld pulse, plus much finer weld pulse control. You can also do voltage/current feedback during the weld pulse. Makes for very precisely controlled and repeatable welds.

Hey Texas, thanks for that.

This would explain the differences in price!!!!

Cheers.

 

[texaspyro]

is the LCD just a standard serial or parallel I/O type?
any idea if you want to sell either bare boards or kits? how much? how much for a pre-programmed MCU?
i have some cornell-dublier 1F caps that should fit that board nicely.
rick

The touchscreen lcd/microcontroller is a (semi-)commercial item. See http://www.mega-donkey.com Runs $159. I chose that device since, well, uhhh... I designed it and am intimately familiar with its mega-awesomeness. The board has a built in bootloader and can be programmed via a serial port.

We have not decided on what/how to sell it yet. In small quantities the parts add up to around $600 :
CPU/LCD $160, caps $140, FETS $80, misc parts $50, charger supply $70, PCB $100. Depending upon the quality of your closet 'o crap and scroungeability you might be able to shave a couple hundred bucks off that. For instance with a little lucky Ebay action, I have purchased sets of caps for under $60. You could also back off on the number of caps or use cheaper fets (but risk offending the welder gods and burning forever in cheapskate hell)

One could conceivably build a LCD-less CPU board with an ATMEGA-2561 CPU and run it off a laptop for quite a bit less... but definitely not worth the hassle for a one shot deal. The code does fit in the space for a Mega-128 chip, but it is getting to be a tight fit without doing some modification. I suspect one could cut the parts cost of a more commercialzed unit by half.


As far as the Cornell caps are concerned, what is the spacing of the terminals? Audio caps are on 1.25" centers. If they are a different spacing you would need to cable them in or drill a new mounting hole in one of the cap busses. Also, what is their ESR? I have checked a LOT of caps, audio and otherwise, and it is hard to beat the Rockford/Monster/Scosche caps.

 

[rkosiorek]

they are computer grade. specs say ESR is 1mR of better. i have never measured them. the hole spacing is 1.25" using 1/4-20 screws. original price was megabucks. i have charged and discharged these several times to check their capacity and they do measure at just over 1F. original spec was -10/+20% and they live up to that. i have no reason to doubt that the ESR will also be as claimed.



i've been collecting parts since 1967. i don't have a car and a 2 car garage is too small. i've got about 50 IRFB2907's left over from something. as for the charging supply i've got a 24VCT 15A transformers that i could build into a decent 0-15V@30A variable supply. so i think the major expense would be purchasing the Donkey, LCD and FET board.

i've built a couple of the SCR based simple welders in the past using the same capacitors. traded those for lord knows what at a hamfest or swapmeet. those worked well enough for the NiCad an NiMh packs i was doing at the time. but i have always thought i wanted something better and more versatile.

now i still need some convincing. i'd like to see some pictures of your unit in action. or pictures of the resulting welds.

rick

 

[texaspyro]

Looks like the caps should do nicely...

No need for a variable supply. The welder does that for you. It can adjust the cap voltage from 0 to whatever the charger supply puts out. Regulates the voltage to around 30 mV or so.

A 24 VCT transformer should do over 16 V peak with a full-wave (non-bridge) Schottky rectifier (I used a CPT30060, 300A 60V module). You could charge the caps to 16V which would be around 400 watt-secs. There are some nice brand new 250-400 watt toroids on Ebay for under $40. Made by http://www.antekinc.com. They can supply just about anything voltage/current you want for very reasonable prices.

I've run the thing off a small 15V wall wart... it just takes a while to charge the caps. Sucks the poor thing down to around 2V starting out. It's amazing that the charger fets can even be turned on at that voltage. I do have to run the MegaDonkey off it's own supply when I do that. Also works great off a 12V/8Ah SLA battery.

 

[rkosiorek]

i have used some Antek toroids in some tube amps. they make some nice output transformers. but i could just use what comes to hand.



those are 12 inch calipers open to 5.25" the transformer is 24VCT @10A. the rectifiers are 1N4589 rated 150A continuous and 2100A surge. of course that heatsink is just pure overkill. but just add a power cord and there is a decent unregulated power supply.

rather than a resistor, did you consider using a 12V 25W brake light as the limiting resistor for charging the caps?

rick

 

[rkosiorek]

of course i would only be using two of the rectifiers. but i do have some spares.

rick

 

[texaspyro]

rather than a resistor, did you consider using a 12V 25W brake light as the limiting resistor for charging the caps?

The resistor is used to discharge the caps... nothing limits the charge current except wiring/charge fet/transformer resistances.

Ideally the resistor is actually a precision resistor. I measure it's value (plus the discharge fet and PCB trace resistances) down to the milliohm. The welder has a capacitance measurement function. By timing the capacitor discharge through that known resistance, I can calculate the capacitance down to the millifarad level... essential in the whacky world of fraudulent audio crapassators.

The main operating mode of the welder is to specify the weld pulses by energy. You need to know the welder capacitance to do that. During a weld pulse, the capacitor voltage is checked every 100 microseconds. Energy = 0.5 * C * V*V Since I know C and V, I calculate the energy that is on the cap in real time (down to millijoule resolution). A pulse ends whenever the target energy level on the cap is reached (or a specified time interval has passed)

Specifying the weld parameters in energy gets around a lot of the weld variations due to lead resistance, probe changes, voltage, etc differences. Once you know what energy parameters work for a particular weld, other people can use that data to repeat your welds despite substantial differences in their machines.

Using simple time based weld pulses is rather crude and not repeatable between different welders or even on the same welder. A subtle change in your electrodes can greatly affect the weld current, etc.


I used Schottky rectifiers in the charger since they have half the voltage drop of silicon diodes... mine does not get even slightly warm and I get another 0.5 volts out of the charger. Not using a bridge rectifier saves you another diode drop and cuts the losses in half again. Suitable devices can be had on Ebay for around 10 bucks.

Here is my charger supply... no fans, no heatsinks. Just a transformer, diode module, and filter cap. Diode is bolted directly to the filter cap.

 

[texaspyro]

You mentioned that you would like to see some welds. Here is a photo of a few:

Everything is sitting on two pieces of 18 gauge (0.050") sheet steel welded along the seam. Can't break the pieces apart by hand. It has an oiled surface and no prep was done before the welds. The welds are "series" welds with one electrode on the same side each piece.

That nickel on the right is not going anywhere... I broke a few razor blades and bent a screwdriver tip trying to get it off.

Two new thirty cent pieces to the rear. Can't separate the nickel and the quarter by hand with pliers and vice grips. Three dimes held together with 0.003" nickel.

Five mil copper with 4 mil nickel strips.

In the foreground is 0.020" inconel wire loop welded to 10 mil stainless.

That little nub on the lower left of the stainless is 0.020" tungsten. Wire broke before the weld did... but tungsten is rather brittle.

That wispy thing coming out of the center of the "U" in the 10 mil stainless is 0.005" titainium wire welded to the 18 gauge steel.

What doesn't show up is the 60 micron (thinner than a non-red human hair... if you know what I mean ) gold plated tungsten wire welded to the stainless. It is so thin that a 9 volt radio battery will instantly vaporize the stuff...

Try that with an SCR welder...

 

[belgianbeer]

Hi texaspyro nice CD welder board you have there.. can you share the schematics with us? Especially how did you handle the mosfet gate drive? I guess that is the most important part of any good CD welder design.

I'm also in the process of building my CD welder and I already bought an overstock of 400pce 47000µF/40V EPCOS (B41456-B7479-M) capacitors with ESR of 5.9mohm as you can see in the picture below. I plan to connect 20 of them together so I have 0,94F/40V with 752 Ws of energy and cap ESR of 0,295 mohm.
I think that is sufficient for good battery tab welding.