There is a problem with rebuilding these battery packs however. Joining the cells with a thin nickel strip is impossible without a spot welder. You might say “What about soldering the pack together?” Well, heat from the soldering process would significantly decrease the life of the cells, if it didn’t kill them outright. “So just buy a spot welder!” Not going to happen. I can justify spending a few hundred dollars on components to build a spot welder, but not the $1200 or so it would take to buy one. “So break out a soldering iron and an oscilloscope and build one!” OK, I think that is doable but its going to be a complex process. This will be the most complex electronics project I have undertaken to date.

The design that I will describe here is derived mainly from these three sources:

• http://www.pittnerovi.com/jiri/hobby/electronics/welder/
• http://www.pocketmagic.net/2012/07/capacitor-discharge-microspot-welder-cutter/#120208
• http://www.turtlesarehere.com/html/cd_welder.html

The heart of the design is a bank of four 330,000 µF 16 V electrolytic capacitors connected in parallel. This should give an equivalent capacitance of 1.32 F and should hopefully lower the capacitors ESR to a usable level for this application. The capacitors shown here were sourced from eBay and are made by Nippon Chemi-Con. The data sheet for these capacitors does not list an ESR but lists a ripple current. Ripple current really is not useful to us, and I do not have an ESR meter, so we will have to evaluate the capacitors in the testing phase.

All of that stored energy has to be delivered to the weld joints with some sort of flexible cable, for which I will cut up an old welding lead I have lying around. I am currently leaning on TIG welder tungsten electrodes for the actual contact tips, but that is currently up in the air. Copper buss bars will connect the capacitors together and will connect to the cable and switching silicone.

Switching this much power is quite tricky and has led to failures in other people’s projects. Simple spot welder designs use a SCR to switch the load, which is certainly durable enough to handle the abuse, but latches in the conducting state until the voltage drops low. A design like this relies on adjusting the capacitor bank voltage to tune the weld. This is OK but does not provide much fine control. More advanced designs use MOSFETs to switch the current on and off in short pulses. This allows for tuning the welder more accurately and allows a “pre-heat” pulse to clean the weld site before the main welding pulse. This is the design I will attempt to use.

The MOSFETs I chose for this design are International Rectifier IRFB7437. This is a fairly beefy device with a low on resistance and a fairly low gate capacitance. I plan on using six of these devices in parallel to provide some safety factor.

I will drive these with a dedicated low side driver IC under microcontroller control. The MOSFETs can be seen in the picture to the left in the tube. I purchased extra in case I destroy parts in testing. Also in the photo are the protection diodes and TVSs I plan on using to protect the capacitor bank, and a 4×20 LCD display to provide a user interface for the microcontroller. I plan on driving this project with a 40 pin PIC micro.

To provide power for the project I plan on using a 1u server power supply. This simplifies things since the PSU provides almost all of the voltages this project requires. It will also allow me to incorporate a  standby feature that may be useful.

I know this is not much in the way of a formal design, but I am back-of-the-napkin designing this project. I did run a few simulations to reassure myself that the expected currents would probably not destroy the MOSFETs. Now all I need to do is build up and test the circuitry and write the control code. I’ll keep you posted.

Interspersed with this project, I have been trying to complete a long-term project for a local company located in an internet dead zone. the companies only direct service options are dial-up which is too slow, satellite with its ridiculous ping times and bandwidth caps, and a fiber optic line which would cost about $20,000 to install and would set them back $500 a month. To get around this I used a long distance wireless link to another site which has a reasonably priced and decently fast cable connection. The link distance is about two and a half miles. For line of site in southern Illinois, this requires elevation at both ends. The company in question happens to have a lightly used 100 foot free-standing tower at its site, and we were able to install 50 feet of Rohn 25 tower at the other end. I currently do not have pictures of the 50 foot end, but may post them later. I also do not have images from the latest climb on the 100 foot side, but I did find some of the photos from the last time I went up to change a router on a different link.  The tower appears to be a Rohn SSV which  in this configuration is rated for about 15 square feet of antenna area in a 70 mph wind zone. This is quite a bit more than what we actually use.

Each section of the antenna that you can see bolted together, is 20 feet long. The router and dish seen here sits at about 80 feet. The new dish sits at about 90 feet. The new link uses a pair of Ubiquiti NanoBridges in the 5 GHz frequency space. The bridges alignment is somewhat critical, with only a few degrees of beam width to play with. Luckily, due to the local geography, the two bridges are at almost the same altitude, which simplifies dish aiming somewhat. The link is capable of 40-50 Mb data rates and is quite stable. Initially the link was unstable, but lowering the output power of the radios seemed to improve stability greatly.

I have done quite a few odd jobs for this company in the past. The great thing about working with them is they only bring me interesting problems to solve, and they have reasonable budget expectations. Here is some of the work I have done in the past:

The image to the left is a network rack wall mounted in a mechanical room. Not terribly interesting, but It worked out well for this installation. On the right is a NEC DSX telephone system. This was not something that I had done before, but was not terribly complex. The two worst parts were terminating all of those cables and programming the system.

Well, that’s it for now. In the near future I will be documenting a battery pack spot welder build and a small software project that a friend and I are working on.