Ultracapacitor Driven, Dual Pulse Spot Welder, a Novel Iteration by wubwub89

Last crawled date: 3 years ago

UPDATE (consolidated): Hi everyone. I have yet to find time for a video, mostly due to life pressures, etc, I have zero free time for anything that isn't paying. Please, as I stated in multiple places of this post, you must pay close attention to your primary resistances! Very small bits of resistance accumulating will negatively affect the operation of your welder. I have had more people write me saying the design doesn't work, and they include a picture of a total rats nest using 12 gauge wire. It will never, aside from elevated voltages, work right without really low sub-ohm resistances across the machine's primary circuits. If you have any questions, please read through the guide again before you write me. If you send me a message asking about something that is clearly specified in the guide, I may not be polite. If you have questions about something that is NOT specified in the guide, by all means feel free to write me or comment.
I do not plan to change this design at this time, however, in the near future I MIGHT switch the unit over to a toroid transformer on a 2x15v secondary in parallel and down-regulated for a big boost in charging current. At this time though, I am happy with the conversion efficiency of the large switched supply (it's really just a lightly modded laptop charger), and I am happy with the dwell time between firing, usually around 1-1.5 seconds depending on voltage used.
I have been using the welder consistently in building my own cell packs of all shapes and sizes, along with some other weird spot welding, like ground strapping on projects with metal cases. Everything is working great and consistently without any failures (exception being one full blow-out on an aluminum case cell, my fault for even trying a resistive weld on that!) at about 7500+ individual welds (I don't remember what the actual counter reads exactly).
I have had a few cell packs with mild issues, but not due to the welder. Apparently I picked up a batch of "pure nickel" tabs that were just nickel alloy coated really high iron content steel, and tons of them corroded to hell. I checked, and sure enough, tons of sparks coming from that batch of the grinder wheel. Some of the China direct sellers on various sites will sub in crap parts because they either don't know because they're just a salesman pushing widgets, or don't care. It's probably both, but regardless of all that make sure you check your source for nickel tabs.
UPDATE 2/2/18: I am having some hardware trouble with my camera recording in HD that will hopefully be solved in the next 30 days. I have gotten a few messages about my design not working, well, then you built it incorrectly or your components are out of spec! It's critical to minimize the primary circuit resistance properly! Please note, this is a critical engineering spec design with very little room for error or changes. Just using crimped lugs over soldered lugs may create too much resistance, circumstantially, and you MUST BE USING 4/0 GAUGE PRIMARIES OR LARGER, PERIOD [Edit: To clarify, I am saying this just to make sure anyone who builds the unit with a long primary circuit run doesn't encounter resistance issues]. I have NO trouble creating strong, deep coalescence through multiple boundary layers at low duty. IF you are getting blowouts, turn your primary pulse time and/or voltage down! Once again, this design does not allow for much error, including capacitor/bank quality from previous use and abuse.
UPDATE: After some more testing and production applications, it's working great for certain. However, I am struggling with some certain metals due to their resistance, notably copper and aluminum. I actually blew a hole through the casing of an aluminum LFP battery! (Oops, tossed it in a bag, then an empty paint can, and in the backyard to be safe, it's not reacting, if it was Li-Ion or it would have started thermal runaway and out-gassing potentially). That being said, it's demonstrably safe to use on steel casings as long as you don't lose the reigns on the pulse width and voltage. After investigating later on (when ambient temps were below 40 deg F), approximately a 0.5mm hole was found in the casing.
Hey everyone!
Firstly, This post is a work in progress! While the device is constructed, thoroughly tested, and working great, I will update and expand upon this post over the next month when I have time. Please like and collect to follow! I will be providing a very clear and advanced schematic for the electronics in this configuration. How you choose to build your enclosure and power supplies is totally up to you, but I will show you how to build and operate this type of unit safely. Again, this post will be reformatted, edited, and expanded over the next month. There will also be a detailed YouTube video regarding this methodology. Thank you!
This unit was designed and built by myself, for the purpose of spot welding battery tabs, spot welding small metallic projects, and generally, for providing very large amounts of current in a short period of time when needed.
This product's design is protected under Creative Commons Attribution, please give credit where credit is due!
DISCLAIMER and WARNING
This device utilizes both line AC voltage and extremely dense energy storage through the capacitor bank. AC 85v+ (probably less under the right conditions) with a big current supply is LETHAL when a human being acts as a ground path for the current available. Do not EVER, EVER touch, dismantle, assemble, test, look funny at it, or otherwise work with line voltage unless you know what you are doing! Seriously! The capacitor bank presents a very real, but different threat. The capacitor bank builds a reasonably safe DC voltage current source, which DC voltages below 24 volts can be considered safe, there are still some serious consequences to mishandling. DEAD SHORTS ACROSS HIGH-DENSITY ELDC UNITS WILL CAUSE A FIRE, EXPLOSION, SEVERE PERSONAL INJURY AND PROPERTY LOSS! Seriously, you do not want to dead short one of these units! ELDC shorts will vaporize box wrenches, destroy low impedance circuits, raise your car payment, and make your wife leave you! All kidding aside, please be careful when handing high-current devices, make sure you understand how to prevent traces from forming where they shouldn't, and avoid improperly constructed conductive parts, including wiring, IC's, anything that has a low electrical resistance is potentially an exploding fireball with shrapnel when working with super-high current devices.
So, now that my readers know they can accidentally kill themselves or lose everything in the process of building this device, let's get on with it!
ABSTRACT
The purpose of this article is to define the systems and subsystems of the aforementioned device and their working principles in detail, along with providing a clear, informative, and confusion-free schematic of the system and its subsystems. We will also be discussing material science involving spot-welding with this type of device. This document was written by an engineer, from the perspective of a DIY need to weld battery tabs and small metallic objects that are sensitive to heat. The author is not a professional welder. The writer was simply frustrated with the poor write-ups found on the web about these devices and did not approve of their safety or design effectiveness.
HOW IT WORKS
In principle, this device will adjoin two metallic objects together without the introduction of material (such as a welding rod etc), and generally, without the use of flux. In the case of welding battery tabs, the current can be passed through the tab, into the substrate, back out through the substrate, and back into the tab, in a closed loop, or it can be passed through the tab, and into the substrate. Of course, the polarity can be reversed by changing the probe location and orientation. The results of either method in either polarity configuration will vary greatly based on the resistance of the tab used, the resistance of the target substrate, in addition to overall performance attributable to the overall resistance of the devices current path, along with the ESR of the capacitor. Tab material types and construction will greatly affect the welding procedure, as the current path will differ between different types and methods employed. It's important to familiarize yourself with the type of materials you will be welding together, generally, nickel and steel are my primary targets. Other metals will require more voltage and a longer pulse time in many cases, but this requires further investigation on my part.
WHAT'S IT MADE OF?
This device has a 5.1A digitally controlled and selectable constant voltage/current supply that controls the UC(Ultracapacitor) bank primary voltage, a microcontroller that controls the ON/OFF pulse time for both a small pulse to 'preheat' the materials and build an (ideally) gapless current path, then a larger pulse to complete the weld. Current paths with gaps will cause material breakdown and substrate erosion, which usually occurs when there is a loud pop, and very small pieces of molten metal go flying. In parallel with the primary welding circuit, there is a switchable 5-ohm resistor array (two 100w, 10-ohm resistors in parallel) with thermal insulation active cooling. The entire device is powered by a salvaged and reconfigured 19.5v+ laptop power supply, with each component individually voltage regulated and monitored as follows: the microcontroller and MOSFET array controller is powered by a dedicated L7812cv (12v+) regulator circuit, and the case cooling fan is regulated down to 9.5v+ with a variable CV regulator. Also, on the front-right of the device, you may notice a LED cluster, along with a LED light bar labeled 'Problem', both of these operate in parallel with the primary circuit, but they vary in how they are hooked up, we'll go over the data acquisition later on.
All of the wires in this unit is stranded 20, 18, 16, 12, 8, 6, or 4/0 AWG wire, with PVC or Silicone insulation. All of the high-current lugs are soldered instead of crimped. This is a personal preference, and since the device is to regarded as immobile and we're not expecting to experience multiple G's of vibrations, also, our high current flow will be well below the time needed to get the joints flowing again. This comes from professional knowledge and experience, where as long as the thermal loss from resistance remains lower than the thermal dissipation, to keep it simple, this device will not heat up terminals fast or high enough. Also, we don't want to worry about moisture oxidizing our copper wire, and potentially causing a complete or partial break down the line, as a free-swinging wire ready to carry up to 1400 amps is a terrible situation (See the section about exploding fireballs). We want to build a device that will, optimistically, have properly designed subsystem failure points that will prevent a catastrophic failure. Just about everything besides the primary circuit is fused, just in case of a short. The dissipation of the UC bank through a small gauge wire will almost instantly start a fire. ( I don't take this type of energy storage lightly, and neither should you!)
In my application, the Maxwell 58F UC(lovingly nicknamed, FireBoom9000) was chosen due to its low ESR (approximately 0.22mΩ), consisting of six 2.7v 500F supercapacitors placed in series on a protection board (which prevents overcharging, promotes voltage potential balance, and hopefully, reverse voltage spikes) and allows for a peak storage voltage of 16.2v, which we will never reach due to both energy requirements, and the 14v limitation of the TVS diode on the MOSFET array (more on that later). Also, the Maxwell 58F unit comes in this nicely packaged blue/gray ABS case for safety. Other types of storage could have been chosen, but I preferred this unit for its size and electrical characteristics. It's entirely unnecessary to use 6 3000F BoostCaps or something equally ridiculous in size. At about 12v, with a 25ms primary pulse, and a 15% short pulse (15% of 25ms, variable on the controller), we get a voltage drop of around 0.43v per firing. With the 5.1A supply set to 4A @ 12v, it takes about 1.5-2 seconds to get back to 12v in the UC bank. I finished my first 18650 Li-Ion 4s1p pack in about 20 minutes, WITH a BMS, without any bad welds.
Now, let's talk about the pulse control unit. When I started this endeavor, I wanted to build my own, write my own board files in Eagle CAD, send them out to one of the manufacturers, wait 30-60 days for my professional 4 layer boards to come back to me, write the software for an ATMEL 348p, prototype, break it, rinse and repeat. Then I found a wonderful source for a unit that was already produced, one with almost the exact specification I came up with (in my head) for this device. I decided that it was in my best interests and time conservation to purchase a manufactured unit from http://malectrics.eu , sold by Marc Shoenfleiesh for around $105USD (Shipped to Chicago). After talking with Marc through a few emails, we both were in agreeance to the function of this unit utilizing a UC bank in place of a lead-acid battery without trouble (He is a nice guy, very polite and helpful, I recommend his products). I also purchased his preassembled welding probes since they were around the same price to build them here, they're very nice. The control unit is powered by an external L7812cv without a common line attached, utilizing the module primary common for that circuit. This unit utilizes eight IRFB7430 MOSFET IC's in parallel to manage the current pulses. (Datasheet here: https://www.infineon.com/dgdl/irfb7430pbf.pdf?fileId=5546d462533600a4015356169bcd1e53) for current switching. This array, when closed, provides around 150 ohms resistance, allowing a handful of my subsystems to function in their own unique ways depending on their placement on the common bus. In my design, there are two common busses, one is not switched, the other is 150 ohm to the primary circuit when MOSFET array is in an 'open' state. The unit also includes a small OLED screen and a rotary encoder for control selection. There is also an 'Auto-pulse' function, but I will not use it with this device, ever. I utilize a (very obnoxious and unnecessarily large, thanks, Amazon Prime) footswitch as a trigger for this device. It's removable and uses a TS 1/4" bulkhead on the front of the case.
The outboard assembly includes the welding probes themselves, which are 6 gauge wire, terminated to thick copper electrodes cut to points. They work great, you can even see them dance from the current flow!
This one is important, to protect EVERYTHING from flyback current, I built a rather beefy TVS diode array. That's the orange shrink-wrapped guy labeled "Anti-Death Diode Array" on the two current bulkheads. This is a reverse facing diode that will allow inductive current to pass backward at the head of the circuit after the MOSFET array closes, causing an extreme reversal of current. Think of it as analogous to "water hammer" in the plumbing of your home; when you turn off the water flow rapidly, you hear a bang in the wall. This is the tendency of the water to try to keep flowing, attempt to compress, and immediately flow backward, causing cavitation in the pipes. When that vacuum bubble collapses, we hear that bang, at risk of gross oversimplification. By placing our very large diode array in this position, we are protecting all of our circuitry from flyback current by allowing this very strong inductive pressure to "flyback" from one probe to the other. The specifications of this unit will be listed later on.
On to the boring stuff. However, it's still fairly important.
The case is made from two acrylic drawer organizers that can be found on Amazon Prime, I cemented a 12" acrylic hinge, and an acrylic latch onto these, and made a decent, non-conductive case for pretty cheap. Both the primary cc/cv unit and the pulse control unit are on custom cut nylon standoffs from the top of the case. These are secured with vanadium steel M3 cap-heads of varying length, which long enough to provide a secure hold to the case.
Now, we're going to talk about an extremely sensitive subject for myself; glue. I don't like glue. Glue doesn't like me. It's a necessary evil in budget assemblies and time constraints. There are only two types of adhesives used in this entire project; methyl ethyl ketone (MEK) cement, and high-temperature hot glue. Both are entirely non-conductive and cheap. MEK cement is a solvent cement designed for plastics, it works extremely well at binding acrylic, among a few other things. It is very tough and almost entirely rigid after curing, too. This was utilized for all of the plastic to plastic bonds in this project. Hot glue was used for all of the wiring positionings at the edges on the case, some added protection in certain areas against catastrophic shorts, and for a few of the more temporary mountings of components. I don't want to permanently cement a power supply in place to have it fail and need replacing. You may also note that I stayed away from silicone based products. I do not always approve of room temperature vulcanizing rubber for use in electrical projects, due to the insulative variance that seems to occur with air bubbles within the RTV bead. Unless it needs to be permanent for certain, and you're pressing the parts together firmly (e.g., hand-tightening a water pump while the RTV starts to vulcanize) I do not recommend a silicone, and for this project, absolutely no silicone (that does not include silicone wiring insulation, different process).
That's all for now, I will update with a complete electrical schematic and a detailed video mirroring this insight when I get around to it.
Thanks for reading!
-Alex