Charles Builds Stuff |
I happen to have acquired four 1000-foot rolls of 18AWG automotive wiring.
What else was I supposed to do?
Small problem with the plan: Capacitors suck.
The capacitors that I have on hand are of unknown provenance and dubious origin. They are also of unknown DC ESR, and equally confusing ESL.
On top of that, there's another issue.
Building a coilgun with 18AWG wire poses a challenge. Since it is automotive wiring, not magnet wire, you lose a lot of coil packing efficiency. Which means that you need current where you would otherwise be able to get away with more turns of copper. However, because of the girth of the wire in question, you can't just keep adding wire without reaching diminishing returns, and actually, diminishing results. So, instead, you need more current.
RLC discharge curves ruin everything. With 10 of the capacitors on hand in an array for each stage, assuming 170 windings on a 35mm, 10mm-ID spool, the current pulse would have time to peak and peter out in the course of a 14 ms firing pulse.
This would save a lot of hassle with flyback suppression, as none would be needed, but at the cost of significantly lowering the power.
Lithium ion batteries are cheap, readily acquired, and capable of some truly amazing things if treated right. More importantly, high performance batteries are effectively ideal voltage sources. The batteries in use on my RC car project can output 200A at 24V for a period of ten seconds while remaining within operating specifications.
200A is, applied correctly, enough juice to get a steel armature moving pretty darn quick, pretty darn quickly. It's the "24V" part that could be a problem....
LiPO batteries are categorized (generally) by the number of cells in series that they are composed of. This value ranges from one to six, and are laballed as "1S" through "6S". 6S is the largest LiPO battery size that is readily bought. It operates at approximately 24 volts, ish, nominally. Two of these packs in series, then, would provide a magnificent 48V DC, with an ESR measured in single-digit milliohms.
With a correctly designed coil, that could make things happen. Hole-in-ceiling things.
The Waterloo Engineering 3D print centre just so happens to have a promotion running whereby one can get a free custom milled PCB every two weeks. So, with a plan in hand and a hobbyist DipTrace license, I submitted a design and waited.
And waited.
And was soon informed that the PCB mill was out of service for a little while. So I took some guesses as to what the issues would be, and ordered the PCBs from JLCPCB without trying out a prototype board
The plan was to have a Teensy 3.2 running on a controller board with a set of open-collector outputs set up connected to a wiring harness running down the barrel and connecting the stages together along with photogate feedback returning. The open-collector output would be used to drive an opto-isolated gate driver (FOD3182) which would switch the power MOSFET. A TO-220 Schottky diode in series with a resistor would be used for flyback suppression, and a 1 milliohm shunt would be installed inline with the coil to permit current measurement.
You might then ask, how does the isolated gate driver drive the MOSFET's gate? Well, luckily, I thought that through: A linear regulator! It's a simple part, gets us a nice reliable 15V supply so long as we have our main power lines, and is cheap and readily installed.
Now, going from 50V to 15V is perhaps ill advised, but, with some capacitors, the average power dissipation should be very small. It's only a MOSFET gate!
Once the boards and parts arrived, I soldered up the controller board and a pair of stage switching boards, and tried to set everything up.
And it worked, first try!
Now, the issues began where the current meets the LDO. Because, while the average power dissipation on the 15V regulator was very small, the pulsed current was massive - probably around 9A for a few nanoseconds. The capacitors, according to my understanding, should have helped out with this.
Unfortunately, after the explosion, I decided that perhaps that wasn't going to work out. So it was back to the drawing board as I considered the next steps forward.
This all took place during the opening half of my 2B term of university. As stress ramped up, the project fell to the wayside.
Until, of course, co-op applications ramped up.
A particularly interesting position opened up: Hacksmith Entertainment Ltd, or The Hacksmith, was hiring a co-op student. I found the position, looked it over, and declared to myself that I was going to get this damn job. I polished my résumé, wrote a cover letter, which mentioned the coilgun project, and sent in the application package.
With four days left in the interview window, I received an invitation from the company. After gleefully punching the air, the realization settled in on me: I had to nail this interview.
And how better to do that than by showing up with a working project that is right up their alley?
Now, unfortunately, this was in the midsts of midterm week. So, in between studying for one exam a day, I had to figure out a prototype chassis, and fix the small explosion problems. The device was going to be only two stages instead of six, and assembled as hastily as possible.
But, sure enough, two hours after my last exam, I finished off the coilgun, got dressed for an interview, and ever so delicately placed the assembled device in the passenger seat of my car.
The intewview went quite well, and I landed the position. The rest of the term wandered past without much opportunity to work on the device. While taking MTE220, I learned how to do photogates properly (The 2021 mechatronics class will remember the words "Transconductance Amplifier" for as long as we live) and played about with using an operational amplifier to provide readily adjustable feedback control.
Midway through the work term, Ian informed me that we had secured a sponsorship to produce a video about a coilgun.
Which was an odd way to pronounce "It's go time."