I need a 4:1-6:1 tranny that has a flat-ish response from about 100Hz-10kHz to interface a thingie to my computer’s soundcard. A Neutrik NTE4 goes for about $12 and would work fine, but there is shipping and waiting for it to arrive, AND I don’t really need anything that nice, so… back to the wonderful local thrift store to grab some DIY supplies. In the back of the store, there are bins of 99 cent electronics that have everything from RCA cables to antennas to old routers to cell phone chargers. So I grabbed a few wall wart adapters to cannibalize. I want to reuse the core and the magnet wire, so I’m looking for something small with low current and voltage ratings to be sure that the laminations and wire are as thin as possible.
FIRST, LET’S GET AN ADAPTER:
This one will do fine:
OKAY, now let’s salvage our parts. First we need to crack it open and see what we have inside. Usually it is easiest to use a hacksaw to cut into the corner and then pry it apart with a screwdriver. Then we split it open and pull out the guts:
Most wall warts these days have slide-in windings, so we can just pop them out and measure the wire. I don’t have any calipers accurate enough to measure magnet wire, so I wrap 20 winds around a small metal ruler, divide the measurement by 20, and then consult an AWG chart to see what it is. In this case, I got 40 AWG wire (about 500 feet or so, which is more than enough), 36awg wire, and about 30 small silicon steel laminations with 1/8″ wide side legs… perfect:
I use a razor blade to crack apart the laminations. Somethings they are stubborn depending how well they are lacquered, so be careful. Usually you have to mangle the first few laminations out with pliers, or tap them out with a screwdriver and hammer to give yourself a bit of wiggle-room. The rest should pop apart pretty easily, but be careful! If you are uncomfortable doing it with a razor blade, you can use a metal putty knife, which is what I do for larger trannys:
We’re going to convert this to a quasi-toroid/c-core/whatever shape because I have to squeeze the tranny into a metal shield tube, so we’ll cut off one of the legs with wire cutters, and just use the middle and one side:
LET’S DESIGN OUR TRANNY:
Since the low end response is mainly set by the primary coil inductance, we can calculate what we need via 2*pi*Fz*L if we know our source impedance, which is a few kOhms in this case. Our freq is 100Hz and our source we will just call 2k Ohms worst case scenario since our load impedance is a 20k sound card input, so if we rearrange our formula, we get 4000/(pi*200)= ~6.4 Henries.
NOW… to know the EXACT number of turns, we will need to know the AL value of the core, but we don’t since we are salvaging parts, so we’ll just assume a few numbers for generic silicon steel laminations, plug it into an equation, give a healthy error margin to fudge some numbers, and do the hand math. I’ll save you that painful process (there are some calculators online you can use, or just look up the equation on just about any site that sells transformer laminations) and say that I’ll need about 2700 turns (which is an estimate that is preeeetty rough, so I’ll shoot for 3000), which is totally do-able. We’ll need about 500 on the secondary, which is also fine for the core winding window we have.
ONTO THE WINDER:
The “bobbin” We’ll be using is… dramatic pause… a drinking straw! I cut them a bit long and then use rubber grommets to prevent the coil from unraveling:
I wound the secondary over the primary, separating them with a few layers of paper medical tape.
When the coil is wound, I use clear nail polish on the ends to keep the coil formed, and then put a few drops of nail polish drying drops on it, and then hit it with electronics freeze spray… because I hate waiting for crap to dry:
Next, we will put out coil on the core. I crammed the u-shaped laminations in, alternating directions, and then capped each layer with one of the I laminations cut to length (like a typical EI core), hit the edges with nail polish, and solder small copper tabs to the ends of the coils to give alligator clips something to bit into for testing:
Finally, I taped it up with aluminum tape to hold it together and shield the coil a bit from EMI. The finished product, under test:
TESTING: THE MOMENT OF TRUTH.
I am under no illusions that this will be a high-performance transformer: there is a reason that Jensen/etc charge so much for their trannys, BUT I am pretty sure this will do the job. I won’t post any square wave responses or harmonic data, but here is a simple frequency sweep (under load, line level):
NOT TOO BAD! Especially for under $1 and an hour of time. Somehow, we managed to hit the min-spec right on the nose. A little more low-end would be nice, but it is only down -1dB @ 100Hz, so that works. This is for testing, but even at -3dB @ 70Hz, it is usable for actual musical signals like guitar and vocals (although the transient response is probably poor). The high-free response is better than I expected.
Since I cut the E laminations down to make a ‘core style’ configuration, the efficiency is not great with the winds on one side only. To up that bit, you could cut the thicker leg down to the size of the other leg, and split both the primary and secondary in half, winding half on each leg and wiring the like coils in series, like this:
(Alternatively, you could just use the unmodified EI laminations and go with a shell type transformer if space is not a concern)