On 03/08/14 09:11, Stuart Longland wrote:
In the interest of science, I gave it a shot just then. With a 100nF
capacitor in series to block the DC, I wired it to a DIN5 plug (all my
radios have been set up with adaptors to DIN5 headset jacks) and tried it.

It did work, but without any amplification or impedance matching, the
modulation is well down. I might try winding a small transformer and
see what that does.

Well, I did some math, to transform a 32ohm load to ~600 ohms, I need a
turns ratio of ~4.3:1.

math.sqrt(600.0/32.0)
4.330127018922194

I played with this a little, and so I thought I'd add a couple more,
make it a nice round number. 4.5:1 would give me a 648ohm load on the
HV side.

I had a few L8 toroid ferrite cores laying around, these to be precise:
http://www.jaycar.com.au/productView.asp?ID=LO1230

Not being sure about the number of turns, I made a guess at 45:10, and
started winding.

Just tried it and now the audio is completely dead. Transformer loss is
too high for the feeble signal out of the "microphone".

My guess is that the turns count is waay too small, and that maybe
450:100 might be closer to the mark, but I really don't feel like
winding that many turns on a toroid. 45 felt like a marathon.

So for this to work, I need an in-line amplifier of some sort. My
challenge is to RF-harden it, and have it compatible with both dynamic
microphone inputs as well as electret: as the same headset will probably
be used with both. (If the phone rings and I'm on the bike, I will
sometimes plug the helmet into the phone to answer it. Right now all
headsets are interchangeable with all devices, a feature I'd like to keep.)

I'm thinking possibly a FET, since that's what's embedded in the
electret capsules, however I suspect this will be prone to the same problem.

Looks like I might be getting out the oscilloscope and making some
measurements with the PTT down, try to figure out where the RF is
getting in.