Hi,

I had exactly the same issue a few months back. Winding so many turns
on a FT50-43 is not realistic. Actually, winding the same number of
turns on any core, even the pot cores recommended by other people are
not so good for this application. The issue is the parasitic
capacitance. In my case, I needed ~2mH inductors. After I wound the
required number of turns on FT50-43, which was about 60 (IIRC), then
measured the parasitic capacitance, it was too much. The self resonance
was somewehere in the below 1 MHz. The 60 turns just about fits onto
the core. If you go for second layer (and beyond), the capacitance will
skyrocket.

A diplexer is supposed to provide close to real, near 50 Ohm impedance
to the mixer. At the same time, it needs to work as a low pass filter,
to keep the RF out of teh audio stages. If you do some simulations of
the common audio diplexer topologies and take into account the
parasitic capacitance of the inductors, you'll find out that the match
is close to 50-Ohms real only at audio, maybe up to LF. Completely
useless at HF and above.

So what's the solution? A few options:
1/ Try a material with higher permeability. Toroids are best, because
they don't need shielding and they don't break as easily as pot cores.
I'd recommend FT50-J. This is I think the highest permeability that
Amidon sells. You should end up with much smaller number of turns. It
may improve things in your case.
2/ Forget diplexer. Use an attenuator, followed with a low pass filter.
This will increase the noise figure & gain of the mixer-post-amp combo,
so you'll have to make it up with more gain ahead of the mixer. But
you'l have higher IP3. You could set the amount of attenuation to
balance the noise figure against IP3. Where the optimum is, I have no
idea.
3/ Since you are mentioning BFO & IF, I assume that you have a
superhet. The mixer at the end of the IF strip does not have such hard
requirements, since you've already filtered the 'bad guys' out. For
SSB, you would not care about intermod performance after IF filter too
much; well maybe except for audio distortion. Actually, for a product
detector in a superhet for something like SSB, don't even use a diode
mixer. Too much power and not necessary. Switch over to an active
mixer, like the NE602.

Anyway, your mileage may vary. The above should work for common cases.
If you really need a proper broad-band RF diplexer which also works as
an audio low pass filter, then you're in for an interesting journey...

Hope this helps,
Fil
VK3FLP

wrote:
hello,

After completing my AF amp and BFO, I'm in the process of doing IF/BFO
diode ring mixer.
I understood that it is a good idea (or even mandatory?) to follow DBM
with a diplexer.

I found one from http://www.qrp.pops.net/dip2.htm

My question is: how on earth should I wind those 10mH inductors? My
calculations show that I should wind 487 turns on FT37-43 (which is
obviously impossible), or 437 turns on FT50-43 (which is hard..but
should be possible).

Any other ideas how to make 10mH inductors for AF diplexer?

thanks!