On 28 Oct 2005 09:44:00 -0700, "K7ITM" wrote:



OK, maybe you're beginning to understand. Q can be calculated as
reactance (at resonance) divided by the effective series resistance, or
as effective parallel resistance divided by reactance at resonance.
For a loop where you know the series resistance, it's easiest to use
that first relationship. If you put your 10 ohm receiver input in
series with your 10 ohm reactance loop, you've ruined all that effort
to get to a very low loop conductor resistance and obviated the need
for high-Q capacitors. And we're all having a very hard time seeing
how you will couple your 10-ohm receiver input to EITHER the
parallel-tuned loop OR the series tuned loop, without having nasty
consequences for your holy-grail Q.


It appears to me that a 10 ohm receiver input impedance is dead center
with regard to any loop configuration I could come up with...it
doesn't work with anything! It's easy to see now::

You might think it's best to impedance match ("conjugate match") to
your load, so you transfer the most power to the load. However, that
may not be optimum from a system design standpoint. If you already
have enough signal (along with atmospheric noise) that the receiver
doesn't contribute significantly to the overall SNR, then you may be
better off by intentionally mismatching so that the Q remains high, if
that's important to you. (I personally think you've overrated it, but
that's up to you to decide.) But even if you're wanting to get the
lowest noise contribution from your electronics, the appropriate match
is generally not the conjugate impedance match that results in highest
power transfer. For example, an MMBT2222 NPN transistor running at
about 100uA collector current in a common-emitter configuration with no
feedback will have a low-frequency (e.g. 60kHz) input resistance around
50kohms, but the optimal source resistance from a noise standpoint--the
source resistance which will yield the lowest noise figure for the
amplifier--will be about 2kohms. At optimal source resistance, you can
get a noise figure well below 1dB from an MMBT2222--and from many other
bipolars.


Is the MMBT2222 the same as a 2N2222, which I already have in my junk
box? Also have 3904's, maybe they are just as suitable?

Should I tune the output, or rely on a modest input tuned filter in
the front end and just do a no tune in and out common emitter?

Assuming a 50 ohm receiver input impedance, is it ok to take the
output across a 50 ohm emitter resistor?


One reason that people like to use FET amplifiers across their
"parallel-tuned" loops is that the amplifier input resistance is quite
high, but (using appropriate FETs) the noise contribution of the
amplifier is negligible. And with proper design, the distortion
contribution can be considerably lower than the distortion of your
detector. For high source impedances, JFETs can give noise figures
that are a small fraction of a dB.


JFET's sound good too.

Richard, which type of buffer amp has a lower distortion and better
linearity in the presence of (potentially) large out of band signals?
If I do use a buffer to preserve the Q, it has to be 'clean' in order
to preserve the performance of the QSD receiver. Appreciate any
enlightenment along these lines.

I'm not opposed to using an op amp IF it provides cleaner output
signals and if the power consumption is manageable.

I must say it seems much cheaper (and probably more practical) to use
a buffer to preserve the existing Q than it is to build higher Q into
the loop...only to have it cut in half by the addition of a receiver
front end.

Regards,

T