The distortion analyzer has a tracking bandpass filter. I'm using the
HP8903E, but for this purpose, even the 334B would do the trick. You
can use the analyzer to display the signal level of the sine wave
inpedentent of noise to the degree the bandpass filter can remove it.
[The 8903E reads amplitude, distortion, and frequency.]

I never reached the -60db point, because the signal was lost in the
noise. But you have to ask yourself, if the signal is in the noise, you
do care about the rejection?

In the dark ages, I was designing switched capacitor filters for modem
products. When you sold the chip as a Bell standard filter, you had to
implement that exact filter. However, the reality was the noise floor
and THD totally negated the effectiveness of the filter. [Power supply
rejection is a factor too.] So yeah, I could attenuate the alternate
(out of band) channel, but the filter noise was more significant than
what I was attenuating. Once the product became integrated, nobody
could see the filter implementation. The amount of filtering was
reduced to a level where the alternate channel (out of band) signal
level was minimum by picking a compromise between the number of poles
of filtering and the noise level of the amplifiers. Less poles of
filtering meant fewer all passes were required.

Now I'm not happy the noise floor of the radio crapped out at around
-45db. I don't know if measuring at the line outputs would have been
better. I need to do the measurement again.

It is quite possible that you can use your sound card to measure the
amplitude of the signal independent of the noise. BTW, I should have
mentioned I turned on the 1khz modulation on the RF signal generator.

My goal was to test a crystal filter I picked up at the flea market and
installed in my radio. [A good deal at $10.] One thing I learned is the
coupling capacitors make a difference in terms of the signal level. I'd
have to open the radio to see what I used, but I recall if the caps
were too small, the signal level was attenuated.


Michael Thorpe wrote:
On 11 Jan 2006 14:14:01 -0800, wrote:

http://www.lazygranch.com/images/radio/cwfilter.gif
I did this test as follows:
1) radio in AM
2) AGC turned off
3) RF generator connector to antenna input
4) distortion analyzer connected to earphone output

Pick a frequency. I used 10Mhz. Set the RF generator and radio to that
frequency. Leave the radio dial alone. Sweep the RF generator manually
and measure the audio level from the distortion analyser. Plot the
results. This was a crystal filter, so you will need a RF generator
with 1Hz steps.

This is an in-situ measurement, i.e. this is not the same as just
testing the IF filter by itself. You need the distortion analyzer to
get the audio signal level independent from the noise.


Many thanks, this is interesting. What I am still unsure about is how
you determine the -6dB and -60dB points using the distortion analyzer:
When you say "measure the audio level", what exactly do you measure,
the audio signal fundamental (i.e. filtered at the modulation
frequency)?

So in fact you are only using the distortion analyzer to filter the
audio and to measure the level? I don't have a distortion analyzer,
but I guess I could simply use an oscilloscope or an audio spectrum
analyzer (PC sound card based).

Michael