[DSP-10] AA1YN bandpass flatness & birdies

Mon, 25 Mar 2002 19:50:46 -0800

Hi Lee and others interested in these problems!

As W7SZ has identified, the changing noise with tuning is associated with
crystal filter flatness.

First for the theory side of thing. The crystal filter, along with virtually
all modern filter designs, are "synchronously tuned." This means that each
loop of the filter is tuned to the center frequency, with the bandwidth and
response coming from the interaction between the resonators. In our case,
the synchronous tuning is pretty much set by the crystals and the coupling
caps. The two tuning adjustment coils are really parts of an 'L' networks
used to step up the impedance to that needed by the filter. But, they can
present a reactance, and when all is tuned properly this reactance will be
gone. Also, this will be the condition that will minimize the insertion
loss. That is why the tuning for max signal will be very close.

Fig 5 in Sept QST is an actual measured filter.

Now for an alternate alignment procedure (this is from last Dec - I thought
I had posted it then. If so, I'm sorry for the duplication!):

---Procedure--------

1-First, the best test setup for playing with this is a noise input and
watching the spectral response. The handiest noise source is a preamp or
converter with a gain of 15 dB or more.  No antenna on the input. 50 Ohms is OK.

2-Now what to watch. Put the DSP-10 in LSB (this can be done in USB, but the
frequencies get hard to follow. Activate LSB and use it). The hit Alt-A to
see the frequencies. The DSP-10 IF#2 shown is the frequency corresponding to
zero-beat or the left edge of the spectral display. It will range from 10251
Hz to 15250Hz as the radio is tuned.

3- Set the SA width to 4800 and the IF filter to wide (Alt-$). The spectrum
from about 500 to 3500 Hz will now be flat, except for the response of the
crystal filter. Ignore everything outside of the 500 to 3500 range. Maybe
put a piece of tape over it!  Set the SpecAve to 36 or so, to smooth the
noise response out. Put the display in 2 dB/div.

4-Tune the DSP-10 to 146.159 700 and the IF#2 will be 10300. The region from
500 to 3500 Hz corresponds to the IF#2 of 10800 to 13800. Then tune the
DSP-10 up 100 Hz to 146.159800. The IF#2 is now 15200 ad you will see 15700
to 18700 inside the 500 to 3500 range. These two frequencies are good enough
to see the response of the crystal filter.

Concentrate on the lower response, from 10800 to 13800. It is much more
sensitive to tuning than the upper one. First tune L12 and L13 for maximum
response. You can use the Alt-2 and Alt-3 to keep the display on the screen.
Tune slowly because of the Spec-Ave of 36. Next, tune L12 and L13 in small
steps to make the response as flat as possible. After the bottom part of the
filter  (10800 to 13800) is good and flat, look at the upper part (15700 to
18700). If it is not flat, try tuning it to correct this. On the ones I
worked with, this was not needed.

On mine, I was able to get the response flat to 1 or 1.5 dB peak to peak
across any 3 kHz band.

--------------end procedure -------------

Lee, on the birdie, keep in mind that the 2nd I-F (10 to 20 kHz) is
inverted. If you are in USB on 2-meters, you are in LSB at the 2ndd I-F so
the birdie tunes backwards.

There is a 5 kHz signal in U104, so this may be picked up by L32. Be sure
that nothing in the way of wires is close to L32. Check the frequencies of
the 2nd I-F DDS (variable BFO) by using an ALT-A followed by a frequency
change. See if they make sense for being the harmonic of the 5 kHz signal.
Also, if this fails, give us your settings and let me dig for a
corresponding birdie.

Good luck and thanks to Larry.

73, Bob  W7PUA

>
>At 09:51 PM 3/23/2002 -0500, you wrote:
>>After much thinking about the wave of noise, I've come to a conclusion. I 
>>may be so far off with my thoughts that I need someone like W7PUA to set 
>>me straight. - Anyway, here goes....
>>
>>The audio window is just under 5khz wide and we are moving that window 
>>5khz for a total of about 10khz.  We are asking the crystal filter to be 
>>flat for that 10khz.  If we adjust the filter coils L12 and L13 only for 
>>max signal, we may indeed be peaking that filter with a flat top width of 
>>about 2khz.  If we start measuring the noise at a frequency just above the 
>>5khz step and then slowly move up in frequency, you will see the peak of 
>>the noise at the high end of the audio window and it will look like a wave 
>>as we move up in frequency until this wave is at the low end of the audio 
>>window.  Once we reach that 5khz step point, the noise wave will then move 
>>back up to the top of the window.
>>
>>Unless I am wrong, we really need a procedure for adjusting the crystal 
>>coils L12 and L13 so the crystal filter is as flat as possible for the 
>>full 10khz and not just centered in the 10khz window with it's 3db points 
>>at +/- 5khz.  What I tried was to set the frequency to the low end of the 
>>5khz step and adjust L12, set the frequency to the high end of the 5khz 
>>step and adjust L13.  By doing so, I was able to see 2 noise waves which 
>>kind of backs up what I've been describing. At least in my mind.
>>
>>If the local osc was continuously variable and we weren't moving an audio 
>>window, you would never see this phenomenon and we could happily set up 
>>the crystal filter for a 3khz bandwidth and be pretty much flat across it.
>>
>>Anyone else have any thoughts on this???
>>
>>As to the 5khz birdie I have, I have ran heavy grounding straps from the 
>>board to the chassis, from the board to the DSP box, I drilled 6 small 
>>holes in the top of the DSP box and soldered 14awg wire to tie the top and 
>>bottom foils together next to each screw. I have double checked all 
>>grounding wires and touched up any which might be less than perfect.  I 
>>think I've reduced this birdie about 1db so it is now only 4db in 
>>height.  I did manage to lower the overall noise of the receiver from 
>>-149dbm to -152dbm with a 50 ohm termination on the antenna so there was 
>>an improvement there.  The birdie still moves up in the audio frequency 
>>when you move up in receiver frequency.  Since there are no actual 
>>frequency changes within the 5khz step, I still feel this has to be 
>>internal to the DSP program.
>>
>>One other ground point which eliminated my digital noise was to run a 
>>ground strap from pin 5 of the serial jack to its chassis mtg screw.  I 
>>had only grounded one end shields before at the DSP box.
>>
>>Hope my ramblings aren't too far afield.
>>
>>Lee Scott - AA1YN
>>
>>_______________________________________________
>>DSP-10 mailing list
>>[email protected]
>>http://mailman.qth.net/mailman/listinfo/dsp-10
>
>
>_______________________________________________
>DSP-10 mailing list
>[email protected]
>http://mailman.qth.net/mailman/listinfo/dsp-10
>
>