[DSP-10] Filter Sweeps - new results and questions from n5bf/6

Courtney Duncan cbduncan at earthlink.net
Tue Aug 21 01:34:59 EDT 2007


This year got a little busy.  In February my mother had both her 
knees replaced and in May my first daughter graduated from college. 
Now I'm back having a little DSP-10 fun again.  Recent results are 
posted at

http://cbduncan.110mb.com/HamRadio/Dsp10/Dsp10.html

(For those of you with old links, I've moved my website to this new location.)

I have a couple of questions for the group about this filter sweep data:

The first two plots cover 30 KHz, that is, six cycles of 5 KHz, 126 
MHz synthesizer switching in 100 Hz points.  I'm interpreting these 
data as the ripple or some passband shape of the 19.665 MHz first IF 
filter.  (I think hams call this the "roofing filter" in contest 
rigs, is that correct?)  Peak to peak, the passband varies by about 2 
dB which seems about 1 dB more than it should be.  Have I installed 
something wrong or, in any case, is there something I can/should do 
about this?  Or is it about right for this case and I shouldn't worry?

On the third plot, I've swept the whole band (10 KHz points) and it 
looks about like you'd expect except that it seems to roll off 
excessively at the bottom of the band where all the interesting 
activity is.  For instance, I did most of my EME2 tests below 144.1 
and was therefore losing a dB or two just due to operating down there.

Would it help to re-peak the filter chain at something like 146.5 or 
146.0 rather than 147.0 or am I seeing some other effect here, like 
amplifier device or filter rolloff?

I picked the power setting so that all of this would be right around 
what I think is the 1 dB compression point for the Brickette output 
(about 3 watts).  If anything, getting into compression should 
flatten a plot like this shouldn't it?

I guess most DSP-10 users are using the box as an IF to microwave 
frequencies.  In that case the band-sweep issue might not be so 
important, but it seems like the 19.665 MHz filter shape could be.

As always, any insight would be greatly appreciated.

73, Courtney, n5bf/6


More information about the DSP-10 mailing list