[hpsdr] ALEX - Call for Comments - III

Ray Anderson ray.anderson at xilinx.com
Thu Mar 8 11:57:23 PST 2007



As Peter mentioned, NPO and COG type dielectrics are pretty good from a
signal distortion point of view even with large signals. X7R, X5R and
relatives have a much higher capacitance to voltage change. With large
voltages they kind of act like varicap diodes (not very good ones at
that), that can lead to non-linear behavior and distortion.

Also, as mentioned, you need to pay attention to any ferrites or
powdered iron components in the filters. The material will magnetically
saturate at a certain power level and the inductance will then change.

Long ago I was involved in satcom work where we had built up a system
for injecting broadband noise into a bandlimited system to establish
different Eb/No levels for measuring bit error performance. Our 70 MHz
wideband BPF looked just great with just our low level CDMA signals
running through it, however as we increased the level of the injected
Gaussian noise, the total  noise power in the filter bandwidth (low
level noise integrated over the BW of the filter) eventually reached a
level that it began saturating the tunable powdered iron core inductors
and caused our nice Chebychev filter shape to degrade into something
that didn't even look like a filter.

73,	Ray   WB6TPU






Pieter, N4IP wrote:

Most NP0 and C0G capacitors have extremelly low distortions. I have 
performed tests on these type caps with as many as 132 carriers and 0dBm

each and the performance is very very good. As you start going up in
value, 
the materials required to increase the capacitance per unit area become
more 
and more unstable and non linear. Some of these materials exhibit a 
capacitance over voltage characteristic. So as the voltage on the cap 
varies, so does the capacitance. This tends to created mixing products
since 
a change in capacitance (modulated by one carrier's voltage) causes the 
whole filter/resonator to be modulated in phase and amplitude.
An even bigger problem are the ferrite beads that are used in the high
value 
chip inductors. These can have very non linear characteristics specially
if 
the core is conductive and the enamel coated wire touches the core.
The powder iron cores from Micrometals are also extremelly good as well
as 
the wirewound surface mount inductors that have no ferrite.

 




 1173383843.0


More information about the Hpsdr mailing list