Message	Date	From	Subject
1735	2008-06-21 21:53:17	la3pna	Expected loss in TTC
I have over the past time had some focus in 2M transceivers, and this has lead to a lot of tests on different filters. In an triple tuned circuit based on the paper that acomplies EMRFD, i did with some tuning get aprox. 3dB loss, this filter is designed for 133.6MHz, to be used after the local oscilator. This overall filter loss sems to be way to low, 3dB is not what I did expect, more like 6dB. Is this an indication that there is somthing wrong? 73 de Thomas LA3PNA
1737	2008-06-22 11:26:19	Wes Hayward	Re: Expected loss in TTC
Hi Thomas, and group, Alas, it seems that there always more loss than we expect, or less gain in an amplifier or mixer, etc. In that earlier lifetime when I was designing integrated circuits, we used to call this the "phantom dB." It was always there, it seems. The filter design software that is on the CD that comes with EMRFD (DTC and TTC) includes an estimate for loss. Generally, that is bogus and it not very accurate. I have eliminated the "feature" from my own versions of the software. I urge you all to forget about that detail. The 1998 QEX paper dealing with a triple tuned circuit that I developed for use with our spectrum analyzer described a filter that had a loss of around 3 dB. This estimate came from the original simulations which were based upon measurements of resonator unloaded Q. I got good correspondence between the simulations and the measurements for this particular case. But the filter had the same components in it that were in the original resonator that was measured for Qu. The programs that I've written and the usual calculations that I've done assume a finite Qu for the inductors and then assume that the capacitors are ideal, completely without loss. This is often a valid assumption for simple HF filters. But it is not so good at VHF. The losses in capacitors can be a major problem, especially with the usual SMT capacitors that we tend to use these days. If you want really low loss SMT capacitors, expect to pay a premium for them. The other thing to remember is that loss is directly and intimately related to bandwidth. If you make a filter narrower, the loss will increase. Conversely, a wider filter will have less loss. If you want the least loss, use a Cohn design. That is, use equal coupling. The k and q design parameters are in Chapter 3. The Cohn design is also known as the "minimum loss" filter. Remember that a Cohn Filter refers to a particular set of k and q values and NOT to either a crystal filter or a particular LC preselector that was popular back in an earlier time. It is hard to obtain good control over loss in a VHF filter where the coupling is obtained through gimmick capacitors or apertures between resonators, etc. While these coupling methods are quite valid, they are harder to characterize than the simple discrete coupling capacitors or inductors that we used at HF. But if you are careful with actual coupling and loading measurements, you can still get good correspondence between simulati

Hi Thomas, and group,

Alas, it seems that there always more loss than we expect, or less
gain in an amplifier or mixer, etc. In that earlier lifetime when I
was designing integrated circuits, we used to call this the "phantom
dB." It was always there, it seems.

The filter design software that is on the CD that comes with EMRFD
(DTC and TTC) includes an estimate for loss. Generally, that is
bogus and it not very accurate. I have eliminated the "feature"
from my own versions of the software. I urge you all to forget
about that detail.

The 1998 QEX paper dealing with a triple tuned circuit that I
developed for use with our spectrum analyzer described a filter that
had a loss of around 3 dB. This estimate came from the original
simulations which were based upon measurements of resonator unloaded
Q. I got good correspondence between the simulations and the
measurements for this particular case. But the filter had the same
components in it that were in the original resonator that was
measured for Qu.

The programs that I've written and the usual calculations that I've
done assume a finite Qu for the inductors and then assume that the
capacitors are ideal, completely without loss. This is often a
valid assumption for simple HF filters. But it is not so good at
VHF. The losses in capacitors can be a major problem, especially
with the usual SMT capacitors that we tend to use these days. If
you want really low loss SMT capacitors, expect to pay a premium for
them.

The other thing to remember is that loss is directly and intimately
related to bandwidth. If you make a filter narrower, the loss will
increase. Conversely, a wider filter will have less loss. If you
want the least loss, use a Cohn design. That is, use equal
coupling. The k and q design parameters are in Chapter 3. The Cohn
design is also known as the "minimum loss" filter. Remember that a
Cohn Filter refers to a particular set of k and q values and NOT to
either a crystal filter or a particular LC preselector that was
popular back in an earlier time.

It is hard to obtain good control over loss in a VHF filter where the
coupling is obtained through gimmick capacitors or apertures between
resonators, etc. While these coupling methods are quite valid,
they are harder to characterize than the simple discrete coupling
capacitors or inductors that we used at HF. But if you are careful
with actual coupling and loading measurements, you can still get good
correspondence between simulati