Cohn filters, a question of W0IYH
Lawrence Stoskopf
Just finished reading another discussion of the Cohn filters in the Jan
97 Radio Communications, Technical Topics, p. 62 and have a couple of
questions about the topic of receiver front ends in general to kick
around. Where better than here.
Sabin discussed Cohn filters in QST July 1970, I think there was a
similar filter used by Squires in an early, high dynamic RX using the
7360 tube as a mixer.
There was some discussion in TT about the Cohn filters being very lossy
if not tuned correctly and this is addressed to some extent in the past
TT discussions.
RF Design had a article comparing reed (I think) relay switching vs.
saturated transistor switching and found the relays much inferior.
My homebrew 160 RX is a 15 year old Martin design using a MCL SAY-1 high
level mixer with 23 dBm LO drive (a Gumm osc based on Hayward's Solid
State Design for the Radio Amateur...it works better than Rhode's later
simulation seems to indicate). This followed by a diplexer at 9 MHz, an
early FET, the VMP-4 running grounded gate at about 150 ma current, into
one of the KVG 8 pole filters, thru several Plessy gain/control stages,
another KVG filter (14 poles total) and out. And old design.
I have the capability of switching in a 3 and 6 dB pad on the input and
normally switch in whatever it takes to just have the noise level rise a
bit when the input is switched from the termination (used during
transmit) to the antenna. Then I manually adjust the IF gain so that
the AGC doesn't bring the AF noise level too high when no signal is in
the passband. (I never got around to getting away from the Plessy AF
derived AGC so that method calms the AF output a bit.)
Input filtering? The antenna tuner. I'm setting here looking out of
the window at 4 towers 3 miles away used by an AM station on 1150 and
it's no bother. I don't even get a 2nd harmonic of a station on 910
about 8 miles away...probably a testament to their filtering!
After this lengthy introduction, questions:
Suppose I add the minimum-loss bandpass filter from the Jan TT, 4
sections, 4 dB loss on 160 with 66 KHz BW. This loss is about what I
normally add. I'm apparently not bothered by the BC band signals and it
certainly does not help much for contests as the whole contest band is
essentially within this passband. What do I gain by adding this filter?
(Can't get the passband much narrower without using a Q-multiplier type
system or the loss gets too high for even 160.)
So possibly a different approach? Sabin (or was it Squires) used an
impedance inverting circuit to present a reasonable load to an active
mixer in the desired passband, but which thru the inverter presented a
low impedance to the mixer (as the filter impedance rose out of the
passband) lowering the gain outside of the passband (Av = gm * RL). You
can't do this with a ring mixer as the reflected power circulates back
thru the mixer causing all kinds of grief. (This approach must present
all kinds of interesting situations to the filter which is normally
designed to work with some fixed driving and load impedances.)
I throw this out mainly to get a thread going. As a long time RSGB
member, Pat Hawker fan, I keep reading about the high RF levels in
Europe, as presented by Chadwick and others, and high dynamic range
efforts by all. I still can't see how a 20 M, 3 dB loss filter with a
510 KHz BW can do much more than lower the total signal level to the RX
by 3 dB and help the blocking level, not dynamic range, that way. There
seems from here to be a few high power signals in that passband. 40 m.
looks worse.
Perhaps I'll take one of the pocket SW radios on my Sept/Oct trip to
Amsterdam, Moscow, and St Petersburg to see what the fuss is all about.
Puzzled from Kansas,
N0UU
William E. Sabin
Lawrence Stoskopf wrote:
>
>
> Sabin discussed Cohn filters in QST July 1970
Correct.
> I think there was a
> similar filter used by Squires in an early, high dynamic RX using the
> 7360 tube as a mixer.
That I am not aware of (don't think so).
>
> There was some discussion in TT about the Cohn filters being very lossy
> if not tuned correctly and this is addressed to some extent in the past
> TT discussions.
I am not aware of this. A Chebyshev filter, improperly tuned, would have
the same problem. The min-loss filter is just a little better than the
Chebyshev. The idea of the Cohn (min-loss) is that for a given insertion
loss it has the best skirt selectivity. The tradeoff is that the passband
is not "flat". It needs to be properly terminated at both ends. At the
output end, instead of a lossy resistor, a "dynamic" load, such as the
input impedance of a grounded gate FET, or the input of a diode quad
mixer, would be best.
The book "Single-Sideband Systems and Circuits" by Sabin and Schoenike,
Second Edition, McGraw-Hill 1995 has a chapter on preselectors that
discusses the Cohn "min-loss" filter design in detail.
>
> RF Design had a article comparing reed (I think) relay switching vs.
> saturated transistor switching and found the relays much inferior.
>
I have used relays with great success. But they have to be high quality.
>
>
> Suppose I add the minimum-loss bandpass filter from the Jan TT, 4
> sections, 4 dB loss on 160 with 66 KHz BW. This loss is about what I
> normally add. I'm apparently not bothered by the BC band signals and it
> certainly does not help much for contests as the whole contest band is
> essentially within this passband. What do I gain by adding this filter?
Probably not much if you are not having any problems.
> Sabin (or was it Squires) used an
> impedance inverting circuit to present a reasonable load to an active
> mixer in the desired passband, but which thru the inverter presented a
> low impedance to the mixer (as the filter impedance rose out of the
> passband) lowering the gain outside of the passband (Av = gm * RL).
Sabin, QST, July 1970.
You
> can't do this with a ring mixer as the reflected power circulates back
> thru the mixer causing all kinds of grief. (This approach must present
> all kinds of interesting situations to the filter which is normally
> designed to work with some fixed driving and load impedances.)
True for passive mixers, but a useful technique for FET mixers.
> I keep reading about the high RF levels in
> Europe,
Europe is a high level environment. Studies confirm.
Bill W0IYH