Topband
[Top] [All Lists]

Re: Topband: Series LC to notch AM broadcast ?

To: topband@contesting.com
Subject: Re: Topband: Series LC to notch AM broadcast ?
From: Michael Tope <W4EF@dellroy.com>
Date: Mon, 21 Feb 2022 12:58:18 -0800
List-post: <mailto:topband@contesting.com>
Hi Jim,

When I lived in Melbourne Florida, I had an 920 KHz AM station with a four tower array a short distance from my house (I forget the exact distance, it was probably on order of 1/2 mile). The daytime pattern beamed toward my house with a TPO of 5KW. At the time I was using a Ten-Tec Paragon. During the day I would get all sorts of cross-mod when listening on my 80/40 meter fan dipole.

After some trial and error, I discovered a little external attenuation would cure the problem, but the Paragon's internal attenuator wouldn't help. It turned out the signal pickup from AM 920 was so strong it was overwhelming the bias applied to the PIN diodes used to switch coverage segments in the Paragon's front-end filters The Paragon used relays to switch in low-pass filters which were in both the receive and transmit paths, but the high-pass filters were only used for receive and were switched with these PIN diodes. I built a small low-Q notch for 920 KHz on a perfboard and installed it in the portion of the Paragon's receive path between the T/R relay and the PIN diode switched filters.

The notch provided enough attenuation to completely eliminate the cross-modulation. There was a marked decrease in sensitivity around 920 KHz, but it didn't render the Paragon receiver totally dead in that portion of the BC band, so I never bothered to make the notch relay switchable.  I just left it in all the time.

Later on I got an ANC-4 noise canceller and discovered a different problem. A simple high-pass eliminated overload of the ANC-4 receive path from 920 KHz AM, but it turned out that there was another local AM transmitter on 1580 KHz. This one was further away, thereby not strong enough to cross-modulate PIN diodes, but plenty strong enough to overload the front-end of the ANC-4 receive path. For this I built a three section high-pass with two series C/shunt-L sections and one parallel tuned notch section set to 1580 KHz. This did the trick. This was also an RX only filter. I believe I ended up adding a standard ICE high-pass to the transmit path, but I don't recall exactly why.

This work was all done circa 1995/1996 before I moved to California. Recently when this topic came up on the reflector, I went hunting for the schematic of this filter, but was unable to find it. I did, however, find the filter in drawer full of RF odds and ends. It's built inside of an enclosure made of pieces of double-sided circuit board material soldered together. When I have some spare time, I need to sit down and sweep it on a network analyzer and create a schematic for it. Same goes for the notch inside the Paragon. I still have the rig, but would need to take it apart to document the schematic. Both schematics may exist somewhere, but there is no guarantee so I figure biting the bullet and reverse engineering the filters will be faster and much less annoying than trying to find written documentation that may no longer exist 😉

As far as what values to use, you should be able to simulate this with LTSpice and/or QUCS. With ideal components you can make the notch as deep and narrow as you want, but reality bites when you throw in the temperature stability of the components and their finite unloaded Q (in particular the inductor). There are also voltage extremes that come into play with very high loaded Q series notches that are placed in the transmit path (and current extremes for parallel resonators with high loaded Q). I suspect if you fiddle around with an inductor unloaded Q of 100 in simulation, you'll get a good feel for what is practical. To nail multiple interfering stations, you might try looking at an elliptic filter design (i.e. Cauer) which has an "equal ripple" stopband (https://en.wikipedia.org/wiki/Elliptic_filter). The elliptic filter includes transmission zeroes in the form of LC resonators which put deep notches at various points in the filter stopband.

Here is an example of a QUCS elliptic filter simulation:
https://rf-tools.com/lc-filter/description.html#qucs (click on "Example 1: S-Parameters").

73, Mike W4EF............

On 2/20/2022 9:41 AM, jim.thom jim.thom@telus.net wrote:
Has anybody tried using a simple series L-C to notch out ONE offending AM
broadcast station ?  I'm talking about wiring from hot side of coax...to
chassis / ground....like via a T connector etc.

On paper, it should work. Did some minor research, and one comment was that
by using higher values of L would result in  higher Q..and a deeper notch.
Another comment stated to use some initial values, like what spits out on a
L-C  online calculator for practical values..... then  multiply one value
by the other...then take the square root of the result.   Then you ended up
with 2 x numerically equal values of L + C..... and supposedly the greatest
notch depth.

On software, I tried several values..from one extreme to the other, and
they all resonate on the same freq.   Also tried in software,  using 2 x
numerically same values..and it too, also resonates on the same freq.

The rationale behind all of this is... in some cases, there is only one
offending AM broadcast station.  Typ  HP  filters offer little rejection
towards the top end of the AM broadcast band..... like  1200-1710 khz.

I would like to try it, but am still confused as to which combo (using
practical values) will result in the deepest notch.  It would have to be
wide enough to remove the 20 khz wide AM signal.  A fixed coil + variable
cap, or padded variable cap could be used to fine tune the notch freq.

Perhaps   2 or more LC filters could be used in parallel, to notch out 2 or
more offending stations ?


Jim   VE7RF
_________________
Searchable Archives: http://www.contesting.com/_topband - Topband Reflector

_________________
Searchable Archives: http://www.contesting.com/_topband - Topband Reflector
<Prev in Thread] Current Thread [Next in Thread>