500 - 515 kHz Preselector

The 500 - 515 kHz preselector described here was built for use ahead of the station receivers at WE2XGR/2 - primarily to reduce mixing/intermodulation products from AM broadcast stations. Although no interference was noted when using the high Q Marconi transmit antenna for receive, it was a different matter when using broadband, receive-only antennas such as the LF/MF K9AY. Small receive only antennas typically require a preamplifier in order to hear weak signals. While the robust, low-noise broadband preamplifier in use showed no signs of overload, the additional gain from the preamplifier caused mixing/intermodulation problems in the receivers that followed - especially when operating adjacent to the AM broadcast band. This filter is used between the Marconi transmit antenna or amplified receiver only antennas and the receivers.

The filter is a 3-section Butterworth design with a 3-dB bandwidth of about 10 kHz and is tuneable over the 500 to 515 kHz range. Details of the filter can be found at 500-515 kHz Preselector. Insertion loss is about 5.4 dB and the spectral analysis displays and the table below show the filter response. Good rejection of the broadcast band signals was achieved and all mixing/intermodulation problems have been eliminated. An insertion loss of 5.4 dB isn't a factor since the Marconi transmit antenna requires about 20-dB attenuation before being applied to the receiver and the amplified K9AY about 6-8 dB.

Spectrum analyzer/tracking generator measurement of the preselector. Horizontal scale is 150 kHz/division and vertical is 10 dB/division.

Close in spectrum analyzer/tracking generator measurement of the preselector. Horizontal scale is 5 kHz/division and vertical is 10 dB/division.

10 kHz bandwidth is about as narrow as one can make this filter since inductor unloaded Q is the limiting factor. The inductors used here are wound on 61 type material which yields the best inductor Q of any of the ferrite materials. The Q of these coils measure approximately 300. Don't substitute different core materials or prewound coils unless you can verify that inductor Q approaches 300. A three-section, 365-pF broadcast variable is the main tuning element although only a small amount of its tuning range is actually used. These capacitors are fairly common items and are still available new from several sources.

In addition to the broadcast variable each section of the filter has an adjustable padding capacitor to allow easy alignment of the filter. 20 pF padders are shown but 50 pF would probably be a better choice. Inductors that are wound on ferrite cores typically have a wide spread of inductance for the same number of turns because of material inconsistency and the high ratio of inductance per turn. If you have a means of measuring inductance it should be possible to get them quite close and smaller padders would be adequate. If not, wider range padders will come in handy when aligning the three sections. All fixed value capacitors should be high Q types with silver mica or polystyrene preferred.

The filter was constructed in a homebrew enclosure made from double-sided pc board. This material can be bought inexpensively on ebay and makes excellent custom-sized, rf-tight enclosures. BNC connectors were used for the input and output connections. No attempt was made to shield individual filter sections as this particular broadcast variable didn't have shielding between the sections. Filter stopband performance may suffer slightly at the higher frequencies because of this but filter performance was deemed adequate as is.

Since the broadcast variable will tune the filter beyond the design frequency it should be noted that performance falls off outside the intended frequency range. On the low side insertion loss will increase and the filter bandwidth will become tighter. On the high side insertion loss will decrease and the the bandwidth will be wider.