[Boatanchors] Re: IF Filter testing...

[Carl Huether]

I'll add my .02 here also since I typically spend a good part of a day in
front of network analyzers.

When testing any filter use a 3-5 dB pad at both input and outputs to
minimize any impedence irregularities in the test equipment. I am assuming
known 50 Ohm filters in this instance. This is very important if you are
using a receiver as the frequency selective voltmeter.

If the filter impedence is known and not 50 Ohms you can use a resistive
network known as a minimum loss pad to transform to 50 Ohms.

At all times keep the input signal level to the minimum needed to map a
curve. Filters can be driven too hard and they will produce their own
spurious; mechanical in particular are easily destroyed by overdriving.

If you are trying to see skirts 80-100dB down keep in mind that only the
better instruments have that range capability.

Carl
KM1H


----- Original Message -----
From: "Barry L. Ornitz" <ornitz@tricon.net>
To: <boatanchors@mailman.qth.net>
Cc: <jordana@nucleus.com>; <Barockteer@aol.com>
Sent: Tuesday, December 17, 2002 6:59 PM
Subject: [Boatanchors] Re: IF Filter testing...


> Jordan Arndt, VE2SWL, asked:
>
> > Hi.. is there a simple method of testing IF filters for
> > throughput and resonant frequency..??
> > Specifically mechanical filters, ceramic filters etc..??
> > Anyone out there familiar with out of circuit testing of
> > these...? Please  any advice would be greatly appreciated...
> > 73 de Jordan...
>
> Tony, K1KP, presented an excellent response but did not
> mention one important thing.
>
> In a completely unknown filter, finding the proper frequency
> can be difficult.  Mechanical and crystal filters shown a very
> large number of spurious responses outside the passband of the
> filter.  Finding the correct response is sometimes difficult.
>
> I had access to a Hewlett-Packard 3577A network analyzer (5 Hz
> to 200 MHz) when I tested about a dozen "hamfest specials".
> The only ones that were trivial to measure were those marked
> with their frequency - and one special unit from a NMR machine
> (4 kHz center frequency, passband 2 Hz).  If you scan the
> filters over a wide frequency range, you have to tune VERY
> slowly or you will miss the response.  Even once you know the
> center frequency, you have to scan very slowly in frequency to
> get the real response curve. [Remember that what you are
> exciting the filter with is really a frequency modulated
> signal, and FM signals have an infinite number of
> sidebands spaced at the modulating frequency (scan rate).]
>
> Determining what is a real and what is a spurious response can
> be difficult.  Usually the desired response characteristic
> will have the lowest attenuation and a passband shape that is
> most "square."  The filter terminations are important here too
> since if the filter is not terminated properly, you will not
> see its true response.  On new filters, the manufacturer will
> often provide data on the location of the spurious responses
> in frequency, but with an unknown hamfest filter - who knows?
>
> With the Collins mechanical filters, the center frequencies
> tended to be relatively standard, 250, 300, 450, 455, and 500
> kHz - except the filters designed for telephone services where
> a large number of frequencies below 100 kHz were used.  For
> crystal filters, the best places to look for responses are
> between 2 and 6 MHz, and around 9.0, 10.7 MHz and 21.4 MHz.
>
> If the filters are currently installed in surplus equipment,
> the resonant frequencies of any transformers or tuned circuits
> used with the filters are a good hint of the filter frequency.
>
>         73,  Barry L. Ornitz     WA4VZQ     ornitz@tricon.net
>
> _______________________________________________
> Boatanchors mailing list
> Boatanchors@mailman.qth.net
> http://mailman.qth.net/mailman/listinfo/boatanchors
>