Ham Radio Input Filter Question

[Veramacor]

Hello, I'm looking to come up with some input filters for a direct conversion receiver for Amateur bands :

80 meters (3.5Mhz)
40 meters (7 Mhz)
20 meters (14 Mhz I have in attached image)
10 meters (28 Mhz)

Attached is a filter that is going into a transistor then a SA602 mixer chip 
full article:
http://mightydevices.com/?p=227

My questions are:

1) Does this filter have a name?  ie Colpitts etc.

2) How can I calculate the required inductor and capacitor values for the above bands?

I've tried LTSpice, but not very confident in the program.

Basically trying to duplicate the above band pass filter for 80,40, and 10 meters.

Thanks in advance to anyone who can shed light!

Jeff

[w2aew]

It could be a Butterworth, Chebyshev, Elliptic (Caur), Bessel, etc. One of the nicest free passive LC filter design tools around is this one:
http://www.aade.com/filter.htm

[G0HZU]

Probably not what you want to hear but I can see a few issues with your circuit.

Basically, it can't work as drawn because the base of the transistor is tied to ground via the 3.9uH coil. There's a DC blocking cap missing somewhere.

Also, I'd question if this amplifier section has ever been tested because I would expect that the input capacitance would be very high (Miller Effect) so if you want my opinion I think the original designer added this amplifier to an existing circuit as an afterthought and drew it in (wrongly) and never tested it. The input capacitance of the transistor ought to severely detune the BPF that feeds it due to Miller Effect. It's a very poor amplifier design even if you ignore miller Effect.

So even if you add the DC blocking cap you will have problems with the high input capacitance of that BJT amplifier stage because the capacitance will be too large to absorb into the BPF capacitance. Especially at 20m and 10m. Also, the amplifier design is a poor one generally.

But the BPF topology looks like a Top C coupled BPF with a capacitively tapped input.

[G0HZU]

You shouldn't need an amplifier here if it feeds into an SA/NE612 mixer but if you really do want to fit one then maybe add another transistor and make a cascode amplifier. This would cure the input capacitance issue wrt Miller Effect.

[gregariz]

My questions are:

1) Does this filter have a name?  ie Colpitts etc.

2) How can I calculate the required inductor and capacitor values for the above bands?

As G0HZU has noted the designer of this has missed a low value series decoupling capacitor before the base of the amplifier circuit. It will likely be similar in size to c4 so that it doesn't load down the filter.

As far questions go:

1. This is a cascaded Bandpass filter (ie in this case 2 parallel tuned circuits set to the 20m band ~ 14,2MHz). If you could afford another inductor its better to avoid the low value of C4 and place a series tuned circuit in this leg in order to make a bandpass PI filter. This will improve insertion loss. Regardless what is presented works and is a commonly used version of this filter.

2. Use the parallel tuned circuit equation

ie 1/(2pi*sqrt(LC)

the equivalent series capacitance of C8&C13 = C12 = 27pF

C8 & C13 exist because the designer is trying to convert from the low antenna impedance (maybe 50Ohms) to the high impedance across the top of the tank circuit (maybe 10K -dependent on Q of your inductors). ie the reactance of C13 is much lower than C8 so provides better power transfer into the tuned circuit from the antenna. Most receivers of this variety use an input transformer instead (a Toko KANK was once common but these are now hard to get) but this arrangement also works.

IF you are going to use a small telescopic whip antenna to listen to local stations you would be better to remove/replace C8 and C13 with a single capacitor of the same value as C12 (27pF) and then attach the telescopic whip to the top of this capacitor (ie not the ground side). This would provide a better match than the low impedance point set up by the capacitive divider.

For a similar reason C4 is determined experimentally in order to not lot cross-load the tuned circuits and will be very small 1-10pF. As it is a very low value it will make this tuned circuit somewhat lossy so its generally experimentally adjusted as high as possible in order to maintain selectivity (reduced loading) while maintaining a decent insertion loss.

Probably the most widely built version of this class of receiver (they all look much the same) is G3RJV's sudden

https://archive.org/details/73-magazine-1991-10

[G0HZU]

I think the BPF as drawn is OK if the BJT wasn't included. Then it could be fed via a blocking cap direct into the SA602 with its 1500R // 3pF input impedance. I think the BPF would tune up somewhere close to the 20M band.

But I think the BJT will cause all kinds of issues even if the DC blocking cap were included and the values of all the components in the BPF and the blocking cap were optimised to suit the BJT. I think the designer was lazy here and just added the BJT amplifier and forgot to include the DC blocking cap and also forgot to readjust C4 and maybe a few other values in the BPF in order to get a decent response in the filter when the BJT was inserted.

The SA602 is a very weedy mixer in terms of input IP3 and adding maybe 20dB of gain ahead of it will cause overloading issues in an already weak mixer. The SA602 has a low noise figure and reasonable gain so there should be no need to fit an amplifier ahead of it.

[gregariz]

The SA602 is a very weedy mixer in terms of input IP3 and adding maybe 20dB of gain ahead of it will cause overloading issues in an already weak mixer. The SA602 has a low noise figure and reasonable gain so there should be no need to fit an amplifier ahead of it.

Yes I agree with that. ie the sudden and others put an attenuation pot on the front the NE602 for the reasons you mention. Its hard to know what was the reasoning of an amp with no gain control but it may have just been experimentally arrived at with some horrid aerial attached.

[T3sl4co1l]

The filter circuit is a coupled-resonator type.  Therefore, it has a bandpass characteristic.  The pass and skirts can be sharp or flat, with any of the standard named filter shapes.

The general form is, a resonator (LC parallel resonant) coupled with series capacitors of small value, or inductors of high value, or some E/M coupling due to proximity.  The capacitor divider at the input is another form of this, as is using a tapped coil.  Lots of ways to configure the same thing, so don't be intimidated by the formats.

This unfortunately needs Java, but (if you can get it running) shows how coupling works.
http://www.eee.bham.ac.uk/edt/filter/FilterDesignPage.html

Tim