HIGH PASS FILTER FOR OPTIMUM HF RECEPTION

[Roger]

can you please let me know what it is that you used to generate the signal so you cold test your filter?

The low-cost noise generators sold on ebay come in several different versions. Some of these do not have an output which is flat across the HF band unless they are modified (some owners have done that). In particular, the popular BG7TBL noise source has limitations and requires caution when testing wideband filters in the HF band. If used with a spectrum analyser or an RSP a correction table needs to be applied to the output numbers in order to compensate for this. You can read more about the limitations of the BG7TBL at https://owenduffy.net/blog/?p=11193

For narrowband filter testing or when using them with a return loss bridge (for antenna testing etc.) the lack of a flat noise output is not as much of a concern.

[glovisol]

If used with a spectrum analyser or an RSP a correction table needs to be applied to the output numbers in order to compensate for this. You can read more about the limitations of the BG7TBL at https://owenduffy.net/blog/?p=11193

Since I do not wish to open a digression (which would only confuse matters) on merits/demerits of low cost noise generators in this thread, only devoted to filters developed for front end protection of RSP receivers, I am opening elsewhere a dedicated thread on this subject.

I shall only add that the so much vituperated Noise Generator BG7TBL, as we shall see in the separate tread, is perfectly adequate to characterise HF low pass wideband filters, this being the implied question posed to me by illllim.

[glovisol]

HIGH IMPEDANCE FILTER CHEBYSHEV N=9 / 0.1 dB RIPPLE / 560 to 560 OHM #10

Filter practical construction - the multi-coil technique

As previusly stated, the main problem with the filter's construction was that the size of the coils developed in the prototype (#7) were too big to fit the entire filter in the die cast utility box. As nothing is really new, a technique I had seen in German surplus equipment of WW-2 came to my mind. You wind all coils on a single dielectric tube with a significant reduction is size and much easier winding of the enamelled copper wire. This is the multi-coil technique. I used a 20 mm dia. PVC electrical conduit pipe, made a drawing of the in-line positions of the 5 coils and drilled 10 holes of 2 mm diameter. The holes hold pins that are forced in and are the coil terminals.

To construct the filter, you push the first pin in, solder the coil beginning, then you wind the first coil. At the end of the winding you push another pin in and solder the end of the first coil and so on. Figures 1 and 2 give a better idea than one thousand words. The beauty of this is that, once you know the number of turns, you wind the entire filter in less than half an hour. As we have seen already, coil Q's are in the order of 150/200, so no problem with filter performance.

With this type of assembly all wiring is done on the tube itself: each coil beginning is connected to the next beginning wth bare copper wire while every coil end joins the next coil end trough a capacitor: this becomes very clear looking at the schematic (look at #7, several posts ago, to have an idea).

Another advantage of this technique is that, by replacing the wire jumpers with capacitors you have a truly balanced filter, equisymmetrical to ground. Figure 3 shows the complete filter being tested with Noise Generator/RSP1-A - RSP SAS Spectrum Analyser. Once you see that the filter is working, e.g. you did not make mistakes with coil turns, you glue the coils in place with cyanoacrylate liquid.

More in the next post.

[sdrom33]

Hi glovisol, thanks for your reply in the other thread on noise gen. In figure 1 I see many holes below in the tube but pins are above. You use these holes for what?

[glovisol]

Hi Sdrom33,

The holes below were made before I had a definite idea on how to proceed with the preparation of the tube. They are not necessary at all and serve no purpose.

[glovisol]

HIGH IMPEDANCE FILTER CHEBYSHEV N=9 / 0.1 dB RIPPLE / 560 to 560 OHM #10

Filter testing

With the proposed filter construction it is possible to completely assemble and test the filter out of the shielding box, as shown below. Bandstop and bandpass testing is done at 50 Ohm level, using the Noise Generator (50 Ohm) connected to the filter's input via a 50 to 560 Ohm wideband transformer and the RSP-1A ( 50 Ohm) driven by the RSP-SAS software and connected via a 560 to 50 Ohm transformer. The return loss /SWR measurements are done at the 560 Ohm to 560 Ohm level because the AA-30 Network Analyser can accept any impedance. So the generator is the AA-30 with an output adapter made with a double SMA male and a PC board female connected to the filter's input and the load is a simple 560 Ohm resistor. These setups are shown in the Figures 4 & 5 below.

[glovisol]

HIGH IMPEDANCE FILTER CHEBYSHEV N=9 / 0.1 dB RIPPLE / 560 to 560 OHM #10

Filter schematic and data

This high impedance filter is more critical than the 50 Ohm unit as far as components are concerned. Coil winding must be done very carefully to ensure the exact number of turns specified. Coil inductance measurement helps, if the facility is available. The same holds for capacitors and exact values can be selected from a batch of several. This is easy for the 100 pF standard value. For the 91 pF value use a parallel of selected 39+33+22 pF.

[glovisol]

HIGH IMPEDANCE FILTER CHEBYSHEV N=9 / 0.1 dB RIPPLE / 560 to 560 OHM #10

Completed filter

Below pictures of completed multi-coil filter mounted into the die cast aluminium enclosure and one test screen showing bandstop/bandpass transition with lid closed. All tests to be uploaded in next posts done with the built-in 560 to 1000 Ohm wideband low noise toroid transformer, e.g. the filter as shown is ready to be connected between a long wire antenna and the HI Z input of the RSP receiver.

More in the following posts.

[glovisol]

HIGH IMPEDANCE FILTER CHEBYSHEV N=9 / 0.1 dB RIPPLE / 560 to 560 OHM #10

Completed filter

More SWR plots

[glovisol]

HIGH IMPEDANCE FILTER CHEBYSHEV N=9 / 0.1 dB RIPPLE / 560 to 560 OHM #10

Amplitude/attenuation testing and discussion on the Noise Generator levelling shortcomings.

In the following screens we can see the amplitude performance of the completed High Pass filter and the explanation of the effect of the levelling imprecision of the Noise Generator, as mentioned in the relevant thread: "A close look at low cost Noise Generators".