BUILDING SELECTIVITY PROTECTION INTO THE RSP FRONT END IN LF/HF

General discussions
glovisol
Posts: 662
Joined: Thu May 10, 2018 6:42 pm
Location: Piedmont, Italy

BUILDING SELECTIVITY PROTECTION INTO THE RSP FRONT END IN LF/HF

Post by glovisol » Thu Jun 27, 2019 3:37 pm

BUILDING SELECTIVITY PROTECTION INTO THE RSP FRONT END IN LF/HF

Go to next post.

Reason: No reason

glovisol
Posts: 662
Joined: Thu May 10, 2018 6:42 pm
Location: Piedmont, Italy

Re: BUILDING SELECTIVITY PROTECTION INTO THE RSP FRONT END IN LF/HF

Post by glovisol » Thu Jun 27, 2019 3:42 pm

BUILDING SELECTIVITY PROTECTION INTO THE RSP FRONT END IN LF/HF


1. GENERAL
The RSP Processor is, by definition, a wide band reception device, sporting up to 10 MHz of accessible bandwith. This characteristic is essential when looking at spectrum monitoring and it is not by chance that very many RSP’s are used as Panadaptors to complement the operation of narrow band transceivers. But the RSP is also an excellent communications receiver and in this mode of operation, the reception of weak signals in presence of strong interference is best accomplished by increasing the front end selectivity by some sort of pre-selection.

An input pre-selector is a low pass, high pass, band reject or band pass filter. A band pass filter attenuates all frequencies above and below the received signal as much as possible to minimize the action of noise and of adjacent signals. RSP class processors are designed for maximum receiver bandwith and this feature allows the operator to see at a glance all stations transmitting in the selected frequency range. Wideband operation is a special feature of these processors, but has a significant disadvantage in low noise weak signal reception.
What happens is that in a wideband receiver AGC reduces RF/IF gain by measuring the level of received signals. If in the same wideband frequency range weak signals and strong signals are present at the same time, the AGC will automatically reduce the gain, pushing the weak signals down, below the received noise level. The pre-selector prevents this negative action by creating a window as narrow as possible around the wanted signal and thus attenuating all strong signals, nearby and far away in frequency. Thus, theoretically, the AGC action is controlled by the level of the wanted received signal only.

With single tuner RSP’s we are trading bandwith for sensitivity, thereby defeating their wideband feature, but with double tuner RSP’s, like the RSPduo, we can have the best of both worlds, by using TUNER #1 protected by a preselector as a low noise receiver and TUNER #2 wideband, as a built-in panadaptor.


L/C preselectors are of five kinds:

(a) Wide Lowpass filters, eliminating all frequencies below the cutoff frequency. Two examples, for improving LF reception by attenuating the strong signals of the MW Broadcast band have been published in this Forum.
(b) Wide Highpass filters, eliminating all frequencies above the cutoff frequency. Three examples have been published in this forum for improving reception in the 160 m band and above by attenuating the strong signals of the MW Broadcast band. Also have been published in this Forum.
(c) Wide Bandstop fixed bandwith/frequency filters. For example the Airband starts at 118 Mhz and below it, in the 88 – 108 MHz frequency range, extremely strong FM Broadcast stations tend to overload the RSP front end, causing a very high noise level. The design for an effective 88 – 108 MHz bandstop filter, with complete interference elimination, has already been presented in this Forum
(d) Wide Bandpass fixed bandwith/frequency filters covering a limited frequency band. For instance the 40 m band extends from 7,000 to 7,200 KHz and a wide preselector would usefully protect the entire band from adjacent very strong signals. In the 40 m band, above 7,200 KHz, extremely strong AM Broadcast signals are present, sometimes as strong as -35 dBm at night. These not only can take control of AGC action, but also bring about ADC overload, compelling the operator to manually reduce front end gain. A wide filter attenuating these unwanted signals would allow optimum reception of weak ham signals below 7,200 KHz at maximum gain.
(e) Narrowband tuneable Bandpass filters covering a band as narrow as possible. Here the sharpest selectivity that can be accommodated is limited by the physical characteristics of the inductor available. This subject and related information, with explanation on the significance of unloaded Quality Factor Qu, has already been discussed here:
https://www.sdrplay.com/community/viewt ... 6&start=10

More detail in the next post.

Reason: No reason

glovisol
Posts: 662
Joined: Thu May 10, 2018 6:42 pm
Location: Piedmont, Italy

Re: BUILDING SELECTIVITY PROTECTION INTO THE RSP FRONT END IN LF/HF

Post by glovisol » Mon Jul 01, 2019 7:46 am

2.1 NARROW BAND PRE-SELECTORS - TUNEABLE BANDPASS FILTERS @ 3.8 MHz

In the frequency range 3500 - 3800 KHz, Iron Powder material N2 toroids, have measured ( by me) Qu's of 400 or higher. We must now consider the problem of the tuning capacitor. Using an inductance of 50 uH, easily obtainable with a Micrometals core T-80-2 (20 mm external diameter) tuning capacitance values are as follows:

3,500 KHz - L=50 uH - C=41 pF
3,650 KHz - L=50 uH - C=38 pF
3,850 KHz - L=50 uH - C=35 pF

Thus the total capacitance variation is of 6 pF approx. which can be obtained with varicap diodes or with a hand made cylindrical double tuned capacitor.

Even the best varicap diode (BB104 / BB204) will realistically have a Qu= 100, at most. Our preselector parameters are therefore as follows:

Qu=100
Ql=20
Insertion loss Il=2 dB/single tuned & 2.8 dB/double tuned.
Coefficient of coupling p=1

And calculated responses are shown below, RED for single tuned and BLUE for double tuned. From these responses we see that past the 80 m band edges (+/- 200 KHz) the attenuation is +/- 10 dB and goes to 30 dB @ +/- 600 KHz. In fact this response is so large we could just tune the pre-selector at mid-band and get full 80 m coverage and this could be a choice for some, by using fixed capacitors and a small air trimmer with lower Qu coils.

But we can do much better with variable capacitor tuning.
Attachments
3.65 MHz pre-selector varicap tuned 1.png
3.65 Varicap pre-selector narrow band frequency response
3.65 MHz pre-selector varicap tuned 1.png (17.66 KiB) Viewed 4327 times
3.65 MHz pre-selector varicap tuned 2.png
3.65 Varicap pre-selector wide band frequency response
3.65 MHz pre-selector varicap tuned 2.png (21.83 KiB) Viewed 4327 times

Reason: No reason

glovisol
Posts: 662
Joined: Thu May 10, 2018 6:42 pm
Location: Piedmont, Italy

Re: BUILDING SELECTIVITY PROTECTION INTO THE RSP FRONT END IN LF/HF

Post by glovisol » Mon Jul 01, 2019 8:05 am

2.2. NARROW BAND PRE-SELECTORS - TUNEABLE BANDPASS FILTERS @ 3.8 MHz

Using a double gang easily self made variable capacitor, our system Qu rises to 400 or better and the new parameters are as follows:

Qu=400
Ql=80
Insertion loss Il=2 dB/single tuned & 2.8 dB/double tuned.
Coefficient of coupling p=1

The calculated responses are shown below, RED for single tuned and BLUE for double tuned, in the two last pics. For the same insertion loss, the capacitor tuned pre-selector affords a significantly better selectivity (35 dB @ +/- 200 KHz) offering real protection to our RSP front end. By changing the coupling coefficient p (easily done by altering the value of the coupling capacitor as will be shown in the practical filter implementation next) narrow band selectivity can be changed at will.

The first pic shows the result of cascading two double tuned sections using a low noise amplifier in between, something I am going to test this coming fall.
Attachments
3.65 MHz  variable cap pre-selector 2 sections.png
This is what happens with two HI Q sections in series with a low noise amplifier in between
3.65 MHz variable cap pre-selector 2 sections.png (18.7 KiB) Viewed 4321 times
3.65 MHz pre-selector variable capacitor 1.png
3.65 variable capacitor pre-selector narrow band frequency response
3.65 MHz pre-selector variable capacitor 1.png (18.52 KiB) Viewed 4324 times
3.65 MHz pre-selector variable capacitor 2.png
3.65 variable capacitor pre-selector wide band frequency response
3.65 MHz pre-selector variable capacitor 2.png (24.48 KiB) Viewed 4324 times

Reason: No reason

glovisol
Posts: 662
Joined: Thu May 10, 2018 6:42 pm
Location: Piedmont, Italy

Re: BUILDING SELECTIVITY PROTECTION INTO THE RSP FRONT END IN LF/HF

Post by glovisol » Tue Jul 02, 2019 4:17 pm

3.1 WIDE BAND PRE-SELECTORS - FIXED BANDPASS FILTERS @ 3.8 MHz

In this post we are going to look at another alternative: build a fixed pre-selector protecting the entire 80 band, 3500 to 3800 KHz. The design procedure is based on the transformation of a Cauer Low Pass prototype into its bandpass equivalent. In the example shown I give all the formulas for the simple calculations involved, so that anyone interested can design his own with Excel or with a hand calculator.
1. Cauer N=3 Bpass filter data.png
FIGURE 1 - Frequency trasformation of band-pass data into low-pass data
1. Cauer N=3 Bpass filter data.png (16.7 KiB) Viewed 4173 times
Since we know the center frequency and the passband edge frequencies, we need to put an approximate figure for the band-stop frequencies and find, by calculation, the EQUIVALENT LOWPASS PROTOTYPE. Looking at Figure 1 the impedance level set at 3 KOhm may seem a bit extravagant, but later on we shall see the reason for this choice. What we wish to obtain is a filter as plotted in Figure 2.
4. Cauer N=3 Bpass plot.png
FIGURE 2 - CAUER CC 03 50 30 Bandpass plot
4. Cauer N=3 Bpass plot.png (20.04 KiB) Viewed 4173 times
So our filter must pass with no attenuation from 3500 to 3800 KHz and attenuate at least 35 dB @ 3300 KHz and below and @ 3900 KHz and above. From these data we compute the normalized cutoff of the lowpass prototype, which is 2. We choose a passband ripple of 1.25 dB, equivalent to 50%, (the higher the ripple, the higher the stopband attenuaton at the price of a small insertion loss) and finally we compute the required Qu of our coils and are happy to see it is only 243! This means that the filter can be realized with material N2 toroids for super performance, but also with modest size air coils. So this is surely a big advantage over narrow band filtering.

Low pass filter normalized component data is shown in Figure 3 below. This is tabulated data easily avaliable, as we shall see next. In the following post we shall see how to transform this data into the actual component values of our filter.
Attachments
2. Cauer N=3 prototype data.png
FIGURE 3 - CAUER CC 03 50 30 normalized component data
2. Cauer N=3 prototype data.png (5.58 KiB) Viewed 4173 times

Reason: No reason

glovisol
Posts: 662
Joined: Thu May 10, 2018 6:42 pm
Location: Piedmont, Italy

Re: BUILDING SELECTIVITY PROTECTION INTO THE RSP FRONT END IN LF/HF

Post by glovisol » Fri Jul 05, 2019 8:57 am

3.2 WIDE BAND PRE-SELECTORS - FIXED BANDPASS FILTERS @ 3.8 MHz

It should be evident that all data & calculations shown in the previous Figures 1 to 3 come from an Excel spreadsheet. Unfortunately I can only post pics of Excel, as the Forum software does not allow me to post the Excel file, otherwise all Forum members could be able to do these designs using the file directly, instead of writing their own. In all formulas SQRT() or RADQ() mean square root of.

This filter has more components than the simple bandpass double tuned narrowband filter, in fact is uses 4 coils, but the required Qu is significantly lower, at 243, for the same or lower insrtion loss.

We come now to the filter impedance level. The lowpass prototype and its corresponding bandpass filter can be de-normalized at any impedance level, as set in Figure 1. But we must recall that we must try and use the best Qu possible and by experiment we have seen that the best Qu values are in the range of approx. 2 to 50 uH, whether we use Powder Iron N2 material or reasonable size air coils. Now this inductance range and reasonable capacitance values, with these filter prototypes, can be obtained at a relatively high impedance level, which is also easily scalable to lower values by using link coupling. The advantage of having all calculations on one Excel spreadsheet is that we can quickly change parameters and immediately see the effect.

With links we can accommodate 50 to 50 Ohm, or, for instance, 450 Ohm (Beverage antenna) to 1 KOhm HI Z RSP balanced input. There is no problem of leakage inductance here, limiting the coupling action, for the simple reason that the filter is narrow band. If we were implementing a very wideband filter, then the leakage inductance variation over frequency would distort the filter response, as we have seen in the highpass filter thread.

Finally, the frequency response plot shown in Figure 2 is exactly that of the bandpass filter: I presented it before, just to give the ida of what we were hunting for. The two side nulls shown in Figure 2 are of course at the resonant frequencies of the two parallel L/C circuits placed in series, at 3322 and 4003 KHz.
Attachments
Filter transformation schematic.jpg
FIGURE 4. Prototype and transformer bandpass filter schematics
Filter transformation schematic.jpg (58.92 KiB) Viewed 4046 times
3. Cauer N=3 component computation.png
FIGURE 5. Bandpass filter components computation
3. Cauer N=3 component computation.png (15.07 KiB) Viewed 4046 times

Reason: No reason

glovisol
Posts: 662
Joined: Thu May 10, 2018 6:42 pm
Location: Piedmont, Italy

Re: BUILDING SELECTIVITY PROTECTION INTO THE RSP FRONT END IN LF/HF

Post by glovisol » Sat Jul 06, 2019 3:08 pm

3.3 WIDE BAND PRE-SELECTORS - FIXED BANDPASS FILTERS @ 3.8 MHz - PRACTICAL IMPLEMENTATION

Figure 5 below shows the detailed practical implementation of the 3.65 MHz pre-selector, with air core coil technology, compared to Powder Iron toroid technology. At this frequency, due to the higher Qu's, Powder Iron is the winner and the tested filter provides the calculated performance with less than 2 dB insertion loss, compared to the 4.5 dB loss of the air core coils. Because of the presence of two coupling links with the inherent impedance matching and common mode isolation capability, no extra transformers and/or baluns are necessary. If care is taken in link positioning on the coils to minimise parasitic capacitance, the filter will also provide excellent rejection of locally generated noise.

It must be stressed at this point that this pre-selector, if properly constructed in a well shielded box, will let the RSP "see" its pass-band to the exclusion of any other signal from zero to al least 30 MHz. At higher frequencies the limitations of the parallel resonant circuits placed in series become evident and their parasitic capacitances will make higher frequency signals "jump" across. If desired this limitation can be eliminated by placing a very simple lowpass filter (with a cutoff around 20 Mhz) in series with the pre-selector.

To demonstrate the flexibility of application of modern filter synthesis, a low impedance (25 Ohm) filter, as opposed to the 3 KOhm impedance of the presente unit, will be shown next.
Attachments
Pre-selector Wide Band 3.65 MHz schematic.jpg
FIGURE 5. Practical wide band pre-selector filter implementation
Pre-selector Wide Band 3.65 MHz schematic.jpg (115.24 KiB) Viewed 3949 times

Reason: No reason

glovisol
Posts: 662
Joined: Thu May 10, 2018 6:42 pm
Location: Piedmont, Italy

Re: BUILDING SELECTIVITY PROTECTION INTO THE RSP FRONT END IN LF/HF

Post by glovisol » Sun Jul 07, 2019 6:25 am

3.4 WIDE BAND PRE-SELECTORS - FIXED BANDPASS FILTERS @ 3.8 MHz - DUAL PROTOTYPE PRACTICAL IMPLEMENTATION

The Cauer low pass filter prototype comes in two versions: the shunt input version shown in Figure 4 and the series input version, where a shunt capacitor is replaced by a series inductor and the parallel tuned L/C circuit placed in series is replaced by a series tuned circuit placed in shunt, as shown below.
Dual prototype.jpg
Dual prototype.jpg (6.94 KiB) Viewed 3889 times
By using the dual prototype, reasonable and advantageous values for coils and capacitors are obtained at a low impedance level (25 Ohm) as shown in Figure 6, below. The series input version has the advantage of being low impedance, thus shielding requirements are less critical. Furthermore there is less chance of leaks at high frequencies, because the effect of parasitic capacitances is less marked with this topology.

The disadvantage of this filter is that additional transformers are needed to convert the 25 Ohm impedance to the required levels. Figure 6 shows transformers based on ferrite binocular core BN-73-202. These transformers are good for lab filter testing, but are unnecessarily wideband for the job and additionally introduce a relatively high level of parasitic capacitance with its unwanted common mode noise leakage. Better low capacitance transformers can be implemented with small toroids of 73 materal wound with non-overlapping primary and secondary windings.
Attachments
Pre-selector Wide band 3.65 MHz schematic (2).jpg
FIGURE 6. Practical wide band pre-selector filter implementation, dual prototype
Pre-selector Wide band 3.65 MHz schematic (2).jpg (129.01 KiB) Viewed 3889 times

Reason: No reason

glovisol
Posts: 662
Joined: Thu May 10, 2018 6:42 pm
Location: Piedmont, Italy

Re: BUILDING SELECTIVITY PROTECTION INTO THE RSP FRONT END IN LF/HF

Post by glovisol » Mon Jul 08, 2019 11:49 am

3.5 WIDE BAND PRE-SELECTORS - LOW PASS FILTER TABULAR DATA

Low pass filter tabulated data is available from many sources. Uploaded below the C 03 50 table, taken from Reference 1, the most complete I have seen, where the data used for the 3.65 bandpass filter has been taken from. Partial data can also be obtained by carefully looking on Internet sites on Filter Synthesis.

I apologise for not being able to present laboratory and practical test data as usual, but I am on holiday at present, far from the lab, so I can only carry on theoretical and calculation work for the time being.

REFERENCES:

1. Handbuch zum Filterentwurf by Rudolf Saal, AEG-TELEFUNKEN, Berlin, 1979.
2. Simplified modern filter design, by Philip R. Geffe, John F. Rider Publisher, New York, 1963.
3. Handbook of Filter Synthesis, by Anatol I. Zverev, John Wiley & Sons, Inc., Hoboken, N.J., 1967
4. Normalized filter design tables, taken from Filter design handbook and downloadable as PDF here:
https://www.google.com/search?q=normali ... e&ie=UTF-8
Attachments
SAAL Filter table 1.jpg
SAAL Filter table 1.jpg (115.02 KiB) Viewed 3788 times
SAAL Filter table 2.jpg
SAAL Filter table 2.jpg (142.74 KiB) Viewed 3788 times

Reason: No reason

mike0agner
Posts: 23
Joined: Sat Jun 01, 2019 2:42 am

Re: BUILDING SELECTIVITY PROTECTION INTO THE RSP FRONT END IN LF/HF

Post by mike0agner » Sat Jul 20, 2019 9:55 pm

Here are 2 passive preselectors - the first won't work below 1.6 Mhz, the other is much more broad banded...

http://www.mfjenterprises.com/Product.p ... d=MFJ-1046

http://www.crosscountrywireless.net/preselector.htm

Mike

Reason: No reason

Post Reply