LOW PASS FILTER FOR OPTIMUM LOW FREQUENCY RECEPTION

[glovisol]

MEDIUM WAVE REJECTION WITH A CAUER C 05 15 40 PASSIVE LOW PASS FILTER

With the wide input bandwith of the RSP class receivers, low frequency reception from 1 KHz to 500 KHz is very often ridden with intermodulation and higher order mixing products caused by very strong Medium Wave broadcasting signals 600 to 1800 KHz.

Continued next post for text amendment purposes.

[glovisol]

MEDIUM WAVE REJECTION WITH A CAUER C 05 15 40 PASSIVE LOW PASS FILTER

With Medium Wave signal strengths in the order of -50 to -40 dBm reception in the LF band, with the antenna connected to the HI Z input, most spurious products can be eliminated by using the LO IF setting and the MW notch filter, see:

https://www.sdrplay.com/community/viewt ... 742#p12742

and referenced white paper. For Broadcasting stations coming in stronger than -40 dBm (bringing the front end to a near overload condition and/or with RSP receivers without the MW notch filter) a simple solution is to use a sharp low pass filter between antenna and the HI Z input. The filter must pass frequencies up to 500 KHz, but attenuate frequencies above as much as possible. A good choice are elliptic type filters which have nulls in the stopband, close to the cutoff frequency and hence provide very sharp transitions. These filters are tabulated and all parameters, including theoretical response curves, can be calculated by an Excel spreadsheet, as shown in Figure 1.

The filter passes all frequencies, with an in-band ripple of 0.5 dB, up to 500 KHz, then provides an attenuation of 45 dB at the design frequency of 780 KHz, the sharp slope being caused by two nulls (notches) at 810 and at 1220 KHz. Overall theoretical response is shown in Figure 2.

More detail in the next post.

[glovisol]

MEDIUM WAVE REJECTION WITH A CAUER C 05 15 40 PASSIVE LOW PASS FILTER


ERRATA CORRIGE DATED APRIL 12, 2019
The plots shown below, while showing the correct frequency response, picture a high level of interference due to noise radiated by the switch mode PSU of the Noise generator. Clean plots, uploaded today, are shown in the last post of this thread.

Uploaded below theoretical performance of the low pass filter in the passband, Figure 4 and actual filter measurement with noise generator / SDR Spectrum Analyser, with an RSP-1A working as Spectrum Analyser front end, Figure 5. At very low frequencies the most critical component is the balanced ferrite transformer, whose reactance decreases more and more as frequency goes down. The 50 to 1000 Ohm ferrite transformer used for the measurement has a secondary inductance of 0.7 mH @ 100 KHz which is on the low side (should be over 3 mH at least). This shows in the roll-off at very low frequency.

[glovisol]

MEDIUM WAVE REJECTION WITH A CAUER C 05 15 40 PASSIVE LOW PASS FILTER - DIFFERENCE BETWEEN TESTING AND OPERATING CONFIGURATION

In order to be able to test the filter with just one transformer, it was necessary to calculate it at an impedance level of 1 KOhm and to use a 50 Ohm (noise generator side) to 1 KOhm (filter side) transformer, as shown in Figure 6. With this test configuration the transformer is before the filter, while in the receiver operating configuration the antenna is connected to the filter directly (unbalanced configuration) and the filter's output is connected to the P and N balanced receiver input by means of a balanced transformer.

So in the receiver operating configuration we must design:
(a) the filter for an input/output impedance of 500 Ohm to accommodate the long wire/Beverage antenna output impedance and
(b) transformer for PRI = 500 Ohm (filter output) and SEC = 1 KOhm (HI Z RSP receiver input P & N).

The new filter calculation with Z = 500 Ohm is shown in Figure 7. A keen observer would notice that, halving the impedance, capacitor values have doubled and inductance values halved.

Final filter configuration in next post.

[glovisol]

MEDIUM WAVE REJECTION WITH A CAUER C 05 15 40 PASSIVE LOW PASS FILTER - TRANSFORMER (BALUN) DESIGN

The transformer is a critical and most important component of the receiver system. It insulates the receiver's delicate front end from dangerous transients and minimizes noise pick up. The transformer must match the 500 Ohm filter impedance to the RSP receiver balanced input, Z = 1 KOhm. If the desired VLF coverage range is 100 - 500 KHz, the secondary must have an inductive reactance Xl @ 100 KHz of at least 2 Kohm (4x) therefore:

L = Xl/(2*Pi*f) = 2000/(2*3.14*100*10^3) = 1/(314) = 0.0032 H = 3.2*10^6 nH (nanohenries) which is a substantial inductance requiring a relatively large ferrite core. The Siemens Siferrrit E core type M30, with no air gap, using ferrite material N30 is suitable. The nominal Al of this core is 2500 nH (+30/-20%), but I found Al=2350 measuring the material I had in the bin. To calculate number of turns N:

Ns = SQRT (L/Al) = SQRT(3.2*10^6/2350) = 37 T

Primary turns Np = 37/SQRT(1000/500) = 37/1.41 = 26 T

For a 50 Ohm primary we would have : Np = 37/SQRT(1000/50) = 37/4.47 = 8T

ENAMELLED COPPER WIRE O.3 MM DIAMETER ALL WINDINGS.

Transformer & cores shown in Figure 8.

[glovisol]

MEDIUM WAVE REJECTION WITH A CAUER C 05 15 40 PASSIVE LOW PASS FILTER - OPERATIONAL FILTER DETAILS

The filter schematic, with component values for Z=500 Ohm, is shown in Figure 9. Using a generator and an oscilloscope it is possible to adjust C2 and C4 for obtaining broad nulls at 810 and 1220 KHz, however, if the component values are accurately measured prior to assembly, no adjustment is necessary. The filter must be enclosed in a metal box, otherwise the components will pick up strong broadcast signals directly. As shown in Figure 9, ground must be connected to the box, as well as the low leads of the filter components, C2, C4 and primary return of T1. This ground must be connected to terminal GND of the RSP receiver. The two secondary leads of T1 go to terminals P and N of the RSP receiver. Care must be taken to keep leads to the balanced input as short as possible, see Figure 10. Figure 11 shows filter construction.

[glovisol]

MEDIUM WAVE REJECTION WITH A CAUER C 05 15 40 PASSIVE LOW PASS FILTER - FILTER PERFORMANCE

Filter works as calculated and predicted. Attenuation measured on the very strong Local RAI broadcast station at 900 KHz, coming in at -34 dBm (causing overload so that RF gain had to be kept at minimum) is attenuated to -87 dBm, e.g filter attenuation at this frequency is -53 dB. Filter action is shown in Figures 11,12 & 13. With the filter inserted no intermodulation and/or higher order products can be detected in the 1KHz to 600 KHz frequency band, even in zero IF mode.

Coil construction in the next post.

[glovisol]

MEDIUM WAVE REJECTION WITH A CAUER C 05 15 40 PASSIVE LOW PASS FILTER - COILS & CONCLUSION


Coil data is as follows: NOTE all coils use enamelled copper wire 0.3 mm diameter.

L2 coil data added on 15/02/2019

L1: 165 uH - 4 1/2 turns on Amidon binocular core material 73 type: BN-73-202. Coil wound around one external wall, the 1/2 turn by inserting the wire terminal in the other hole.
L2: 22 uH - 20 turns on Micrometals T 16-26, yellow-white 16 mm dia. toroid.
L3: 264 uH - 6 turns on Amidon binocular core material 73 type: BN-73-202. Coil wound around one external wall.
L4: 63uH - 3 turns on Amidon binocular core material 73 type: BN-73-202. Coil wound around one external wall.
L5: 134 uH - 4 turns on Amidon binocular core material 73 type: BN-73-202. Coil wound around one external wall.
Exact value capacitors impossible to find: they must be selected and/or combination made with more capacitors in parallel. Capacitors should be mica type for excellent temperature stability. I selected the maximum coil prototype, instead of the maximum capacitor prototype in order to minimise the number of capacitors used.

To conclude, a 50 dB lowpass filter should always be used when operating the RSP receiver at frequencies below the MW broadcasting band, this is because, in general, the MW stations and signals are the strongest that are usually present at the antenna terminals. This filter provides the necessary attenuation to avoid receiver overload and spurious generation in LF and VLF bands, is mandatory with the original RSP-1 /RSP-1A and RSP-2 Spectrum Processors and very useful with the other RSP class receivers, allowing to use the built-in MW notch filter for additional suppression.

REFERENCES:
- Handbook zum filterentwurf - Rudolf Saal unter Mitarbeit von Walter Entenmann - AEG-TELEFUNKEN 1979.
- Ferrites and Hardware - Data Book 1983/84 - Siemens.

[someYguy]

Very good and very well done! Thank you!

For about the last 3 years, I've been using a surplus commercial 500khz lowpass filter for NDB searching etc. and it works wonders - vastly improves the RSP1 here at those frequencies.

[glovisol]

After reading the excellent thread:

https://www.sdrplay.com/community/viewt ... 784#p12784

I revised the filter schematic to emphasise the earth/ground connections earth stake to box, internal common ground returns and proper connections to the RSP receiver.