SDRplay Community Forum

This post to allow future amendments. Go to next post.

The very high power FM Broadcast signals in my area overload my RSPduo front end to the point of saturation and reception of Airband, 2 m Amateur Band and Marine band signals is obstacled by overload noise. The RSPduo internal notch filter, while very efficient in removing FM band overload, still provides significant attenuation outside the 88-108 MHz spectrum. For this reason and spurred by Steve Andrews' new Ver. 1.0 Spectrum Analyser sofware, I started looking at VHF Band-Stop filtering and the first result is shown below.

This band-stop filter is derived from a Chebyshev prototype and has the following characteristics:
N= 5
Passband ripple: 2 dB
Zin = Zout = 50 Ohm
Low cutoff frequency = 83 MHz
High cutoff frequency = 113 MHz

Performance of the breadboard prototype is shown below. Insertion loss outside the stopband is negligible and this allows troublefree scanning of the entire Airband, with no danger of front end overload and no incoming signal attenuation.

Further details in future posts.

Shown below is the filter prototype built on an Arduino breadboard. At VHF/UHF filter layout is very important. Here below we can see that the prototype is good just about to 220 MHz, then insertion loss starts to increase: this problem will be eliminated with proper layout, when enclosed in its shielding box.

Also coils look like an horror story: calculation yelded too many turns and they must be re-sized in the next prototype.

Tuning of an N=5 filter can be quite critical and tricky, so I resorted to a simpler, N=3 Chebyshev which is much easier to set up and tune. This filter yelds 25 dB attenuation at 88 & 108 MHz and 35 db at 98 MHz, while taking negligible insertion loss. If more FM band rejection is needed, two such filters in cascade can be used and I am going to test this solution next.

For measurement and tuning it is advisable to insert a 10 db attenuator between Noise Generator and filter under test and this, once again, proves the fact that the higher the output of the Noise Generator, the easier and better the measurement.

Finally the "prototype layout shortcoming" mentione in the previous post is not real, but caused by an unexplained drop in sensitivity of the Spectrum Analyser at higher frequencies, as shown in the pic. below.

The N=3 prototype filter works so well and tunes so easily it is unnecessary to use a more complex and more critical Chebyshev design type: this is the reason why the N=5 filter was discarded. If more stop-band attenuation is required, two N=3 cells can be assembled in series, with a 3 dB resistive attenuator in between. The attenuator isolates the two cells and makes filter tuning (done by delicately pruning the coils) quick and easy. For easiest tuning, the two sections may be pruned separately and then joined together by soldering in the attenuator resistors.

The final prototype using two N=3 cells plus the attenuator, its performance and comparison with the single cell filter are shown below.

In the next posts filter details will be given.

Here below Band-Stop filter data. This filter is very easy to build and tune and is very useful in completely removing RSP Spectrum Processor overload caused by High Power FM band 88-108 transmitter signals, while allowing full receiver sensitivity in the nearby Airband frequency range.

I am waiting for the die cast aluminium shielding box and will upload final results after having shielded the filter.

3 dB ATTENUATOR ELIMINATION - FILTER VER. 2

The 3 dB attenuator loss can be eliminated by replacing it with a simple coil. Tuning is slightly more critical, but insertion loss improvement is very significant. New schematic uploaded below,

COMPARISON BETWEEN DESIGNED FILTER AND COMMERCIAL FILTER

The uploaded picture compares the Chebyshev 2xN=3 designed and developed filter to a commercial filter readily available on the market.

Insertion loss @ 1 dB, ultimate attenuation @ 40 dB and frequency response on the low side @ 78 MHz are the same for both filters. The designed filter has some advantage on the high frequency side and works better for the Airband. At 118 MHs the commercial filter still loses 12 dB, while the Chebyshev loses only 4 dB, while both providing the same attenuation of 40 dB @ 108 MHz.

The commercial filter, built with surface mount components, has the advantage of using approx. 1/4 of the physical volume taken up by the hand made filter.

A lot of interference in these filters as well. You sure nothing wrong with the measurement setup?

Sdrom33 you posed a very difficult question. Now, the plots for the filter I built of course have strong FM stations coming in, because, as I have written, my prototype filter was NOT shielded when measurements were done, as I was waiting for its die cast box, which eventually arrived on Friday and I shall be able to assemble and measure it propely shielded next week.

The fact is that also the commercial filter shows interference within the Analyser plot and this filter seems to be well built in its metal enclosure, so the question is: how are these signals getting in? They cannot get in tru the filter, because we see an attenuation of 40 dB in the measurement plot, so they must come in AFTER the filter. I eliminated the interconnecting cable between filter output & RSP-1A input and replaced it with a double male (very short adapter) and then re-tested. The FM interference was still there, inside the plot. This made me suspicious.

I removed the double male and the filter from the RSP-1A and terminated the input with a coaxial SMA 50 Ohm load and switched SDRuno (Ver.1.3 tuned to 96 MHz) on. The result is shown in the uploaded pic: some FM signals are so strong they go through the shielding box of the RSP-1A. The signal at 98 MHz comes in at -93 dBm!! If I connect any metal object to the ground side of the SMA socket, the feedthru signals come in even stronger.

So we must understand that, altough excellent, shielding of the RSP-1A cannot be perfect and very strong signals at VHF and beyond may come in. The important thing is to be aware of this limitation.