About bandpass filtering and filters to improve noise?

Discussion in 'General Technical Questions and Answers' started by KE0EYJ, Jul 24, 2017.

ad: L-HROutlet
ad: l-rl
ad: Subscribe
ad: abrind-2
ad: Left-2
ad: L-MFJ
ad: Left-3
  1. N1OOQ

    N1OOQ Ham Member QRZ Page

    In the olden days, the circuits within your receiver were not all that linear - they generated a lot of intermodulation (IMD) products. This is what everyone was concerned with from about the mid 1970s to about the 90s... "IMD dynamic range", often just called "dynamic range".

    Defined, it is the range from the minimum discernable signal (MDS - the smallest signal your receiver can pull out of the thermal noise background) to the largest signal your receiver can process without creating IMD products that act like real signals (i.e, the IMD "junk" begins to rise above the noise floor at this high signal level).

    Back in the day, the difference between the two was smaller than it is now... receivers were plenty sensitive (nice, low MDS), but the IMD junk started causing problems at a not-very-high signal level. Today's receivers are something like 20 to 50 dB (about 3 to 8 S-units!) less sensitive to this. This is good - it means you hear less stuff that isn't really there, and loud nearby signals don't stomp on your receiver and make it go deaf when TC2W is responding to your call.

    These days we have synthesized receivers... the oscillators are all generated by digital circuits rather than LC circuits. It has apparently been quite a lot of trouble to get these to be as quiet as a simple LC oscillator. The term you see is "phase noise", specified at a certain offset from the oscillator center freq... like "-120 dBc at 100 kHz", or whatever. That little bit of noise 100 kHz off your oscillator's freq mixes with any signal that happens to be there and produces results that you hear. I suspect this is the problem you are having.

    Ok... that was my best shot at a quick explanation... how did I do Zedders?
    KE0EYJ likes this.
  2. VA3VF

    VA3VF Guest

    I thought dynamic range was the ability to receive a weak signal, in the vicinity of a much stronger signal, XX kHz away. I don't have any technical review here to see the value of XX. The weak signal being a real one.

    I suggest everybody download the ARRL document explaining every item they test in their reviews. It's available on their web page, not sure if you have to be a member to download this particular document.

    73 de Vince, VA3VF
  3. W6RZ

    W6RZ Premium Subscriber QRZ Page

    This is one of the big advantages of the SDR technology used in the IC-7300. There is no local oscillator(s) to cause phase noise issues. There is the sample rate clock of the ADC, but that's at fixed frequency and can be made quite good with a crystal oscillator. This can be seen in the RMDR numbers of the IC-7300 that I posted previously.

    Reciprocal mixing dynamic range: Not specified. 14 MHz, 20/5/2 kHz offset:
    preamp off, IP+ off: 114/107/101 dB;
    preamp off, IP+ on, 114/108/102 dB

    The FT-891 is significantly worse.

    Reciprocal mixing dynamic range: 14 MHz, 20/5/2 kHz offset: 98/82/71 dB

    So I'm going to have to disagree that this is the problem. KE0EYJ clearly prefers the FT-891, which should have more RMDR/phase noise problems not less. I'm still in the better noise reduction on the FT-891 camp.

    RMDR definition:

  4. N1OOQ

    N1OOQ Ham Member QRZ Page

    So it direct samples the RF to do this magic? No IF? I am definitely not up to speed on the latest DSP stuff...
  5. VA3VF

    VA3VF Guest

    Yes, it's called DDC - Digital down conversion.
  6. KE0EYJ

    KE0EYJ Ham Member QRZ Page

    Well, the 7300 is the better sounding rig, by a longshot (FT-891 is flat and somewhat lifeless, in sound) but it does dig out the signals, when DSP helps, in a noisy environment. For daily use, I prefer the 7300, but am sad that it seems weaker in a high noise environment.

    As for the request to see how strong my local broadcast is, above 7.200, here are sone signals in the early evening, last night. The station closer to 7.2 was not there? Not sure why, but it usually is. These are about +40 to +60 db, at peaks.

  7. KE0EYJ

    KE0EYJ Ham Member QRZ Page

    I meant the 891 digs them out, not the 7300.
  8. SM0AOM

    SM0AOM Ham Member QRZ Page

    There are many definitions of dynamic range, and even more interpretations of the definitions.
    When designing receiver systems, we can see it from at least three viewpoints;
    1. The single signal dynamic range. This is easy, only one signal exists in the frequency range of the receiver, so the dynamic range is simply between the noise floor and when there is excessive distorsion. Same reasoning can be applied when there are many signals, but approximately the same strength.
    2. Two-signal dynamic range. Here one weaker wanted signal is evaluated in the presence of one interfering signal. Here we can express the dynamic range in several ways, one is when the S/N is reduced by oscillator noise or reciprocal mixing (RMDR), Another is when the gain of the wanted channel is reduced (gain compression or "blocking").
    3. Three-signal or multi-signal (IMD3) dynamic range. Here the wanted weak signal is surrounded by two or more much stronger interferers. The primary degradation mechanism is intermodulation Products, and a secondary is reciprocal mixing. The "Noise Power Ratio" or NPR test is multi-signal dynamic range testing taken to the extreme, when the whole RF passband is filled with white noise, except for the wanted signal channel. This is a very demanding and conclusive test for receiver strong signal performance.

    Modern receivers have both two-signal and three-signal dynamic ranges that by far outperform the inherent dynamic ranges (signal to distorsion products) of the RF environments found today.

    The absence of close-in synthesizer noise in the SDR architectures makes two-signal and three-signal dynamic ranges approach each other, which previously only was found in e.g. man-pack equipment where their use with small antennas made IMD3 dynamic ranges of secondary importance.

    From the spectrum plots, it seems that the BC signals above 7200 kHz are not stong enough to create problems on their own.

    If the graduations on the scale can be trusted, the strongest signal is about 50 dB above the ambient noise, with a carrier level of S9+20 or -50 dBm. The system noise floor would then be -100 dBm (-135 dBm/Hz) which corresponds to an antenna noise figure of at least 40 dB. This is such a high ambient noise floor that the receiver sensitivity or noise floor itself becomes quite irrelevant, and IMD3 dynamic range limitations would not be approached before receiver overload (ADC clipping) occurs.

    It must be understood that SDR:s have a very different behaviour with respect to overload compared to analogue receivers. The onset of ADC clipping comes abruptly, and creates a multitude of discrete spurious signals which are easily distinguished from other signals. In analogue receivers, the onset of overload is more gradual, with a quite slow reduction of wanted S/N either from oscillator noise or IMD products, as the interfering signal levels increase.

    Putting the numbers into their context, a modern receiver should permit a three-signal dynamic range exceeding 80 dB or an IP3 of at least +5 dBm in the laboratory environment, where the IMD3 products are falling just above the thermal noise floor for typical receiver noise figures.

    When operating in high-noise level environments, a noise figure of about 30 dB would, which is a 20 dB attenuator in front of the previous receiver raising the IP3 to +25 dBm, be sufficient. This still equals to an ambient noise limited dynamic range of more than 80 dB.

    Statistics gathered and analysed by G3RZP point in the direction that 80 dB would cater for at least 90% of the operating requirements. The quest for 100 or 110 dB three-signal dynamic ranges is therefore quite moot from the overall performance aspects, but give "bragging rights" to the user.

    The previous IP3 numbers are representative for "mediocre" receivers with today's standards. Already in the mid-80's there were analogue receivers with IP3 numbers of more than +40 dBm (Telefunken E1800A) on the market which had so good strong-signal properties that they were extremely difficult to evaluate properly. The manufacturer even had to issue a booklet describing the proper measurement procedures and interpretation of the results, to avoid claims that the receiver did not live up to specifications.

    The bottom line of all this is that the receiver claimed performance has to be put in perspective with all other system parameters. If the ambient noise is high enough, even mediocre receivers give all the operational performance that may be obtained at this particular location.

    The reasons for why some receivers perform better than others in man-made noise, have to be searched in how their signal processing actually handles the particular statistical properties of the noise.

    KE0EYJ and OH2FFY like this.
  9. SM0XHJ

    SM0XHJ Ham Member QRZ Page

    I agree that with the signal strength these broadcast signals show, the dynamic range (with any definition) can not be the liming factor in this case.

    As a side note, it is however interesting that ARRL and Sherwood measured so completely different RMDR values on the IC-7300. One fault ARRL has done is of course to measure noise floor with IP+ off, and most of the other values with IP+ on. It is not possible to use the receiver in that way in the real world. But they have specified RMDR (at 14 MHz only) with both IP+ on and off, and there is not significant change in those numbers. That is in itself a bit weird, since that is exactly the type of parameter the IP+ function is supposed to improve and indicating they may not be measuring it correctly.

    SDR's are in no way immune to phase noise issues. Although it may be easier to make a stable LO for the ADC as the frequency is fixed, it is also very easy to introduce jitter in the ADC clock and FPGA/DSP clocks if not designed properly. A simple thing like leaving an input open in the same I/O block in the FPGA can completely destroy the phase noise figure.
  10. SM0AOM

    SM0AOM Ham Member QRZ Page

    My previous comment assumed proper design and implementation, but the fact remains that the internal Clocks, "NCO:s", present in the DDC signal path of an SDR are derived by frequency division processes from the master clock.
    So, if the master clock has low jitter and the internal workings of the FPGA and other DSP parts of the SDR are properly done, the close-in phase noise performance will surpass any realisable PLL-based oscillator.

    There is a paper by dr Rohde that I saw cited in course documentation ages ago about the performance limits of PLL-based systems, and there the HP8644B signal generator (two-man burden and $50000) was stated as a state-of-the-art example, with -90 dBc/Hz at 100 Hz and -120 dBc/Hz at 1 kHz offset. Now the equivalent Close-in sideband noise of e.g. the pioneering Perseus SDR is about 20 dB better, allowing it to distinguish a 50 Hz hum sideband in a crystal oscillator at -95 dBc. No tuneable PLL synthesiser based receiver LO can even approach this.


Share This Page