Forward on behalf of Leif SM5BSZ: _______________________________ Hi All,
I am not a member on the Elecraft list, but I did work quite some time on this issue: > I, too, would be very interested in seeing how "entire HF spectrum digitized > at once" receivers would fare in the real world. I am *very* skeptical. > Although the Perseus (with that architecture) receiver fares relatively well > in the recent ARRL test, the testing completely ignores what happens when > there are not *two* but thousands of strong signals in the passband. With > this new architecture, it's imperative that ARRL and other labs find other > ways of evaluating real world performance as a simple two-tone test becomes > completely irrelevant. You are right. The "simple two-tone" test is irrelevant for units like the Perseus or the SDR-14. Have a look here (QEX Nov/Dec 2006): http://www.sm5bsz.com/dynrange/qex/digital-imd.pdf and here (for real life testing on 7MHz): http://www.sm5bsz.com/digdynam/practical.htm There are two properties one has to characterize when testing the dynamic range of a radio receiver. 1) The ability to receive a small signal in the presence of a single dominating interference. What has been published by ARRL is totally wrong and should just be discarded. The procedure adopted by the ARRL is intended to measure the point where the front-end goes 1 dB into saturation (or rather any stage preceeding the bandwidth-defining final filter.) The FT1000D is not at -154 dB for example: (QEX Mar/Apr 2006): http://www.sm5bsz.com/dynrange/qex/bdr.pdf Even in those cases where the ARRL Lab BDR is correctly measured (to give 1 dB gain compression) the result is not at all the blocking dynamic range as understood in the professional world where it refers to the ratio (difference in dB) between the strongest and the weakest signals that can be simultaneously present without loss of readability of the weakest signal. Surely the bandwidth, modulation type, frequency separation and other things have to be specified, but generally the mechanism is reciprocal mixing at narrow frequency separations (or some other phenomrnon that behaves in a similar fashion. e.g. AM modulation in preamplifiers.) I like to call this the "two-signal dynamic range" (one strong and one weak.) 2) The linearity of stages preceeding the bandwidth-limiting filter. Conventionally this is measured with "a simple two-tone test" Such measurements are often incorrect for various reasons, but in recent years they are generally more reliable since a better understanding for the problems is now widespread in the ham community. I prefer to call this test "the three-signal dynamic range" One weak plus two (equall) strong signals. The point of 1 dB compression is closely related to IP3 in those cases (large separations) where the attenuation for all the test tones would be the same. There is yet another way to measure the same thing and that is by measurement of cross modulation. (How strong can a 60% modulated AM signal be before it modulates the desired weak signal with x% AM modulation?) As it turns out, worst case is with only two signals so the best measurement is a two-tone test. One has to follow the interference all the way to the noise floor. (And below in case the modulation mode is operated below the noise floor of the main filter. As an example, CW is typically operated down to 15 dB below the noise floor since the human ear has an effective bandwidth of less than 50Hz) The link above, QEX Nov/Dec 2006, shows that the level of the interference at the frequency of third order intermodulation is independent of the signal level. It stays constant at about -100 dBm for a SDR-14 which is high above the intrinsic noise floor. There are also signals at a similar level corresponding to IM5, IM7, IM9, IM11, ... The mechanism is feedback from the 3 volt p-p signals at the digital side. Linear has cips LTC2207 and others that reduce this feedback by 15dB according to their own information if I remember correctly. What I can measure on the Perseus is that IM3 flattens out at about -125 dBm: http://www.sm5bsz.com/perseus/perseus.htm The Perseus has NF=22 dB while the SDR-14 has NF=12 dB. The intermodulation-free dynamic range is 104 dB for the Perseus, but only when the A/D converter is operated at the largest possible signal level. For the SDR-14 it is 82 dB. In real life there are more than two strong signals and that improves the performance drastically since the phase-stability of the signals on the digital side is destroyed (dithering.) A radio like the Perseus would probably outperform a conventional radio like the IC-7800 in a A/B test on e.g. 7 MHz with real antennas and many BC stations provided a location can be found where many BC stations are strong enough to produce false signals on any of the receivers. The IC7800 has a dynamic range of 104 dB according to ARRL but it would not be as favourably affected by a large number of strong signals. For close range intermodulation (5 kHz and below) the Perseus would outperform the IC-7800 with a large margin. Provided, of course, that the antenna signal is strong enough. For optimum performance, the operator must know how to place a suitable attenuator or amplifier between the antenna and the receiver to place the dynamic range at the correct posuition. In a real life blocking test, the Perseus outperforms the IC-7800 with a wide margin. Someone actually tested and wrote about it on the Perseus mailing list. > > It would be rather interesting to see how the concept used by the K3 (24 > > bit digitalization after IF) compares in the real world to the early 14 > > bit digitalization as used by the ADT-200A (and afaik by the Perseus RX > > from Microtelecom), especially with the chips available currently. The K3 strategy should not have any problems with noise pick-up from the digital side so the level where "intermodulation" flattens out should be very low. My own design, WSE2500+Delta44 with NF=17 dB flattens out at -140dBm (QEX JanFeb 2004): http://www.sm5bsz.com/linuxdsp/qex/040102qex020.pdf On the other hand, the blocking or rather reciprocal mixing test on the K3 will be limited by the VCO which is -120dBc/Hz at 2 kHz and 141dBc/Hz at 20 kHz (QST April 2008) This converts to blocking dynamic ranges of 93 and 114 dB respectively for K3. The Perseus is MUCH better, 115 and 123 respectively. (ARRL Lab made a mistake in the December issue which I expect they will correct in a coming issue.) In real life, better reciprocal mixing is not yet important because ALL amateur transmitters have too much sideband noise, keying clicks and other impurities that dominate over the performance of a really good modern receiver. Soon someone will start to sell the inverse of Perseus for our transmitters and they will allow drastical improvements on the transmit side. Same band, multi-transmitter operations from field-days and other activities will become much easier in the future:-) 73 Leif / SM5BSZ _______________________________________________ Elecraft mailing list Post to: Elecraft@mailman.qth.net You must be a subscriber to post to the list. Subscriber Info (Addr. Change, sub, unsub etc.): http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/subscribers.htm Elecraft web page: http://www.elecraft.com
