At 10:22 AM 1/23/2007, Jeff Anderson wrote: >Hi Tom, > >I believe the bandpass will depend, in part, on the >input impedance of the device to which it's attached.
Particularly the interaction of parasitic L and C... >Many months ago I measured the passband of one of >these devices for a friend who had terrible hum with >her stereo and wanted to use one of these, and it was >flat out to 50k (which was the limit of my dsp-based >measurement system) into the fairly-high impedance of >the measurement system and using a low output >impedance white-noise generator as the source. > >I just took a look at the low frequency performance on >my sdr1k with the RS isolator and my Delta 44. I fed >the SDR with white noise (from an old GR noise >generator that I use expressly for this sort of >thing), and I expected to see a "notch" on the >display corresponding to the device's low frequency >cutoff. Suprisingly, even at max Zoom In (6 >Hz/pixel?) and with the display shifted so that the IF >0 frequency was centered in the display, I think I >*might* have seen the *hint* of a notch. Of course, this measures only the power spectrum (magnitude) and not the phase response. Although phase errors will manifest themselves as poor image rejection, etc., they tend not to be so obvious in a power spectrum of broadband noise.. The image being 20 dB down (which is quite poor) which results from a substantial phase error (5-6 degrees), would be hard to see in the power spectrum (1% power error is, what, 0.04 dB or thereabouts), especially if you're looking at the noisy "grass". >Another negative that I did see, though, were some >distortion products that became visible as one sweeps >a generator (I used my 8640b) through the IF "0" >frequency. perhaps not "distortion" per se, but poor image rejection. A CW tone swept through the pass band will find those phase imbalance errors quite effectively. > I did not see these when the isolator was >*not* in-circuit. But they seemed to disappear pretty >quickly as I moved the generator away from the IF 0 >frequency, so these may not be a significant issue, >or, perhaps, only an issue for signals near the (now >much attenuated) "hump." The low frequency roll off (basically having to do with the low frequency properties of the transformer) could easily be slightly different between transformers.. slightly different parasitic C, maybe a different turns count or arrangement of the windings, etc. That would lead to the phase shift being slightly different, which would affect the image rejection. > But this raises a point >worth noting: additional devices in the audio path, >such as the gnd-loop-isolator, could add some amount >of distortion to the receive signals. Indeed.. On the other hand, in theory, this sort of thing is able to be calibrated out. Last year, there was some discussion about providing a means to calibrate the audio path at more than a single point. All it takes is figuring out the frequency domain response needed, and combining that with the other frequency domain filters. No additional CPU horsepower needed. Jim, W6RMK _______________________________________________ FlexRadio mailing list FlexRadio@flex-radio.biz http://mail.flex-radio.biz/mailman/listinfo/flexradio_flex-radio.biz Archive Link: http://www.mail-archive.com/flexradio%40flex-radio.biz/ FlexRadio Homepage: http://www.flex-radio.com/ FlexRadio Knowledge Base: http://kb.flex-radio.com/