E4000 based receiver raw samples => broken direct sampling support ?

Quick update thanks to the long week end in France which left enough time to test all combinations of hardware and software: 1/ I tried applying the patch provided by Steve on this list on librtlsdr and compiling gr-osmosdr for running in gnuradio-companion. I know I am running the right source block linked with the right library (added debug messages) => no signal on a R820T2 dongle connected to a signal generator (I+), the other pin (I-) either grounded or decoupled with a 100 nF cap. 2/ doing the same on a E4k based dongle also provided only noise and no signal, a strange result since in the very beginning of my experiments (2012) I had used this same dongle for direct sampling. Being a zero-IF, I could just bypass the E4k mixer and record the raw samples fed to I+ and Q+ (I- and Q- being grounded). Surprisingly, running the same gnuradio-companion flowgraph (osmosdr source -> Qt time sink) with NO direct sampling option does yield correct samples, so the hardware is sound and functional. 3/ E4k no longer being available, I have to find a way to compensate for the R820T IF: I checked using no option in osmosdr source that I could get some relevant signal when tuning the signal generator around the IF frequencies (4.8 MHz ? 5 MHz ? difficult to say with all the aliases), while the option "rtl=0, direct_samp=1" always returns noise. I am not familiar enough with the RTL2832U configuration registers to mess with them myself, but I believe direct sampling in not functional at the moment in librtlsdr. I might be wrong of course. Thanks, JM -- JM Friedt, FEMTO-ST Time & Frequency/SENSeOR, 26 rue de l'Epitaphe, 25000 Besancon, France