R820T upper frequency ?

[friedtj]

I am struggling with using some of the R820T-based dongles I bought (http://www.dx.com/p/170541).
I am wondering where the specs from http://sdr.osmocom.org/trac/wiki/rtl-sdr comes from: although
I am told that the operating frequency range is 24 - 1766 MHz, I am struggling to lock
at 1574.2 MHz. From the datasheet (http://superkuh.com/gnuradio/R820T_datasheet-Non_R-20111130_unlocked.pdf)
we are already well above the "official" operating frequency of this receiver frontend, but
I nevertheless attempted to reproduce with this dongle the GPS decoding scheme I successfully implemented
with E4k based dongles (for teaching purposes, I'd need a reliable supply of dongles but I only have
a couple of E4k based dongles left, one of them probably damaged since the frontend is overheating).
I did try to replace L9 (RF input), L2 + L6 (RF filters) and most significantly the pump charge
RC circuit R8/C16 from the datasheet with high quality RF-grade components, but only managed to
extend by ~60 MHz the locking range of the PLL, from 1.500 to 1.560 GHz. I am unable to to gain the
last 20 MHz of reliable locking to receive GPS !

The way I measure is trivial: a frequency synthesizer generates a single tone (-35 dBm) and I look
at the gnuradio FFT output. After a couple of seconds, the PLL unlocks and the visible single tone
vanishes from the FFT. At 1.5 GHz the measurement is unreliable, lasts for about 1-10 seconds
at 1.55 GHz, and a second at most at 1.56 GHz. The dongles never lock at 1.57 GHz, let alone 1.5742 GHz.
I attempted the experiment with 3 dongles (of the 12 I have) so something is definitely wrong with
either the layout of the PCB, or the information on the web page.

Thanks for any info or which component to upgrade to extend the operating range (by only another 30 MHz !), 
JM

-- 
JM Friedt, FEMTO-ST Time & Frequency/SENSeOR, 32 av. observatoire, 25044 Besancon, France