R820T frequency tuning issue

As briefly mentioned on the #gnuradio IRC mailing list, I am facing an
issue with tuning a R820T
DVB-T dongle at the GPS carrier frequency. I had qualified the various
dongles at 1.5 GHz, and had
not considered that the conclusions from this measurement would not
apply at 1.57 GHz. This issue
only exists when using a R820T and not with a E4000 dongle (which
works fine at all mentioned
frequencies, and whose datasets allow for processing GPS signals).

I have completed the following experiment:

  • a frequency synthesizer (Rohde & Schwartz SMA100) is set to 1570 MHz,
    output power of -10 dBm,
    lab-made antenna (ie, dangling wire)
  • R820T DVB-T located next to the synthesizer
  • either running rtl_sdr -f 1570000000 -n 100000 file.bin and plotting
    the resulting log in GNU/Octave
    (f=fopen(‘file.bin’);d=fread(f,inf,‘uchar’);plot(abs(fftshift(fft(d)))))
  • or in a simple gnuradio-companion flowgraph a gr-osmosdr source
    connected to the WX GUI FFT sink,
    in both cases sampling at 2048 kHz

In the latter case I cannot get any signal, hinting at an unlocked or
misconfigured PLL frequency. In the
former case (using rtl_sdr), I get for example
http://jmfriedt.sequanux.org/rtlsdr.png in which the
carrier was shifted by 100 kHz steps below 1.57 GHz when recording each
curve.

Unfortunately I don’t know how to tackle the issue other than by filing
this bug report.
gnuradio and gnuradio-companion are 3.7.3
gr-osmosdr is v0.1.1-1-g37b09bc5 (0.1.2git)

The receiver is R820T+RTL2832U, sampling at 2048 kHz.

Thanks, JM


JM Friedt, FEMTO-ST Time & Frequency/SENSeOR, 26 rue de l’Epitaphe,
25000 Besancon, France

As briefly mentioned on the #gnuradio IRC mailing list, I am facing an issue
with tuning a R820T
^^^^^^^^^^^^
channel

[…]

I have completed the following experiment:

  • a frequency synthesizer (Rohde & Schwartz SMA100) is set to 1570 MHz, output
    power of -10 dBm,
    lab-made antenna (ie, dangling wire)
  • R820T DVB-T located next to the synthesizer
  • either running rtl_sdr -f 1570000000 -n 100000 file.bin and plotting the
    resulting log in GNU/Octave
    (f=fopen(‘file.bin’);d=fread(f,inf,‘uchar’);plot(abs(fftshift(fft(d))))) …
    ^
    d=d(1:2:end)+i*d(2:2:end)

sorry for the mistakes.

JM


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

As a quick followup to my previous post, I confirm that the R820T is
well suited for GPS
signal decoding. I was not expecting the huge frequency offset (> 100
kHz at 1.57 GHz) and
was not searching far enough from the expected carrier frequency during
the acquisition phase.
However these results were acquired by runnig rtl_sdr -s … -f …
while gnuradio-companion
running gr-osmosdr is unable to lock at 1.57 GHz and barely locks (for a
few seconds !) at
1.56 GHz. I have analyzed the various version of librtlsdr (used by
rtl_sdr and by gr-oscmocom)
as well as the set_frequency functions and am unable to track any
difference between the two
codes (so far).

JM


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

On 03/28/2015 10:23 AM, [email protected] wrote:

The R820T is generally flaky above about 1550MHz. Maybe in your
rtl_sdr tests, you just got lucky.

gr-osmosdr uses the SAME underlying library as rtl_sdr, and at the
level of Gnu Radio, all of the low-level stuff like
“is the PLL locked” is utterly invisible to the Gnu Radio layer.

On 03/28/2015 11:29 AM, [email protected] wrote:

Sorry for the trouble, will make sure now the issue is indeed solved, and let
the measurement run until
monday morning.

Thanks, JM

Your results are consistent with what I’ve heard about the R820T above
1500MHz or so. It starts to become
“deaf”, and then as you move higher, the PLL won’t lock at all.

On 03/28/2015 10:56 AM, jmfriedt wrote:

(http://sdr.osmocom.org/trac/wiki/rtl-sdr) when the datasheet mentions a 1 GHz
upper limit.

I should mention that the above statement has been tested on a random sampling
of all 10 dongles
I have bought for teaching, and conclusions have held reliably whichever
peripheral was used
(all bought from the same supplier, probably same batch, all running the same
R820T frontend, all
with the same very poor 60-80 ppm offset).

Thanks for the reply, JM
So, if you just use something like

osmocom_fft -f 1575.420e6 -g 30

And feed it a low-level signal-generator CW signal, you’re saying that
it doesn’t work?

Your results are consistent with what I’ve heard about the R820T above
1500MHz or so. It starts to become
“deaf”, and then as you move higher, the PLL won’t lock at all.

I loose 10 dB sensitivity from 1.50 GHz to 1.575 GHz ! At 1.5 GHz the
R820T was more sensitive
than the E4k, but now I’ll have to review this statement at 1.57 GHz !

Thanks for the help, JM


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