[USRP-users] N210 DDC gain

Josh Blum josh at ettus.com
Wed Sep 5 16:14:19 EDT 2012



On 08/29/2012 05:31 AM, Andrew Senior wrote:
> Hi all,
> 
> earlier this year I did a calibration on my N210, feeding a sine signal
> at -20 dBm into one input of a Basic Rx and recording the 16-bit output
> data with a subdevice specification of "A:A" and a gain of 6 dB. The
> device sample rate was 6.25 MHz.
> 
> I found that the recorded samples had an amplitude of ~1332 counts and
> since the input amplitude is ~30 mV, this corresponds to a sensitivity
> of 44400 counts/V. This surprised me a bit since with 6 dB gain, I'd
> have expected something closer to 65536 counts/V. These measurements
> were done with UHD 3.3.2 on the host and the firmware images for 3.3.0
> on the device.
> 
> However, when I repeated the calibration recently, I was surprised to
> find that it had changed. I now get only ~818 counts for the same input
> or a sensitivity of ~27000 counts/V. This was with UHD 3.4.2 (both host
> and device).
> 
> I can see from the repository that there's been quite a bit of
> re-working of the DSP in the device between these versions which
> probably explains the change, but I'm still puzzled by the gain
> apparently being so low now. Can someone explain the reasoning behind this?
> 

There is some scaling compensation for the DSP chain thats between 1.0
and 2.0 that was taken care of by an extra set of multipliers in the
older UHD.

Those multipliers have been re-purposed with the addition of the sc8
stuff, and the compensation is done in the data type conversions in the
host.

However, when you ask for sc16 type and the wire type is also sc16,
there is no host scaling adjustment, which explains the gain discrepancy.

-josh




More information about the USRP-users mailing list