A project that I am working on is using a USRP to generate pure tone
signals and measure those signals after they have passed through our
"black box" application. We need to measure down to DC so we have
modified the TX and RX boards with "granddaughter" boards.
We have run into a problem which can be seen on both the original
daughter boards and our modified ones. There seems to be bleed through
from the TX boards to the RX board, at frequencies up to 8kHz.
Test setup:
- Running the USRP with one TX board and one RX board
- Nothing is connected to the SMA connectors
- One TX DAC is outputting a pure tone generated by gr.sig_source_s
- One RX ADC output is being mixed with a gr.sig_source_c using a
gr.multiply_cc. The output is then filtered with gr.fir_filter_cc
(response graph, of original RX & TX boards, here
http://www.kamieniecki.com/krys/gnuradio/AC_front_end_cropped.png) It
looks like the response of a lowpass filter to me.
Does anyone have any ideas before I get together with our EE?
BTW:
We are getting nice strong clean signals at 9kHz, when we connect the
TX to the RX.
For TX we use an AD8057, and for RX we use an AD8138.
I'm also looking into releasing schematics for these "highly advanced"
modifications.