Usrp - gnuradio agc

GNU Radio
1

Hi at all,

I need to implement an AGC to adapt the signal to the ADC’s full
dynamics.
I work with an USRP N210 with CBX-40USRP DAUGHTERBOARD…

I investigated that :

There is a programmable gain amplifier (PGA) before the ADCs to amplify
the
input signal to utilize the entire input range of the ADCs, in case the
signal is weak.
This PGA is software programmable.

Auxiliary Digital I/O Ports:
The USRP motherboard has high speed 64 bit digital I/O ports.
These digital I/O pins are connected to the
daughterboards interface connecters (RxA, TxA, RxB and TxB).
These signals can be controlled from software by reading/writing to
special
FPGA registers and each can be independently configured either as
digital
input or digital output.
Some of these pins are used to control specific operations on the
installed
daughterboards such as
controlling the selection of receiving RF input port, in automatic
transmit/receive mode,
controlling power supply feeding for different TX and RX parts,
synthesizer
lock detection,…etc.
It can be also used to implement AGC processing and can be very helpful
in
debugging FPGA
implementations when connected to a logic analyzer.

So, using UHD USRP SINK, the PGA is modified just acting on the field
Ch0:
Gain(dB) ???
Second, there is some block that allow to extract these Auxiliary
Digital
I/O Ports ???
Practically I want extract analog data before the PGA to process it
software…

2

Hi Simone,

to make this short: Yes, you just set the gain using that field, or at
runtime, using the set_gain() method of the USRP source.

Everything you can do with a USRP is usually wrapped in the USRP
source/sink blocks, which are basically “just” wrappers that take the
UHD library (which is the driver for the USRPs) and adapt them to GNU
Radio.
As such, there is no “direct” way to modify the digital outputs. This is
by design – these are largely routed to the daughterboard, in your case
the CBX, to communicate settings and commands (e.g. setting the gain,
changing frequency and so on), so as a user you usually “shouldn’t” mess
with them.
You can, nevertheless, just modify the UHD driver, add a daughterboard
sensor interface or the like (by changing the daughterboard interface
source code) that would, upon querying that sensor, issue some change to
the IO ports. I don’t recommend this – a) you’ll be on your very own
and b) what would you do with that? most of the lines are routed to and
terminated on the CBX [1], so you’d have to modify the hardware. If I
remember correctly, there is one daughterboard that has an always-on
AGC, the TVRX2, which is a bit of a special-purpose daughterboard, as
non-adjustable AGCs only fit signals that fit the AGC – ie. which
symbol rate is far enough from the AGC loop bandwidth but which
experience fading at a rate that allows the AGC to do its job.

Getting the signal “before” PGA means you don’t want to use the mixer
and filters; that basically means you will have to implement spectrum
sensing in an adjustable pass band yourself, at which point you’re
building something that is effectively nearly as complex as the CBX
itself, unless you go for something very simple.

To be honest, unless you have very special needs, using a fast energy
estimator that sets the daughterboard’s gain in software on a PC would
clearly be my first step, and modifying/adding hardware would only take
place if that system couldn’t fulfill the receiver’s needs.
We had a discussion on [email protected], not too long ago,
about implementing something like that in the FPGA, which would reduce
latency; I can’t quite remember the name of that thread, though…

Greetings,
Marcus

[1] http://files.ettus.com/schematics/cbx/CBX.pdf

3

Hi Simone,

On 25.07.2014 15:55, Simone Ciccia S210664 wrote:

Hi Marcus,
thank you for your answer…

Ok, suppose that we are in reception,
I said before the PGA thinking that at this stage the signal is already in
baseband ( MIXER -> FILTER -> PGA -> ADC ), so in this way I will have
compared signal level before PGA with the full scale of ADC.
Actually, compare the schematics I linked to: the variable attenuation
comes before the mixer, so
Antenna->LNA->PGA->mixing->ADC is the order.
But your are right, comparing energy is more simple!
Just another question…
Using the USRP as receiver, we set the gain as “variable” in UHD source
block “options”.
Now, the changes of this variable take effect on the USRP during runtime???
or just set the initial value and we need to use set_gain() ???
I suppose we’re talking about GRC flow graphs now :slight_smile:
Yes, there is a callback that will automatically set the gain when you
change the variable at runtime.
You can use set_gain() directly, too.

Greetings,
Marcus