Maximum input signal power for WBX


What is the maximum recommended input signal power for the WBX
daughterboard? Where can I find this in the documentation or online at My understanding is that the LNA is very sensitive and it is
easy to overload it and damage it. I thought it was 0 dBm, but I have
seen several numbers on this mailist, such as -10 dBm and -15 dBm, so
I’m making this post to get a definitive answer. Thank you.

Steve McMahon

On 11/27/2010 11:02 PM, Steve M. wrote:

Discuss-gnuradio mailing list
[email protected]


The first stage in a WBX is an MGA-62563, which has a specced max Pin of
+20dBm and
22dB gain, the next stage device has a specced max Pin of +13dBm.
Which means that
with the attenuator stage set to minimum, the max Pin in the first
stage would be about -9dBm,
in order not to seriously overload (and possibly damage) the second
stage, which is an
MGA-82563, “fronted” by a digitally-controlled attenuator with a
minimum insertion loss
of roughly 1.5dB (HMC472LP4). This is all from the published
schematics in the
“Downloads” section of the Ettus website.

The reality seems to be that the GaAs LNAs are sometimes more delicate
than their spec
sheets would imply. If it were me, I’d never put more than -25dBm into
a receiver chain that
had a GaAs LNA in it. Such a signal would generally be regarded as
“thundering strong”,
“earsplittingly loud”, etc :slight_smile:

One needs to keep in mind that the daughtercards with active Rx chains
on them (that is
all of them except LF_RX and BASIC_RX) are designed for over the air
use, which means
that they generally expect signals to be arriving no stronger than
-30dBm, and usually a lot
lower, like -60dBm or lower. A good receiver generally has a
minimum-discernible-signal, for
narrowband signals, below -115dBm. Injecting signals into such a
sensitive receiver that are
over 100dB stronger than that is asking for trouble. Best case is
that you end up in non-linearity
territory, and worse case is that you damage the first-stage LNA (or
possibly subsequent
stages). The maximum input power of the motherboard ADC is roughly
+10dBm, but I’d
generally de-rate that by about 2-3dB.

Gate insulators on microwave GaAsFET transistors are thin. Really,
really thin. Bias them
the wrong way, that insulator goes “pffffffft”. Put too much power
into them, that insulator goes
“pfffffffft”. Give it a shot of ESD, and that insulator, yup, goes
“pffffffft”. There’s not a lot you
can do to protect them, either. Anything you put in front of an LNA
generally degrades the
noise figure, often severely. Limiter diodes? That would badly screw
up your impedance matching,
and kill your noise figure. Gas discharge tubes? Fine at low
frequencies, but the shunt capacitance
will start giving you headaches at higher frequencies, and won’t
protect you from somebody putting
in +10dBm. Input circuit with a DC-path to ground? Sometimes works,
but sometimes, your
“impulse” ends up getting stored in the shunt inductor, and when the
field collapses? “Pfffffftttttt”.
Again, that won’t protect against too much input power in “normal”

My understanding is that SiGe LNA parts are a little more forgiving, but
not by a whole lot, and they
generally have poorer noise figures than their GaAs counter-parts
(although that is changing

Bottom line? Sensitive microwave receivers are, well, sensitive. If
it were my lab, I’d make
certain that my techs knew never to inject more than -25dBm into any
receiver, unless they were dead-certain that it could survive the

Principal Investigator
Shirleys Bay Radio Astronomy Consortium