[USRP-users] Amplifiers for USRP Transmission

Sylvain Munaut 246tnt at gmail.com
Mon Jul 6 16:49:31 EDT 2015


> My questions are, what do I need to consider when looking for an amplifier
> to use with a USRP, are there any special problems a beginner would be
> likely to miss when trying to use those 2 things together?

Filtering !

SDR outputs are very often not all that clean and have plenty of
harmonics and images and stuff like that.
When you just use the raw output at low power you can often just
ignore them because they're going to be 10+ dB lower than the main
signal and so that's low power enough to be ignored.

However as soon as you start amplifying, those imperfections will get
amplified too and will start interfering with other transmissions.

To reduce the images you want to TX to the USRP at a higher sample
rate than needed and apply sharp digital filtering to your signal
(sharper than what the USRP itself can do in the FPGA).

You might also want to configure the analog bandwidth as small as
possible (although I don't think that's applicable to the USRP1+SBX
since there is no variable analog bw there).

To reduce the intermodulation products you want to not drive the
components too hard. So make sure to only make your signal swing
between +- 0.8 and not 1.0 for instance. You also want to not put all
the gain stages at maximum. Back off a bit, like 80% or so, so that
you keep all of them in their most linear regions.

Finally to get rid of the harmonics, you want an external band pass
filter that's center around your frequency of interest. Minicircuits
has those too. You can for example use two of them, one between the
SDR and your external amplfier and then one after your amplifier (to
get rid of the harmonics added by your amplifier). If you only have
one, then place it after the amplifier.

> Also, if an
> amplifier's gain is listed at 20 dB, would that be 20 dB on top of the 31.5
> the daughterboard is capable of outputting, or would the amplifier be
> useless in that it is capable of providing less power than the daughterboard
> itself?

1) Your daughter board is definitely not outputting 31.5 dBm ...

2) The gain will de added to the daughter board output as long as
you're not saturating the amplifier. See the P1db rating for a
guideline of maximum power (Output power at 1dB compression). For
instance, the zx60-v63+ you linked to has a P1db of 17.8dBm at 2G ...
so it's even less than 100 mW.

3) You _DEFINITELY_ want to keep the total output power of the
amplifier like > 3 dB below the rater P1db so that you're well within
the linear region and your signal isn't distorted and you don't create
too much harmonics.

Note that (3) actually depends on the modulation and external
filtering quality. For some modulation like GMSK (used in GSM), you
can use amplifiers in their non-linear region and rely on external
filter to get rid of unwanted harmonics. This is why GSM amplifiers
are cheap to make. OTOH when working with high PAPR modulation (like
OFDM), then you actually need to backoff way lower than the P1dB just
because during the peak of the signal you need your amplifier to still
be linear. And that's why amplifying LTE is much harder and you get to
advanced and crazy stuff line pre-distortion ....

> And third, how is a device such as this one:
> http://www.minicircuits.com/pdfs/ZX60-V63+.pdf powered?  The data-sheet
> gives the numbers 5 V and 69 mA but I don't see an obvious way to power it
> aside from just soldering wires onto the marked bumps that say +5 and
> ground, is that the actual intended way to do it?

Yes, it is.

Some other can also be powered by providing a DC bias on their output
port using a separate bias-tee. That's not the case for this one
though AFAICT.



More information about the USRP-users mailing list