Re: [Discuss-gnuradio] USRP Power Supply Noise

[John Orlando]

On Tue, Aug 24, 2010 at 9:32 AM, Jeffrey Lambert <address@hidden> wrote:
> Hello,
> I am attempting to build an analog front-end to a USRP 1 device.  I have
> selected a VCO that will suit the frequency range we are looking at: It is
> an RFVC1800 and has a tuning range of approximately 7.4 - 12.4 GHz over a 13
> volt tuning voltage.  As this VCO is very sensitive to external noise, some
> power filtering will be necessary to minimize jitter and phase noise.  What
> I would like to know is whether or not there are any other adverse effects
> from using the simple switch mode supply that is included with the USRP?

Are you referencing the power brick that comes with the USRP (and
provides the raw +6V DC input)?  If so, you'll definitely want to do
some significant filtering/cleaning up of this supply, and want to
think carefully about the gnd paths to your VCO as well.  IIRC, we saw
non-trivial ringing on the gnd that had a period of ~5 uS and occurred
every 20 mS (don't quote me on the exacts here, as my memory is a bit
fuzzy, though this is what I sketched into my engineering notebook at
the time).  We ultimately determined we were simply seeing the
residual effects of the switching power supply in the power brick and
had to live with it or use a better external power suply.

Most daughterboards take the raw +6V passed to it right from the power
brick, and the use LDOs on the daughterboard to generate the voltages
necessary for the sensitive RF components.  But this is typically
stepping down to 5V/3.3V, which, with a 6V input, provides enough
margin for an LDO to work and still provide excellent power supply
ripple rejection (PSRR).  This is key when generating a stable supply
for feeding a VCO, as any variation (i.e. ripple) will show up as
variation in the output of your VCO.

If I were you, I'd start with a good oscilloscope and begin probing
around the power supply lines and the gnd lines of the USRP to see
what the supplies look like, and from there you should be able to
assess whether or not you'll be able to use this voltage as a starting
point for your design.  You'll also want to make sure the gnd
reference point for the scope is as close to the measurement point as
possible to ensure that you aren't picking up stray
signals/interference.

Good luck, and keep the list posted of progress.  Sounds like quite a
challenging project  :-)

-- 
John Orlando
CEO/System Architect
Epiq Solutions
www.epiq-solutions.com