[time-nuts] 35601A as stand-in for 11848A ?
bruce.griffiths at xtra.co.nz
Sat Jan 5 19:34:53 EST 2008
SAIDJACK at aol.com wrote:
> Hi Bruce, John,
> thanks for the hints, I don't have much time right now, but some feedback to
> the questions:
> I am using resistive 50 Ohms termination out of the IF port, then followed
> by an inductive input 1.5MHz LC filter.
A capacitive input filter will be better.
Unless you want to examine the phase noise at high offsets disconnect
the 50 ohm termination.
This increases the phase detector gain.
Reflecting the sum frequency back into the mixer (eg with capacitive
termination) lowers the noise.
> I will try to AC couple the 50 Ohm termination and see if that helps.
> The PLL works fine, and the noise floor doesn't change when I connect it to
> the OCXO or if I leave the EFC input to the reference disconnected and the
> OCXO EFC input just terminated, so I am quite sure that's not causing the issue.
> The reference is a Wenzel Ultra low Noise OCXO with a noise floor
> I am trying the setup using an E1938A DUT. The noise floor out of the Mixer
> is about -120dBc/Hz, way too high. Tried other sources with known noise
> floors of about -160dBc/Hz as well, similar results.
> Will write more later,
A phase noise calibration standard would be very helpful.
The simplest method is to add the outputs of an OCXO with a gaussian
noise source using a hybrid combiner.
This produces equal amounts of AM and PM noise in the same way as
thermal noise does albeit at a much higher level.
The AM noise will be rejected by the phase detector if its operating at
the 90 degree point or is saturated.
For low frequency offsets a pseudorandom noise source may be useful.
Alternatively one can use a JFET opamp to amplify the thermal noise of a
150K resistor to a suitable level.
At higher offset frequencies a calibrated zener noise source can be used.
In actual use the OCXO signal is split into two parts, one part is
combined with the gaussian noise source and drives one phase detector
(mixer) input port, the other input port is driven by the clean OCXO
signal. The resultant noise spectrum at the phase detector output will
reflect the phase detector system frequency response including the
effect of the PLL.
This is the technique used by NIST to caibrate phase noise measurement
More information about the time-nuts