Datasheet

LTC6957-1/LTC6957-2/
LTC6957-3/LTC6957-4
31
6957f
For more information www.linear.com/LTC6957-1
Phase Noise Measurement
Additive (also called residual) phase noise can be particu-
larly challenging to measure. Figure 12 shows a typical
laboratory set-up for testing the LTC6957-1 phase noise.
The LTC6957-1 has the lowest broadband phase noise of
the various dash numbers (equal to that of the LTC6957-3/
LTC6957-4) and the lowest close-in noise with a corner
frequency below 2kHz, so it presents the most challeng-
ing case.
The various components and their role will be discussed
as this will illustrate both the care that must be taken
to realize the full performance of the LTC6957, and the
demanding nature of making phase noise measurements.
The signal starts with a 122.88MHz CW tone from the
Agilent 8644 synthesizer at a fairly high power level of
12.5dBm. Tw o series LPFs at 150MHz cut out all the high
frequency noise components that would otherwise con-
tribute noise because of the aliasing caused by the limiting
action of the LTC6957. A signal splitter then separates the
signal in two; one path will propagate through the DUT and
the other won’t, a common method used for measuring
residual phase noise.
applicaTions inForMaTion
In theory, all the phase noise in the signal source will be
rejected with the reading reflecting only the difference in
noise between the two paths. However, the rejection is
not perfect, particularly at very high offset frequencies
where the phase difference between the two paths pro-
gressively increases, thus the successive lowpass filters
on the signal source.
The Agilent 5505 measurement system uses the N5500A
front end, which includes a mixer to compare the signal and
reference phases. For amplifier noise, it is appropriate to
feed the DUT path to the signal input, but for clock buffers
that create fast clock edges, it is usually advantageous to
use the reference input, which seems to be sensitive only
to the edges and not noise throughout the period. This is a
reasonable thing to do because the LTC6957 is designed
to drive ADC encode inputs or mixer ports which have the
same qualitative properties.
Both the signal and reference inputs to the test set need to
be fairly large (15dBm to 20dBm) to realize the best noise
floor, so both signal paths include Mini-Circuits ZHL-2010+
low noise amplifiers to boost the signal. The LTC6957-1
was operated from 2V/–1.3V supplies so it could drive a
Figure 12. Setup for LTC6957-1 Phase Noise Measurement Using Agilent E5505
6957 F12
REF
N5500A
50Ω TERMINATION
COUPL
OUT IN
MINI-CIRCUITS
ZHL-2010+
MINI-CIRCUITS
ZHL-2010+
MINI-CIRCUITS
ZFBT-6GW-FT
MINI-CIRCUITS
ZX10-2-12-5
MINI-CIRCUITS
ZFDC-20-5-5+
MINI-CIRCUITS
ZFDC-20-5-5+
6dB
ATTENUATOR
3dB
ATTENUATOR
3dB
ATTENUATOR
SIG
10dB
ATTENUATOR
COUPL
LINE STRETCHER
ARRA L9428A
IN OUT
10dB
ATTENUATOR
SPUR
INPUT
CAL TONE
MONITOR
2V
–1.3V
MCL
LFCN
–150
MCL
LFCN
–150
AGILENT 8644
122.88MHz
12.5dBm
DUT
1
2