13
Agilent 16950B and 16951B Specifications and Characteristics
State (Synchronous) Analysis Mode 300 Mb/s State Mode 667 Mb/s State Mode
tWidth* [1, 2] 850 ps*, 550 ps typical 850 ps*, 550 ps typical
tSetup 0.5 tWidth 0.5 tWidth
tHold 0.5 tWidth 0.5 tWidth
tSample range [3] –4 ns to +4 ns –4 ns to +4 ns
tSample adjustment resolution 80 ps typical 80 ps typical
tSample accuracy, manual adjustment ± 300 ps ± 300 ps [4]
Maximum state data rate 300 Mb/s (DDR) 667 Mb/s (DDR)
600 Mb/s (Dual sample) 1066 Mb/s (Dual sample)
Maximum channels on a single time base 340 – (number of clocks) 306 – (1 clock)
and trigger [5]
Memory depth – 16950B [5] Option 001: 1 M samples Option 001: 1 M samples
Option 004: 4 M samples Option 004: 4 M samples
Option 016: 16 M samples Option 016: 16 M samples
Option 032: 32 M samples Option 032: 32 M samples
Option 064: 64 M samples Option 064: 64 M samples
Memory depth – 16951B [5] 256 M samples 256 M samples
Number of independent analyzers [6] 2 1
Number of clocks [7] 4 1
Number of clock qualifiers [7] 4 N/A
Minimum time between active 3.33 ns 1.50 ns
clock edges* [8]
Minimum master to slave clock time 1 ns N/A
Minimum slave to master clock time 1 ns N/A
Minimum slave to slave clock time 3.33 ns N/A
* Items marked with an asterisk (*) are specifications. All others are characteristics.
[1] Minimum eye width in system under test.
[2] Your choice of probe can limit system bandwidth. Choose a probe rated at 1066 Mb/s or
greater to maintain system bandwidth.
[3] Sample positions are independently adjustable for each data channel input. A negative
sample position causes the input to be synchronously sampled by that amount before
each active clock edge. A positive sample position causes the input to be synchronously
sampled by that amount after each active clock edge. A sampling position of zero causes
the input to be synchronously sampled coincident with each clock edge.
[4] Use of eye finder is recommended in 667 Mb/s state mode.
[5] In 300 Mb/s state mode, with all pods assigned, memory depth is half the maximum
memory depth. With one pod pair (34 channels) unassigned, the memory depth is full. One
pod pair (34 channels) must remain unassigned for time tags in 667 Mb/s state mode.
[6] Independent analyzers may be either state or timing. When the 667 Mb/s state mode is
selected, only one analyzer may be used.
[7] In the 300 Mb/s state mode, the total number of clocks and qualifiers is 4. All clock and
qualifier inputs must be on the master modules.
[8] Tested with input signal Vh = +1.125 V, Vl = +0.875 V = 1 V/ns, threshold = +1.0 V,
tr/tf = 180 ps ± 30 ps (10%, 90%).