Sulzer 2.5C quartz oscillator and power supply
The Sulzer 2.5C is a very high-quality double oven 2.5 MHz
AT-cut crystal frequency standard from the early 1960's.
It is exceptional even by today's standards, easily outperforming
modern OCXO such as the HP 10811A in stability and frequency drift.
Short-term stability of parts in 10-13 and
drift rates of parts in 10-11 per day
are not uncommon.
The oscillator electronics included dividers producing
1 MHz and 100 kHz outputs in addition to the primary 2.5 MHz output
(or should we say 2.5 Mc ;-).
The 2.5C crystal oscillator often came in an open 19" rack
which included the model 2.5P 24 VDC NiCad battery backup power supply.
See Sulzer 2.5P schematic for details.
Some of these oscillators were labeled Tracor after Sulzer was acquired.
After 30 to 40 years (were the 60's that long ago!)
some old Sulzer oscillators either no longer work
or they have excessive phase noise
or frequency instability
(e.g., the proportional oven circuits have problems).
The one below appears to have survived the years quite well
both cosmetically and electrically.
Measured Performance Summary (see below or click to enlarge images)
Oscillator Short-term Performance (large images)
The four stability plots were obtained with a
Time Interval Analyzer frequency stability measurement system
and a homebrew scheme to produce online webcopy images
rather than printer hardcopy plots.
With 0.1 picosecond (100 fs) resolution and an embedded PC
the 5110A calculates and displays
auto-scaled log-log Allan Deviation plots in real-time
for tau as short as 0.01 second and as long as you are patient.
The noise floor of the 5110A is on the order of 10-17
and a Hydrogen Maser was used for the reference channel
so the measurement system is typically orders of magnitude more stable
than the unit under test.
Thus for all tau up to a day or week
the plots below can be assumed to be absolute
rather than relative to the reference oscillator.
- Allan Deviation
is a standard log-log plot of frequency stability.
Tau ranges from 0.01 or 0.1 seconds out to
104 or 105 seconds
(depending on how long the measurement run was).
The plots are true Allan Deviation when the sampling mode is
and Modified Allan Deviation when the mode is "Averaged Phase".
The mode is indicated on the lower left of the plot.
- Allan Deviation Table
are the data points, including uncertainty, for each tau plotted.
- Phase Difference
is the raw phase difference between
the reference and the oscillator under test.
After an hour into a run this plot is less interesting for quartz standards
(since the 5110A does not remove frequency drift during a run).
- Frequency Difference
shows the computed two sample difference in frequency
at one second sample times,
plotted with 60 seconds per grid line,
for approximately 10 minutes full-scale.
The plot shows the most recent 10 minutes
of differences regardless of the duration of the run.
This plot graphically reveals oscillator performance
from about 1 to 1000 seconds:
the magnitude and character of the wiggles within grid lines
reflect stability on the order of seconds to tens of seconds
while the overall character of the band across the plot shows stability
on the order of hundreds of seconds.
The Allan Deviation for tau between 1 s to 100 s are
statistical summaries of this information, of course,
but sometimes seeing the raw wiggles of a phase plot
can give insight into oscillator behavior.
Oscillator Long-term Performance
Return to Museum of aging oscillators index.
Return to LeapSecond.com home page.
Email stories/comments/questions to tvb.
Page last modified 21-July-2001.