[time-nuts] Datron 4910/11/12 LTZ1000 Reference Board (finally)
wje at quackers.net
Sun Feb 8 14:28:47 UTC 2009
I've finished retracing the LTZ ref board I promised months ago. While
it somewhat resembles the LTZ1000 data sheet example, it differs
significantly in a number of details. The heater control is far more
complex, using a PWM scheme. Take the schematic with a bit of a grain of
salt... tracing it is difficult because there is no access to the back
of the board, at least not without more surgery than I'm willing to do
on my secondary reference! In particular, the internal resistor
connections in the Vishay precision resistor network are just guesses.
If anyone has any info about Vishay TO5 canned precision networks from
the 80's, please share. I'm confident about the heater side circuitry,
but less so about the output side. You will also need to refer to the
Datron cell schematics already posted, since significant portions of the
heater control as well as the output buffer amp are on that board. If
anyone notices any highly questionable topology, let me know and I'll
try to confirm it. I haven't spent the time to do much of a sanity check
One final note - although not apparent from the schematic, the ref board
itself makes extensive use of guarding. This is critical for
ultra-precision refs, since even picoamps of leakage in the wrong place
can easily lead to multi-PPM errors. I would estimate that fully 50% of
the total trace length on the board is guard traces.
As a bonus, I've also scanned the 10V reference from my Solartron 7081
8.5 digit voltmeter. This is an HP 3458 class voltmeter. The interesting
(and amazing) thing about this reference is that it does NOT use an
LTZ1000, or even an ovenized reference. It uses a 1N829 Zener with some
clever linear and non-linear temp compensation. My meter routinely
achieves a 1-year drift of less than 1ppm compared against my
(calibrated, NIST-traceable) Datron ref.
They are all on my FTP site. Look for DatronRef.PDF,
SolartronReadme.txt, SolartronRef-1.PDF, and SolartronRef-2.PDF.
login: ftp at quackers.net
or via a browser:
They said 'Windows or better'
so I used Linux.
More information about the time-nuts