[time-nuts] HP5065A C-field current is temperature sensitive

[Poul-Henning Kamp]

--------

I had this morning, and put it into action right away:

I disconnected the A15R8 resistor from the tempco afflicted A15CR5 zener diode,
and hooked it up to my Fluke 732A's 10V output (which is speced to 10mA).

I had expected to see a vast improvement in C-field drive stability, but
while the improvement is measurable, it is a lot less than I had expected.

Having eliminted that one, there are two influences left:

* Noise from the +20 V supply leaks through

The data evidently shows this to be the case, but the correlation does
not explain all the C-field current variability.

* Tempco in the constant current driver circuitry

I'm increasingly leaning this way, because even though A15R8 is fed
from the rock-stable Fluke, the voltage at the A15Q6A base still
varies quite a lot with temperature, but the short term jitter is
much better.

I don't have enough data yet to conclude if the MVAR floor is improved,
but it looks quite promising so far:  The curve has very confidently
dipped below 1e-13 at 5000 seconds.  It used to flatten at 3e-13.

Next experiment is going to be driving the C-field solenoid directly
from the Fluke 732A through a suitable low tempco current-limiting
series resistor.

Anybody want to bet what *that* will do to the MVAR floor ?

I'll write this up on my web-pages once I have more data.

-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.