[time-nuts] Replacement A9 boards for the HP 5065A
phk at phk.freebsd.dk
Mon Feb 26 17:05:52 EST 2018
In message <CAD2JfAjxVwO3coVTivD-Yp=XZqd78WKBDwKD0KA9fc8dq6GeOQ at mail.gmail.com>, paul swed writes:
>I reverse engineered the system and guess what cap was bad? The integrator.
>So not being all that smart, I hooked 2 X 10 UF caps in series. Been
>working like a champ for 18 years.
That capacitor isn't nearly as important as people think.
The input signal to the integrator is continuous and jump-free and
the relevant timeconstant is sub-second. Dielectric absorption
doesn't matter when there are no voltage jumps.
We all tune the EFC of the Xtal to set the meter to zero CONTROL
voltage, which means there is no voltage for the capacitor to leak,
so that doesn't matter either,
If you pick a capacitor with a couple of hundred volts rating, its
leakage current will be less than the air and the PCB near it anyway.
When I experimented, I could hardly find *any* property that mattered
for that capacitor, not even the exact capacitance, because the
adjustment procedue handles that.
The only thing I could provok, was that almost all capacitors I
tried were sensitive to touch.
I didn't establish if this was mechanical (and if so if it was the
capacitor or something else on the board) or if it was thermal
(capacitors have astounding tempcos).
HP tied two O-rings around the capacitor they choose, I pressume
that is a clue that they found something similar.
The biggest issue is probably that most of the relevant capacitors
are square blocks, like for instance TDK/KEMET B32774D8505K.
TDK/KEMET C4GAJUD4500AA3J could be an option, but I suspect it
is too big to fit in the existing PCB.
Either way, cheap and plenty available.
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.
More information about the time-nuts