[time-nuts] HP Z3801 RB fun
kb8tq at n1k.org
Tue Dec 27 22:26:10 EST 2016
The loop gain (as Magnus mentions) needs to be “correct” for the Rb. Since
the firmware was written for an OCXO with a PPM(ish) trim range, tacking it
on to a PPB(ish) Rb will require everything to scale by 1000:1. That’s 10 bits.
You may (or may not) have enough resolution in the math to handle that wide
a range of gains.
If it really locks, the next challenge is to get it to run out to ~ 4 days or more on
the loop. Again, the firmware may or may not be able to do this. Unless it does,
the Rb really will not lock as well as it could.
With the Rb correctly locked, you still have the issue that the Rb is a 2x10^-11
device (ADEV) at 1 second and the OCXO probably is 2x10^-12 (or better). For
driving a counter with a normal gate, the Rb may not show much (if any) improvement.
There is the slight chance the ADEV will get the firmware confused…..
Since the firmware source is running around somewhere, you *could* dig into
all of this wonderful stuff :)
> On Dec 27, 2016, at 8:33 PM, paul swed <paulswedb at gmail.com> wrote:
> OK so could not resist. The replacement of the HP 10811 with a PIEZO ocxo
> went very well. So in for a penny in for a pound. Why not an RB. Having a
> few around. Choice efratom FRS-c. Lot of thought went into this choice. Not
> really it was there. It has a inverse frequency to EFC voltage just like
> the piezo and opposite the HP 10811.
> Well it connects runs and locks using the simple 1 opamp level ofset and
> and phase inverter.
> But locked is a funny comment. According to HPsatstat and LH. Its pll is
> locked. But both scope and the LH graph shows a very slow frequency drift
> that with the offset I can make positive or negative. A cycle at 10 ns
> takes quite some time about an hour compared to a TBolt and another non
> modified Z3801. At this level of resolution some things start to become
> I almost wonder if the modified z3801 under some level of drift assumes
> everything is locked and fine. Can it be pushed beyond expectations. I
> suspect it can.
> Anyhow as I say a bit of GPSDO fun.
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
More information about the time-nuts