[time-nuts] Re: TSC5110A noise floor?

[John Miles]

> I didn't try it, but it occured to me that there might be some benefit in
> replacing the 7809/7909 pair with low noise equivalents, but maybe there
is
> a second VR stage or just adequate filtering so it wouldn't matter.
> 
> Does anyonw know what sets the noise floor of a DMTD device using ZCDs
> such as the TSC5110A?  Also, I believe that the 53100A has a lower noise
> floor - how was this achieved?

Hi, Jim --

Interesting to see two independent failures in the 78xx regulators.  Any
idea how that might have happened?  From what I've seen, when a
three-terminal regulator fails, it's often because the design places more
capacitance downstream of the regulator than upstream.  At power-down time,
the reverse current causes the pass transistor to fail with a C-E short.
Then, the next time you power the unit up, all of the input voltage is sent
to the load and your day gets worse from there.  Sounds like you got very
lucky in this case.

The 5110A is basically a digital DMTD.  The ZCDs measure a signal that has
been mixed down to a low IF by either one DDS or two DDSes with a common
clock, depending on how far apart the input signals are.   The instrument
floor is largely driven by its noise bandwidth, which is near 450 Hz
according to one paper (Sojdr et al. from 2003 IFCS).  

Not knowing anything about the 5110A's actual circuitry, I'd have to assume
that the ZCD has enough power-supply immunity to reject noise from a 78xx
regulator.  In any event physical proximity wouldn't be the area of concern
as far as noise goes.  They are probably keeping the regulators outside the
ZCD enclosures for thermal stability reasons.  So there is unlikely to be
any room for improvement by using different regulators alone.  Try it,
though!  Maybe you've still got some luck left. :)

Later test sets including the 5115A/5120A/5125A and 53100A are entirely
different animals, in which the ZCDs are replaced with RF ADCs.  You can
think of the 5110A as a narrowband time-interval counter and the newer
instruments as narrowband radios.  The latter can support measurement
bandwidths 10x to 100x lower than the 5110A, and they have some other
advantages such as additional channels for cross-correlation. 

-- john