[time-nuts] HP 5316B External Reference

[Richard (Rick) Karlquist]

On 7/11/2020 3:51 PM, Wes wrote:
> As I understand it this counter has an external reference input that 
> isn't used directly as the time base but injection locks the internal 
> time base.  Does anyone here know how well this works. Is using a GPSDO 
> as a reference a worthwhile accuracy improvement?
> 
> Wes Stewart  N7WS
> 

I worked in Frequency Counter R&D at the HP Santa Clara division
around the time the 5316B introduced, but didn't know about
this detail.  AFAIK, most counters we made did NOT work this way.
Whatever external reference you fed in (warts and all) was used
to clock the counter.

OTOH, most RF instruments (other than counters) used a PLL
(not injection locking) to lock the internal oscillator to the
external reference.  IMHO, this architecture only made sense
on sig gens and spec ans, not on counters.

I will point out that the 5316B was a cost reduced version of the
5316A.  I was the project manager for the 5334B, which was a
cost reduced version of the 5334A.  I believe there was a third
model where the B version was cost reduced from the A version,
but I don't remember which one.  As a group, these were called
the "killer B's" because they would supposedly kill off Racal
Dana sales or some silly marketing nonsense to that effect.

It's possible that for some obscure reason, a scheme was employed to 
save money that superficially resembled injection locking.
I'm guessing a multifunction chip was used that had a
built in oscillator with an external crystal, but no access to the 
connection from the oscillator output (on the chip) to the downstream 
circuit.  Are you sure that the 5316B actually uses injection locking, 
rather than having the external reference simply drive the crystal pins
on the IC?  This would be easy to verify by feeding
in an external reference that was say 100 ppm off frequency
and seeing if the counter still worked OK.  No injection
locking scheme AFAIK would ever pull that far.  If it actually
is true injection locking, the problem is that whether it worked
would depend on the relationship between the free running
frequency and the external reference frequency, of course.
IOW, it might work in some cases and not in others.

Rick N6RK