[time-nuts] HP counter basic oscillators
Richard (Rick) Karlquist
richard at karlquist.com
Wed Dec 28 00:48:38 EST 2016
On 12/27/2016 11:48 AM, GandalfG8--- via time-nuts wrote:
> All the 5313x counter basic oscillators are indeed extraordinarily poor,
> and I've always assumed that their only purpose was to demonstrate that the
> unit was basically functional.
I don't know exactly what is in the 5313X counters, but it might be
like what was in the 5334A counter that I inherited when I was
project manager on the 5334B counter. Namely, it was an oscillator
utilizing an ECL line receiver as the active device. The same
design had been used at my previous employer (Zeta Labs). The
design is championed by engineers who know nothing about designing
crystal oscillators. Unlike designs using a transistor, the gate
oscillator requires no skill to design and will pretty much
oscillate without fail, and will never squegg. There is also no
need to convert a sine wave to a logic level signal.
What's not to like about that? A lot, it turns out. The temperature
stability and phase noise are horrible, you have no control over the
amount of drive to the crystal, and it burns a lot of current for no
I think the usage model breaks down into 3 categories:
1. You are using the counter for some relatively crude measurement
for example of the frequency of a 555 oscillator, so even the Mickey
Mouse oscillator is good enough.
2. You need real precision and should move right up to a 10811.
3. You have an external 10 MHz reference. For example, at HP,
we had a cesium standard that was distributed around the plant.
This was called the "house standard". At Zeta Labs we had
an HP105 for the house standard.
More information about the time-nuts