Low-noise frequency divider

[kpjamro]

What does it mean for a frequency divider (aka pre-scaler) to be "low noise?  The output is stated as being "rectangular".  The output wave form is attached.  This is dividing 100 MHz down to 10 MHz.  Is only phase noise relevant here?  I cannot measure phase noise directly.  Would a MDEV measurement be meaningful?


Ken

 

[Benta]

One can only guess, but it sounds a bit like a marketing pitch.
1: low noise input for low amplitude analog inputs.
2: low irradiated noise from the power supply lines.

[eugene]

What could possibly go wrong at 100 MHz? 

[MegaVolt]

Low Noise = Low Jitter

[jonpaul]

Comparateur or Schmitt has jitter, creating phase noise

Jon

[tom66]

Usually phase noise, which is typically a result of Johnson noise on the input buffer into the divider block, or the internal logic within, which will cause jitter on the output.

The result of phase noise is that the fundamental at 10MHz is 'smeared' over a larger range, it is no longer a sharp, singular point.  Of course, in reality, you can never have such a perfect input frequency to begin with, and barring some devices which 'clean' input frequencies, the phase noise is going to be determined by the worst component in the chain.

The consequence of this is that it reduces timing margins for downstream components, degrades ADC and DAC performance, and annoys audiophools who believe they can hear picoseconds of jitter.

[jonpaul]

Jitter is  critical in scopes trigger, time interval  measurement, GPS, nuclear research, at pico..femtosecond scale.

Digital Audio ADC DAC clock, see Dr Steve Harris 1986 AES / Crystal Semiconductor paper on effects of Jitter on digital audio.

We used wideband follower, buffers and MECL in 1970s.

See classic HP time interval counters 5370A, 5372A etc.

Jon

[kpjamro]

Thanks. 

It looks to me like the ragged top edge of the waveform is not random noise but rather a "feedthrough" of the input 100 MHz signal.  Since the output signal is perfectly synchronized with the input signal (it's a frequency divider) then this "noise" will not affect the phase of the 1->0 transition of the output and not contribute to phase noise.

In other words the raggedy top of the waveform itself does not contribute to phase noise and maybe does not matter.  It could be removed by passing the signal through a "squaring amplifier" or Schmidt trigger IC.  But you cannot tell by eyeball if that 1->0 transition is moving around in time.  I suspect it would cause a small variation in frequency (less than 1E-8) and so it might show up in an MDEV measurement. That's why I asked.