[time-nuts] 33120A purchase off epay

jimlux jimlux at earthlink.net
Mon Mar 6 10:22:50 EST 2017

On 3/5/17 9:14 PM, Christopher Hoover wrote:
> My HP 33120A Function Generator died. (I'm still trying to fix it, but it
> isn't obvious what's wrong ... rails check so I'm suspicious the EEPROMs
> gave up the ghost.)
> I bought a new one on epay week befor last for a fair price.
> It arrived and I was surprised/excited to find it had the external
> reference option. Just what a TimeNut needs.
> The external ref option wasn't mentioned explicitly in the listing,
> although I can see it (in retrospect) in one of the pictures.

I don't know about on the 33120, but on the 33620 series, the external 
reference doesn't actually phase lock the internal clock for the ARB. 
It's more of a frequency reference, and the output is steered to match 
the frequency of the external reference. (counters and NCOs)  The phase 
noise of the 36620 is entirely determined by the internal oscillator, 
and the ADEV is a bit tricky - it's like the ADEV of a disciplined 
oscillator, not a phase locked loop.

There are ways to sync multiple 33620 and 33120s (and the earlier 3325) 
so that you have multiple outputs with a phase coherent relationship. 
You have to do it manually.. feed them all the same reference, then 
adjust til they're synced, then they'll stay in sync.

Here's Agilent's ap note on externally syncing 33120s

about your option 001

service manual and schematic of your Opt001

(From Keysight about the 33620 series:
When using an external 10 MHz time base with this instrument, a PLL 
syncs together the internal oscillator with that external signal, as you 
have surmised. To my understanding, because of this however, the 
stability of the function generator’s time base is still dependent on 
the internal oscillator. If an OCXO is being used in the 33600, the 
instrument will be limited to a input frequency range of 10 MHz ± 1Hz. 
Therefore, even if an external input signal at 10 MHz was stable to 10 
ppb, for instance, the instrument would still be limited to a 100 ppb 
stability because of its’ own internal oscillator.  This means feeding a 
more stable reference signal into the function generator, will likely 
not improve performance. Again, this is to my own understanding of how 
this instrument works and I am going to double check this with our 
factory to see if they agree or disagree. I will update you when I have 
that confirmation.

More information about the time-nuts mailing list