[time-nuts] another Ebay mixup, 5370

Dr Bruce Griffiths bruce.griffiths at xtra.co.nz
Sun Jun 10 10:04:47 EDT 2007

Tom Van Baak wrote:
>> Measurements taken from 3325B at 1 MHz:
>> 5370B reference:       INT          EXT (same 5087A as 3325B uses)
>> Square wave, 2v pk:    15-20 ps     11-12 ps
>> Sine wave, 2v pk:      50-70 ps     30-35 ps
>> Square wave, 1v pk:    20-25 ps     12-13 ps
>> Sine wave, 1v pk:      60-80 ps     40-45 ps
>> Square wave, .1v pk:   40-60 ps     40-45 ps
>> Sine wave, .1v pk:     600-900 ps   500-600 ps
>> -- john, KE5FX
> Very nice data. Thanks for staying up late to make the measurements.
> What trigger levels and input configuration did you use for this?
> Based on these results can someone calculate how much jitter is
> due to the 5370B input circuits alone and how much is due to the
> rest of the circuitry of the unit?
> /tvb

Since, at various times the HP5370 used a discrete ECL gate crystal 
oscillator, an HP10544A, and an HP10811A as the internal reference, it 
would also be useful to indicate which of these a 5370 uses as its 
internal reference.
The time interval mode (+TI only or +-TI) should also be stated togeter 
with input impedance settings and attenuator settings.

Taking the external 1MHz sinewave (EXT reference for HP5370) and 
assuming input attenuator set to 1x, and the 2 inputs are connected in 
parallel to the START BNC then the corresponding trigger circuit noise 
for 0.05V pk input (input signal is attenuated by 2 at the trigger 
circuit inputs for 50 ohm Zin ) is about 123uV rms for both the START 
and STOP channels which seems plausible.This gives about 50ps noise 
induced jitter at 1V peak amplitude. However since the time interval 
mode isnt specified nor are the number of samples taken, these 
calculations are somewhat uncertain.
Also the intrinsic jitter of the source is not known (to me at least). 
At 2V peak the noise induced trigger jitter could be 25ps with other 
noise say 10-12ps.

The SR620 has a wider input trigger circuit bandwidth and can thus be 
expected to be somewhat noisier. Both counters use ECL and/or similar 
performance logic for the high speed sections.


More information about the time-nuts mailing list