Hi,
I have a workplace question. I have a multi clock system where there are
master and standby clock sources feeding various PLLs. These clock
frequencies are in the range 10MHz to 125MHz and are either LVDS or
PECL. I would like to check the performance of the system when the clock
source to the PLL switches so what perturbation, down to the ps level. I
do have at my disposal a 1GHZ 4 chanel DPO3000 real time scope, a 54750A
sampling scope and a spectrum analyser. It did also cross my mind that
the old Modulation Domain analysers of the past might be useful, I have
also used a CNT-91 in the past. Would a 53230A be up to it, just time
stamp and then post process?
For the high-bandwidth stuff, the HP 54750A should be my weapon of
choice, the eye-measurement can measure jitter below what CNT-91 and
53230 would be envious off.
To measure the jitter of the trigger point you need a delay line
somewhat longer than the 22 ns delay from trigger point. What you do is
that you mount a good wideband power-splitter on the trigger input and
then the delay line to the channel input. This way you can measure your
trigger jitter and also the jitter of the subsequent transition, and
then you can subtract the first jitter from the second by first squaring
both, subtract and then take the square root.
Now, next you can go on to measure the other signal with the base as a
trigger. This will only work well if they have relatively simple
relationships to each other.
For lower frequency noises, the CNT-91 and TimeLab would be my choice of
weapon.
Depending on how good or bad your system is, you might need a better
system or this will work. Regardless, this will be a start and I think
you can make good use of the tools you have to get a start and a feel.
Cheers,
Magnus
Hi,
On 12/31/2017 07:50 PM, time_nuts@jsalisbury.clara.co.uk wrote:
> Hi,
>
> I have a workplace question. I have a multi clock system where there are
> master and standby clock sources feeding various PLLs. These clock
> frequencies are in the range 10MHz to 125MHz and are either LVDS or
> PECL. I would like to check the performance of the system when the clock
> source to the PLL switches so what perturbation, down to the ps level. I
> do have at my disposal a 1GHZ 4 chanel DPO3000 real time scope, a 54750A
> sampling scope and a spectrum analyser. It did also cross my mind that
> the old Modulation Domain analysers of the past might be useful, I have
> also used a CNT-91 in the past. Would a 53230A be up to it, just time
> stamp and then post process?
For the high-bandwidth stuff, the HP 54750A should be my weapon of
choice, the eye-measurement can measure jitter below what CNT-91 and
53230 would be envious off.
To measure the jitter of the trigger point you need a delay line
somewhat longer than the 22 ns delay from trigger point. What you do is
that you mount a good wideband power-splitter on the trigger input and
then the delay line to the channel input. This way you can measure your
trigger jitter and also the jitter of the subsequent transition, and
then you can subtract the first jitter from the second by first squaring
both, subtract and then take the square root.
Now, next you can go on to measure the other signal with the base as a
trigger. This will only work well if they have relatively simple
relationships to each other.
For lower frequency noises, the CNT-91 and TimeLab would be my choice of
weapon.
Depending on how good or bad your system is, you might need a better
system or this will work. Regardless, this will be a start and I think
you can make good use of the tools you have to get a start and a feel.
Cheers,
Magnus
