here's a link to "Test and Measurement World" article on the measurement of jitter:
http://email.tmworld.com/cgi-bin2/flo/y/eLfG0GxWoA0FxQ0Bvb60AH
http://email.tmworld.com/cgi-bin2/flo/y/eLfG0GxWoA0FxQ0Bvb60AH
Most of these products might be OK for some digital audio clock
measurements, but none of them particularly impress me.
They are all oriented to looking at jitter in digital circuits --
buses, etc. If you want to look at just one really good
clock signal they fall short.
I've seen several of these systems in the past few weeks.
The Wavecrest has a lot of s/w bells/whistles. However
actual jitter measurement is not all that great.
Agilent scope is nice; didn't have a chance to check it against
a good reference.
Tek will be seen tomorrow; we'll try to check out one
of Guido Tent's very nice clock oscillators
What one really needs is a really good phase noise
measurement system.
measurements, but none of them particularly impress me.
They are all oriented to looking at jitter in digital circuits --
buses, etc. If you want to look at just one really good
clock signal they fall short.
I've seen several of these systems in the past few weeks.
The Wavecrest has a lot of s/w bells/whistles. However
actual jitter measurement is not all that great.
Agilent scope is nice; didn't have a chance to check it against
a good reference.
Tek will be seen tomorrow; we'll try to check out one
of Guido Tent's very nice clock oscillators
What one really needs is a really good phase noise
measurement system.
this was on the ham newsgroup last night
From KE9OA, discussing spectrum analyzers:
"One way around this problem is with a low frequency analyzer, such as the
3582, using a clean signal source with a mixer to translate the frequency
down. The HP 70000 Phase Noise test system that we used at Rockwell-Collins
used this setup. A 3585 was used out to a 40MHz offset, while the 3582
would do the close-in stuff. The system could measure down to -154dBc. We
used to use this system for characterizing OCXOs when I was working in the
Synthesizer Group. I know that it is cumbersome, but it could work, if you
have a clean signal generator to use as an LO source. I think one of the
best generators available was the HP 8642 series. They used that modulated
fractional N divider
(MFD, Sigma-Delta Modulation), ending up with a phase noise spec
of -153dBc.
Sorry about the rambling! A 3580 would do the trick in your case, along
with a Mini-Circuits SRA-3 mixer.
From KE9OA, discussing spectrum analyzers:
"One way around this problem is with a low frequency analyzer, such as the
3582, using a clean signal source with a mixer to translate the frequency
down. The HP 70000 Phase Noise test system that we used at Rockwell-Collins
used this setup. A 3585 was used out to a 40MHz offset, while the 3582
would do the close-in stuff. The system could measure down to -154dBc. We
used to use this system for characterizing OCXOs when I was working in the
Synthesizer Group. I know that it is cumbersome, but it could work, if you
have a clean signal generator to use as an LO source. I think one of the
best generators available was the HP 8642 series. They used that modulated
fractional N divider
(MFD, Sigma-Delta Modulation), ending up with a phase noise spec
of -153dBc.
Sorry about the rambling! A 3580 would do the trick in your case, along
with a Mini-Circuits SRA-3 mixer.
Hi,
Measuring jitter for audio is pretty difficult. Some people claim you need an clock oscillator in your CD player with jitter as low as 1 Ps. Measuring that low? And is that really needed? I wonder. It is possible to measure down to 1 Ps with modern and very expensive Tek TDS scopes.
For DIY you are left with the technique of measuring side band noise and calculating RMS jitter from it. That involves a mixer and a “jitter free” reference oscillator indeed. The down converted spectrum can be analysed with a 24bit-soundcard and FFT software. But for the reference oscillator you do not you need a top-end reference signal generator. Simply build two identical clock oscillators. The side band noise of these two is not correlated then. So if you mix these oscillators you end up with a 3 dB higher side band noise spectrum compared to a single oscillator mixed with a “jitter free” reference oscillator.
Calculating jitter from side band noise is rather cumbersome. But is that needed? In general the lower the side band noise and the smaller the bandwidth of it, the better.
Measuring jitter for audio is pretty difficult. Some people claim you need an clock oscillator in your CD player with jitter as low as 1 Ps. Measuring that low? And is that really needed? I wonder. It is possible to measure down to 1 Ps with modern and very expensive Tek TDS scopes.
For DIY you are left with the technique of measuring side band noise and calculating RMS jitter from it. That involves a mixer and a “jitter free” reference oscillator indeed. The down converted spectrum can be analysed with a 24bit-soundcard and FFT software. But for the reference oscillator you do not you need a top-end reference signal generator. Simply build two identical clock oscillators. The side band noise of these two is not correlated then. So if you mix these oscillators you end up with a 3 dB higher side band noise spectrum compared to a single oscillator mixed with a “jitter free” reference oscillator.
Calculating jitter from side band noise is rather cumbersome. But is that needed? In general the lower the side band noise and the smaller the bandwidth of it, the better.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.