Pico Technical Support and Discussion Forum

Wavelength · User Joined: 01 Apr 2008 Posts: 4

Gang,

I need to test the noise on power supplies for clock oscillators. The typical method is to use a spectrum analyzer at the lowest frequency base with AC inputs and get a spectrum of the noise using average and log X axis.

The problem with R5 is that while you can do this and for low frequencies (0-73Hz) the problem is the software does not show anything below -110dB.

With a 12 bit signal analyzer like the HP 356X series they usually go to -130dB and -150dB for 14 bit.

I tried using the ADC-216 which does allow up to -150dB on R5 but it has it's own set of problems.

With the 3224 and R5 I can first do a shorted probe measurement to get a baseline for noise. Then merge that with the UUT and get a better result. Unless the noise floor is below -110 for which then it just dies.

Another problem which I noticed yesterday is that everytime the R5 software does an average it opens a new window. I could not understand why R5 kept crashing until I realized this. I expected it to keep averaging over the same window.

Anyways... I am trying to get some pretty low measurements as I have some new discrete regulators in the 33nV-100nV range. So if anyone can tell me how to use the 3224 to accomplish better testing I would appreciate it.

Thanks
Gordon

picojohn · Site Admin Joined: 11 Jun 2007 Posts: 311

Hello Gordon,

To achieve a merge using PS6 would require that you export the data in spreadsheet format and perform the manipulation there.

Saving a spectrum view as a .csv file is problematic with the current version of PS6, but will be resolved in the next release.

You may be kept informed of new releases by subscribing to our newsletter at http://www.picotech.com/cgi-bin/enewssubscribe.pl

I hope that this is helpful.

Regards
John

Pico Technology

Pico Technical Support and Discussion Forum