This is getting pretty intense for a Sunday!! Both Mr. Javor and Mr Novak make excellent observations. Both center on the method of measurment as a point of concern.
To measure the voltage ripple I used a high badwidth (>1GHz) sigle ended probe with very short leads. In order to establish the error margin, I used the null experiment technique. (I don't have a diff probe with sufficient bandwidth to hand). I then used a piece of coax with very short leads (just like Mr Javor recommends) and a 50ohm resistor in series with the signal to feed the same voltage ripple to a SA. Clearly there is a voltage division here but thats easily accounted for. My concern is that the impedance of the planes is so much lower than the 50ohm input of the SA and I wanted to "match" that as much as possible. I belive I have taken care of as much of the measurement problem as possible. Still, the maximum voltages between the two measurments do not come close at all. From: owner-emc-p...@majordomo.ieee.org [mailto:owner-emc-p...@majordomo.ieee.org]On Behalf Of Ken Javor Sent: Sunday, July 27, 2003 10:27 AM To: istvan novak; Charles Grasso; Emc-Pstc Subject: Re: Q on Correlation of Votage ripple with a Spectrum Analyser Mr. Novak makes some excellent points. I was under the assumption that the phrase "voltage ripple" implied conducted emission measurements at a LISN port. Hence my comments on mode separation. Across a spectrum of even 30 MHz, any normal scope probe I know of (1 or 10 M Ohm in parallel with 5/10/20 pF) will present a varying load to the measured waveform. I know there are some broadband active differential probes with low shunt capacity, but I have no experience using them. The scope should be made to look like a flat 50 Ohm load to help correlate to a spectrum analyzer. Instead of using long leads to make measurements from one point of the Vcc plane to another, I would make coaxial measurements from one point on the Vcc plane to chassis ground or the reference/image plane if one exists, and then make the same measurement at another point on the Vcc plane, and compare the waveforms, perhaps using a built-in "math function" if your scope has that. Of course this technique requires the reference/image plane/chassis ground to be an equipotential plane... By coaxial measurement I mean using coax rather than a scope probe, and terminating the shield at the reference point, while extending the center conductor just far enough to connect to the point of interest on the Vcc plane. Theoretically, an even better technique would be to have a place on the board where the power plane reference was available circumferentially around a Vcc via, and connect the coax shield to the reference plane and the center conductor to the Vcc contact. Given a 50 Ohm load at the other end (with a blocking cap to protect it), and an FFT capability with enough memory, I believe you could achieve correlation with the individual spectral components measured by a spectrum analyzer (which of course would also need a blocking cap with this config). > From: "istvan novak" <istvan.no...@worldnet.att.net> > Date: Sun, 27 Jul 2003 11:02:02 -0400 > To: "Charles Grasso" <cgrassospri...@earthlink.net>, "Ken Javor" > <ken.ja...@emccompliance.com>, "Emc-Pstc" <emc-p...@ieee.org> > Subject: Re: Q on Correlation of Votage ripple with a Spectrum Analyser > > Charles, > > Doing this kind of correlation is very difficult for the following reasons: > - unless you measure a very simple and dummy system, hardware today is > so complex that you cant predict for sure its activity; it is a strong > function > of time. > -tThe spectrum analyzer and scope will look at the same signal > in different ways: analog spectrum analyzers have a seep time and settling > time determining the frequency and aperture of visit each frequency. > If you have a spectrumn analyzer used for compliance tests, > probably the CISPR filter is on. Scopes on the other hand (digital scopes) > undersample the signal, whether it is called real-time or not. Memory > and displey refresh rate does not allow scopes to display and process all > data points of high-frequency signals. Real-time scopes do it for a > given time window, but it is usually way less than the time constant of > a CISPR filter on the spectrum analyzer. > - connection to the source makes a big difference. I assume when you > calibrated the reading with a sine wave, a coaxial cable with coax > connectors > at both ends was used. Presumably the product does not have a coaxial > connector on the Vcc plane, so you have to make your own connection or > use a hand-held probe. This is very extra noise usually gets in the path, > and > the scope reading becomes unrealistically high. I have found no active > scope probes so far, which would give a correct reading in a noisy > environment. > We hopefully should not see noise on the Vcc planes more than a few > hundred mV. In contrast, many scope probes can pick up spikes as big as > volts from the environment. If you want to measure noise levels below > 100mVpp, > double-shielded coax is necessary in noisy environments. Here the 'noisy > environment' refers to the close vicinity of the point you test. The > simplest > test is: take your present probe, and hook up a good double-shielded coax > to the same points. Check both readings on the same scope at the same time, > and compare. > > I hope this helps. > > Best regards, > Istvan Novak > SUN Microsystems > > ----- Original Message ----- > From: "Charles Grasso" <cgrassospri...@earthlink.net> > To: "Ken Javor" <ken.ja...@emccompliance.com>; "Emc-Pstc" > <emc-p...@ieee.org> > Sent: Saturday, July 26, 2003 9:11 PM > Subject: RE: Q on Correlation of Votage ripple with a Spectrum Analyser > > >> >> Hi all, >> >> Actually I was using a good ole Spectrim Analyser >> so I sidestepped the windowing issue/software issues >> altogether. >> >> What I was(am)trying to do was match the max voltage >> as measured on a scope with the value as measured >> on a SA. >> >> I first calibrated myslef using a known source - a sine wave. >> The amplitudes fell in just as theory predicted. Encouraged, >> I then probed the Vcc plane on a product I was working on >> and was not so happy!! >> >> Any ideas? >> >> -----Original Message----- >> From: Ken Javor [mailto:ken.ja...@emccompliance.com] >> Sent: Saturday, July 26, 2003 4:19 PM >> To: Charles Grasso; Emc-Pstc >> Subject: Re: Q on Correlation of Votage ripple with a Spectrum Analyser >> >> >> I presented a paper on that very subject about a decade ago at one of the >> EMC T&D magazine EMC symposia. I used a Fluke Scopemeter and some FFT >> software that came with it. The Fluke interfaced to the PC through an >> optically isolated RS-232 protocol. It worked quite well from a >> pre-compliance or troubleshooting point-of-view. You could use time >> windowing to separate the signals deriving from leading and falling edges >> from the signals deriving from the pulse itself. I used LISNMATE and >> LISNMARK mode separation devices to show that the rising/falling edge >> signals were common mode, while the pulse itself generated differential > mode >> signals. >> >>> From: "Charles Grasso" <cgrassospri...@earthlink.net> >>> Reply-To: "Charles Grasso" <cgrassospri...@earthlink.net> >>> Date: Sat, 26 Jul 2003 16:18:36 -0700 >>> To: "Emc-Pstc" <emc-p...@ieee.org> >>> Subject: Q on Correlation of Votage ripple with a Spectrum Analyser >>> >>> >>> Hi All, >>> >>> >>> Has anyone tried correlating the voltage ripple >>> as seen on a scope with the amplitudes measured >>> on a Spectrum Analyser? >>> >>> I tried doing that the other day with ..umm. minimal >>> success. I think that due to the comples convoltions >>> that would have to occur when FFT'ing an irregular >>> voltage shape. >>> >>> Charles Grasso >>> Echostar Communications. >>> >>> ------------------------------------------- >>> This message is from the IEEE EMC Society Product Safety >>> Technical Committee emc-pstc discussion list. >>> >>> Visit our web site at: http://www.ewh.ieee.org/soc/emcs/pstc/ >>> >>> To cancel your subscription, send mail to: >>> majord...@ieee.org >>> with the single line: >>> unsubscribe emc-pstc >>> >>> For help, send mail to the list administrators: >>> Ron Pickard: emc-p...@hypercom.com >>> Dave Heald: emc_p...@symbol.com >>> >>> For policy questions, send mail to: >>> Richard Nute: ri...@ieee.org >>> Jim Bacher: j.bac...@ieee.org >>> >>> Archive is being moved, we will announce when it is back on-line. >>> All emc-pstc postings are archived and searchable on the web at: >>> http://www.ieeecommunities.org/emc-pstc >>> >> >> >> ------------------------------------------- >> This message is from the IEEE EMC Society Product Safety >> Technical Committee emc-pstc discussion list. >> >> Visit our web site at: http://www.ewh.ieee.org/soc/emcs/pstc/ >> >> To cancel your subscription, send mail to: >> majord...@ieee.org >> with the single line: >> unsubscribe emc-pstc >> >> For help, send mail to the list administrators: >> Ron Pickard: emc-p...@hypercom.com >> Dave Heald: emc_p...@symbol.com >> >> For policy questions, send mail to: >> Richard Nute: ri...@ieee.org >> Jim Bacher: j.bac...@ieee.org >> >> Archive is being moved, we will announce when it is back on-line. >> All emc-pstc postings are archived and searchable on the web at: >> http://www.ieeecommunities.org/emc-pstc >> > This message is from the IEEE EMC Society Product Safety Technical Committee emc-pstc discussion list. Visit our web site at: http://www.ewh.ieee.org/soc/emcs/pstc/ To cancel your subscription, send mail to: majord...@ieee.org with the single line: unsubscribe emc-pstc For help, send mail to the list administrators: Ron Pickard: emc-p...@hypercom.com Dave Heald: emc_p...@symbol.com For policy questions, send mail to: Richard Nute: ri...@ieee.org Jim Bacher: j.bac...@ieee.org Archive is being moved, we will announce when it is back on-line. All emc-pstc postings are archived and searchable on the web at: http://www.ieeecommunities.org/emc-pstc This message is from the IEEE EMC Society Product Safety Technical Committee emc-pstc discussion list. Visit our web site at: http://www.ewh.ieee.org/soc/emcs/pstc/ To cancel your subscription, send mail to: majord...@ieee.org with the single line: unsubscribe emc-pstc For help, send mail to the list administrators: Ron Pickard: emc-p...@hypercom.com Dave Heald: emc_p...@symbol.com For policy questions, send mail to: Richard Nute: ri...@ieee.org Jim Bacher: j.bac...@ieee.org Archive is being moved, we will announce when it is back on-line. All emc-pstc postings are archived and searchable on the web at: http://www.ieeecommunities.org/emc-pstc
