RE: Dithered clocks and EMC - BEWARE
Paolo, Thank you for the information. I suspect that you are correct as to the critical nature of the frequency modulation method. The design engineer picked the device, probably as the cheapest vendor. The device did not behave as I expected, based on some Internet research of dithered clock technology. We were very pressed for time, but I hope someday to spend some more looking at this technology. Perhaps my experience explains the digital TV set interference, since all such devices are obviously not the same. Scott Lacey -Original Message- From: paolo.ronc...@compuprint.it [SMTP:paolo.ronc...@compuprint.it] Sent: Monday, March 20, 2000 5:21 AM To: emc-p...@ieee.org Subject:RE: Dithered clocks and EMC - BEWARE I have a very succesfull experience with SSCG (Spread Spectrum Clock Generation) that is a patented technology used to intentionally spread the energy of oscillators and clock signals and any signal derived from the clock. I am sure Keith Hardin knows it first-hand because he is one of the developers ! We have been using an SSCG chip since 1995 on most of the electronics of our printers - most of them class B devices, with oscillator frequencies in the range 24 - 32 MHz. We even managed to get significant cost reductions by switching from 4-layer to 2-layer boards and still passing CISPR22 and FCC class B limits for radiated emissions. One point that seems not to have come up in this discussion is the kind of dithering we are talking about. In SSCG the main clock frequency is modulated, but just ANY frequency modulation is not enough ! Key is the wave-shape of the modulating function, and that is the crucial point in the SSCG patent (Keith, correct me if I'm wrong). The trick is to maximize the rate of change of the function representing the frequency modulation, so that the oscillator spends the minimum time on any given frequency within the modulating range, thus avoiding additional peaks in the emission spectrum. The modulating function in the SSCG technique does just that. This is best explained in the first (to my knowledge) publication on the subject (here, too, Keith knows much more !!) : K.B.Hardin, J.T.Fessler, D.R.Bush (Lexmark Intl.): Spread Spectrum Clock Generation for the Reduction of Radiated Emissions - 1994 IEEE Intl. Symposium on EMC, August 1994 Chicago (Symposium Record page 227) I suspect an explanation to Scott's problem may be that he didn't use the SSCG technique but another form of dithering that was not controlled in such a way as to get the flat frequency distribution that you get with SSCG. Also the amount of frequency deviation can be important. The attenuation you get in quasi-peak (QP) readings is dependent on that. I hope Scott can give us more details. As to the problems with digital TV, I honestly don't have any knowledge about that, but I remember the same inventors of SSCG did a study on interference potential of this technique : K.B.Hardin, J.T.Fessler, D.R.Bush (Lexmark Intl.): A Study of the Interference Potential of Spread Spectrum Clock Generation Techniques - 1995 IEEE Intl. Symposium on EMC Atlanta ( Symposium Record page 624). Hope this helps. Paolo Roncone Compuprint s.p.a. - Italy --- This message is from the IEEE EMC Society Product Safety Technical Committee emc-pstc discussion list. To cancel your subscription, send mail to: majord...@ieee.org mailto:majord...@ieee.org with the single line: unsubscribe emc-pstc For help, send mail to the list administrators: Jim Bacher: jim_bac...@mail.monarch.com mailto:jim_bac...@mail.monarch.com Michael Garretson:pstc_ad...@garretson.org mailto:pstc_ad...@garretson.org For policy questions, send mail to: Richard Nute: ri...@ieee.org mailto:ri...@ieee.org --- This message is from the IEEE EMC Society Product Safety Technical Committee emc-pstc discussion list. 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: Jim Bacher: jim_bac...@mail.monarch.com Michael Garretson:pstc_ad...@garretson.org For policy questions, send mail to: Richard Nute: ri...@ieee.org
RE: Dithered clocks and EMC - BEWARE
I have a very succesfull experience with SSCG (Spread Spectrum Clock Generation) that is a patented technology used to intentionally spread the energy of oscillators and clock signals and any signal derived from the clock. I am sure Keith Hardin knows it first-hand because he is one of the developers ! We have been using an SSCG chip since 1995 on most of the electronics of our printers - most of them class B devices, with oscillator frequencies in the range 24 - 32 MHz. We even managed to get significant cost reductions by switching from 4-layer to 2-layer boards and still passing CISPR22 and FCC class B limits for radiated emissions. One point that seems not to have come up in this discussion is the kind of dithering we are talking about. In SSCG the main clock frequency is modulated, but just ANY frequency modulation is not enough ! Key is the wave-shape of the modulating function, and that is the crucial point in the SSCG patent (Keith, correct me if I'm wrong). The trick is to maximize the rate of change of the function representing the frequency modulation, so that the oscillator spends the minimum time on any given frequency within the modulating range, thus avoiding additional peaks in the emission spectrum. The modulating function in the SSCG technique does just that. This is best explained in the first (to my knowledge) publication on the subject (here, too, Keith knows much more !!) : K.B.Hardin, J.T.Fessler, D.R.Bush (Lexmark Intl.): Spread Spectrum Clock Generation for the Reduction of Radiated Emissions - 1994 IEEE Intl. Symposium on EMC, August 1994 Chicago (Symposium Record page 227) I suspect an explanation to Scott's problem may be that he didn't use the SSCG technique but another form of dithering that was not controlled in such a way as to get the flat frequency distribution that you get with SSCG. Also the amount of frequency deviation can be important. The attenuation you get in quasi-peak (QP) readings is dependent on that. I hope Scott can give us more details. As to the problems with digital TV, I honestly don't have any knowledge about that, but I remember the same inventors of SSCG did a study on interference potential of this technique : K.B.Hardin, J.T.Fessler, D.R.Bush (Lexmark Intl.): A Study of the Interference Potential of Spread Spectrum Clock Generation Techniques - 1995 IEEE Intl. Symposium on EMC Atlanta ( Symposium Record page 624). Hope this helps. Paolo Roncone Compuprint s.p.a. - Italy --- This message is from the IEEE EMC Society Product Safety Technical Committee emc-pstc discussion list. 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: Jim Bacher: jim_bac...@mail.monarch.com Michael Garretson:pstc_ad...@garretson.org For policy questions, send mail to: Richard Nute: ri...@ieee.org
Re: Dithered clocks and EMC - BEWARE
Keith Hardin wrote: The output spectrum is a series of stationary harmonics spaced at 30kHz apart This is very similar to a problem I had with PLL in a product once. Major spike at the main frequency coming through with a picket fence on either side spaced out by the freq. of the PLL. Poor board layout was the culprit if I remember correctly. Didn't we have a gentleman from Lexmark who started the spread spectrum technique do a presentation at the Santa Clara EMC Society? I seem to remember there being some constraints to do doing it - the dither being less than some rather low percent of the freq being dithered? --- This message is from the IEEE EMC Society Product Safety Technical Committee emc-pstc discussion list. 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: Jim Bacher: jim_bac...@mail.monarch.com Michael Garretson:pstc_ad...@garretson.org For policy questions, send mail to: Richard Nute: ri...@ieee.org
RE: Dithered clocks and EMC - BEWARE
I never used SSCG, but following the thread, my humble opinion is the following: If the interference just comes from the oscillator, it is feasable that the use of a SSCG will even increase the interference, as the circuit elements inside the SSCG that create the clock signal will also have a current consumption that will contribute to the interference emanating from the clock generator. You will probably get better results from the use of a SSCG, if the interference not just comes from the oscillator itself, but mainly from the circuits driven by the SSCG. - Keith Hardin (in the mail shown below) wrote: ... The output spectrum is a series of stationary harmonics spaced at 30kHz apart To my understanding, SSCG produce a FM modulated clock. I dimly remember that FM does not produce a series of stationary harmonics spaced at the modulation frequency, but the harmonics change dependent the modulating signal. - I think the SSCG does not fool the QP Detektor, it fools the bandwidth of the test receiver. The effiencency of fooling depends not only on the modulation frequency, but also on the frequency offset. In the example, where the use of the SSCG did not help to reduce interference from the second harmonic, the frequency offset was probably less than 60kHz (peak to peak) - For analog TV you will minimize the perception of interference on the screen if the modulation frequency of the SSCG is 39kHz or 55kHz, as these frequencies have the maximum offset from the line frequency harmonics. Furthermore, for analog TV the assessment of the level of acceptable interference is decided by the eye and the brain of the beholder. In digital TV it is decided by the software on the chip, and that these days is still inferior to the brain of a beholder. Where I live, there is no digital TV available. Maybe digital TV is not suitable for a terrestial air link. Dipl.-Ing. Rene Charton Manager EMC Services _ TUV Rheinland Taiwan Ltd. TAIPEI HEAD OFFICE Spring Plaza Building 14F, No.6, Min Chuan E. Rd., Sec. 3, Taipei 104, Taiwan, R. O. C. Tel. (02) 2516 6040 Ext. [ 086 ] e-mail: r...@twn.tuv.com khar...@lexmark.com on 03/18/2000 03:30:51 AM Please respond to khar...@lexmark.com To: emc-p...@ieee.org cc: m...@california.com, sla...@foxboro.com (bcc: Rene Charton/TUV-Twn) Subject: RE: Dithered clocks and EMC - BEWARE Scott, Sorry to hear about your experience but we have had great success using Spread Spectrum Clock Generation (SSCG) since 1994 and have shipped many millions of units. Most PC's built over the last 2 to 3 years also use this technology. For back ground information, see http://www.lexmark.com/sscg/ and http://developer.intel.com/ial/scalableplatforms/sdt.htm. Included in this information is what SSCG is and its effects on some other devices. I suggest one read these references as a basis for further discussions. Experimental data has clearly shown is that the QP and Peak readings are exactly the same for SSCG signals (No Fooling). An SSCG output properly designed will create a clock that is frequency modulated at ~30kHz. The output spectrum is a series of stationary harmonics spaced at 30kHz apart. Finally, this method will attenuate the clock and system as compared to the same system with the modulation turned off. All other EMI reduction techniques will also reduce emissions further. Our design philosophy is to bring all good low cost techniques to bare on reducing the emissions and SSCG is just one of them. Keith Hardin Lexmark International For the last time this came up, please see the append on the emc-pstc archives at http://www.rcic.com/ with subject Clock Oscillator Re: Spread spectrum clock oscillator (1) 20-Feb-97 . slacey%foxboro@interlock.lexmark.com on 03/17/2000 11:18:46 AM Please respond to slacey%foxboro@interlock.lexmark.com To: macy%california@interlock.lexmark.com cc: emc-pstc%ieee@interlock.lexmark.com (bcc: Keith Hardin/Lex/Lexmark) Subject: RE: Dithered clocks and EMC - BEWARE Robert, and the group, Although I have only limited experience (1 instance) with dithered clocks, I thought that I might share that experience with the group. I had a product that was failing radiated emissions at one particular frequency with a vertical antenna orientation. I tracked the problem down to a particular cable and circuit card. Use of a near field probe identified the oscillator package as the source of the emissions (the second harmonic). The device was located near the card edge, insufficient decoupling, etc. All the textbook layout errors. A check of the oscillator specifications showed that it was a standard TTL device with a fanout of 10 inputs. I had already found some fixes that would reduce emissions, clamp-on ferrites, additional shielding, etc., but thought it made more sense to reduce the emissions at the source. I suggested to the design engineer that a low
RE: Dithered clocks and EMC - BEWARE
Lacey others, Interesting. But with only the 2nd harmonic being your problem, it's possible the dithered clock's energy envelope was mostly or all still within the 120 kHz passband of the EMI receiver, depending upon the amount of dithering. Was this the case? I'm curious about the digital TV receiver report. I understand that the entire 6 MHz of a digital TV channel's frequency allocation appears to be filled with content as opposed to the old analog TV systems which basically had audio video sub-carriers. This suggests the digital receiver front end designs are different (more vulnerable?). Hopefully, there are others on this forum that can explain. Jack Cook, Xerox Corp -Original Message- From: Lacey,Scott [mailto:sla...@foxboro.com] Sent: Friday, March 17, 2000 8:19 AM To: 'Robert Macy' Cc: 'emc-p...@ieee.org' Subject: RE: Dithered clocks and EMC - BEWARE Robert, and the group, Although I have only limited experience (1 instance) with dithered clocks, I thought that I might share that experience with the group. I had a product that was failing radiated emissions at one particular frequency with a vertical antenna orientation. I tracked the problem down to a particular cable and circuit card. Use of a near field probe identified the oscillator package as the source of the emissions (the second harmonic). The device was located near the card edge, insufficient decoupling, etc. All the textbook layout errors. A check of the oscillator specifications showed that it was a standard TTL device with a fanout of 10 inputs. I had already found some fixes that would reduce emissions, clamp-on ferrites, additional shielding, etc., but thought it made more sense to reduce the emissions at the source. I suggested to the design engineer that a low-power TTL oscillator, with a fanout of 2 inputs, would reduce currents through the offending etches. Someone else suggested a dithered clock device instead. When we tested the dithered clock, emissions were actually worse. We had simply spread the problem over a wider spectrum. I have heard some success stories for these devices, but results in this case were disappointing. I still think a lot depends on the layout. I strongly suspect that real-world disruption to nearby devices is generally going to be worse, since the idea behind these devices is to fool quasi-peak measurements in order to pass. Caveat Emptor! Scott Lacey -Original Message- From: Robert Macy [SMTP:m...@california.com] Sent: Friday, March 17, 2000 9:38 AM To: emc-p...@ieee.org Subject:Fw: Dithered clocks and EMC - BEWARE Of interest, so I forward this to the group: - Robert - Robert A. Macy, PEm...@california.com 408 286 3985 fx 408 297 9121 AJM International Electronics Consultants 619 North First St, San Jose, CA 95112 -Original Message- From: John Woodgate j...@jmwa.demon.co.uk Newsgroups: sci.engr.electrical.compliance,sci.electronics.design Date: Tuesday, March 14, 2000 11:55 PM Subject: Dithered clocks and EMC - BEWARE About a year ago we had a thread on this subject, concluding that some research was needed to see whether dithered clocks were better or worse in terms of conforming to EMC requirements. I learned very recently that some reliable but as-yet unpublished research has found that digital TV receivers are some 40 dB (!!) more sensitive to dithered clock emissions than to unmodulated carriers. This is likely to lead to changes in EMC limits within maybe as little as three years, since the authorities certainly don't want to be deluged with complaints of interference from people who have just opted for digital TV. So, if you are thinking of using a dithered clock, think again! -- Regards, John Woodgate, OOO - Own Opinions Only. Phone +44 (0)1268 747839 Fax +44 (0)1268 777124. http://www.jmwa.demon.co.uk I wanted to make a fully-automated nuclear-powered trawler, but it went into spontaneous fishing. --- This message is from the IEEE EMC Society Product Safety Technical Committee emc-pstc discussion list. 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: Jim Bacher: jim_bac...@mail.monarch.com Michael Garretson:pstc_ad...@garretson.org For policy questions, send mail to: Richard Nute: ri...@ieee.org --- This message is from the IEEE EMC Society
RE: Dithered clocks and EMC - BEWARE
Scott, Sorry to hear about your experience but we have had great success using Spread Spectrum Clock Generation (SSCG) since 1994 and have shipped many millions of units. Most PC's built over the last 2 to 3 years also use this technology. For back ground information, see http://www.lexmark.com/sscg/ and http://developer.intel.com/ial/scalableplatforms/sdt.htm. Included in this information is what SSCG is and its effects on some other devices. I suggest one read these references as a basis for further discussions. Experimental data has clearly shown is that the QP and Peak readings are exactly the same for SSCG signals (No Fooling). An SSCG output properly designed will create a clock that is frequency modulated at ~30kHz. The output spectrum is a series of stationary harmonics spaced at 30kHz apart. Finally, this method will attenuate the clock and system as compared to the same system with the modulation turned off. All other EMI reduction techniques will also reduce emissions further. Our design philosophy is to bring all good low cost techniques to bare on reducing the emissions and SSCG is just one of them. Keith Hardin Lexmark International For the last time this came up, please see the append on the emc-pstc archives at http://www.rcic.com/ with subject Clock Oscillator Re: Spread spectrum clock oscillator (1) 20-Feb-97 . slacey%foxboro@interlock.lexmark.com on 03/17/2000 11:18:46 AM Please respond to slacey%foxboro@interlock.lexmark.com To: macy%california@interlock.lexmark.com cc: emc-pstc%ieee@interlock.lexmark.com (bcc: Keith Hardin/Lex/Lexmark) Subject: RE: Dithered clocks and EMC - BEWARE Robert, and the group, Although I have only limited experience (1 instance) with dithered clocks, I thought that I might share that experience with the group. I had a product that was failing radiated emissions at one particular frequency with a vertical antenna orientation. I tracked the problem down to a particular cable and circuit card. Use of a near field probe identified the oscillator package as the source of the emissions (the second harmonic). The device was located near the card edge, insufficient decoupling, etc. All the textbook layout errors. A check of the oscillator specifications showed that it was a standard TTL device with a fanout of 10 inputs. I had already found some fixes that would reduce emissions, clamp-on ferrites, additional shielding, etc., but thought it made more sense to reduce the emissions at the source. I suggested to the design engineer that a low-power TTL oscillator, with a fanout of 2 inputs, would reduce currents through the offending etches. Someone else suggested a dithered clock device instead. When we tested the dithered clock, emissions were actually worse. We had simply spread the problem over a wider spectrum. I have heard some success stories for these devices, but results in this case were disappointing. I still think a lot depends on the layout. I strongly suspect that real-world disruption to nearby devices is generally going to be worse, since the idea behind these devices is to fool quasi-peak measurements in order to pass. Caveat Emptor! Scott Lacey -Original Message- From: Robert Macy [SMTP:m...@california.com] Sent: Friday, March 17, 2000 9:38 AM To: emc-p...@ieee.org Subject: Fw: Dithered clocks and EMC - BEWARE Of interest, so I forward this to the group: - Robert - Robert A. Macy, PEm...@california.com 408 286 3985 fx 408 297 9121 AJM International Electronics Consultants 619 North First St, San Jose, CA 95112 -Original Message- From: John Woodgate j...@jmwa.demon.co.uk Newsgroups: sci.engr.electrical.compliance,sci.electronics.design Date: Tuesday, March 14, 2000 11:55 PM Subject: Dithered clocks and EMC - BEWARE About a year ago we had a thread on this subject, concluding that some research was needed to see whether dithered clocks were better or worse in terms of conforming to EMC requirements. I learned very recently that some reliable but as-yet unpublished research has found that digital TV receivers are some 40 dB (!!) more sensitive to dithered clock emissions than to unmodulated carriers. This is likely to lead to changes in EMC limits within maybe as little as three years, since the authorities certainly don't want to be deluged with complaints of interference from people who have just opted for digital TV. So, if you are thinking of using a dithered clock, think again! -- Regards, John Woodgate, OOO - Own Opinions Only. Phone +44 (0)1268 747839 Fax +44 (0)1268 777124. http://www.jmwa.demon.co.uk I wanted to make a fully-automated nuclear-powered trawler, but it went into spontaneous fishing
RE: Dithered clocks and EMC - BEWARE
Robert, and the group, Although I have only limited experience (1 instance) with dithered clocks, I thought that I might share that experience with the group. I had a product that was failing radiated emissions at one particular frequency with a vertical antenna orientation. I tracked the problem down to a particular cable and circuit card. Use of a near field probe identified the oscillator package as the source of the emissions (the second harmonic). The device was located near the card edge, insufficient decoupling, etc. All the textbook layout errors. A check of the oscillator specifications showed that it was a standard TTL device with a fanout of 10 inputs. I had already found some fixes that would reduce emissions, clamp-on ferrites, additional shielding, etc., but thought it made more sense to reduce the emissions at the source. I suggested to the design engineer that a low-power TTL oscillator, with a fanout of 2 inputs, would reduce currents through the offending etches. Someone else suggested a dithered clock device instead. When we tested the dithered clock, emissions were actually worse. We had simply spread the problem over a wider spectrum. I have heard some success stories for these devices, but results in this case were disappointing. I still think a lot depends on the layout. I strongly suspect that real-world disruption to nearby devices is generally going to be worse, since the idea behind these devices is to fool quasi-peak measurements in order to pass. Caveat Emptor! Scott Lacey -Original Message- From: Robert Macy [SMTP:m...@california.com] Sent: Friday, March 17, 2000 9:38 AM To: emc-p...@ieee.org Subject:Fw: Dithered clocks and EMC - BEWARE Of interest, so I forward this to the group: - Robert - Robert A. Macy, PEm...@california.com 408 286 3985 fx 408 297 9121 AJM International Electronics Consultants 619 North First St, San Jose, CA 95112 -Original Message- From: John Woodgate j...@jmwa.demon.co.uk Newsgroups: sci.engr.electrical.compliance,sci.electronics.design Date: Tuesday, March 14, 2000 11:55 PM Subject: Dithered clocks and EMC - BEWARE About a year ago we had a thread on this subject, concluding that some research was needed to see whether dithered clocks were better or worse in terms of conforming to EMC requirements. I learned very recently that some reliable but as-yet unpublished research has found that digital TV receivers are some 40 dB (!!) more sensitive to dithered clock emissions than to unmodulated carriers. This is likely to lead to changes in EMC limits within maybe as little as three years, since the authorities certainly don't want to be deluged with complaints of interference from people who have just opted for digital TV. So, if you are thinking of using a dithered clock, think again! -- Regards, John Woodgate, OOO - Own Opinions Only. Phone +44 (0)1268 747839 Fax +44 (0)1268 777124. http://www.jmwa.demon.co.uk I wanted to make a fully-automated nuclear-powered trawler, but it went into spontaneous fishing. --- This message is from the IEEE EMC Society Product Safety Technical Committee emc-pstc discussion list. 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: Jim Bacher: jim_bac...@mail.monarch.com Michael Garretson:pstc_ad...@garretson.org For policy questions, send mail to: Richard Nute: ri...@ieee.org --- This message is from the IEEE EMC Society Product Safety Technical Committee emc-pstc discussion list. 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: Jim Bacher: jim_bac...@mail.monarch.com Michael Garretson:pstc_ad...@garretson.org For policy questions, send mail to: Richard Nute: ri...@ieee.org