The derivation I saw (over 25 years ago) for the equation BW = 0.35/Tr assumed a critically damped 2 pole response when measuring risetime between 10 and 90%. Different factors other than 0.35 can be proven for different reference percentage (20-80% which I think the 0.5 come from) and different responses (damping and number of poles). For filters the -3dB frequency indicates where the output power of a test frequency applied to the input is less than 0.5 of the input. For example, the simple RC network mentioned will output low frequency components with less than 3dB attenuation but any component above the knee frequency will be attenuated by 3 dB or more. The knee frequency has nothing to do with the highest frequency present in the signal but tells the designer where the roll off in amplitude begins. For a single pole response the amplitude will roll off -20 dB per decade, 2 pole will roll off at -40 dB per decade above the knee frequency. For signals there is a similar concept for energy content. A step with 100 psec risetime will have more energy in higher frequency harmonics, (out to 3.5 GHz BW) than a step with 100 nsec risetimes (3.5 MHz BW). To pass a 1 MHz square wave with 100 psec edges without degrading the rise time implies the bandwidth of the transmission medium needs to be 3 to 5 times the bandwidth of the step. SystemRisetime = SQRT(Steprisetime^^2 + Mediumrisetime^^2). From inspection if both the medium and stimulus have the same risetime (100 psec) the output risetime will be 140 psec. If one considers bit error rate then the significance of the bandwidth of the medium becomes clear. As the bandwidth of the medium decreases then the rise and fall times of the signal at the output becomes slower and slower closing the eye in the eye diagram. Tom Dagostino Modeling Manager Mentor Graphics Corp. SAS tom_dagostino@xxxxxxxxxx 503-685-1613 -----Original Message----- From: Clewell, Craig [mailto:cclewell@xxxxxxxxxxxxxx] Sent: Thursday, June 13, 2002 6:00 AM To: 'Anand.Kuriakose@xxxxxxxxxx'; si-list@xxxxxxxxxxxxx Subject: [SI-LIST] Re: 3dB or Knee Frequency Anand, I believe the t = .35/BW comes from a unit step into an RC circuit. If you have a series circuit consisting of a unit step source u(t) with a resistor and capacitor than the equations fall out like this... Vc = 1-e^(-t/RC) ~~~ Therefore, t = -RC * ln(1-Vc) and since BW = 1/(2pi RC) you get... t = [-1/(2pi * BW)] * ln(1-Vc) If you use a rise time of 10 - 90% and substitute... t = [-1/(2pi *BW)] * (ln(0.1) - ln(0.9) ~~~ Therefore t = 0.35 / BW. Regards, Craig Clewell -----Original Message----- From: Anand.Kuriakose@xxxxxxxxxx [mailto:Anand.Kuriakose@xxxxxxxxxx] Sent: Wednesday, June 12, 2002 11:47 PM To: si-list@xxxxxxxxxxxxx Subject: [SI-LIST] 3dB or Knee Frequency Hi All, When we talk about high edge rate signals, we talk about two terms namely 1> 3 dB frequency, which is related to the edge rate as approximately 0.35/Tr and 2> Knee frequency (mentioned in Howard Johnson's famous book and many SI-list mails) which is given as 0.5/Tr. Questions: 1> What is Knee frequency? Does it relate to the highest harmonic present in the signal with significant energy? How relevant it is in High speed board design as against 3dB frequency? 2> To determine the decoupling caps, we have to determine the highest significant frequency involved in the switching signals. Now the question is should we consider the 3dB frequency ( mentioned in High Speed Digital Design by Stephen Hall) or the knee frequency ( in Howards Johnson's Black MAgic) as the best approximation? Let us consider Tr=1ns. Then we have 3dB frequency = 350Mhz and Knee frequency = 500 Mhz. Regards, Anand. ------------------------------------------------------------------ To unsubscribe from si-list: si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field or to administer your membership from a web page, go to: //www.freelists.org/webpage/si-list For help: si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field List archives are viewable at: //www.freelists.org/archives/si-list or at our remote archives: http://groups.yahoo.com/group/si-list/messages Old (prior to June 6, 2001) list archives are viewable at: http://www.qsl.net/wb6tpu ------------------------------------------------------------------ To unsubscribe from si-list: si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field or to administer your membership from a web page, go to: //www.freelists.org/webpage/si-list For help: si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field List archives are viewable at: //www.freelists.org/archives/si-list or at our remote archives: http://groups.yahoo.com/group/si-list/messages Old (prior to June 6, 2001) list archives are viewable at: http://www.qsl.net/wb6tpu ------------------------------------------------------------------ To unsubscribe from si-list: si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field or to administer your membership from a web page, go to: //www.freelists.org/webpage/si-list For help: si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field List archives are viewable at: //www.freelists.org/archives/si-list or at our remote archives: http://groups.yahoo.com/group/si-list/messages Old (prior to June 6, 2001) list archives are viewable at: http://www.qsl.net/wb6tpu