In addition to this very instructive mail: very similar to Reverse Pulse Technique is the work of Brian Casper at al. on Peak Distortion Analysis. However, the latter considers sampled response built from the input or cross-talk channel bit stream. However, this approach is assumed aperiodic. Arranging the input current steps properly in time and alternating their signs, one can theoretically make the voltage drop arbitrarily 'bad' provided that the current steps can follow arbitrarily dense in time. In practice, they probably can't and therefore the total voltage over/undershoot is limiter. The term "reverse pulse techniques" follows straight from the fact that the voltage is expressed by a convolution integral (on tau), where there is a product of the 'input' current excitation (I(tau)) on the impedance response reversed Z(t - tau): V(t) =3D INT[from -inf to t] on (Z(t-tau)*I(tau)*d_tau) =20 -----Original Message----- Msg: #1 in digest Date: Sat, 18 Aug 2007 09:51:50 -0400 From: istvan novak <Istvan.Novak@xxxxxxx> Subject: [SI-LIST] Re: Reverse Pulse Technique method? Hi Pras, The authors of the paper may be out of town, so let me give you a quick=20 summary until they chime in: The Reverse Pulse Technique is based on the same powerful method that is used in high-speed passive channel simulations to get the worst-case eye in one pass, without simulating billions of bit transitions in the time=20 domain. The method makes use of the fact that the Impulse Response or=20 the Step Response contains all information that you possibly ever need=20 about the linear system. You excite the PDN impedance with the=20 specified fastest step current, record the step response, and process=20 it. The processing: you start backward (hence the name reverse) from=20 the DC solution, and record sequentially the voltages and times of=20 maxima and minima until you arrive to the time of excitation. If you=20 excite the actual PDN with alternating step-up and step-down current=20 steps at the relative times when the step response showed the maxima and minima (in reverse order), you are guaranteed to get the worst case=20 one-sided transient noise. You dont even need to do the time-domain=20 simulation with the worst-case excitation pattern: the transient part is the difference of the sum of maxima and sum of minima in the step=20 response. A further advantage of the method is that you can get the=20 step response by simulating or measuring the impedance-versus-frequency=20 curve and do the transformation into step response. Hope this helps. Istvan Novak SUN Microsystems ------------------------------------------------------------------ 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 technical documents are available at: http://www.si-list.net 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