Been detached for a while as I relocated to a new state that for the time being has power all the time, but is somewhat low on water. I haven't been following this entire thread, and may have missed the essence of the discussion, but I assume that it is more specifically targeting CPU's and high power ASIC's with relatively complex ceramic or organic package substrates, large currents with transient event times on the order of > ~0.5 nS. If I understand the arguments it seems these devices would need boatloads of localized on device decoupling to manage transient activity at > 100 MHz? Curious if this same methodology would apply to something like a Sonet TIA in a CSP, or a RF PA in a flip-chip package? In designs utilizing these devices I generally see fairly small value components on the order of 100's of pF utilized for on-board decoupling. To a first order I would imagine it is intended to provide power decoupling at frequencies that could exceed 100MHz, but I could be completely off my rocker. Any comments welcome. -Brett I find the water shortage issue somewhat perplexing as I had been told it rains all the time in Oregon. As an interesting bit of trivia I recently discovered a rainfall map of US the for the summer months that showed Oregon having on the order of 10" of rain between May 1st and October 1st. Conversely my home state of Minnesota receives about 2.5-3 times that amount in the same period. I guess 'rains all the time' like many things is matter of perspective? -----Original Message----- From: Chris Cheng [mailto:chris.cheng@xxxxxxxxxxxx] Sent: Thursday, July 12, 2001 12:36 PM To: 'si-list@xxxxxxxxxxxxx' Subject: [SI-LIST] Re: decoupling believe me i have the best &^%$* package. you are confusing signal traces that have pf of loading and 1nh of inductance with core that has uf's of loading and ph's of inductance. orders of magnitude difference in resonance. i have done a few ghz processes with >100W power packaging with the best flip chip substrate. you can put a million low impedance cap solder right at the pin of the package and it won't make a &^%$* difference to the core noise >100MHz. what ghz processes have you done ? -----Original Message----- From: Ed Priest [mailto:Ed_Priest@xxxxxxxxxxxxxx] Sent: Thursday, July 12, 2001 10:09 AM To: 'si-list@xxxxxxxxxxxxx' Subject: [SI-LIST] Re: decoupling You can pass out a +ghz signal single endedly and if you think you can't get > 100Mhz noise out of a package you are using some really &^%$* packages or are using die that don't have >100Mhz switching activity. In a package with coupled planes - the power planes look like transmission lines with distributed poles and zeros. Good flip chip packages have very low impedance between PCB and the die. I guarantee you can get high frequency noise down to the PCB - I've help troubleshoot a lot of boards for people who thought they couldn't. If you have a bunch of chips on a board with 400Mhz switching, spending time lowering the impedance of your power distribution at 400Mhz is time well spent. Ed ------------------------------------------------------------------ To unsubscribe from si-list: si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field 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