Hi Dale, Only Pins D+/D- are bga pins. The B+/B- pins are actually two pins on an 8 pin soic. ________________________________ From: Dale Rasmusen [mailto:dale.rasmusen@xxxxxxxxxxxx] Sent: Monday, April 18, 2005 4:36 PM To: Jean Bratton Subject: RE: [PCB_FORUM] Diff pair phase tolerance I tried the forum, but I cannot tell if my messages are getting sent. Jean, I would look at it in a different way. Intrinsically, the signals are basically at the same level on the die of the BGA. When they get out to a ball of the BGA one is 50 mils longer than the other. If you go straight out of the BGA to the next component and the difference is 50 mils, then you are exactly matched to the length. Dale Rasmusen, CID Teknovus 1351 Redwood Way Petaluma, CA 94954 dale.rasmusen@xxxxxxxxxxxx (707) 665-0400 x133 (707) 665-0491 fax -----Original Message----- From: icu-pcb-forum-bounce@xxxxxxxxxxxxx [mailto:icu-pcb-forum-bounce@xxxxxxxxxxxxx] On Behalf Of Jean Bratton Sent: Monday, April 18, 2005 1:18 PM To: icu-pcb-forum@xxxxxxxxxxxxx Subject: [PCB_FORUM] Diff pair phase tolerance Hi all, We're running into something and would like to know if there's a better way of handling this... We have a diff pair going to 4 pins. Lets call the pins A+, B+, C+, D+ on the one net and A-, B-, C- and D- on the other net of the pair. The pin escape lengths at the A+/- C+/- and D+/- pins are identical in length on each net, but the pin escape on B+ is 50 mils shorter than that of B- due to the location of the pins on the part and the desire for their vias to be adjacent so that the overall transmission line is smooth. (I've tried to sort of draw a picture below, but don't know how much it'll help.) The phase tolerance we're really concerned with is the over all distance from A+ to D+ compared with the distance from A- to D-, so the extra pin escape length at B- really doesn't bother us. However, it seems like the system is forcing the overall length of each segment to match. Is there any way to define a pin pair that the phase tolerance will look at, and only that pin pair? Our current solution is to jack up the phase tolerance number so it doesn't holler, then create a pin-to-pin match group of just these two nets to force the overall lengths to match. Is there a better way? We have a bunch of these and this seems kind of kludgy... Thanks for any feedback. Jean A+ B+ C+ D+ | | | | *------*------------*------* *------*------------*------* | | | | A- | C- D- | B- Jean Bratton Sr. Printed Circuit Designer Freedom CAD Services, Inc. 603-864-1300 x1349