The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> High-Speed I/O Design >> min metal width in high speed digital logic?
https://designers-guide.org/forum/YaBB.pl?num=1194100054

Message started by chase.ng on Nov 3^rd, 2007, 7:27am

Title: min metal width in high speed digital logic?
Post by chase.ng on Nov 3^rd, 2007, 7:27am

Hi all,

I have a question regarding laying out a high speed custom digital block. Using minimum metal will minimize the metal to substrate capacitance therefore theoritically can help in high speed digital. However, in circuit like a clock buffer that is to have a large fan out (that needs to be operating at high frequency), the charging and discharging current of the parasitc capacitance could be very large is a very short duration. What is the general practice here? Should I layout with min width metal (to reduce parasitic)? Will this actually lower my circuit's reliability (due to EM)? Anyone has experience regarding this?

Thanks,
chase

Title: Re: min metal width in high speed digital logic?
Post by joel on Nov 7^th, 2007, 10:27am

I like to use 1.5x min width and space, and double vias. If you set up some test cases, I think you'll find that this gives a nice improvement of the RC characteristic of the wire without too much area expense. The dual-damascene metalization structure results an a resistivity penalty at minimum width beyond linear ohms/square. You're generally not worried about substrate capacitance for interconnect, rather area and sidewall capacitances to other traces. Since sidewall is greater than area at minimum pitch, the 1.5xmin-width/space provides a net reduction in capacitance even though the area-cap has increased.

Current density is of concern on highly loaded nets. You mention clock trunks. If possible, put these on top-level thick metal. You'll find explicit current density rules in your process documentation. AC rules are much easier to satisfy than DC. In fact current-density violations are the industry's dirtly little secret at the moment.

I'll add the comment that the bulk of CMOS is routed at minimum pitch. So it's definitely reasonable to expect decent results with that configuration.

Title: Re: min metal width in high speed digital logic?
Post by chase.ng on Nov 10^th, 2007, 7:42pm

Hi joel,
Thanks for the advice.