• Previous message: Jeff Sondeen: "Re: standard cell library DRC errors"
• In reply to: cfk: "mosis bondpads"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

cfk writes:

 > PadGnd.mag, PadVdd.mag, PadIO.mag and a very important one called
 > Lib_Pads.mag. Lets invoke magic with its -w argument and an interesting

don't forget to read the README in that directory.

 > technology file. I believe this technology file to be .25 micron (although
 > things start getting a little fuzzy soon).
 > 
 > magic -w -TSCN5M_DEEP.12.TSMC

when magic starts up on that techfile, it echos a message that tells
what Mosis processes a techfile is for.

 > 
 >  At this point, it should be possible to use the File dropdown window to
 > load Lib_Pads.mag which has 10 pads instantiated and they are all unexpanded
 > by default.
 > 
 >     The next step is to select PadGnd by clicking with the box tool inside
 > the PadGnd and typing e to expand it. Next select the metallization itself
 > at the top of the pad and zoom in until the screen is almost full.
 > 
 >     At this point, put a um grid using the command "grid 10um 10um" and one
 > can see the metallization of the ground pad is about 73 microns square. If
 > one then switches to the wiring tool and clicks on the same ground pad, then
 > magic says it is 600 units wide.
 > 
 >     So I did this to make a point and then make a suggestion.
 > 
 > 1. Magic reads in the bondpads file properly from Jeff Sondeens web site and
 > runs with the SCN5M_DEEP.12.TSMC technology file, which is, I believe
 > .25micron, 5metal, Taiwan Semi and their are no DRC errors with this file
 > (thats really good news for both the mosis file and magic)
 > 
 > 2. Its really tough to figure out dimensions. I had to jump through a couple
 > of hoops and not believe the wiring tool when it is telling me "units". I
 > really think that magic would be improved by creating a ruler capability in
 > microns (or even say, femto-cubits) that makes it easy to figure out just
 > how large (or small) something is.

to figure out how big your bond pad metal is, just multiply by lambda,
which is .12um for SCN5M_DEEP.12.TSMC.  so a 750x750 lambda box
would be 750*.12=90umx90um.

/jeff


 > 
 > With highest regards to magic hackers, Charles

• Next message: cfk: "minor magic-7.2.48 issue"
• Previous message: Jeff Sondeen: "Re: standard cell library DRC errors"
• In reply to: cfk: "mosis bondpads"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]