With transmission speeds of 10 Gbps, you have to be careful not to have too much line-to-line skew in the differential channels. The glass weave is a big contributor here. When you go to 25 Gbps, this is a MAJOR issue that you NEED to address.
Two recent blog posts over on the DesignCon community highlight this and I highly recommend you read through this if you are doing 10+ Gbps boards today or anytime soon. You may be sitting on a time bomb.
First Eric Bogatin talks about the glass weave skew issue and how a new (expensive) material from Isola may solve the problem by having glass and epoxy with the same DK.
Then Lee Ritchey explains how he uses (cheaper) 3313 weave glass to achieve the same thing, but also points out that not all 3313 material is the same.
Read the DesignCon paper referenced by Lee Ritchey, and learn one key thing:
If you specify “Isola 3313” on your stack-up today, you should change that to “Isola 3313-MS” (same goes for 1067).
The difference (-MS) meaning Mechanically Spread. It turns out that Isola manufactures 3313 in two different ways in different factories. The “-MS” version has both the glass fiber fill-yarn and warp-yarn spread out to be very flat before the weaving process. In the other version this only happen for one of the two yarn directions. BIG difference – that may/will hit you depending on how many 10+ GBps channels you build.
This topic is part of what we cover in the 3-day SI course.
Leave a Reply