Sierra does have preferred stack-ups and I would image other fab houses do, too. I asked our design team to give you more details. They will do so tomorrow.
Thank you, Lucy!
Much appreciated,
As a general rule, we recommend only one prepreg style per laminate type; that is, say you have a multilayer that is all FR4 type material then we would only want to use one prepreg type. If you had a board with, for example, with Rogers material on the top and bottom layers and FR4 for the internal layers then you would most likely want two different prepregs. The only reason I know of to not follow that rule is that occasionally you cannot get the exact thickness you’re after using available sizes so you switch to a slightly different style to get the thickness needed.
All our Standard stackups are made considering 50 Ohm SE, 90 Ohm diff, and 100 Ohm Diff pair. they give manufacturable trace width for most of the common impedance values.
75 Ohm single-ended is not a common value but still can be done using the our standard stackups.
we see 75 Ohms for RF applications. Antennas, and all but not very common.
the trick is to not take adjacent copper layer as reference but to skip it and take the next one. this is to get a good trace width. As taking adjacent references will give you very low trace widths for 75 Ohms SE which may not be manufacturable.
being an RF application requirement also having Coplanar traces along with 75 Ohm single-ended trace is recommended.
Hi james,
the two examples that you mentioned are right. thanks for those inputs,
The broadside is used for smaller impedances as the trace widths become large when the impedance required is low. but hardly used nowadays. edge coupled is only used.
it is difficult to obtain high accuracy due to registration errors.
- This registration error can be mitigated by using a core between the broadside coupled traces and not a prepreg. As a core is already cured and once etched the traces will not move during lamination to have an error due to registration.
- Also core thickness is common throughout the length and is not dependent on the amount of resin loss and copper percentages on adjacent layers