Question re: ADCLK905 demo board

The Gerbers for the EVAL-ADCLK905 board show 17-mil traces for the 50-ohm lines going to and from the chip, and the Gerbers call for RO4003C material.  The traces on the actual board, though, are about 8 mils wide.  That makes me think the RO4003C layers are 4 mils thick, but RO4003C doesn't seem to be offered in 4 mils (although RO4350B is.)  Can you clarify the actual layer stackup that was used to fab this board?  Did your board house switch to RO4350B?

  • 0
    •  Analog Employees 
    on May 13, 2015 7:56 PM

    The  50 ohm trace width is 17 mils, the art gerbs I have are 17 mils, there are some short 10 mil traces to get off the DUT and make the path to the fan out where it’s changed to 17 mil width wide. The pic attached, shows the 10 mil trace width coming off the dut, that is determined by the pin width / spacing on the dut, then it fans out to 17 mil width. Ok so now back to the 8 mil question?  Hopefully you see what I see, and the short trace length getting off the dut is not an issue, the 50 ohm trace width is 17 mils wide.  I feel sure what the FAB does, is the GND plane is 4 mils above the pwr plane and used IS410 for that, to help build capacitance on the pwr plane.  Then used the R04003 on the top/bottom outer layers for the 17 mil trace, impedance match back to layer 2 & 5 with whatever the FAB house determined for the height spacing on that. Then used a filler between layers 3 & 4 to get the board up to 62 mils thick.


    I attached the *.pdf fab/drill dwg they used to fab this board, it does show the R04003 material for the outer layers top/bottom

    Hope this helps

  • Right, the Gerbers show 17 mil traces as you're saying, but if you measure the traces on the actual PCB with a caliper, they appear to be much narrower than that.  Not only are they not necking the traces down for the DUT connection, as the Gerbers call for, but they're actually a bit fatter near the DUT:

    The return loss between J9-J10 isn't too bad:

    ... although it's more like 40 ohms than 50.

    So I'm having trouble figuring out what the board house actually did in this case.  I'm working on a board with somewhat stiffer VSWR requirements, and while your demo board is working fine with 20 dB return loss, I'd like to understand how  to avoid a similar discrepancy in my instructions to the fab. 

  • 0
    •  Analog Employees 
    on Aug 2, 2018 3:00 PM
    This question has been assumed as answered either offline via email or with a multi-part answer. This question has now been closed out. If you have an inquiry related to this topic please post a new question in the applicable product forum.

    Thank you,
    EZ Admin