¿ªÔÆÌåÓýThe resistors are likely going to have an alumina substrate with a dielectric constant of about 9.8. You¡¯re gaps will need to actually increase at the resistors to reduce the capacitance to maintain 50 ohms.
You¡¯ll also want to place your resistors closer to Y junction, in fact nearly at it, if possible.
With a back side ground plane, and the specified board thickness, you¡¯re going to have a hybrid mode with some microstrip and some coplanar waveguide modes each carrying current.
You probably don¡¯t need dozens of vias to the ground side, but at least should have some strategic ones near the Y junction, after the resistors, and maybe at midpoints and connector.
Sonnet has a free version of their simulator that should do enough to simulate this. You¡¯re probably going to have to do some tricks because you¡¯ll have to insert two non-ideal components, or simulate the resistors with a second metal layer and
a dielectric brick, or exceed four ports. I don¡¯t recall the exact license limitations but you may have to work around or more of them.
On a final note, the lower dielectric constant substrate is a bigger problem with the lower Er of RO4003.
Stripline layout with pockets milled for the resistors and center pins shouldn¡¯t be overlooked as another way.
Any way you cut it, the huge N connectors with a plane that abruptly ends with Teflon and center pin presents some challenges.
It might be worth, if you had a VNA, placing a prepared connector adjacent to a sheet of PCB material with one side grounded and an open on the other, and then measuring the same on a through line and trying to figure out the effective capacitance
and inductances by measurement. I don¡¯t (yet) have N adapters for my 8510C or I¡¯d offer to help with that.
I¡¯m interested in making up a satisfactory substitute for an 11667B with 3.5mm connectors. I¡¯ve been passively lurking.
From: [email protected] <[email protected]> on behalf of Jared Cabot via groups.io <jaredcabot@...>
Sent: Sunday, November 12, 2023 11:55:05 AM To: [email protected] <[email protected]> Subject: Re: [HP-Agilent-Keysight-equipment] DIY 11667A Power Splitter, resistor placement? ?
Don't worry, I haven't given up!
I just need to remember to take one step at a time rather than try to do it all at once and overwhelm myself with the details... Here is an idea for a PCB design so far. It's a 1.9mm wide, 1oz (35um) thick trace on 1.524mm thick Rogers 4003, a coplanar waveguide with clearance of 0.245mm (except at the 1206 resistors that tightens to 0.18mm to maintain 50ohm impedance) The center pins of the N connectors is 3mm diameter, so the solder join will hopefully 'funnel down' to the 1.9mm wide PCB pads. The other pads around the perimeter are for grounding the N connector shells. I also haven't put vias anywhere yet. Still to do once layout is finalised. I'll probably just stick 0.1% thin-film resistors in for now, mounted upside down. The pad width has been tightened to the same width as the resistors too, so there will be minimal step. It shouldn't be a problem as they will be hand soldered anyway. The high frequency RF resistors are much smaller which makes impedance matching much harder with my basic skills.. Let me know how that looks so far... :) Jared. |
