N-Way resistive splitter
I am trying to do a N-way (N between 8 and 12) resistive splitter that will work up to 4 to 5 GHz. I have the math worked out and it works perfectly in a schematic simulation. However, when I do a momentum simulation and include the PCB trace length I get very bad performance. I am doing half of the connectors on the top layer and the other half of the bottom layer of the pcb. The top and bottom inputs are connected with vias. I have the 50 ohm trace width calculated, so that is probably not the problem. I believe the main problem is the point where all the inputs come together.
Does anybody have experience with this that they can share or can anyone point me to a resource on a resistive splitter layout.
Thanks.
Hey sarajlija - just a quicky as I pass by my PC...
Good web resource:
http://www.microwaves101.com/encyclo...ower-splitters
If you build it with a star topology (so the resistors are like the radial spokes of a bike wheel) it should be fine to very high frequencies. Beware passing the RF through the PCB with vias though - if your board is thick, the series inductance can impose all sorts of weird resonances/peaks/troughs in your return loss! Stay single sided above a plane (with coplanar waveguide/microstrip) and all will be good :)
Can you post a photo of your layout?
Theoretically a resistive splitter should be broadband, but making one yourself with very flat response up to 6 GHz could be very challenging.
Have to take care about all the layout and component parasitics, as in the picture attached.
For example when I build a simple broadband 3-way splitter (star topology) for high frequencies, I place on each arm of the star three 50 ohms SMD resistors in parallel (one on top of the other), resulting a 16.6 ohms resistor.
The resulting resistor have 3x less parasitic series inductance than using only one 16.6 ohms resistor.
