Parasitic Capacitance of the Isolation Resistor of a Wilkinson combiner

Hi everybody.

I am designing a Wilkinson Combiner to sum two signals in phase of 75[W] each, in 6.3 [GHz], in AWR software.

I selected an isolation resistor of 60[W], but in the datasheet mentions 2[pF] of parasitic capacitance. Therefore, i added a 2[pF] capacitor in paralell to the resistor in the simulation.

That capacitor, degrades so much the isolation between inputs ports (S32 or S23 parameters, being the port 1 the common port), from -50[dB] to only -6[dB]. This is logic, because 2[pF] in 6.3[GHz] are 13 [ohm] of reactance, connected between the ports.

I can't find a resistor with better capacitance, except for some very expensive diamonds resistors, out of my reach.

Is there any other alternative to improve the performance?

I have been trying other combiner configuration, like the hybrid branchline and rat race, which have the resistor connected to ground, but if i add the parasitic capacitance in parallel to the resistor to the simulation, i always get a poor isolation between input ports.

Can someone help me with this problem please?

Thanks in advance.

Best regards,

Vevo.

Maybe you can connect two series 25 ohm resistors (assuming the isolation resistor is 50 ohm) and decrease the effective capacitance.

Palpurul said:

Maybe you can connect two series 25 ohm resistors (assuming the isolation resistor is 50 ohm) and decrease the effective capacitance.

Click to expand...

Thanks for the suggestion. I tried that, but it seems that i need less than 0.25 pF to keep the isolation to at least -15dB. Thus, i would need four or more resistors in serie, and it is very impractical.

Anyway, thanks for your reply Palpurul.

Vevo.

Have you ever checked this manufacturer ??
http://www.vishay.com/resistors-fixed/high-frequency/
[url]http://www.emc-rflabs.com/Passive-Components/Resistors/Tab-Cover
[/URL]

I presume the problem is brought up by requiring high power rating (60W) for the resistor which is incompatible with the assumption of an ideal (lumped) resistor. It's not only the parallel capacitance but also the resistor size related to wavelength that disturbs the coupler design.

I presume, you need to model the actual resistor as distributed component with its parasitic properties in the EM solver and adjust the geometry for best overall transfer function.

FvM said:

I presume the problem is brought up by requiring high power rating (60W) for the resistor which is incompatible with the assumption of an ideal (lumped) resistor. It's not only the parallel capacitance but also the resistor size related to wavelength that disturbs the coupler design.

I presume, you need to model the actual resistor as distributed component with its parasitic properties in the EM solver and adjust the geometry for best overall transfer function.

Click to expand...

Thanks for the suggestion! I did not use the EM simulation of AWR, because i do not know how use hehe, i am learning about that right now. I am only use the clasical circuit schematic simulation using the microstrip components and lumped components.

I found a quote of this resistors http://www.emc-rflabs.com/Rflabs/me...ts/Resistors/Chip/CR0505DXXX,5TB.pdf?ext=.pdf which have 0.1 pF, U$S 45 unit, it is expensive, but i assume it saves me a lot of headaches.

Should i use the EM simulation if i use a resistor like that?

Thanks for your help!

Vevo.

Should i use the EM simulation if i use a resistor like that?

Click to expand...

They are rather small, so you can probably work with a lumped resistor idealization.

Are you aware of the heat sink requirement? The other side of using such a smaller resistor is a very high power density.

FvM said:

They are rather small, so you can probably work with a lumped resistor idealization.

Are you aware of the heat sink requirement? The other side of using such a smaller resistor is a very high power density.

Click to expand...

I think to use a big aluminum heatsink below the resistor, but i am not calculated yet the size.

How much should measure (in wavelength terms) a component to be considered a lumped element or a distributed element?

Because in a splitter i designed before, i used a resistor size 2512, which have 6.25mm (it is nearly 1/4 wavelength). I attach a picture for you here.

I'm afraid that will causing me problems when I arm and measure it in the real life.