## RF trace simulation in ADS

• 0

We are trying to simulate 42mm RF trace at 77GHz in ADS. The values that were obtained for return loss were +6.4 dB. But when the same trace was simulated for a length of 25mm, we are getting good return loss (-17 dB). We are using CPWG structure for RF traces. Is there any relation between the length of RF trace and return loss? Kindly let us know if there is any possible way to eliminate this. Thank you.
• 0

1) Is that return loss or |S11|? For a passive circuit, return loss is positive and |S11| is negative. If it is going positive to negative you have something fundamentally wrong with your simulation.

2) I assume this is EM simulation in Momentum? I don't use ADS, but maybe there are equivalents of Microwave Office's port impedance and passivity; check those. You have to be careful for port setup of CPWG. Typically the ground ports are "negative numbered" and of course you need two of them to evenly split the current. What do the currents look like? Do you have a proper mesh? Do the results converge as the mesh density is increased?

3) Finally, return loss does vary with electrical length. That's a fundamental property of a transmission line that is not exactly terminated with it's characteristic impedance. You can also have dispersion, but again I think you have something fundamentally wrong.
• 0
Dave replied the topic: RF trace simulation in ADS

4) As well as port setup, be careful with via and/or ground strap spacing in the EM simulation; spacing relative to each other and distance from the first strap/vias to the ports can cause unexpected resonances in the result.
• 0

1.It is the S11 parameter.
2. It is the mometum simulation in ADS. We are getting proper S11 and S21 when the length of the RF trace is within 25mm. But if we increase the length beyound 25 mm, we are facing this issue. We are having two adjacent ground and one bottom ground and they are defined as negative in the port setup.Current distibution is as shown in the attachment. I have taken this snapshot from momentum visualization window. Mesh formed takes the colour of the ground conductor. Cells/Wavelength is 20 for this simulation.
3. These results are obtained from the layout simulation. Is it required to terminate it with 50 ohm in schematics?
• 0

Thank you Dave.
Is there any specification or formula to calculate the distance of the pin to the first via?
We have maintained a constant spacing between vias in this simulation.
• 0
Dave replied the topic: RF trace simulation in ADS

I don't know of any specific formulas for the distance from pin to first via. There might be some - check out Coplanar Waveguides and Circuits by Rainee N Simons. I haven't read it, but it's on the MW101 recommended books page: www.microwaves101.com/encyclopedias/book...icrowave-engineering I suspect in reality most people just treat it as an EM optimisation problem.

In the screenshot you posted, what's the dielectric constant and what's the line length? 77 GHz in free space is 3.9 mm, shorter once you take into account dielectric constant and filling factor, and I think you'll want the ground vias to be closer than that. Lambda/10 is a good rule of thumb at lower frequencies but almost impossible to fabricate at mm-wave.
• 0

Thank you Dave for the link. The dielectric constant is 3.1 and the length of the trace is 42 mm. The maximum via seperation that the board fabricator can provide is 0.4mm.
• 0
Microwaves101 replied the topic: RF trace simulation in ADS

A long line that is mismatched will have a periodic response in S11, lots of dips and valleys. If you just look at one frequency you will have no idea what is going on. If you plot the trace across frequency and look at the maxima (-6 dB or whatever), then subtract 6 dB from that number (-12 dB in my example) that is the return loss that matters. The dips are caused by constructive interference due to the two mismatches, one at each end.

I hope I didn't misunderstand the question...

Steve
bold
italic
underline
strike
code
quote
ulist
image
url
mention
reveal