Click here to go to our main page on microwave terminations
New for September 2020. A "dot" termination is usually a thin-film resistor in a circular pattern, fed from a transmission line. The combination of the resistance and the capacitance to ground of the dot make it behave like a termination over huge amount of bandwidth, and you don't need a via ground. You could also think of it as a resistive open circuit stub. Typically, the component uses 50 ohm/square resistive material, but we've never seen a good analysis of using the sheet resistance as a design variable. Come to think of it, the "dot" does not have to be round, and we've never seen an analysis of other shapes. Perhaps a better shape might be the ace of clubs, or the ace of spades? If anyone has a gds file of those shapes send it our way and we'll do an EM simulation!
You can learn about the dot termination that works to 110 GHz in Matt Morgan's CalTech thesis which is available on-line
http://www.submm.caltech.edu/~jonas/Theses/Matt_Morgan_03.pdf
We contacted Matt, and his response was
"It always amazed me how little-known this little gem is."
We agree!
Here's a variation of the dot termination that adds via grounds (what a cheater) that was patented in 2003:
https://patentimages.storage.googleapis.com/6d/4e/f1/3eefb4cf2db836/US20030048146A1.pdf
GaAs dot termination example
There are no good studies of dot terminations on the web that we can find, so we made our own using Axiem EM analysis. Note that once you have picked a substrate dielectric, its thickness and the ohms/square, you might not really have a say in how well a dot termination works over a specific band.
We created four dots of resistive material on 100um GaAs, each with a short fifty-ohm line input that we de-embedded off, as shown below. The diameters are 400, 600, 800 and 1000 microns. These are some beefy loads, one of these days we will examine their power handling. For sure they will handle much more millimeter-wave power than any 50-ohm resistor that you DC-ground with a via hole. We double-dare you to try to prove us wrong.
Different Diameters
Here's the same dots, with mesh applied at 100 GHz.
Different Diameters with Mesh
Here are dot termination predicted return loss (we accidentally called them pill loads, which is a different thing) versus frequency, with 50 ohms/square sheet resistance on 100um GaAs The 800um diameter load has great bandwidth (10 to 100 GHz with 15 dB return loss) but who can afford to waste that much GaAs?
100um 50 ohms/sq
Here are the predictions on a Smith chart. The red line is nearly a bullseye, but you can imagine that some clever tuning with the access line could make it even better...
100um GaAs 50 ohm/sq on Smith chart
What if we were able to change the sheet resistance to 40 ohms/square, either intentionally (not likely) or accidentally (happens all the time.) The performance suffers a little, but it is probably OK for most applications.
100um 40 ohms/square
What happens at 60 ohms/square? You could argue it improved the performance.
100um 60 ohms/square
What about on a thin (50um) substrate? With 50 ohms/square, The performance definitely degrades.
50um 50 ohm/sq
There probably is no 150um process available, but what happens with the substrate is thicker? It looks good at low frequency, but there is some weird mode going on around 70 GHz.
150um 50 ohm/sq
Conclusion: GaAs dot terminations may not be very useful, unless you were looking for a monolithic high-power termination. But you can imagine that on other substrates where you can choose dielectric constant, substrate thickness and sheet resistance, you can make a nicely matched, affordable high-power broadband load.