matching connector to microstrip question

Thread Starter

yef smith

Joined Aug 2, 2020
756

ZCochran98

Joined Jul 24, 2018
304
There's a ton of different taper types, and each one is tailored to a different application. The easiest is a linear taper, while something like a Klopfenstein taper (which is a specialized form of a Chebyshev taper) will provide an equiripple match with arbitrarily-low reflections (determined by you as a design choice) above some minimum frequency (which determines the length). What is most important to you? How much length do you have to work with? Depending on your ripple, frequency, and phase requirements, you might be able to just use a simple linear taper. If you need ultra-low reflections with equiripple characteristics, then a Klopfenstein is good. But if you're limited on space, then exponential is good. If you have all the space in the world to make the transition, a linear taper will work. Try some simulations of each, if you have access to ADS (you can do this in CST or ANSYS too, but modeling anything other than a linear taper is a pain, in my experience).

Also, keep in mind: the transition is not just a coax to microstrip transition: it's a grounded CPW to microstrip transition (due to how the coax connector works). So you'll have to taper your top ground plane away from the microstrip line as well to the point where its space from the microstrip makes it irrelevant.
 

Thread Starter

yef smith

Joined Aug 2, 2020
756
Hello ZCochran98,yes i will try to do linear taper.I do have a CST simulator to test the linear taper.
Based on the step file i see d_in=0.18mm d_out=1mm when i put the PTFE 2.2 dielctric constant i get 70 Ohm.
So is it a 70 ohm connector?

Thanks.

1707134276078.png

1707134074048.png
1707134568237.png
1707133756706.png
1707133700239.png
 
Last edited:

ZCochran98

Joined Jul 24, 2018
304
Based on a simulation of the CPW line it needs to connect to I did in some spare time, I'd estimate the impedance at the tip of the coax is closer to 60 than 70, but either is a good approximation of the impedance you need to taper from.
 

Thread Starter

yef smith

Joined Aug 2, 2020
756
Hello ZCochran , Rin=0.18mm Rout-0.65 epsilon=2.2 this is my 50 OHM
when i put it on top of my microstrip line i get very large reflections.
what it the theory of matching the coax to microstrip transition?
Thanks.
 

ZCochran98

Joined Jul 24, 2018
304
Matching the end of a coax to a microstrip is similar to the transition between CPW and microstrip. Where you connect your coax connector it needs to be a CPW (or, rather, grounded CPW) line in order to continue the coax wave from one medium to another (preferably impedance-matched). Keep in mind, where your connector contacts the board there is a space where it is air between the pin and the shield, and not PTFE. Use that section of the coax to calculate your necessary impedance for CPW to match there. Then, do your taper/transition from whatever impedance that is in CPW to whatever you need in microstrip (50 Ohm). The gap between the top ground plane of the CPW and the center line will likely have a higher "slope" than the line taper itself. You'll end up with something similar to the (very crudely drawn) sketch below:

1707163445667.png

The sketch, for the record, is supposed to be symmetric across the x-axis, but I'm not good at drawing in MS Paint.
The centerline is the center conductor for the CPW and microstrip, while the shapes above and below it are the ground planes for the CPW section. The center conductor at the left side is matched in width (maybe slightly wider) to the coax pin, and the GND is spaced apart so the (grounded) CPW impedance matches the impedance at the tip of the coax connector. Then, the taper occurs, and the ground plane also tapers, but faster (so you need to properly tune both transition angles) to go from (grounded) CPW to microstrip. If you don't tune these parameters together correctly, then you'll have large reflections.
So, you have a few parameters to tune:
  1. Center conductor width where the coax makes contact
  2. Ground spacing between that same center conductor and the top ground
  3. The microstrip width (usually fixed for whatever the 50-ohm width is)
  4. The transition length
  5. The top ground plane transition angle (usually "faster" than the center conductor width transition)
The center conductor at the coax contact point should be about as wide as the coax center conductor pin, and the ground plane should be spaced close enough that the coax shield can make contact with it. The transition length and the ground plane angle then need to be adjusted/tuned.
 

Thread Starter

yef smith

Joined Aug 2, 2020
756
Hello Zcochran98,I am trying to follow your words .
I have a coax cable with exposed pin touching the surface of the microstrip as shown bellow.(dielectric layer is invisible so you would see the VIA shielding)
The outer conductor of the coax is touching the GND of the microstrip.
The exposed pin touched the surface of the microstrip.
In real life we have soldering connecting the outer pin with the microstrip do i need to make the pin plunged inside the connector or the surface of the pin needs to touch the surface of the microstrip?
Thanks.
1707209723246.png
 

ZCochran98

Joined Jul 24, 2018
304
Can you send a front view and a side view of how you have the coax connected? It looks like the pin is hovering above the line and the shield isn't making contact with your top ground plane. In addition, your port on the end of your microstrip line intersects too deeply with the top ground - its edges should just barely make contact with the top ground and bottom ground (top spacing of port is fine).
 

ZCochran98

Joined Jul 24, 2018
304
That picture is clearer. The 0.18 mm cylinder is the center pin, and goes where you have it (on the CPW center conductor). The match you have isn't bad, if what you've modeled is the coax-to-CPW transition with the 0.18 mm diameter pin. As I expected, the CPW at that point is about 60 ohms, so you'll have to taper it up in width so the output is 50 ohms. The return loss you've plotted is almost spot-on with the return loss in the datasheet, so you're probably not going to get much better than what you have there.
 

Thread Starter

yef smith

Joined Aug 2, 2020
756
Hello , In my simulation the centre pin is 0.3 .
I don’t know where to put the 0.18mm. I got this good result with 0.3 . What do you recommend ?
Thanks.
 

ZCochran98

Joined Jul 24, 2018
304
The 0.18 mm is the center pin, according to the datasheet. The center pin is 0.18 mm, and the inner diameter of the shield is 0.3 mm, if I remember the datasheet right
 

Thread Starter

yef smith

Joined Aug 2, 2020
756
Hello ZCochran98,you showed a very good tip i intent to folow.
could you please give a transmission line intuition regarding these parameters.
so i'll know what RF effect lies behind it andknow what to expect on the smith chart?
  1. Center conductor width where the coax makes contact
  2. Ground spacing between that same center conductor and the top ground
  3. The microstrip width (usually fixed for whatever the 50-ohm width is)
  4. The transition length
  5. The top ground plane transition angle (usually "faster" than the center conductor width transition)
 

ZCochran98

Joined Jul 24, 2018
304
  1. The center conductor and top-ground-to-conductor spacing are intrinsically linked in reasoning. Because they work together to form a (grounded) CPW, you can match the width of the center conductor to the width of the connector pin for alignment purposes. Then, you can choose the spacing of the top ground plane from the center conductor to match the impedance of that center pin (about 60-70 Ohm from before). If you normalize the smith chart at this point to that impedance, you choose those parameters (center conductor width and ground spacing) to be as close to the center as possible. If you normalize the Smith chart to 50 ohms like normal, then the endpoint will be off to the right a little.
  2. The microstrip width: at the end of the transition between the two, you said you have a microstrip, so that needs to be matched to 50 ohms (usually, anyway). So you choose your width appropriately to match to the center of the smith chart there.
  3. The transition length is less of a direct design parameter as it is a consequence of other design choices. Depending on your chosen requirements for ripple and phase, this will vary. How long it is is tied to the transition between the connection point and the microstrip you're trying to go to and the transformation method you chose. In the long run, it'll contribute partially to the path that the impedance transformer will take along the smith chart to transform the impedance at one side to the other (circular, elliptical, or some other path depending on how the taper is designed). It'll also contribute to how far away from the center impedance between the endpoints the transformer will take the match during the transformation. For instance, a 90-degree stepped impedance set to 54.8 ohms (the center frequency between 50 and 60 ohms) will take you on a half-circle between the center of the chart to the point on the chart corresponding to 60 ohms (so 1 and 1.2 once normalized to 50 Ohms, corresponding to \(\Gamma = 0\) and \(\Gamma = 0.05\), respectively), with a small radius. But a shorter line that's tapered may take you in an elliptical path instead.
  4. The top ground plane transition angle is closely tied to the transition length and the transformer type. The taper between the end where the pin connects and the end where the 50-ohm line begins will be of your choosing (how you do it, with linear being the easiest). However, your top ground plane needs to eventually stop or be far enough away from the microstrip line so as to make it a microstrip line and not another GCPW. Thus, why the ground plane needs to transition away from the taper, and swiftly enough so the distance between the center conductor and the plane grows enough so the microstrip line is actually a microstrip line and not a CPW. How quickly depends on a lot of things, but it works with the taper type and length mostly to define the shape of the transformation path across the Smith chart as well (basically, working with the transition length).
All-in-all, when you have your smith chart in the design process, you can do one of two things and then monitor the responses:
  1. Set the ports to be the impedances you know are at the ends, then design the taper so each port sees a perfect match (matched to the center). So if Port 1 is on the side that the connector is on, set its impedance to 60 (or 70) and if port 2 is on the microstrip side, set its impedance to 50. Then, adjust everything (optimizer?) so each port sees a perfect match at the desired design frequency
  2. OR Set the ports both to 50 ohms, and the on Port 1 (connector side), you should see the impedance bet 60 (or 70), so you'd see an imperfect match (\(\Gamma = 0.05\), or \(S_{11} \approx -13\) dB, with the curve passing through 1.2 [or 1.4] on the normalized Smith chart at your design frequency), but on the Port 2 side (the microstrip side) you'd see a perfect match at your design frequency (so \(S_{22}\) would be a null).
Hope this helps!
 
Top