Assymetric Stripline confusion...

[Axemaster]

Hey guys, I just took a look at some asymmetric stripline calculators and I have some serious reservations about the results I'm getting...

For an ordinary microstrip:
Trace Width = 8 mil
Substrate Height = 0.12 mm
Er = 4.4
----> 50 ohms

For an asymmetric stripline:
Trace Width = 8 mil
Substrate Height = 0.12mm (1st GND plane), 1.2mm (2nd GND plane)
Er = 4.4
----> 73 ohms??????

This result seems profoundly counterintuitive. So much so that I seriously doubt it could be true. I checked with multiple online calculators and they all behave similarly. What the **** is going on here?

**broken link removed**

 

[SLK001]

I don't see why you would be confused about the results. The µstrip is half air, half substrate. The stripline is completely buried in the substrate. What were you expecting your results to be?

 

[Axemaster]

I'm confused because I expected the impedance to decrease, not increase. By sandwiching the trace between GND planes, the capacitance per unit length has increased, which should make the impedance go down. I expected the impedance to drop from 50 ohms to 35-40 ohms or something like that.

If the impedance is supposed to go up, then I don't understand the mechanism involved.

 

[KlausST]

Hi,

Characteristic_impedance is different to the usual known impedance.

On the other side ... doesn't it make sense that one wire has an impedance...and thus two wires have twice the impedance...

In the asymmetric case the "load" is connected between stripline and GND plane. But the GND plane shoukd be seen as very low impedance.
But in the symmetric case the load is connected between both lines. The return path is not the low impedance GND plane anymore, but the other stripline with it's impedance.

Klaus

 

[FvM]

The linked asymmetric strip line tool is apparently flawed. I get reasonable results (about 42.5 ohms) with Saturn PCB toolkit and other solvers.

 

[Axemaster]

I did some more investigating, and as FvM said above, it seems like many of the online calculators use a flawed equation. Specifically, they model the distance between layers as:

h_effective = \[\frac{h1+h2 }{2}\]

So essentially they are just taking the average distance between layers and then calculating as if it were a symmetric stripline. This has the advantage of "looking ok" since it matches the results for symmetric stripline when you compare the two. The disadvantage is that it is completely wrong.

It's particularly unfortunate to see that probably 1/2 or more of the online calculators are using this method.

- - - Updated - - -

Sorry that should have read:

h_effective = (h1+h2)/2

I did it in LATEX format but it didn't display...