HFSS port 'Port1' does not have a valid reference plane above or below it.

Status
Not open for further replies.
promach..
You're trying to simulate your planar structure with wrong tool as HFSS.It doesn't mean that HFSS cannot be used for planar structures but it's really cumbersome tool and it's really efficient FDTD problems.
All you have to do is to find a planar EM simulator such as ADS Momentum even Sonnet Lite and simulate your structure without trouble.If you're student, Keysight and National Instruments may help you to give a time limited but fully functional tools.Sonnet Lite can be used up to 4 ports (16MB memory) and it's absolutely free and it's very easy to use and very efficient for planar structures.
HFSS is very powerful but hard to use tool..
 

BigBoss said:
Sonnet Lite can be used up to 4 ports (16MB memory) and it's absolutely free and it's very easy to use and very efficient for planar structures.
Click to expand...

Sonnet Lite is great, but he will hit the memory limits here. The combination of narrow gap and non-90° lines will need to much memory in Sonnet Lite.
 

volker@muehlhaus said:
Sonnet Lite is great, but he will hit the memory limits here. The combination of narrow gap and non-90° lines will need to much memory in Sonnet Lite.
Click to expand...
Yeah, possible.. But he can also request a evaluation from NI for Axiem. It's also very accurate and I have used few times for serious designs and it has never belied me on measurements.
It's fast and accurate..
 

@volker@muehlhaus

I think I know why HFSS gave completely different result compared to ADS for the same DXF file.

DXF is only 2D.
I suppose ADS automatically added the 3D height for the microstrip layout.

In HFSS, the microstrip layout does not have 3D height added automatically unlike the via.

 

promach said:
I suppose ADS automatically added the 3D height for the microstrip layout.
Click to expand...

I created a stackup to map the 2D conductor layers to the correct z position, with the correct thickness. Post #30, second picture.

promach said:
In HFSS, the microstrip layout does not have 3D height added automatically unlike the via.
Click to expand...

There is something else in your 3D view that looks weird: It seems that you don't have the backside (ground) metalization that is required for microstrip. I added that in ADS myself, because I know it is required for this design. In your DXF data, there was no ground metal.

Can you show a side view of your HFSS model? It is difficult to see if the bottom end of the via is "hanging in the air" or connected to the bottom boundary.

There are two ways how to get metal for ground: you can draw it on the ground layer, or you can make the bottom boundary condition an electric boundary with PEC (perfect conductor). That simulation boundary must hen be at the z-position of the bottom layer (ground).
 
Last edited:

In your post #30 , you did not show the balun phase, S21 and S31


Note that HFSS simulates the PCB with air below and above it, hence the large box enclosing the PCB within


Overall view





Side view

 

promach said:
Note that HFSS simulates the PCB with air below and above it, hence the large box enclosing the PCB within
Click to expand...

You can do that, but then you need to draw the ground plane. From your screenshot, I don't see if you drawn ground plane metal, or if that is the boundary of the FR4 dielectric.

promach said:
In your post #30 , you did not show the balun phase, S21 and S31
Click to expand...

My ADS license is busy with something else now, but I can add that plot later.
 

From your screenshot, I don't see if you drawn ground plane metal, or if that is the boundary of the FR4 dielectric.
Click to expand...

See the layer stackup at post #44
 

promach said:
See the layer stackup at post #44
Click to expand...

The stackup doesn't tell if you have a ground polygon actually drawn on those layers.

We had that "via connected?" topic earlier in port #14/#15 but I'd rather want to see the ground polygon in the 3D view. It was missing in the DXF file, so I wonder if your model has the correct ground. Having a layer definition named "ground" means nothing if there is no metal drawn on that layer.

- - - Updated - - -

Here's the phase that you asked for. Note that all your lengths are different, so we don't expect 180° phase shift.



 
Last edited:
Let me investigate more about the ground layer.



- - - Updated - - -

Ok, I have made the balun works.

The real reason is due to the via connectivity to the ground layer.



- - - Updated - - -

@volker

Do you have a theoretical understanding on why balun needs via ?

I suppose this planar marchand balun works on the concept of EM coupling of two microstrip transmission lines ?
 

Zin=Z0* [ ZL+j*Z0*tan(theta）] / [ Z0+j*ZL*tan(theta）]

Someone told me to use the above expression to understand why balun needs via. Any advices ?

Besides, why arg(S_21) - arg(S_31) alternatives between 180 degrees and -180 degrees ?
 

promach said:
Let me investigate more about the ground layer.
Click to expand...

It's possible that the "negative" flag creates a solid ground plane if you drawn nothing.

promach said:
Ok, I have made the balun works.
The real reason is due to the via connectivity to the ground layer.
Click to expand...

Some mistake with the via is what I suspected starting from post 3.

promach said:
Do you have a theoretical understanding on why balun needs via ?
I suppose this planar marchand balun works on the concept of EM coupling of two microstrip transmission lines ?
Click to expand...

The via change the phase in the coupled line (reflection from open end vs. short), so you can't leave them out.

promach said:
Besides, why arg(S_21) - arg(S_31) alternatives between 180 degrees and -180 degrees ?
Click to expand...

That's only a data display issue, the arg() data has that "jump" from -180° to 180° for each of the two phases. I didn't know how to "unwrap" the phase for continuous values <-180°
 
The via change the phase in the coupled line (reflection from open end vs. short), so you can't leave them out.
Click to expand...

How is this related mathematically to Zin=Z0* [ ZL+j*Z0*tan(theta）] / [ Z0+j*ZL*tan(theta）] ?
 

promach said:
How is this related mathematically to Zin=Z0* [ ZL+j*Z0*tan(theta）] / [ Z0+j*ZL*tan(theta）] ?
Click to expand...

Your equation gives the input impedance for a line terminated into ZL. It makes a difference if ZL=0 (via shorting line to ground) or ZL=infinite (open end, via not connected)
 
Why does the balun phase deviates away from 90 degrees and -90 degrees as HFSS sweeps up the frequency ?

Can we do anything to the balun layout in order to make this issue less serious ?

The zip file attached in this post contains the dxf layout file.



- - - Updated - - -

See post #50, I was referring to the phase drift below 4 GHz.

After this, I will solve the phase drift above 4 GHz
 

Attachments

  • planar_marchand_balun.zip
    5.8 KB · Views: 83

Here are results for your latest DXF layout:



promach said:
Can we do anything to the balun layout in order to make this issue less serious ?
Click to expand...

I would start from scratch and re-design the layout, without all those asymmetries. Equal lengths and a symmetric layout, instead of that 135° turn.
 
Last edited:

volker@muehlhaus said:
Your equation gives the input impedance for a line terminated into ZL. It makes a difference if ZL=0 (via shorting line to ground) or ZL=infinite (open end, via not connected)
Click to expand...

Alright. I will redo the layout. To be frank, that 135 degree turn is to allow space for isolation circuit between port 2 and port 3. Please advise.

Besides, how does the length of the coupled microstrip layout affect the value of theta in the expression ?

Note: if we have a via (ZL = 0) , the expression reduces to Zin = j*Z0*tan(theta)
 

promach said:
is to allow space for isolation circuit between port 2 and port 3. Please advise.
Click to expand...

I have no advice for coupler layouts, because I've spent most of my time in simulation (not design). My idea was just to make it more symmetric. The 135° looks a bit suspicious to me, with possibly large parasitics due to current crowding at the extreme bend backwands, but if the layout is optimized that might work ... I don't know.

promach said:
Besides, how does the length of the coupled microstrip layout affect the value of theta in the expression ?
Click to expand...
The equation is for a single line (not coupled), theta is the electrical length of the line (S21 phase)
 


I have made both the coupled lines matching length and width. However, I am still not getting 180 degrees phase response.

Why ?

I have attached the dxf file as zip at the end of this post



 

Attachments

  • planar_marchand_balun.zip
    5.8 KB · Views: 81

Status
Not open for further replies.

Similar threads

Menu
Log in
Register
Cookies are required to use this site. You must accept them to continue using the site. Learn more…