LNA PCB thickness important

[N.Narayanan]

Hi,
I want to use surface mount LNA in my module.
Vendor had tested the LNA in 10mils thickness board and posted the data for that.
But my module board thickness is 60mils.
Could I use it in 60mils thickness board and what are change could I expect?

Thanks.

 

[Ow@i$]

well increasing the thickness of the board will effect your impedance as it will decrease the coupling b/w top layer and bottom layer...... it is better to use the recommended layout and stack-up for optimum performance.

 

[ferdem]

Impedance and substrate loss should be considered.
As long as trace impedances are kept 50 ohm and your substrate loss is reasonable(comparable with suggested substrate) for your operation frequency, you can use it. I expect to achieve reported performance in the datasheet.

 

[volker@muehlhaus]

I want to use surface mount LNA in my module.

Frequency range?

Thick PCB means wider lines for same impedance, and more inductance in ground vias. It depends on the frequency range if this is an issue.

 

[albbg]

The substrate of your module and that of the vendor are the same ? If you have the same thickess, but different epsilon (and/or losses) of the two substrates the performances will be different.
In some cases keeping the 50 ohm lines matched could not be enough

 

[N.Narayanan]

Frequency range is 1.2GHz-1.4GHz.
Also, the circit has few biasing components.
These components were fixed for 10mils board but now I am gonna use it in 60mils board.
Is this OK?

Thanks.

 

[D.A.(Tony)Stewart]

You can reduce loss tangent effects with a Polyamide PCB and thin is better. If not Polyamide use a lower fiberglass ratio with resin.. Keep in mind Permitivity begins to drop from ratings at lower freq. so check. and tolerance is poor on untested boards for controlled impedance.

 

[vfone]

No, is not OK to use the same component values at those frequencies when the PCB thickness is increased 6x. The circuit might need another tuning adjustment.
To maintain the linearity the circuit might need different bias when the matching components (which are frequency dependent) are changed.

 

[volker@muehlhaus]

when the matching components (which are frequency dependent) are changed.

From the initial question, we don't know if there are any transmission line matching components. My interpretation of the initial question ("surface mount LNA") is that it's an LNA chip with a few SMD around it. At 1.5GHz that is something I would consider rather insensitive to the PCB details, except for maybe ground (via) inductance.

 

[vfone]

The initial 10mils it is relative thin substrate for a lumped circuit design. It happen that right now I am doing a GPS narrow-band LNA design for a particular project made on 60mils substrate.
There are no microstrip matching components, just the usual inp/out ustrip transmission lines, and lumped inductors and caps.
Playing in the simulator just with the thickness of the substrate from 60mils to 10mils, the center frequency shift almost 100MHz, and of course the gain and linearity of the LNA was affected.
6x increase in the substrate is a lot, even for a 1.5GHz design, especially when is narrow-band.