# Low pass filter design - Microstrip

#### NR152

Joined Aug 23, 2011
1
Hello!
I am working on a Third Order equiripple Low pass filter design suitable for Microstrip ( Distributed structure). The cut-off frequency varies from 400-800 MHz and the range of capacitnace values varies from 1-7 pF for the following setup ::
SMA connectors at board edge to filter ports via 50 Ohm Microstrip,
unspecified length.
The Filter comprises of five transmission lines directly coupled
together.
Line 1: shunt, 64.9 Ohm, terminated in VarCap.
Line 2: series, 217.5 Ohm.
Line 3: shunt, 70.3 Ohm, terminated in VarCap.
Line 4: series, 217.5 Ohm.
Line 5: shunt 64.9 Ohm, terminated in VarCap.
All transmission line lengths are 2.5 cm.
A 'diagram' of layout is below:
Cap Cap Cap
* * *
* * *
* 1 *3 *5
* * *
********************************
Feed 2 4 Feed
I have to design this on Microwave Office. The problem is that these measurements were based on assuming the Er of the substrate as 2.2 I would like to know the following things :
1. If the Er of the substrate is changed to 4.3, how will the above values change? On what basis can they be recalculated?
2. How to find the length of the transmission line connecting the feed( with impedance 50 Ohms) to the shunt transmission line (1 & 5)?

#### SgtWookie

Joined Jul 17, 2007
22,230
A 'diagram' of layout is below:
Rich (BB code):
           Cap   Cap   Cap
*     *     *
*     *     *
* 1   *3    *5
*     *     *
********************************
Feed       2     4          Feed

#### SgtWookie

Joined Jul 17, 2007
22,230
Just for clarification for those perusing the threads, Er = dielectric constant of the substrate material.

If the dielectric constant of the substrate is increased, it will increase the parasitic capacitance of all of the microstrip traces, so the trace widths would need to be reduced to compensate; otherwise your impedance will be off by quite a bit. Your return loss (S11) will skyrocket along with your insertion loss (both S21 and S12) as viewed on a network analyzer.

[eta]
You might investigate Rogers Corp. low-loss PCB materials:
http://www.rogerscorp.com/acm/index.aspx
But if you are going to stay with RC networks, there would not be much point in it.

Last edited:

#### SgtWookie

Joined Jul 17, 2007
22,230
I'm curious why you're using resistors instead of inductors?
Usually, you would use RC networks in an active filter, not a passive filter. While it will be easier and less expensive to use RC networks, you will have far lower losses with LC networks; just using resistors on the inputs and outputs to match the impedance.