580 MHz Lumped Element Bandstop Filter

vfone said:

Substituting the ideal components in Genesys with their models is just a simple drag-and-drop situation.

Click to expand...

You misunderstood my point. This was not about circuit simulation libraries.

Sure, you can go the way manually and try to re-tune your filter to spec with real values and parasitics. However, in Filter Solutions that is already part of the filter synthesis! The software will automatically evaluate all filter candidates with real components from your selected manufacturer series (S2P data based), and find the best possible filter implementation using real values including parasitics. That gives a much better understanding what specs are possible, and saves a lot of time finding the best topology under real world conditions.

Ansys NuHertz is the best filter design tool ever. Transferring lumped or distributed filter topologies is possible to AWR, Ansys EDT, Sonnet, etc.

RF filter design is a full time job. A filter syntheses software may help a lot, but is not enaugh. This is the reason that most of the RF filter designers have at least 10 years of experience. I mean guys that develop products that actually are for sale on the market, and not stuck in a IEEE publication...

I installed the Nuhertz filter solutions program. After entering my specifications, I added the s parameters file of the inductors and capacitors I will use. (I am using a coilcraft 0402 inductor and an atc 0603 capacitor.) However, all the program does is round up to the nearest available values. This doesn't work either. Is optimization possible on discrete values in this program?

rf997 said:

Is optimization possible on discrete values in this program?

Click to expand...

I need to check the demo version, if that is limited in capabilities. However it might take until next week before i get the demo version.

Before the tool was sold to Ansys, I had the full unlimited version and if I'm not totally wrong that supported discrete values based on manufacturer libraries. Let me check the demo!

rf997 said:

I installed the Nuhertz filter solutions program. After entering my specifications, I added the s parameters file of the inductors and capacitors I will use. (I am using a coilcraft 0402 inductor and an atc 0603 capacitor.) However, all the program does is round up to the nearest available values. This doesn't work either. Is optimization possible on discrete values in this program?

Click to expand...

Nuhertz replace the ideal valued components with nearest equivalent commercially available components. Optimization is not necessary because while replacing elements, it finds the best available value. As you know, you cannot find any value for any component.
If you still want to optimize the values, you can transfer to AWR ( or Ansys EDT ) then use the optimizer but the PDK must have "Discrete Optimization" feature as in Murata PDK. Not all PDKs have this feature.
See my previous posts. I used optimized available components.

I think because I use very low value capacitors (starting from 0.1 pf) and inductors (starting from 1 nF) in my designs, the filter is completely destroyed when I replace it with the closest values of the vendor models.
What is the minimum pF value of capacitors I should use to avoid parasitic effects and tolerances?

Consider a 70 Ohm microstrip on GETEK ML200C

0.25 mm trace 0.38 nH /mm
0.25mm dielectric 0.076 pF/mm

rf997 said:

What is the minimum pF value of capacitors I should use to avoid parasitic effects and tolerances?

Click to expand...

You can try to avoid such very small values, but cannot avoid parasitics here. Very small L look like an ideal component, but once you create the layout you introduce additional series L. On the other hand, too large L values will have low self resonance frequency and don't behave like an ideal L either.

One "trick" that works in some topologies: you can try to "absorb" parasitics from SMD and layoutpads into adjacent circuit elements. For example if you have parasitic shunt C and your design includes SMD shunt C components left and right, you can reduce your SMD capacitor value by the capacitance that you get as parasitics. The same applies for parasitic series L, if that is in series with actual circuit L.

Parasitics shunt C is influenced by PCB thickness, pads and components on a thin substrate are closer to PCB ground and will have large parasitic C. One possible solution is to create a partial ground layer cutout underneath the pads. Much of this is experience, to create a layout with minimum parasitics.

rf997 said:

I think because I use very low value capacitors (starting from 0.1 pf) and inductors (starting from 1 nF) in my designs, the filter is completely destroyed when I replace it with the closest values of the vendor models.
What is the minimum pF value of capacitors I should use to avoid parasitic effects and tolerances?

Click to expand...

Parasitic capacitances or package parasitics are almost not avoidable. One option is to use smallest package but there are other difficulties like mounting, debugging etc.
You have to simulate the filter's layout with practical components. In order to avoid unwanted coupling and line-to-GND capacitances , a good layout with GND free copper pouring should be preferred. I mean, there won't be any close copper pour around components.
Also components should be mounted as close as possible but not too close to prevent unwanted magnetic couplings.