designing with lumped components parasitics

Hello,

How does a more experienced designer deal with component parasitics. Do you calculate or measure them first and then include them in the design of your impedance matching components? If so How?

First I design a basic amplifier with the simulation tools.
I solder the components on my PCB.
I determine the phase offset for my VNA, then I measure the S parameters of my amplifier.
Using the simulation tools and the S parameters measured I design the impedance matching components.
Unfortunately when I place the lumped components on the board due to parasitics I usually find I have gain at 500MHz as opposed to the 430MHz I designed for. I then have to manually desolder then solder components of slightly differing values until I get the desired gain at the correct frequency. The gain is usually 4db less than predicted.

darcyrandall2004 said:

Hello,

How does a more experienced designer deal with component parasitics. Do you calculate or measure them first and then include them in the design of your impedance matching components? If so How?

First I design a basic amplifier with the simulation tools.
I solder the components on my PCB.
I determine the phase offset for my VNA, then I measure the S parameters of my amplifier.
Using the simulation tools and the S parameters measured I design the impedance matching components.
Unfortunately when I place the lumped components on the board due to parasitics I usually find I have gain at 500MHz as opposed to the 430MHz I designed for. I then have to manually desolder then solder components of slightly differing values until I get the desired gain at the correct frequency. The gain is usually 4db less than predicted.

Click to expand...

Have you verified any of your methodology against a manufacturer's reference design or application notes? One can take the dimensions from the artwork they provide (for their ref design board at their design frequency), code that up into your sim engine (also using S-Parms or large signal Z parms for some Freescale parts), taking account the board material type, and see how close you come to their published results in the modeling?

Jim

I take a more empirical approach. For things like DC block caps, I put it in series on a line and measure S21 and S11 to 20 GHz. If it works well, I remember the part number and use it a lot.

For things like resonators, etc, I place them in a known circuit configuration such as with chip input and output coupling capacitors, then measure the resonance frequency and 3 dB bandwidth, and can then calculate the parasitics of the resonator (like Q or R, stray inductance, etc) by curve fitting on a simulator.

Also, it should be mentioned, not all 0603 or 0402 caps and other components operate at higher frequencies (or power levels!) quite the same way. Those ceramic coupling or bypass caps that work well at audio frequencies may not perform at all at UHF and higher and be very lossy (not all ceramic materials are created equal nor were they designed to be).

biff44 is getting on to this point with his method of characterizing individual components at the frequency of interest; no better method exists for really getting what some call 'ground truth' about a component's performance.

When you mention 'lumped' components, we are assuming surface mount L and C (and R) components; leaded components of yesteryear are even worse when it comes to characterizing and your only salvation there is often the 'empirical' (trial and error) approach.


Jim