Resistors aren't resistors
Resistors aren't resistors - at high frequencies, that is.
Many designers don't realize that parasitic elements in real components affect their values. When frequencies reach hundreds of megahertz, basic components such as resistors, inductors, and capacitors take on non-ideal characteristics. Such changes can become critical when you design filters or attempt to optimize power delivery networks, bypass networks, or bias circuits.
We will discuss capacitors and inductors in later posts. For now, let's talk about the lowly resistor. Here's the plot of an ideal impedance of a resistor, which, as you might expect, is a straight line.
Now let's consider a carbon-composition resistor with short leads. If you add the parasitic inductance of the leads and parallel capacitance between the end caps, you should get this simplified model at high frequencies.
Typical values for the carbon-composition resistor (with 1/4-inch leads) might be 14nH of series inductance and 1-2pF parallel capacitance.
Now, if you plot this simplified model versus frequency, you should see the following idealized impedance plot.
At lower frequencies, the plot would be purely resistive (horizontal line). But as frequency increases, the parallel capacitance dominates, and the impedance starts to drop at 20dB/decade. The resistor now becomes a capacitor. The breakpoint occurs here.
There will be a point where the capacitive reactance equals the inductive reactance. For a brief moment, the impedance will become purely resistive once again (though at a much lower resistance value). This series resonance occurs at this breakpoint.
After this point, the series lead inductance becomes dominant, and the poor resistor becomes an inductor. Its impedance plot rises at 20dB/decade.
To help illustrate this, I measured a carbon-composition resistor with 1/4-inch leads and plotted the following.
The frequency varied from 1MHz to 450MHz, so you can't see the impedance increasing because of series inductance. At 100MHz, however, you can see the impedance of the 1kΩ resistor has dropped to about 730Ω. At 300MHz, it's just 330Ω.
Even when using surface-mount components, where the series inductance is typically 1-2nH and the parallel capacitance is 0.2-0.4pF, this can still affect the measured value of impedance at high frequencies into the hundreds of megahertz.
By understanding that parasitic elements of real components can affect the impedance, you can see why keeping lead lengths and circuit traces short and why surface-mount components are superior for high-frequency designs.
Have you run across cases where series inductance or parallel capacitance changed with increasing frequency? How did that affect the performance of your design?
More about Kenneth Wyatt.