types of resistors and inductance

Question

So while trying to interface a 5v SPI to a 3.3v SPI device, using the inline resistor method outlined below, http://www.sparkfun.com/tutorials/65

I discovered that at high clock speeds (above 200khz), my clock signal was being corrupted by the inductance in the resistor (with an oscilloscope, the clock looks signal looks triangular shaped) I am using carbon film (through hole) resistors for prototyping and I heard that these can have high parasitic inductance and not suitable for high frequency use. Is this information correct?

What types of resistor instead should I be getting (digikey link would be appreciated)? Right now, I'm just working on a protoboard but eventually I would like to manufacture this prototype on a PCB using surface mount components. It seems like from what I read about surface mount resistors, the inductance is not typically a problem for applications under 20Mhz.

Answer 1

In order to observe a clean square wave at 200kHz, the rise-time needs to be around 250ns (about 10% of the pulse width). The sparkfun tutorial shows the clock coupled through a 10k resistor. Now, to get 250ns risetime, you need an RC time constant of 114ns giving a capacitance of about 11pF. So the sum total of the circuit input capacitance, the stray capacitance and critically, your 'scope probe capacitance must be less than 11pF. I reckon this is your problem.

If the waveform looks OK at the driven end of the resistor and the input capacitance of the driven device is low, you might be OK but you won't be able to prove it without a very low capacitance 'scope probe.

Answer 2

_{I'm taking you at your word that inductance is the problem, though I find it dubious at a mere 200 kHz.}

Some axial thru-hole resistors are often created and trimmed by spirally cutting the element over a form which can give it some inductance. This is usually seen in metal film resistors, though possibly in some other film-type resistors as well. Carbon composition resistors, however, should possess very low inductances as it's more of a block of material.

Perhaps more fool-proof, some parts suppliers (Digi-Key does at least) have filters so you could just select "Non-inductive" and be done with it.

You are quite correct that SMT resistors have very low inductances. I'm not sure of the trimming patterns, but whatever it is, it's done in a plane, so at most you'll have a turn with an extremely small area.

Answer 3

When looking at fast signals it is usually beneficial to use the 10x setting on the probe, for the reasons discussed in the other posts.