ESL 0402 resistor

[stcoso]

Hi... a short and probably non simple question...


Can anyone quantify the ESL of a 0402 low value (0.5-2Ohm range) resistor around 100MHz ?  What about 0603?



And about their respective reversed (wide terminal)?


Thank you in advance.

[hamster_nz]

Maybe

0402 0.0019 nH
0603 0.0267 nH

From "Frequency Response of Thin Film Chip Resistors"

https://www.vishay.com/docs/60107/freqresp.pdf

[helge]

the 20-30 fF values mentioned in the freqresp.pdf document above relate to the internal shunt capacitance of the equivalent circuit model.

Vishay CHP / HCHP resistors have a known |Z|/R behavior:



  1 Ohm 0603 flip-chip type reaches |Z|/R ~ 1.4 = sqrt(1²+1²) at ~ 620 MHz, so (1 Ohm) / (2*pi*620 MHz) -> 0.256 nH.
10 Ohm 0603 flip-chip reaches |Z|/R ~ 1.4 at 5 GHz -> 0.32 nH.

I find it rather strange that the 1 and 10 Ohm curves allegedly produce the same ESL  for both wrap-around and flip-chip mounted resistors but that should give a sort-of upper bound for 0603 ESL of 0.32 nH at 380 µm thickness (ZTA substrates can be bought in 0.32mm and 0.38mm thickness, your guess is as good as mine what that means for the actual dielectric thickness - but it might be 320 or 325 µm thick Al2O3 + terminals).

[Siwastaja]

What do you do with ESL in isolation? Layout will affect it more. If ESL is important, layout is part of the analysis in any case. Assuming you approximate ESL by simple geometrical models, just assume the resistor is a wire of the same length and it's going to be close.

For example, if you plug 60-mil length and 30-mil diameter in https://www.eeweb.com/tools/wire-self-inductance-calculator/ , even when this assumes cylindrical shape and the resistive material filling the whole cylinder, which isn't even near the right model, still the output is 0.477nH which is pretty close to what was quoted above.

For anything more exact, any laser etching of the resistor during manufacture is going to affect the number, and so is the width and thickness of the resistive material.

But because the typical layout of a 0402 part will be at least 1nH alone, and any realistic circuit with any other part than the resistor will be possibly tens of nH, does it matter if the resistor is 0.2nH or 0.3nH or perhaps 0.4nH in isolation?

[niconiconi]

Howard Johnson (of High-Speed Digital Design fame) has an article on the ESL of 0402 and 0603 bypass capacitors. He used an interesting method to determine the parasitics - He assumed that it's caused by mostly layout and geometry, not components. This allowed him to avoid doing a delicate microwave VNA measurement or any computer simulation, instead, he made a huge metal block and measured that as a 100x scaled model  - a time-honored technique in engineering. As far as I know, his results are comparable with the numbers from E&M solvers. It's around 0.2-0.4 nH with optimal layout. For resistors the number should be similar.

http://www.sigcon.com/Pubs/news/6_09.htm

But practically, 100 MHz is really forgiving. For RF, you can easily get 20 dB return loss without doing any layout optimization. At a higher frequency like 1 GHz, if you don't have access to an E&M solver, the standard RF/microwave design procedure is basically trial-and-error: do a return loss or TDR measurement, tune the layout at the discontinuity, try again until satisfactory results are obtained.

The actual performance is greatly affected by the PCB stackup (mainly the distance between a trace and its ground plane), or the size of the mounting pad. In a 4-layer digital board, the pads are huge compared to a 50-ohm trace, thus there exists two strong capacitive discontinuities around the resistor, this is often performance bottleneck, not the ESL of the resistor. A ground plane cut-out under the pads is often used as compensation. In a classic two-layer PCB for microwave, the traces are much thicker, and the situation is a bit different. Also, sometimes there can be a measurable inductance reduction if the resistor is mounted up-side down because the resistive ink is closer to the ground plane.

[eugene]

Dumb person here. Why would the ESL of a chip resistor be any different from the inductance of an ideal conductor with identical geometry?

Point being that the geometry is either very simple (a thick film blob) or too complex to predict (thin film serpentine.)