43
\$\begingroup\$

What kind of uses do engineers find for 1 pF or lower-value capacitors?

This is the kind of value one gets with two bits of wire close to each other or two tracks.

\$\endgroup\$
  • 7
    \$\begingroup\$ And would you rather rely on a capacitor to provide that value or on the unknown dielectric properties in that regard of your PCB materials, or the precision of someone hand soldering two bits of wire close together? \$\endgroup\$ – PlasmaHH Dec 9 '16 at 14:31
  • 5
    \$\begingroup\$ Like... Any? RF (look for schematics around). Digital (crystal oscillators circuits). Really, it just looks small. \$\endgroup\$ – Eugene Sh. Dec 9 '16 at 14:37
  • 6
    \$\begingroup\$ High frequency (hundreds of MHz or above) frequently use small picofarad value caps in signal filtering and conditioning. Sometimes they ARE constructed out of copper geometry on the PCB itself rather than using discrete capacitors. \$\endgroup\$ – Wossname Dec 9 '16 at 14:37
  • 9
    \$\begingroup\$ @Wossname: Generally in high demanding/precision microwave circuitry though, where price is not an issue and you have precise control over pcb material, copper and gold plating thickness. But also there you mix real capacitors with distributed elements filters. \$\endgroup\$ – PlasmaHH Dec 9 '16 at 14:39
  • 4
    \$\begingroup\$ @PlasmaHH, sounds like an Answer to me :) \$\endgroup\$ – Wossname Dec 9 '16 at 14:48
76
\$\begingroup\$

The smallest capacitor I've used recently, in a filter in a 6 GHz receiver, was 0.5 pF. There were some 2 nH inductors there as well, and you could argue that those could be made with a few mm of track. However, both were smaller than the equivalent way of implementing them in copper.

Perhaps more importantly than the size, is that they were discrete components. When I wanted to change the capacitor from 0.4 pF to 0.5 pF, to retune the filter, I didn't need to respin the board; I just changed the bill of materials.

\$\endgroup\$
  • \$\begingroup\$ Also, would using track instead of a manufactured component result in a less consistent product? Say for example from lot to lot, the actual characteristics of the board production may change, or even if you decide to switch board manufacturers I imagine. \$\endgroup\$ – MDMoore313 Dec 11 '16 at 0:25
  • 10
    \$\begingroup\$ @BigHomie At these frequencies, you don't change board supplier without a re-spin. You might even include a test track on the panel for pre-population test. \$\endgroup\$ – Sean Houlihane Dec 11 '16 at 14:18
  • \$\begingroup\$ @Neil_UK Could you elaborate on the specific design reasons you needed such capacitance values, as well as what necessitated your change from 0.4pF to 0.5pF? I too am not sure what would prompt such small values, and the smallest values I have used are 22pF caps on a quartz crystal, simply because that is what the crystal's datasheet says to use. \$\endgroup\$ – Ehryk Dec 12 '16 at 4:33
  • \$\begingroup\$ Probably a low pass filter to satisfy Shannon’s theorem for the ADC? \$\endgroup\$ – Michael Dec 12 '16 at 6:42
  • 4
    \$\begingroup\$ I don't see that anyone has explicitly pointed out that you are ALWAYS using the PCB (and the rest of the environment) as a capacitor - it's not "this or that", it's "this or (this and that)". You can specify a 0.5pF cap, but the circuit will see >0.5pF due to all the parasitics, which is why @SeanHoulihane points out that changing board vendors may require as board spin. Not always, though - you might be able to make component value change(s) (per @Neil_UK) or just get lucky. \$\endgroup\$ – Fred Hamilton Dec 13 '16 at 18:04
19
\$\begingroup\$

I use a 0.8 pF capacitor in a photodiode transimpedance amplifier (TIA) across the feedback resistor to reduce op-amp noise gain and I've used select on test capacitors from 0.5 pF upwards to centralize a 400 MHz colpitts based VCO.

I've also used a 1 pF capacitor in a quadrature FM detector for driving the tank so that I get high Q and the necessary phase shift of 90 degrees.

\$\endgroup\$
10
\$\begingroup\$

You'll also find them in RFID reader antenna matching circuits.

Here a good impedance matching between the transmitter and the antenna is essential for good performance, and you'll usually do the fine tuning with capacitors.

A 1 pF mismatch can easily make a 20% output power and thus reading-distance difference.

You don't use 1 pF or smaller capacitors alone. They're usually used in parallel with a bigger capacitor. So if your circuit calls for a 19 pF capacitor somewhere you'll use 18 pF and 1 pF in parallel.

Why not use 10 pF and 9.1 pF in parallel you may ask: The reason is, that it's hard to find 1% tolerance capacitors below 10 pF. Small values come with an absolute tolerance of - let's say - +/- 0.3 pF.

You get a better overall precision if you use a precision 18 pF part in parallel with a not-so-good 1 pF cap.

\$\endgroup\$
8
\$\begingroup\$

I'll sometimes use small caps to help match capacitance in filters. Something like a State Variable filter in the 100kHz range, (not often 1 pF, but 2.2 or 3.3 is not uncommon.)

\$\endgroup\$
7
\$\begingroup\$

In addition to everyone else's answers, discrete capacitors tend to be less lossy than that of an embedded solution. In the case of a C0G or a proper microwave dielectric, the discrete capacitor can be an order of magnitude less lossy than a bog standard PCB material like FR4. Less loss means your filters have lower attenuation and have higher Q which helps in blocking unwanted frequencies or making more stable PLLs etc.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.