Question about ceramic caps readings

[antonis]

Are you worryring for that 63kHz LPF, Phend..??

[Phend]

He'll yes.

[eh la bas ma]

Go here,
https://savvycalculator.com/parallel-wire-capacitance-calculator/

Spacing 15mm  (0.015m)
Diameter 1mm (0.001m)
Wire length 1m
Relative permittivity:  1.1             (guess: air 1, plastic sheath 4)

Calculated capacitance of wires = 23pF

So even a couple of centimetres of close parallel tracks could totally overwhlem a 5pF capacitance, especially if the tracks are narrow.

I'm not sure if i really understood. Capacitance is the designation for "storing" some current in something. In french we use the word "condensateur", like the current is condensed inside something. In a powered circuit, the capacitors aren't the only thing condensing some juice ? There is a small amount of current stored in the pcb traces and in the wires ? Also in the DMM wires ?

I don't get why the capacitance would increase if the wires or the traces are close to each other...

[antonis]

Capacitance is the designation for "storing" some current in something.

Storing "charge"..
Motionless charge isn't "current"..

[Rob Strand]

I'm not sure if i really understood. Capacitance is the designation for "storing" some current in something. In a powered circuit, the capacitors aren't the only thing storing some juice ? There is a small amount of current stored in the pcb traces and in the wires ? Also in the DMM wires ?

I don't get why the capacitance would increase if the wires or the traces are close to each other...

Capacitors store charge.  However (ideally) the capacitance itself doesn't depend on the amount of charge it only depends on the physical dimensions (and shape) of the capacitor.

You can see the capacitance of two plates increases with area an decreases with separation,

https://www.insula.com.au/physics/1279/L4.html


The capacitance of two parallel wires is a little more complicated but basically you get a result that larger diam wires increases capacitance (area) and capacitance decreases with separation.

[Mark Hammer]

I find it perplexing that we can stick several billion transistors in less than a square centimeter of plastic, for memory, but are unable to produce capacitors of similar size with a true value that corresponds to the value printed on the side,  What's the holdup?  Do the two sorts of devices impose different sorts of production requirements?

Capacitors require a significant area to store charge. And this is — the dielectric area.

If we were talking about hundreds of nanofarads, I'd agree with you, but there is all manner of devices that manage to provide many capacitors.  For example, your average 8-pin bucket brigade device has at least a thousand.  Yes, they are small value, but they also need to be fairly precise.  So how come THAT is feasible but a ceramic cap that says 22pf won't necessarily read 22pf?

[R.G.]

If we were talking about hundreds of nanofarads, I'd agree with you, but there is all manner of devices that manage to provide many capacitors.  For example, your average 8-pin bucket brigade device has at least a thousand.  Yes, they are small value, but they also need to be fairly precise.  So how come THAT is feasible but a ceramic cap that says 22pf won't necessarily read 22pf?

It's because the processes and materials are different, as are the economics. Semiconductor processing has improved exponentially (literally!) over the last 60 years, fueled by the insatiable hunger for better and faster chips. The ultra-high-end chips getting hundreds of dollars per chip require ever more accurate chip fabs, to the point that a competitive billions-of-transistors fab costs one to ten billion dollars to bring on line. The old fabs are sold to down-market chip makers for what, in the semiconductor biz are known as "dog-food" chips. Even with this mighty collection of technology, it's difficult to make really accurate devices. What IC processes do very well is getting the ratios from one device to another accurate, so ICs are designed so ratios, not absolute values, is what matters. Yeah, the absolutes matter somewhat, but on the same chip, you don't have to hit any one capacitor (for instance) value on the nose. +/- 40% parts don't matter as much when they are matched within 1% relative to each other in the thousands.
Caps are not made by the ultra-advanced semiconductor fabs. They do improve somewhat by the hand-me-downs of semiconductor processes in thin films, sputtering, metalizations, etc. making advances, but the basics of making individual components where each one has to stand on its own as an absolute value progresses much more slowly than the investment-plumped semiconductor processing biz. Sure it's >>possible<< to make 1% caps first time, every time, but companies that did this would go broke because what sells in the market is 5% parts.

[PRR]

I find it perplexing that we can stick several billion transistors in less than a square centimeter of plastic, for memory, but are unable to produce capacitors of similar size with a true value that corresponds to the value printed on the side,  What's the holdup?  Do the two sorts of devices impose different sorts of production requirements?

The billion-transistor CPU is all ON or OFF, no in-between, and no exact values.

Even BBD does not need exact bucket capacitors because buffering (matches voltage level, not charge quantity).

Ultra small caps CAN be measured and thus sold. But to agree on value you have to quantify the entire state of the world around it. As seen here, leads and hands add uncertain capacitance. So does the metal in the wall, to infinity (with infinitely less effect).

Is a shame you boys are too young to remember Radio. From 1,600kHz to 1,610kHz (adjacent channels at the top of the AM band) is 0.33pFd (on the once-standard 365pFd variable capacitor). Shipboard radio operators routinely made such adjustments (broadcast stations were normally spaced a bit further apart, but a home radio could tune <10KHz if it had to). BTW, adding a tiny-cap to a tuned L-C circuit is a fine way to measure a cap against a frequency reference (typically the sharpest ruler around) despite the square-root in the term.

EVBlog: https://www.eevblog.com/forum/metrology/measuring-small-capacities-around-1-pf/
Nobody here needs the 0.1pFd resolution those geeks seek. Hardly anybody here needs to know 50pFd from 100pFd. With some multimeters(*) you can do this by standardizing the EXACT lead layout and position relative to all conductors (don't move that lamp!) and comparing against known-good 47 and 100 pFd caps. If you find they read 107pFd and 160pFd, say ~~60pFd stray, you can know your 50 from your 100. The only audio use-case I know for smaller caps is compensating 709/301/308 opamps, and then you just buy 33pFd and assume all is well.

(*) "Multimeter" already implies a probability of compromise because no meter can be good for all things. And the more you get into the weird ends of electronics, the more the meter costs, because so few others need such a thing.

[Dormammu]

So how come THAT is feasible but a ceramic cap that says 22pf won't necessarily read 22pf?

To accurately measure small capacities, you need to have a more accurate meter, which will cost much more.
Fortunately, the needs of box builders do not require either precise small caps or super-precise meters.

[Phend]

So in the world of guitar effects, for the average hacker, stray capacitance is not an issue if your circuits wires / traces are 0.1 inches apart. But if you design a board with traces 0.04 inches apart you need to, maybe, consider stray capacitance.

[ElectricDruid]

So in the world of guitar effects, for the average hacker, stray capacitance is not an issue if your circuits wires / traces are 0.1 inches apart. But if you design a board with traces 0.04 inches apart you need to, maybe, consider stray capacitance.

I'm not sure I agree. I think stray capacitance comes into play when/if you start to use these stupidly small values. That 5pF in the Rebote 2.5 Delay is crazy, for example^*. There could easily be at least that much stray capacitance floating about somewhere. If you *reckon* on there being 10pF of stray capacitance on anything you put in, and make sure to design so that that's not a problem, you won't get any nasty surprises. At the frequencies we're working at, that's not at all hard to do. 100pF is already a "very small" cap.

^*The Rebote 2.5 uses an inverting input buffer to avoid the signal being inverted overall, since it also uses an inverting mixer to mix the wet and dry signals. In order to avoid a low input impedance, the inverting buffer uses 510K resistors, and the high resistor value necessitates the small cap value. However, a 18pF or 22pF could be used without changing anything else. Similarly, there's a 51pF across a 24K on the output mixer. That gives rolloff above 120KHz, which is far too high and should definitely be larger. 220pF would be a better value. We don't *want* ultrasonic digital noise getting through the output mixer, so block it where possible!
Alternative solutions would be to stop worrying about the overall signal phase (the delays will be out of phase half the time anyway) and use a non-inverting buffer, or to set the pre-delay filter for full audio bandwidth and then take the dry signal from after the inverting MFB filter.

[Phend]

Looking at TL072 circuit images, I do see 5pF, 18pF and others being used.
pF is small stuff, like 10M resistor is big stuff. As you know I haven't been knowledge in these things. But I do know E-12 and E+7 are way out there.
6.022E23 comes to mind, it is used all the time, so think big think small but don't think of nothing at all.

[Rob Strand]

I'm not sure I agree. I think stray capacitance comes into play when/if you start to use these stupidly small values. That 5pF in the Rebote 2.5 Delay is crazy, for example^*. There could easily be at least that much stray capacitance floating about somewhere. If you *reckon* on there being 10pF of stray capacitance on anything you put in, and make sure to design so that that's not a problem, you won't get any nasty surprises. At the frequencies we're working at, that's not at all hard to do. 100pF is already a "very small" cap

+1

Also, the chips themselves have a capacitance looking into the pins which needs to be added on just like a stray capacitance.   The simplest example I can think of is an NE555 which looks like it has 30pF across the pins.   Best seen directly with a monostable config.  The astable config has got more intricacies due to switching times.   In some cases you can guess the extract the stray capacitance from the datasheet, or plots (freq. vs cap value) in the datasheet.

Wiring to switches is another one.

[Rob Strand]

Even resistors themselves have stray capacitance,

https://www.edn.com/resistors-arent-resistors/

"parallel capacitance is 0.2-0.4pF"

A real circuit is a mass of stray capacitances between each node and from each node to ground.   Under  normal circumstances and at audio frequencies most, if not all, can be ignored.   They rear there heads in high-gain circuits.   For clock circuit's it's not audio so the higher frequencies and naturally smaller cap sizes mean the strays can have an effect.

[Mark Hammer]

Anyone else here remember the "gimmick" cap Craig Anderton used for his Spluffer project?  It consisted of two pieces of #20-22 solid core wire twisted together to form a roughly 5pf capacitance between the non-inverting input and ground.  Makes you think about what stray capacitance one might be unintentionally creating.

[Rob Strand]

Anyone else here remember the "gimmick" cap Craig Anderton used for his Spluffer project?  It consisted of two pieces of #20-22 solid core wire twisted together to form a roughly 5pf capacitance between the non-inverting input and ground.  Makes you think about what stray capacitance one might be unintentionally creating

I make those now and then when I need a small value.   You can also tweak the value by tighter twists or by cutting off the end.   I think the term "gimmick" capacitor came from the RF people.  Somewhere in the forum archives I posted some values (maybe 2017/2018).

[PRR]

So in the world of guitar effects, for the average hacker, stray capacitance is not an issue if your circuits wires / traces are ...

To go beyond what you were thinking: in HIGH gain guitar amps/effects, fractional pFd from output to input CAN make "squeal" trouble. 0.1pFd around a gain of 10,000 (typical guitar tube amp from first grid to last plate is 5,000, before soup-up) makes 1,000pFd effective. The loop gain may (Murphy says WILL) exceed unity, and while Nyquist sets a criteria on phase Murphy trumps Nyquist. It screams uncontrollably.

[R.G.]

So in the world of guitar effects, for the average hacker, stray capacitance is not an issue if your circuits wires / traces are 0.1 inches apart. But if you design a board with traces 0.04 inches apart you need to, maybe, consider stray capacitance.

You're an order of magnitude off. 0.04 is 40 mils/ (thousandth of an inch) I commonly use 10/10 mil trace/space rules for sloppy, easy, trivial layouts. 10/10 means 0.010 traces spaced 0.010 inches apart. I go down to 6/6 if I have to. Somewhere between 10 mils and 6 mils I start worrying about inter-trace capacitance. 
"Worrying about inter-trace capacitance" means identifying high impedance traces (JFET/MOSFET gate, or inputs to JFET/MOSFET opamps and/or 1M biased noninverting inputs) and identifying what signals are next to those traces. After identifying the high impedance traces, I identify the signals next to them and whether those are carrying significant signal voltages that they can transmit by capacitance to the high impedance traces. I then increase spacing or run a ground-connected "shield" trace between them.

[Phend]

Sticking in 0.001 (1mm between tracks), 0.0005 (0.5mm "diameter" track), and 0.02 (2cm track length) comes back with 1.6e-12, so 1.6pF. A track is nowhere near as much conductor as a 0.5mm diameter wire though, so then I reduced that to 0.0001 (0.1mm diameter wire

1mm = 0.03937 (0.04) inch
Spacing on a BB is 0.1 inch (between holes anyway)
2.5 times greater

I haven't designed boards so what the common spacings are am not sure, until now, thanks R.G.

[Phend]

But I have designed posts.
It's an "open air circuit " design.
Two knob Tone Bender, sounds great.