Audio Grade Components

[Rob Strand]

I appreciate the information you're providing (including the stuff I didn't quote), but that page's oscilloscope traces are the result of running a 70 volt audio signal (at 600 Hz) through a 100 nf capacitor (26 mA signal current, he says).

The higher voltage just lets you see the non-linearity.   It's there regardless of the level but it's a matter of how much and that does depend on the voltage *across the cap*  not the signal level voltage.

The data on the web is a bit blurred now because you have through-hole ceramics and smd ceramics.

In the old through-hole days you had High-K, Low-K and NPOs. 

https://linearaudio.nl/cyril-batemans-capacitor-sound-articles

SMD caps are clearly crap in some cases.   They can be 1/10 the capacitance at the *rate voltage*. I wrote some stuff about it in this old thread, which also has some links:

https://www.diystompboxes.com/smfforum/index.php?topic=111219.0

This doc has some measurements show X7R vs COG.    COG is generally problem free.  X7R is the edge of good.  There are ones worse than that.
https://training.ti.com/system/files/docs/1324%20-%20Low%20Distortion%20Design%204%20-%20slides.pdf

[ElectricDruid]

What I find funny is modern electrolytics and olden day electolytics often have the same specs on paper but the failure rates of modern caps are *much* higher in reality.

Got any evidence for that?

It seems to me that failure rates would have a lot to do with usage and how people spec parts for a job. Perhaps in the past, people were much more cautious and gave much more margin. Plus they probably weren't told to design for obsolescence.

I'd be pretty confident that with some decent quality parts you could design a modern device that would last just as long as some of the vintage gear. But I haven't got any evidence for that either!

T.

[lcv]

Hi
I read this many years ago (EW/WW 1995) about an Hi-Fi preamp , at that time the dabate was raging on that magazine...





Since then, for my builds,  I don't care anymore .
I'm not marketing them though...

Regards
Lcv

[Rob Strand]

Got any evidence for that?

It seems to me that failure rates would have a lot to do with usage and how people spec parts for a job. Perhaps in the past, people were much more cautious and gave much more margin. Plus they probably weren't told to design for obsolescence.

Only from general observation on how old equipment was before the caps went bad compared to now.  Like these days it's not uncommon to see caps fail within 2 years.   I don't think I every saw a cap replaced within 7 years in the old days.

I have a few examples but they are samples of 1 which isn't statistically relevant.

In defense of modern equipment the usage side of things isn't exactly an apples to apple comparison.   These days parts are stuck in small boxes (so the heat doesn't get out) and the parts are packed close together (so everything heats up).  Kind of like a life of hell for electrolytic caps.

I'd be pretty confident that with some decent quality parts you could design a modern device that would last just as long as some of the vintage gear. But I haven't got any evidence for that either!

I'm sure you can get good parts and some are better than others.

I used to buy the generic computer power supplies.  The caps would fail within 2 to 5 years, more often on the low side.  I've gone through *heaps* and so has my father.  The last time I bought a PSU I did research on the operating hours of the caps and I bought a PSU with long life caps.  Not the insane price PSUs but probably 2 times the price of the generic PSUs.  It's probably 10 years old and it's still going.  In fact the fan started to seize and I had to get the solid grease out of the bearing and re-lubricate.  My father's failed in about 3 years. 

I guess more what I'm saying is generic parts used to last a lot longer.    If you look at the specs they are often quoted at something like 4000hrs.   Both the crap brands and the good brands quote this.  Some extra crap ones are like 2000hrs.  (The temperature rating is also a factor)  However the crap brands always appear on the lists of evil caps brands which are put up on the web.  When you look at people replacing caps on the web it's always those evil brands.  When I fix stuff I often see those evil brand caps with bulging heads in the equipment.

Edit: I forgot to mention older caps were a lot bigger so if you stress them out with ripple current etc they don't heat up as much as they have more surface area.

[amptramp]

Some manufacturers like to add gold plating to components that don't need it.  For example, we used gold-plated leads on the 2N2222A's we used in the Hermes spacecraft and that was a problem:

1. Gold forms intermetallic compounds with solder that became known as the "purple plague" in the industry because of what the damage looked like under a microscope.  These compounds are brittle and usually are not good electrical conductors.

2. The gold plating is thin (we used 150 microinches but most audiophile gold plating is 50 microinches) and porous, allowing moisture to get to the kovar leads on the transistors.  The kovar then acted as a sacrificial anode and rusted out under the gold.  Kovar is a nickel steel that has the same temperature coefficient of expansion as glass so the leads will fit the glass seals on the metal transistor cases and not crack with changes in temperature.

There is one place for premium components: I see some electrolytic capacitors have leads that are magnetic because you can pick them up with a magnetized screwdriver.  This means you have steel wire leads plated to be solderable but the steel adds inductance to the leads.  Maybe not as critical as the inductance added by the rotary winding of anode and cathode foils but I think electrolytic capacitors are the only place where premium components may be noticeable.  A lot of audiophiles recommended Black Gate capacitors made by Rubycon but since the last Black Gate was made in 2006, no one recommends them anymore.  Their leads were not magnetic.  There are other premium brands now and other technologies like the OS-CON capacitors that may supplant electrolytics.

[teemuk]

What's ironic in "audio grade" is that "audio" isn't a really demanding application at all.

No. No need for components to work at extremely high or extremely low frequencies. Hearing range extends to about 20 kHz, Nyquist maybe expects components to work at 40 kHz at best. No need to operate at range of > 100kHz or (gulb) 1MHz and up. No, no need to operate at DC either.

Distortion? Yep, we expect modest performance but amplifiers aren't precision equipment - like ones USED for measuring distortion of those audio circuits. Tolerable levels of distortion are easy to achieve, and those are much, much lower than our actual auditory sensing of distortion. Oscilloscopes and spectrum analyzers are still WAY better equipment to sense distortion than our ears.

No. No need to work at extreme temperatures or temperature shifts. No need to operate in vacuum or in various gases. No need to operate in an environmment infested with things like radiation, corrosive elements or even modest humidity, extensive pressure, heavy vibration, etc.

Audio is so damn easy application that, IMO, saying something is "audio grade" is basically saying something is actually very mediocre but sufficient enough to work in an application that isn't all too demanding. I don't think that's the real intention though so it's pretty much just "snake oil" to me.

[merlinb]

Yep, we expect modest performance but amplifiers aren't precision equipment - like ones USED for measuring distortion of those audio circuits.

+1.  They don't use 'audiophile' components to make distortion analysers. That tells you everything you need to know!

[Rob Strand]

They don't use 'audiophile' components to make distortion analysers. That tells you everything you need to know!

But you won't see crappy ceramics in there either.

I remember speaking to Neville Thiele (the guy that did the Thiele-Small parameters for loudspeakers). He had a lot to do with the audio and TV industry.    He used say how dare those people in the recording studios use non-gold plated connectors when the Audiophiles have spent all that money - obviously he was having a stab a them.

[reddesert]

I appreciate the information you're providing (including the stuff I didn't quote), but that page's oscilloscope traces are the result of running a 70 volt audio signal (at 600 Hz) through a 100 nf capacitor (26 mA signal current, he says).

The higher voltage just lets you see the non-linearity.   It's there regardless of the level but it's a matter of how much and that does depend on the voltage *across the cap*  not the signal level voltage.

But, suppose you biased the cap by putting say 40 volts DC on it and then ran a +/- 2 volt audio signal through it. You're only going to sample a tiny portion of one of those D-E curves, and in that portion the behavior can be approximated as close to linear.  The slope is different than it is at zero volts, meaning that the value of the capacitor is somewhat different than it is at zero volts, but the deviations from linearity are small.

I don't think the bad behaviors of ceramics have been demonstrated to be an audible problem in low-voltage audio circuits like guitar effects. Tube amps would be another story.

[amptramp]

.

[Rob Strand]

The slope is different than it is at zero volts, meaning that the value of the capacitor is somewhat different than it is at zero volts, but the deviations from linearity are small.

Agreed.

I don't think the bad behaviors of ceramics have been demonstrated to be an audible problem in low-voltage audio circuits like guitar effects.

That's a fair statement.

In general I've seen some very poor looking waveforms on circuits using ceramics.  I discovered this at a young age.   As most of the articles show it is clearly measurable.  Since I have some HiFi background I avoid using ceramics for time varying signal (except RF) - you don't want to add any unnecessary distortion.  In recent times the SMD caps as so bad you have to be careful in other areas as well.

Just where the audibility point is I don't know.   I suspect at 10mV it probably won't matter and at 1V you might think about it.   It would take many hours of testing and that would only apply to a single configuration.    Look here,
http://sound.whsites.net/articles/capacitors.htm#s13
At the end of the paragraph the author states the cap can filter it's own harmonics.  I'm not sure about that.  But if you tested the equivalent high-pass filter version you might get different results.

From a practical perspective, the price difference between films and ceramics was never enough for me to work it out.  For SMD boards there maybe a motivation.   I still see a lot of through-hole film caps on SMD boards for audio.

[ElectricDruid]

What's your opinion of C0G/NP0 caps, Rob?

There was a thread about this recently on the Synth-DIY mailing list. They're pretty well thought of in the synth world. I quote:

"Low capacitance change with voltage and with temperature, low microphony, low loss at high frequency. C0G is the best you can get in small C values, in every respect except maybe size and cost. So no reason not to use it here for DIY unless you're counting every last penny or need to fit hundreds of them into a small space. But just for small cap sizes like hundreds of pF."

Recently I've seen C0G in leaded caps - previously it seemed it was only SMD parts. It opens up a another option, at least.

Tom

[StephenGiles]

What's your opinion of C0G/NP0 caps, Rob?

There was a thread about this recently on the Synth-DIY mailing list. They're pretty well thought of in the synth world. I quote:

"Low capacitance change with voltage and with temperature, low microphony, low loss at high frequency. C0G is the best you can get in small C values, in every respect except maybe size and cost. So no reason not to use it here for DIY unless you're counting every last penny or need to fit hundreds of them into a small space. But just for small cap sizes like hundreds of pF."

Recently I've seen C0G in leaded caps - previously it seemed it was only SMD parts. It opens up a another option, at least.

Tom

You may be able to play faster using them!!!

[Rob Strand]

What's your opinion of C0G/NP0 caps, Rob?

They are a good choice for low distortion stuff.   I've never seen any issues.  Some of the links in this thread have some measurements.

There's:
- Old school through-hole disc capacitors. 
- Radial multilayer ceramics which are the through hole.  Internally they are more like SMD caps.  In the early days they were often blue (also yellow) but you do get packages that look a bit like film caps.
- SMDs

[PRR]

There are (as you know) MANY uses of capacitors.

There's "bulk caps", where as long as you get at least 5 pounds of sugar in the bag, you don't really care if it is 10 pounds.You just want "enough".

At the other extreme, an AM radio must hit a 1MHz carrier with better than 1KHz stability, 0.1%. Because there's trimmers, adjusted in production, the precise value is not important, but the tuning tank MUST hold a stable value.

So the cap needs low-low tempco. But the coil also has a tempco. We can't easily make the coil dead-stable. We can (thanks to magic ceramics) make a cap with a complementary corrective tempco. What we do is put the cap and coil together, measure in a heat chamber, and pick an N1500 or P100 cap to counter-act the coil's tempco. (Coils tend to be positive tempco so caps are made mostly negative tempco, though the range runs from P100 through NP0 (Neg Pos Zero) to N1500.)

These are sometimes listed as Class 1 caps.

Now, this has nothing to do with capacitance stability against voltage or other factors. Except it would be stupid to have a precision tempco and random variation from other factors. So a spec-tempco Class 1 cap is normally a very stable cap.

A 1000pFd Class 1 cap is fairly large. A 10,000pFd Class 1 would be really large. But we don't (usually!) use 10,000pFd in precision-tuned circuits. Often enough a 10,000pFd (0.01uFd) cap is a bypass, a "bulk" cap. Like I need 5 pound bag to keep my plans from blowing-away today. We don't care about the exact value, or how it changes, as long as it stays heavy enough that the circuit node does not wobble.

Again Magic Ceramics come in, but now for super-high capacitance in small (cheap) bags. Throw some saccharine in that sugar, get 50 pounds of sweetness in a 5 pound bag. OK, it isn't sugar and works different in coffee, cake, candy, and other baking. But in miniature electronics, X7R and other super-ceramics are the bee's knees for compact bulk capacitance.

Below 1,000pFd, NP0 is "excessively good" but usually small and cheap. It replaced Mica even in the Jensen 990.

Up to ~~0.5uFd, use any of the Film ("poly") caps. They are more than stable enough for audio.

Over 1uFd, electrolytics are the economical choice, but are sloppy. Design the system so the exact value does not matter.