Wow, cap type makes a difference

[petemoore]

  Series like 10 of 'em then get the gizmo [spectrum analyzer or scope ?] I dunno...some kind of sensor that could maybe measure 10x whatever 1 cap's difference might be.
  ie 10 caps in a row should cause...
  about the same thing as a cap 10x the value...
  So...not knowin' what to look for [distortion I suppose], see what happens at high or low frequencies when many caps are strung together.
  Maybe by testing values which are made low by multiplication or high by division [paralleling many a cap] the differences could be better known.

[dudleydoright]

  I conmansulted with my pal Fletch Munson over at the cya and his take was. WHAT ARE YOUR EARS TELLING YOU? nuff said. yeah I guess.

[Hides-His-Eyes]

  I conmansulted with my pal Fletch Munson over at the cya and his take was. WHAT ARE YOUR EARS TELLING YOU? nuff said. yeah I guess.

I have a much better system: what is my wallet telling me

[DougH]

The invalid straw man conclusion that keeps coming up in this thread is that anyone who happens to notice an audible difference between cap types is somehow trying to rationalize using expensive parts.

The original poster's preference was a 13 cent cap. You can look it up yourselves from his Mouser link.

Hardly a "wallet" issue methinks...

[gmoon]

  I conmansulted with my pal Fletch Munson over at the cya and his take was. WHAT ARE YOUR EARS TELLING YOU? nuff said. yeah I guess.

If i recall correctly, Fletch said "turn it up." 

[zombiwoof]

The invalid straw man conclusion that keeps coming up in this thread is that anyone who happens to notice an audible difference between cap types is somehow trying to rationalize using expensive parts.

The original poster's preference was a 13 cent cap. You can look it up yourselves from his Mouser link.

Hardly a "wallet" issue methinks...

Man, I'm with you!.  Lots of "humbuggers" here.  I go by what I hear, if it sounds good, I use it.  If not, I try something else.

Al

[amptramp]

Dredged this old Steve Bench item up from the wayback machine:

http://greygum.net/sbench/sbench102/caps.html

It shows input on x-axis and output on y-axis and they should be a straight line on a diagonal.  Check the picture of the ceramic caps.  They are piezoelectric, so part of the voltage goes into bending the material and part goes out the other side.  Since they are piezoelectric, they are affected by sound and if the effect is in front of a 100db amplifier, they can give you an unpredictable feedback.

The link was taken from here - lots of good reading on this page:

http://audiophool.cjb.net/Techno.html

[petemoore]

they can give you an unpredictable feedback.
  Since you didn't say "Will give you" I can assume there's hope for achieving [gazuntheit] predictable feedback.
 

[amptramp]

they can give you an unpredictable feedback.
  Since you didn't say "Will give you" I can assume there's hope for achieving [gazuntheit] predictable feedback.
 

Good luck with that.  It is difficult enough to characterize the electrical nonlinearity, let alone the unintended function as a microphone.  And since it is an uncontrolled characteristic, there is no guarantee that the next lot of capacitors will behave the same way.  A little more or less external insulation could make a huge difference.

I keep preaching to the heathen here: a good design should never depend on device parameters being within a narrower spread than they normally occur, nor should it depend on undocumented behaviour.

[DougH]

a good design should never depend on device parameters being within a narrower spread than they normally occur, nor should it depend on undocumented behaviour.

I agree with that. Esp in 9v circuits, I have not heard a single component type change that made a radical difference in sound. Which is why I take issue with boutique-speak that claims something has to have this or has to have that. Baloney. If it's a good design it will sound good regardless. Still doesn't change the fact that in amps I hear differences between cer and mica caps.

[gmoon]

I keep preaching to the heathen here: a good design should never depend on device parameters being within a narrower spread than they normally occur, nor should it depend on undocumented behaviour.

Agree in general. But as with any "rule," I can think of exceptions--The valvecaster (out of spec for filament and way out of spec for plate voltage), and C. Andertons Tube Sound Fuzz (CMOS logic gates in an analog context).

Depends on your definition of "undocumented." Undocumented by the manuf could still be consistent behavior.

[amptramp]

I keep preaching to the heathen here: a good design should never depend on device parameters being within a narrower spread than they normally occur, nor should it depend on undocumented behaviour.

Agree in general. But as with any "rule," I can think of exceptions--The valvecaster (out of spec for filament and way out of spec for plate voltage), and C. Andertons Tube Sound Fuzz (CMOS logic gates in an analog context).

Depends on your definition of "undocumented." Undocumented by the manuf could still be consistent behavior.

This may be true for some parameters and common tubes like 12AU7 and 12AX7 do have published plate characteristic curves that go down to 0 volts on the plate and most manufacturers used the same dimensions for plate, cathode, grid spacing etc which made tubes from one manufacturer similar to others.  Even if the heater temperature is out of spec, there is plenty of historical data to show that the effect would be similar from tube to tube.  CMOS gates have a long history of being used in analog circuits back to the 1970's because they had what the common, cheap op-amps of the time did not have - high input impedance.  I used them in an intersting experiment to see if CD4XXX series CMOS amplifiers would work when they were submerged in liquid nitrogen.  (They did, but there were occasional fluctuations f, possibly from cracking of the silicon.)  The original poster's example of microphonic effects in ceramic capacitors is an example of something which is not controlled in manufacture.

[Ronsonic]

I keep preaching to the heathen here: a good design should never depend on device parameters being within a narrower spread than they normally occur, nor should it depend on undocumented behaviour.

Indeed. Sanity demands repeatability. The stuff we work on is goofy enough already, depending on weird phenomena that may not be present in the next batch of parts is asking for trouble. I've had enough of that.

[R.G.]

Agree in general. But as with any "rule," I can think of exceptions--The valvecaster (out of spec for filament and way out of spec for plate voltage), and C. Andertons Tube Sound Fuzz (CMOS logic gates in an analog context).

Depends on the rule. Mother Nature's Rules are pretty immutable. Our understanding of them changes, but not the Rules - they are Rules not "rules". Rules wth exceptions are not rules - they are simply not fully described or understood.

Depends on your definition of "undocumented." Undocumented by the manuf could still be consistent behavior.

Yep, absolutely.

And some engineers have lost their jobs depending on undocumented things they found in parts when the manufacturer felt free to change anything they didn't promise in the datasheets. Pesky manufacturers. Don't they know that the undocumented things have to remain constant? 

[gmoon]

Agree in general. But as with any "rule," I can think of exceptions--The valvecaster (out of spec for filament and way out of spec for plate voltage), and C. Andertons Tube Sound Fuzz (CMOS logic gates in an analog context).

Depends on the rule. Mother Nature's Rules are pretty immutable. Our understanding of them changes, but not the Rules - they are Rules not "rules". Rules wth exceptions are not rules - they are simply not fully described or understood.

All those colloquial sayings ("rules are made to be broken", "gotta know the rules to break the rules", etc.) are handy. Use 'em all the time.

But you can almost trace the confusion people have with physical laws (thermodynamics, conservation of energy, etc.), and all the hokum out there directly to these, um, cliches. Stuff like HHO and perpetual motion. 

Many a software engineer has been burned by dependence on undocumented "features," too.