[sdiy] Orange Drop Capacitors?

[Jerry Gray-Eskue]

I have been watching this thread and have a few comments.

First every type of component construction has imperfections that deviate
from the Ideal Device.
Higher quality devices tend to be a closer match to the Ideal.
Some of the imperfections described below may not be present or are of such
a low order magnitude that they are lost in the noise floor.

In the case of capacitors accurate modeling or measurement of the device
characteristics is complex.
One of the most important deviations is the leakage current in the
capacitor.
The actual leakage value is somewhat unpredictable but adds a roughly
parallel resistance across the capacitor.
Secondly the Reactance deviates from ideal and is frequency dependant but
the error may not accurately track the frequency.

Consider a spiral wound capacitor with leads at opposite ends of the plates.
Each plate is a resistance strip with distributed capacitance to the other
plate with uneven areas of leakage plate to plate.
Due to the Spiral wrap and leakage there also is some level of Inductance in
the capacitor.
It is also likely especially in polarized capacitors that leakage is
asymmetrical in turn making the Reactance and Inductance asymmetrical.
Additionally frequency response is not always even across the plates due to
the localized capacitance and distributed plate resistance, uneven leakage
areas and inductive characteristics.

Each manufacturing style will tend to have its own sort of unique
imperfections.
If you look close enough every device is unique, but at this point you are
well beyond any practical usage of the characteristic.
When the deviations are significant they can color audio, but it has been
said "a difference that doesn't make a difference isn't a difference"

- Jerry

-----Original Message-----
From: synth-diy-bounces at dropmix.xs4all.nl
[mailto:synth-diy-bounces at dropmix.xs4all.nl]On Behalf Of cheater cheater
Sent: Monday, January 04, 2010 10:03 PM
To: Magnus Danielson
Cc: SDIY
Subject: Re: [sdiy] Orange Drop Capacitors?


The only reason for evaluating electronics items by their physical
(electrical, electronic) properties is to make sure that they are the
same as something else we are using as reference. This can be used in
two ways:
1) we are using some abstract model, i.e. we are building an L-R
filter. We need to know that our parts are the same as the L-R model
depends on, i.e. resistors with no inductance, capacitors with no
resistance, etc.
2) we are not using a model that has theoretical grounds and are just
swapping things. Then if we make sure the items we are switching (e.g.
mustard caps vs foil caps) measure the same, they should act the same.
There is no beating physics.
Of course, this last statement is subject to several situations that
don't let it be a logical razor:
1) very often people forget about some properties or measurements. The
imperfection of human thought is the most common reason for people
wondering why on earth two things sound different where they should
sound the same... they 'should' not, you just forgot to measure
something else you should know about.
2) imperfections in test case replication: items work differently
in-circuit than when being measured; imperfections in testing: test
accuracy, certainty, and repeatability
3) imperfections in the application of the theoretical model: the use
case has exited the domain of correctness of the model and therefore
any laws applicable to that model might not be true anymore. For
example, physics laws change at speeds near light speed.
4) incompleteness of a model: we don't know yet why two things work
differently. Who would have thought that a germanium crystal would
work differently in a certain circuit? If you hear two things sound
differently, even though all properties are the same, you are missing
a variable. Use your ears.

All said, should-be-world capacitors are fully described by their
capacitance which is a constant number. Real-world capacitors are not
described by their capacitance; and capacitance is not just a constant
number.

D.

On Mon, Jan 4, 2010 at 16:11, Magnus Danielson
<magnus at rubidium.dyndns.org> wrote:
> David G. Dixon wrote:
>>>
>>> Exactly.  That's why a previous previous poster urged a metrics based
>>> approach to comparative testing.  I think it's the only way to eliminate
>>> the human subjectivity (provided the metric testing is complete and
>>> repeatable).
>>
>> But isn't this all about "human subjectivity" in any case?  Even if you
>> could measure quantitatively which capacitor sounds better, the
>> interpretation of the measurements would be based on purely subjective
>> criteria.
>>
>> There was a big article on metrics in one of our national newspapers the
>> other day (the National Post), part of a series on important developments
>> of
>> the last decade.  Apparently, there is a growing concern among experts
>> that
>> taking a quantitative approach to everything is destroying our ability to
>> apply our instincts to important problems, and thereby paralyzing our
>> ability to take action as a society.
>>
>> I guess the moral is: "Use your ears!"
>
> This is indeed a valid concern. Just because you can measure something to
> great detail, doesn't mean that you get high quality information out of
it,
> because it assumes that the information you get is relevant to the
> conclusion you are going to make. A problem is that is the ability to
> interprent the measurement data into the problem you want. The myriad of
> measurement approaches and summarized data is a good proof of this.
>
> The old IMD/TIM war is another.
>
> The benefit of even harmonics over odd harmonics distorsion is yeat an
issue
> which prohibits simple and straightforward analysis of goodness.
> Toss in the pre and post filtering on a "good" distorsion mode. Toss in
the
> feedback path path. Etc. etc. etc.
>
> Measurements is good hints, engineering tools to guide us somewhat in the
> right direction, but we need to learn what measurement to trust when and
the
> feedback goes through out ears and we gain experience in what to look for
> when.
>
> We could equalize the frequency responce into +/- 1 dB, +/- 0,1 dB or
> whatever limits we want... so OK, that will not be the main issue... but
> then the main isssue may be some distorsion, noise, transient response...
> but on the other hand, some distorsion of parameters the ear actually
> likes...
>
> Cheers,
> Magnus
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at dropmix.xs4all.nl
> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy
>

_______________________________________________
Synth-diy mailing list
Synth-diy at dropmix.xs4all.nl
http://dropmix.xs4all.nl/mailman/listinfo/synth-diy