• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are daily reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Audible difference in high-end capacitors? - ABX samples

ctrl

Major Contributor
Forum Donor
Joined
Jan 24, 2020
Messages
1,598
Likes
6,006
Location
.de, DE, DEU
As announced long ago the last part on the subject of audibility of capacitor upgrade in crossovers. The comparison ranges from a very cheap, decades-old film capacitor to an expensive high-end capacitor.

As promised, this time there are short audio samples of each capacitor (in a simple crossover) recorded with a microphone via tweeter (test setup see below).
1697403041341.png

The full mini series:
Capacitor upgrade in crossover - Is it audible?
Capacitor upgrade - part two
Capacitor upgrade - part three
Replace resistor by low-inductance resistor - Is it audible?
Different Binding Posts - is it audible?
Audible difference in high-end capacitors? - ABX samples

Lately there have been YouTube videos and discussions here in the forum for the thousandth time about "capacitor sound" (i.e. the audible difference of film capacitors of the same capacitance in loudspeaker XO, but of different price ranges and manufacturers).

In forums, on websites or in videos you can read or hear something like "Even if the capacitors measure identically, very clear differences can be heard in different price ranges and designs - in short, clearly audible difference, you can't measure".

To end these arguments once and for all (don't worry, I don't believe that myself), here is the ultimate test:
The capacitors are not only measured, but for each capacitor there are also two recorded audio samples, once a few seconds of pink noise and a short sound snippet of the well-known test music track "Fast Car".

Now everyone can easily perform an ABX test (e.g. with foobar) or can hear the diff file of two caps (see section 6 for more details) and decide for themselves how much differences are audible**.

Index:
1) About the high-end capacitor
2) Test setup
3) Measurement consistency
4) Tweeter measurement with different caps
5) Audio recording samples - see post#2
6) Samples analyzed with DeltaWave - see post#2
7) Compare impedance measurements of the caps in REW - see post#2
8) Are there differences in the phase frequency response (i.e. in the "time domain ")? - see post#568


1) About the high-end capacitor
The following statements can be found about the high-end capacitor (3.3µF Alumen-Z-Cap from Jantzen) used:

The Alumen Z-Cap will offer a more “neutral” presentation in the overall sound in certain crossover designs, and by “neutral” we refer to the focus will be shifted from what in some systems can be a somewhat overemphasized brightness when using a Super MKP capacitor, being replaced with a more even balance between the depth of the midrange and the presence of the tweeter.
Source: http://www.jantzen-audio.com/alumen-z-cap/

on the Pawnshop recording and swapping between Sup-Z, Silver-Gold and Alumen-Z was quite a surprise. Suddenly the sheen was gone and the overall sound image appeared darker. "Darker" is mostly a good sign when you know frequency response is exactly the same and yet, the sound becomes darker without any smear of detail. You start listening to treble rather than tweeter or tweeter + capacitor or whatever adds its sonic signature. This sheen also sometimes adds to the perceived sense of detail and transparency. The Alumen-Z appear to present a more credible overtone structure to instruments and voices and most importantly, not diminishing the level of transparency.
Source: http://www.troelsgravesen.dk/Alumen_Z-Cap.htm

The Alumen Z-Cap has high resolution in a natural way, it is without glare that you usually get with some of those "high-end" capacitors. There is an adictive balance between warmth on one side and and openness on the other....The Jantzen Audio Alumen Z-Cap is also open and also has lots of contrast but does it with a warmish undertone. ... The combination of and balance between these characters is spot on and make the Jantzen Audio Alumen Z-Caps create music with a great sense of realism.
Source: http://www.humblehomemadehifi.com/Cap.html

So, according to these "expert statements" we can assume that the audible differences between the high-end capacitor and the other two capacitors will be dramatically.
Personally, I can hardly stand the tension anymore, how will the measurements and the recordings turn out? :eek: Just keep reading...

** The recordings were not made in a studio, so I can not exclude that the recordings are possibly distinguishable by background noise (street noise, children's cough, TV sound,...) or editing - if you notice something like that, please report right here in the thread, so that I can possibly exchange the files.


2) Test setup

The test setup is identical to the previous parts of the mini series (see above). To better protect the tweeter, this time an air coil was used to form a second order electrical high pass together with the capacitor.
1697406035487.png
To measure the next capacitor or to make the next sound recording, only one terminal has to be reconnected to the next capacitor.
This is to prevent the position of the measuring microphone and tweeter from changing in relation to each other.

Audio interface: MOTU UltraLite mk4
Measuring amplifier: Marantz PM7001
Tweeter: SB21RDC tweeter from SBAcoustics
Measuring mic: iSEMcon EMX-7150
Measuring software: Arta (dual channel measurements when possible)

Measurements itself:
10s of pink noise, 1s pause and then the measurement (for best possible consistency).
The SPL of the measurements were 72dB@1kHz@1m and 90dB@10kHz@1m, more details see the calibrated HD measurement:
1697414297825.png
This SPL should be above the usual listening volume. The tweeters with ring radiator have slightly higher HD2 (but this HD2 will be completely masked) and very low harmonic distortion of higher order. Only the tweeter with ring radiator showed very good consistency, so it was used.



The most important thing, of course, is that the capacitance of the capacitors is almost identical. I have about 30 pieces of the 3.3µF standard film capacitors (about 1$) in stock, so it was possible to find an almost identical capacitance to the 3.3µF high-end Alumen-Z-Cap from Jantzen - which costs about 45$ per piece (was purchased only for you dear reader ;)).
There were also dozens of very old 1µF film capacitors (about 0.05$ each), of which three pieces together (plus a 0.27µF correction capacitor for 0.5$) make the cheapest and oldest 3.3µF capacitor in the test.

Test cap capacitance and deviation
Jantzen Alumen-Z-Cap (45$): 3.289µF
Standard-Cap
(MKP 400V Arcotronics, 1$)***: 3.288µF (deviation <0.05% to Alumen-Z-Cap)
Old-Cap (MKT 100V TGL, 0.15$): 3.292µF (deviation <0.1% to Alumen-Z-Cap)
(measured with TH2821 LCR meter @1kHz)
*** good standard quality MKP cap probably cost you 3-4$ per piece for 3.3µF


The differences of the capacitors can be measured without any problems. Here the behavior of the capacitors at 1kHz and 10kHz the order is always
Alumen-Z-Cap, Standard-Cap and Old-Cap @1kHz:
1697407494162.png 1697407522630.png 1697407582005.png
Alumen-Z-Cap, Standard-Cap and Old-Cap @10kHz:
1697408107895.png 1697408127236.png 1697408156324.png
It is easy to see that the resistive part of the capacitors are quiet different and that the difference increases with higher frequencies.

This corresponds to what one would expect from a capacitor, since unwanted (parasitic) properties exist in addition to capacitance. The equivalent circuit of a capacitor looks like this:
1697445918263.png

Source: https://www.elektronik-kompendium.de/sites/bau/0205141.htm

The most important unwanted properties are (the frequency dependent) Equivalent Series Inductivity ESL and Equivalent Series Resistance ESR. However, certain properties often only come into play at very high frequencies in the MHz range or above.

3) Measurement consistency
To get an impression how consistent the frequency response measurements of the tweeter and approximately how large the errors of the measurement are, a capacitor is measured twice with time between the measurements.
The first image shows the FR of the two measurements. The second image shows the FR normalized to one measurement with a very high resolution of only +-0.5dB:
1697411401872.png 1697411429027.png
In the frequency range 1-30kHz, the two measurements differ by a maximum of +-0.04dB. Put simply, this is probably roughly our measurement error.

Below 1000Hz the SPL of the test signals is so low that ambient noise affects the measurement significantly, therefore all measurement results of the tweeter with the different capacitors were limited to 1kHz to low frequencies:
1697416959925.png


4) Tweeter measurement with different caps
How do these capacitance and resistive differences affect the frequency response emitted by the tweeter? For this we measure the frequency response of the tweeter with the simple crossover for all three capacitors (see test setup above for details) - smoothing is 1/24 oct:
1697411131178.png
Those who know the other parts will not be surprised, since the capacitors were selected for the best possible capacitance equality, so you can hardly tell the difference in the FR.

To see differences, we normalize the frequency responses to the FR of the Alumen Z-Cap and look how big the deviations of the other two capacitors are to it (attention the scale is +- 0.5 dB, again smoothing is 1/24 oct):
1697412101325.png

Only the ancient film capacitor with the largest deviation in capacitance and capacitive impedance shows deviation larger than the measurement error (+-0.04dB).
The deviations of -0.1dB around 10kHz are extremely small and should hardly be audible.

To get a complete picture, let's examine the multi-tone distortion (these include HD and inter-modulation distortion) of the high-end capacitor with the standard capacitor:
1697413094836.png
As expected, the results are nearly identical.

So, the measurements are almost identical, now how drastically will the differences in the recordings be? See next post...

Update 2023-10-17: Added information about measurement equipment. Added Index. Added information about caps.
 
Last edited:
OP
C

ctrl

Major Contributor
Forum Donor
Joined
Jan 24, 2020
Messages
1,598
Likes
6,006
Location
.de, DE, DEU
5) Audio recording samples

As already said, the recordings were not made in a studio, so I can not exclude that the recordings are possibly distinguishable by background noise (street noise, children's cough - daughter with cold, TV sound,...) or editing mistakes - if you notice something like that, please report right here in the thread, so that I can possibly exchange the files.

During the recordings the microphone was about 2cm (0.8'') away from the tweeter. The maximum sound pressure at the mic was about 106-123dB. The used mic, an iSEMcon EMX-7150, shows >143dB 3% THD and around 125dB 0.2% THD**.
1697442178928.png 1697442786024.png Source: Production Partner 6/2012 review
** It is @125dB about 0.2% HD2 which will be completely masked and 0.003% HD3 according to the review.


During the ABX test in foobar I explicitly did not look for possible errors in the recordings, but tried to hear a difference between the capacitors in an "honest way".

Of course, I myself was extremely curious about the results. In particular, how the ancient film capacitors that I myself had bought ten years ago as bagged goods (at that time you could buy a bag of various bipolar film capacitors for a few Euros), would perform.
We expect, after the reviews of the high-end capacitor quoted in the opening post and the strongly presented statements and opinions of individuals in forums and YouTube videos that the difference will be staggering.

Therefore, in an ABX blind test, all ten of ten (10/10) trials should be correctly identified.
In other ABX tests, for example, I had 8/10 correct, but this turned out to be a coincidence after further attempts. Therefore, in order not to commit self-deception it should be 10/10 or two times in a row 9/10 correct answers.

All recordings are in lossless flac audio format.
As attachments you can find three files starting with file names like "01-Samples_FC... .zip". These are the files with the "Fast Car" track sample recordings, one file for each capacitor.
In the "samples_PN.zip" file are the three pink noise recordings for the three capacitors.

My own ABX results look like this:
Alumen-Z-Cap versus Old-Cap for pink noise recording
1697441573244.png

Alumen-Z-Cap versus Standard-Cap for "Fast Car" recording
1697441671773.png

As one can easily see, the different film capacitors were completely transparent and indistinguishable to me. So for me there is no reason to spend fifty times or more the price for high-end capacitors, compared to normal standard film capacitors.


6) Samples analyzed with DeltaWave

If you don't trust your ears, you can also analyze the audio samples using software. DeltaWave from forum member @pkane is excellent for this.
In post#32 he compared the "Fast Car" samples Alumen-Z-Cap vs. Standard-Cap and Standard-Cap vs. Old-Cap with DeltaWave.

With DeltaWave, however, anyone can perform the analyses themselves. It is only important to note that with the cap samples everything below 1kHz and above 22kHz is ignored (below 1kHz noise distorts the samples, above 22kHz is outside the recording range).
For example, here is the delta spectra of the comparison Alumen Z-Cap and the standard cap for the "Fast Car" and pink noise samples:
1697568647640.png 1697568672810.png
On average, the deviations of the two recordings are clearly below +-0.1dB, even individual peaks hardly exceed +-0.1dB.

Hear the difference between the caps

With the software it is also possible to listen to the difference of two samples. To that end, @pkane provided a recording of the differences between the Alumen Z-cap and the standard cap.
This is the real difference (including all differences of phase, distortion, decay, SPL,...) between the two caps.
You can find the file in flac format as an attachment with the filename "Diff-File_Alumen-Z-Cap_vs_Standard-Cap_FC.zip".

Warning: Please be very careful when listening to the "Diff-File_Alumen-Z-Cap_vs_Standard-Cap_FC.flac" recording, because the differences of the caps are extremely small and you have to raise the volume by 20dB and more to hear anything at all.



7) Compare impedance measurements of the caps in REW


The impedance measurements of the capacitors were performed in Arta-Limp. But I imported the measurements in REW (since I don't use REW, get in touch if something is wrong).
In REW you can then view an equivalent circuit (component model) for each capacitor.
1697750886947.png

The REW file with the three impedance measurements for the film capacitors can be found in the attachment "Cap_Imp_compare.zip".
REW shows you the equivalent series and parallel impedance values from the box in the lower left corner.



Update 2023-10-23: Added section 7)
 

Attachments

  • 01-Samples_FC_Alumen-Z-Cap.zip
    2.5 MB · Views: 274
  • 01-Samples_FC_Old-Cap.zip
    2.5 MB · Views: 161
  • 01-Samples_FC_Standard-Cap.zip
    2.5 MB · Views: 123
  • samples_PN.zip
    2.8 MB · Views: 97
  • Diff-File_Alumen-Z-Cap_vs_Standard-Cap_FC.zip
    946.8 KB · Views: 96
  • Cap_Imp_compare.zip
    47.6 KB · Views: 75
Last edited:

GXAlan

Major Contributor
Forum Donor
Joined
Jan 15, 2020
Messages
3,777
Likes
5,838

GXAlan

Major Contributor
Forum Donor
Joined
Jan 15, 2020
Messages
3,777
Likes
5,838

MAB

Major Contributor
Joined
Nov 15, 2021
Messages
2,081
Likes
4,679
Location
Portland, OR, USA
Jantzen is not really 'high-end' . . . don't plotz when you see the price:

Jupiter Condenser, Copper foil, paper and wax.

Audio Note, Copper foil and Silver foil, Mylar and oil.

There are others with prices that are even more ridiculous . . .
This one is on a hand-selected wood dowel, since all of the sonic differences start with that part... the foil is from Costco and is heavy duty, which will give better bass.;)
1697432367806.png

I'm sure I would need to charge a few hundred dollars for the 3.3uF version. Cryo a bit more.:cool:
 

Mnyb

Major Contributor
Forum Donor
Joined
Aug 14, 2019
Messages
2,476
Likes
3,375
Location
Sweden, Västerås
I strongly suggest reading trough all of the previus treads before engaging in any deeper discussions . they are worthwhile an realy good :) just do that .

I have an amateur's question the resistive component in each cap, is it compensated for. Or is this circuit not particulary sensitive to this change .
It's pleasing to to see the tight match in C , so why not for R ?
I can see that only the hodgepodge frankencap of old stuff has any kind real deviation from a straigth line in the fr response , so maybe this tell it all ?
 

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,524
Likes
10,563
Location
Prague
As announced long ago the last part on the subject of audibility of capacitor upgrade in crossovers. The comparison ranges from a very cheap, decades-old film capacitor to an expensive high-end capacitor.
5 years ago I made a similar test between MKT1813 3.3uF cap and Mundorf M-Cap Zn 3.3uF cap. Not only there was no audible difference, but also no differences in measurements with my 2018 system. Caps were loaded with 4 ohm. Files attached.
 

Attachments

  • MKT1813 5.7V.pdf
    96.6 KB · Views: 210
  • Mundorf 5.7V.pdf
    96.3 KB · Views: 163

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,524
Likes
10,563
Location
Prague
One of the beliefs stated by audiophiles who believe there are audible differences between capacitors is energy storage and release. They say this affects transient response. Is there a way to test that? I am really looking forward to your results! @Amir should promote this to the front page.
Yes, easy to measure, just measure transient response with load. It does not matter what you do, if you show that there is no difference, you will instantly get another stupid suggestion from an audiophile. It is only wasting time to discuss with them.
 
OP
C

ctrl

Major Contributor
Forum Donor
Joined
Jan 24, 2020
Messages
1,598
Likes
6,006
Location
.de, DE, DEU
The sample recordings of the capacitors can be found in the updated post#2.


You should also use DeltaWave to do some null testing IF the ABX testing is positive.
This can be done by someone who is familiar with DeltaWave (the recordings are now online in post#2), because the frequency range below about 800-1000Hz must not be evaluated in the recordings, because the ambient noise disturbs there - for more details see post#1 under section "3) Measurement consistency"


They say this affects transient response. Is there a way to test that?
This could theoretically be seen in the frequency response measurements (certain "fast" signals are not transmitted at full SPL) or in the decay behavior of the tweeter, as long as the influence of the capacitor is comparable or greater than that of the tweeter itself.

Since there are now recordings with a test track "Fast Car" for each capacitor, you can easily decide for yourself with an ABX test in foobar whether you can detect serious differences for "fast transients" in the test track.


Jantzen is not really 'high-end' . . . don't plotz when you see the price:
So that's the thanks I get for spending more money on a single capacitor than I ever have in my life, links to even more expensive capacitors, that's bullying.... ;)
 

antcollinet

Master Contributor
Joined
Sep 4, 2021
Messages
7,242
Likes
12,013
Location
UK/Cheshire
One of the beliefs stated by audiophiles who believe there are audible differences between capacitors is energy storage and release. They say this affects transient response. Is there a way to test that? I am really looking forward to your results! @Amir should promote this to the front page.
Transient response is fully defined by frequency (and phase) response.

At 20Khz it is storing and releasing energy as fast as any music can make it.
 

wwenze

Major Contributor
Joined
May 22, 2018
Messages
1,261
Likes
1,798
Is it too late to add a new test subject?

The mythbuster style, if the result doesn't end up like what the myth says, take it to the extreme until it ends up like what the myth says. Very poor-quality electrolytics. Polarized, if you dare.
 

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,524
Likes
10,563
Location
Prague
Is it too late to add a new test subject?

The mythbuster style, if the result doesn't end up like what the myth says, take it to the extreme until it ends up like what the myth says. Very poor-quality electrolytics. Polarized, if you dare.
They are still good enough. Only high-K ceramics is a problem.
 

Sokel

Master Contributor
Joined
Sep 8, 2021
Messages
5,609
Likes
5,529
No Duelund CAST amongst the samples,I refuse to accept the results with no hiend capacitors tested :p

(only joking of course,thanks for that! )
 

tarikuz

Member
Joined
Nov 18, 2019
Messages
34
Likes
30
IMHO when brand new they are all equivalent.. but one thing all capacitors have in common is their tendency to age—some over very long periods of time, others reasonably quickly depending on their operating environment. Heat is their biggest enemy.

That's why when you buy a vintage equipment you should replace all caps (first of all the electrolytics).

I would sugegst to take a look @ https://www.humblehomemadehifi.com/download/INFO_Martin-Colloms.pdf

Your thoughts ?
 

UliBru

Active Member
Technical Expert
Joined
Jul 10, 2019
Messages
123
Likes
337
Some time ago I have watched and listened to a quite interesting demonstration.
Several crossover capacitors have been used by this demo.
For each test one of them has been connected to the outputs of a power amp in series with a proper protecting resistor (low resistance, high load. Just to avoid a short circuit).
The amp has been fed by a CP player.
And surprisingly different capacitors have started to play music acting like a little speaker driver.
The loudness was of course different. Some capacitors could be heard directly, some required a stethoscope.

The conclusion: depending the tightness of the mechanical structure the capacitors behaved like motors. A very interesting experiment.

Now motors can also behave as generators.
And so I wonder if a capacitor can produce distortions (like another example: knock on a cable connected to an oscilloscope).
So mechanical vibrations (crossover capacitor placed in a speaker) may thus influence the sound.
The aging of a capacitor may also change the mechanical behaviour.

Possibly it would make sense to select this way a best (silent) and a worst (loud) capacitor and then to investigate if they give different results with measurements described in the first post of this thread.
 
Top Bottom