Apples to oranges. Figure 5 is a .1uF 100 volt mylar with a .1uF 200 volt polypropylene cap as reference. Figure 7 is a 1uF 200 Volt polystyrene cap with a 1 uF 50 volt teflon reference cap. RL differed for both tests also. So what? Did they sound different?
So why didn't you scale things way back when you did the original measurements way back in the 80's? How does the resulting measurements correlate to it being audible?
Blurry photos don't reveal much.
Of course, you have to TRY the differential cap measurement as shown, to note the action. Back then, I did not have the most advanced oscilloscope available to make the pictures, but they are accurate enough to compare differences in cap material types. This is the first time that 'bleary photos' was brought up as an argument against this test result.
The 'scaling' question only comes up with the RESIDUAL distortion presented with a comparison between a 1uf TEFLON and a 1uf POLYSTYRENE, that set the baseline for the test itself. Null is null in any case.
The 'scaling' question only comes up with the RESIDUAL distortion presented with a comparison between a 1uf TEFLON and a 1uf POLYSTYRENE, that set the baseline for the test itself. Null is null in any case.
A common refrain.
The problem with SOUND assessment is, in the grossest sense, twofold.
1) ABX and DBT's are extremely difficult to pull off correctly. Need I say more? I can if you are interested. However, it's been written about extensively on the net.
2) It's been my experience doing crossover upgrades for customers and also reading about other's experiences when comparing new caps of different types and constructions, the differences are quite subtle. They are not night and day like you'd most likely have comparing different speaker brands of similar construction.
So, the bottom line is, YMMV. Keep in mind also the important variable, listening tastes. Even when someone does find two caps that sound different, you may like the sound of one the other person doesn't like.
So many variables, so subtle an effect.
The journey - so much fun!
For something like this, no, they're not terribly hard. But... frequency response changes with different caps can be predicted (even if not exactly, but close enough) and easily measured. And if they're above well-established audibility thresholds, why would anyone bother doing a DBT?
There's a reason that hard-nosed and smart engineers like Ken Kantor or Jack Hidley would specify their crossover caps for more than just capacitance and voltage.
I think speakerdoctor is talking a lot of sense here. If you look for the Damping Factor curves for mylar aka polyester capacitors, you'll find it is 1-2% at 20kHz.
For polypropylene it is an order of magnitude smaller. Recall DF is the inverse of ESR and measured as the tangent of an angle quite often.
These are tiny effects. You really will struggle to hear that, especially with some series resistance. There's not really any evidence that it is a non-linear effect, translating to 1 or 2% harmonic distortion, which is what would really worry you.
Non-polar Electrolytics are much more straightforward to hear. I can hear it in crossovers as a slight increase in sibilance. DF can be 10%.
How come all those old amplifiers used polar electrolytics then? A simple oscilloscope test shows their non-linearity.
Answer: They were used with single rail power supplies and the correct voltage bias. When biased or polarised correctly, they are quite well behaved.
For polypropylene it is an order of magnitude smaller. Recall DF is the inverse of ESR and measured as the tangent of an angle quite often.
These are tiny effects. You really will struggle to hear that, especially with some series resistance. There's not really any evidence that it is a non-linear effect, translating to 1 or 2% harmonic distortion, which is what would really worry you.
Non-polar Electrolytics are much more straightforward to hear. I can hear it in crossovers as a slight increase in sibilance. DF can be 10%.
How come all those old amplifiers used polar electrolytics then? A simple oscilloscope test shows their non-linearity.
Answer: They were used with single rail power supplies and the correct voltage bias. When biased or polarised correctly, they are quite well behaved.
I think J. Curl thought he was comparing apples to apples when he did his Real Time Signal test linked earlier. Now, he's told he compared apples to oranges!
......and Curl was using some rather objective tests. Now, insert the humungusly variable human factor into the analysis of 'how does it sound?'.
How many? Do they need training? what music do we use? what switching do we use? and the list goes on and on....
To satisfy the likes of DavidL of course!
Time to have your "speaker doctor" license reviewed
I'm not the one making claims that caps sound different so have at it.
So we have a 30+ year old test of various caps. Subjective say-so that mylar caps sound bad.
Not a single ABX,SBT or DBT testing such caps but people just want to believe what they want. Yep that's a great scientific method.
Last edited:
Ignoring basic analysis and well-established thresholds. I'm getting the feeling that you're more interested in trolling than understanding. I hope I'm wrong and that you move from name-calling and sniping to analysis and experiment.
Until then, I suggest that no one allow themselves to be baited.
No Sy, the point being is that no one has bothered to actually test the audible difference that's supposed to be so evident with mylar caps. All we have is conjecture. Subjective only. Some people see numbers posted and jump to the conclusion that mylars must be awful. Do they also ignore the fact that electrolytics measure worse yet they are also used by major manufacturers? Just because you want to point to some numbers and think that must mean that a cap sounds terrible means you haven't bothered to look further.
Having had a look at the literature for capacitor applications, it becomes clear to me that you should make no assumptions at all about aluminium electrolytics in filters around 100kHz, where the ESR of 0.5ohms and the small ESL cause a significant blocking resonance which limits attenuation to around 20dB. Below that frequency you get significant phase change.
With bass shunts, it is well known that NP electrolytics significantly change damping factor and phase and the behaviour of bass filters. This is usually accounted for by design. At higher frequencies in Non polar tweeter filters you would expect the impedance of a Highpass filter to be higher than expected.
Evidently film capacitors are more ideal overall. To reiterate, polyester or mylar has a Damping Factor of around 1-2% at high frequencies, polypropylene nearer 0.1%. How audible any of this might be doubtless depends on the application, but I can assure you that solid state amplifiers are very sensitive to loading at high frequencies because it affects the phase in the feedback loop, which often has a corner around 100kHz by design.
If I take anything from all this, it is that the quality of the capacitor in a tweeter Zobel and the general high frequency impedance will make a significant difference to the frequencies above 5kHz due to amplifier interaction. Since loudspeaker cables have a characteristic impedance around 100 ohms, you might also consider shunting the tweeter with a resistor of around that value too, which correctly terminates the cable. If you have an attenuator fitted, this becomes moot of course.
You see, the differences you hear may really be coming from the amplifier and cables more than the filter.
With bass shunts, it is well known that NP electrolytics significantly change damping factor and phase and the behaviour of bass filters. This is usually accounted for by design. At higher frequencies in Non polar tweeter filters you would expect the impedance of a Highpass filter to be higher than expected.
Evidently film capacitors are more ideal overall. To reiterate, polyester or mylar has a Damping Factor of around 1-2% at high frequencies, polypropylene nearer 0.1%. How audible any of this might be doubtless depends on the application, but I can assure you that solid state amplifiers are very sensitive to loading at high frequencies because it affects the phase in the feedback loop, which often has a corner around 100kHz by design.
If I take anything from all this, it is that the quality of the capacitor in a tweeter Zobel and the general high frequency impedance will make a significant difference to the frequencies above 5kHz due to amplifier interaction. Since loudspeaker cables have a characteristic impedance around 100 ohms, you might also consider shunting the tweeter with a resistor of around that value too, which correctly terminates the cable. If you have an attenuator fitted, this becomes moot of course.
You see, the differences you hear may really be coming from the amplifier and cables more than the filter.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.