+- Chellos Keyboard Players Club (https://chellos-keyboard-players-club.com)
+-- Forum: CATEGORY 17 (https://chellos-keyboard-players-club.com/Forum-CATEGORY-17)
+--- Forum: ARTICLES (https://chellos-keyboard-players-club.com/Forum-ARTICLES)
+---- Forum: CHELLO`S ARTICLES (https://chellos-keyboard-players-club.com/Forum-CHELLO-S-ARTICLES)
+---- Thread: CAPACITOR TEST - VERY SURPRISING RESULTS (/Thread-CAPACITOR-TEST-VERY-SURPRISING-RESULTS)
CAPACITOR TEST - VERY SURPRISING RESULTS - Chello - 04-26-2026
CAPACITOR TEST - VERY SURPRISING RESULTS
My JBL horns are connected in series, so that the impedance is doubled from 6.6 Ohm to 13.2 Ohm.
This is to reduce efficiency, as they are extremely sensitive compared to the rest of the units used in the horns.
This also has all advantages in terms of sound, as the power amplifier works much easier in 13.2 Ohm compared to 6.6 Ohm, and THD (third harmonic distortion) is drastically reduced.
Many years of experience in building loudspeakers have taught me that the simplest crossover is always the best, if you are very careful to choose units that are good enough to withstand a simple filter and its soft crossover and subsequent greater power allocation to the unit.
Some time ago I acquired some relatively expensive capacitors from Jantzen Audio; a Superior Z-Cap of 18 µF 800 Volts and a MKP Cross Cap of 33 µF 400 Volts, which were connected in parallel, giving a crossover of -6 dB at 240 Hz.
This filter worked brilliantly in terms of effects and everything else, but human voices were a little too shrill at high levels, which is a well-known problem with horn systems, by the way.
This was with my MBL preamplifier, and with my new FM Acoustics preamplifier the tendency was reduced, but I wanted to reduce the tendency further.
So I did the following:
I disconnected the Superior Z-Cap capacitor, so that only the MKP Cross Cap was connected, which gave a crossover of -6 dB at 370 Hz.
The result was clearly better than before, with better reproduction of human voices.
Next, I tried the same with the Superior Z-Cap, and was very surprised that this gave about the same result as with both connected, although the crossover frequency was now raised to 680 Hz.
This reveals a merciless truth:
The MKP Cross Cap has better sound characteristics than the Superior Z-Cap, which is significantly more expensive..!
For several decades I have used Sidereal capacitors before these Jantzen capacitors, with which I have exclusively good experiences.
Today it is possible to get hold of these very cheap from AliExpress, and I will order all relevant values and test these against Jantzen - it would not surprise me if Sidereal are the best.
RE: CAPACITOR TEST - VERY SURPRISING RESULTS - Chello - 04-26-2026
I've done some research and found that others have had the same experiences as I have.
The most popular favorites are the Dayton Audio DMPC and the Jantzen Audio Cross Caps - so I made a good choice.
Large horn drivers like my JBLs are effective from a few hundred Hz and up, which requires special capacitors, and the Superior Z-Cap is not best suited for this task, despite its higher sound rating; these are best suited for tweeters.
RE: CAPACITOR TEST - VERY SURPRISING RESULTS - Chello - 04-28-2026
I have now tested many different types of capacitors with different values, including Sidereal.
The result is unlike anything else I have tested with any other unit - the JBL horns react differently to all others.
Articles found on the internet also confirm my experiences, because it is claimed that horns that are actually most effective in the midrange require a 2nd order (-12 dB) crossover; i.e. a capacitor and a coil to attenuate human voices, as these will otherwise always be a bit too dominant.
A coil for a horn also does not have the negative effect that it will have on a tweeter, but the cost is considerable, as we are talking about many mH, and the price is around 50 dollars per piece.
So far I actually prefer the Superior Z-Cap from Jantzen Audio, even though human voices are a bit too shrill - the details are infinitely better reproduced with this capacitor than any other I have tested.
There is also about only one recording that I wish was better, and that is the test track I always use for human voices - "Someone To Watch Over Me" with Marie Bergman.
This recording is notoriously one of the most difficult to reproduce, and a fantastic tool for revealing deficiencies in a sound system.
RE: CAPACITOR TEST - VERY SURPRISING RESULTS - Chello - 04-28-2026
I have now ordered 2 coils for a new crossover - Jantzen Audio Air Core Coil (6.0 mH).
These are quite large - 61 x 30 mm.
Together with the Superior Z-Cap capacitors of 18 µF, these give a perfect 2nd order crossover with -12 dB at 480 Hz, which should work well.
Theoretically this should minimize the tendency for human voices to become shrill - we'll see.
I ordered them with DHL Express, so I will receive them within 24 hours.
Cost: 75€ including shipping
RE: CAPACITOR TEST - VERY SURPRISING RESULTS - Chello - 05-01-2026
New crossover connected and tested
Not as expected.
First they were more expensive than expected, because DHL Express demanded a customs fee, which was also subject to VAT - 35$ extra.
So there were some expensive coils - over 100$.
Connected up there was only a small difference from just the capacitor (-6 dB) - the capacitor plus coil gave -12 dB, but I actually think human voices became even a little more shrill than before, and the midrange generally clearer (TOO clear..?).
Very, very strange.
The only sensible answer I can find is that I have calculated a crossover frequency that is TOO LOW.
480 Hz is probably too low, and I have to try a higher frequency.
I have used the calculations for a Butterworth filter, which puts the greatest work distribution on the capacitor, because I have the best experience with this type.
However, I know that Linkwitz-Riley filters are far more efficient - these put the greatest work distribution on the coil, but have unfortunate effects upwards in the frequency range.
This suggests that such a filter should work better in my case, because I want a crossover frequency below 1000 Hz.
I can try a Linkwitz-Riley filter that divides at 680 Hz, because then I can use the coil I have, and change the capacitor value from 18 µF to 9 µF.
This should give a better effect; otherwise the only solution is to divide MUCH higher.
RE: CAPACITOR TEST - VERY SURPRISING RESULTS - Chello - 05-02-2026
SERIES CONNECTION = ZERO CROSSOVER EFFECT..?
I did an experiment - I connected the JBL horns in parallel, instead of in series as I had done.
Then something "strange" happened: The filter worked !
I made a mistake with the connection, because series connection changes the connection to the filter.
It can be safely stated that the filter now works, because the sound pressure was reduced when I halved the impedance, while the opposite should happen without a filter.
I have some of the best measuring equipment on the market for such things, which complies with the industry's strictest requirements according to the Thiele Small parameters - DATS.
So then all that remains is to check the impedance again, which I have now done by first recalibrating the DATS, performing an Impedance Sweep, and then a Measure Free-Air Parameters test:
My first measurement of 6.6 Ohm is confirmed with 6.572 Ohm
Fs is found to be 330.1 Hz, which also proves that I have repeatedly chosen a crossover frequency that is too low; such a filter will have no effect.
The red line for the impedance curve also suggests that the ideal crossover frequency is as high as 2 kHz or higher, as the curve has only minimal dropouts from there and upwards.
Some may wonder why I did not make this measurement in the first place - this is because I find great pleasure in working without the use of measuring equipment and making practical experiences, and finally making measurements that can explain my experiences.
I have now gone back to 1st order Butterworth, which only requires one capacitor, and use the Superior Z-Cap of 18 µF, which gives - 6dB at about 1300 Hz.
Human voices are still a bit too shrill - horns are extremely difficult to work with compared to all others.
If nothing else, brushes and the like are now even clearer, and I like that.
RE: CAPACITOR TEST - VERY SURPRISING RESULTS - Chello - 05-02-2026
I have now tested with APO EQ, and found that a strong reduction of -8dB at 900 and 950 Hz greatly reduced the problem of shrill human voices.
There was also an area around 1800 Hz with half as much dropout, which I attenuated by -4dB.
I first tried recommended ranges around 4000 Hz, but this affected other instruments adversely.
When I studied my own measurement with DATS, there is a strong dropout in the impedance around 900-950 Hz, which results in this range being reproduced several dB too loud.
It is nice when theory and practice agree..!
Now I know more precisely where the crossover frequency will work best, and I guessed well at 1300 Hz (or higher; 2-3000 Hz), but need an additional coil to increase the reduction to -12dB.
A new value for the capacitor must also be calculated.
A Linkwitz-Riley filter is probably best.
Then I avoid the actual problem range(s), which is 900-950 and 1800 Hz.
RE: CAPACITOR TEST - VERY SURPRISING RESULTS - Chello - 05-02-2026
I did one more test:
I removed the Planar membrane on the horn, and took the same measurements once more.
With Planar membrane:
Without Planar membrane:
The result was quite different; the impedance curve (red) shows a smoother course with less jagged lines, and the dips are a little stretched out with a softer character.
The impedance magnitude (blue) is markedly different, and this says something about how hard the amplifier has to work to drive the speaker at specific frequencies.
FS (free air resonance) is somewhat higher than with membrane.
So theoretically it seems as if my "patent" Planar membrane worsens the impedance course and the amplifier's working conditions - clearly.
The big (and only relevant) question, however, is:
How do the horns sound WITHOUT Planar membranes compared to WITH ?
SURPRISE...
Human voices no longer shrill.
Details (character) are very different - I hear a delicacy in the voice that I have not heard before.
At the same time, voices appear too thin; they lack fullness.
The double bass and details around the strings as well as the resonance in the bass's box are more pronounced.
Brush and drum details are much clearer.
Piano is also much clearer than before - TOO clear..?.
A strange thing is that I hear negative compression, much like with the Pa5X when too many effects are active at the same time, and it feels like there is too much reverb - human voices don't shrill anymore, but many instruments do; especially the piano.
A VERY surprising and shocking thing:
The perspective is MUCH worse, with a clear tendency towards stereo..!
And the dynamics are worse.
CONCLUSION..?
My Planar membranes are going back on..!
I'm going to try a thinner foil, because the one I used is a bit too thick, but easier to work with.
This foil is VERY difficult to stretch up, but I think this one could be a significant improvement.
My Planar membranes are superb in terms of perspective and dynamics, and I can (and must) tolerate slightly shrill human voices, when everything else works far better than without Planar membranes.
These measurements clearly show that theory and practice do not necessarily match, because ugly measurements can actually sound better, as in this case.
For comparison:
Measurements of a tube amplifier versus a transistor amplifier show that the tube amplifier has significant distortion, but it is still far superior to the other in terms of sound..!
So my invention Planar membranes does great things with perspective and dynamics - the JBL horns without these do not impress..!
A strange incident at the end:
When I removed the foil, compressed air came out - POOFFFHHH..!
I know that the horn driver is called a compression driver, but...
Maybe I should call them Compression Planar membranes..?