[Æ]

Okay just for you guys I just took a few measurements tonight.

I just selected some caps with matching values. One pair of 27uF values measuring 53.91uF. The second pair was a 39uF and a 15uF measuring 54.95uF. That's pretty close.

Any other questions?

All those pretty colors, I'm mesmerized. Actually they are giving me a headache. They kind of confused me a bit too.

Are the ESRs of both bundles nearly the same too? Or did you sneak in a lossy/lousy capacitor?

Which bundle of caps produced which colors on the graph? Which bundle (curve) sounds better?

More than likely the differences in uf between the bundles accounts for the very minor difference between the curves.

[Danny Richie]

Are the ESRs of both bundles nearly the same too? Or did you sneak in a lossy/lousy capacitor?

Which bundle of caps produced which colors on the graph? Which bundle (curve) sounds better?

More than likely the differences in uf between the bundles accounts for the very minor difference between the curves.

All four caps are Erse Pulse X poly caps. So no fast discharging Sonicaps have been slipped into the mix.

The Red, Orange, Yellow lines are the mismatched bundles.

The very small difference in uF does not account for the difference in phase shift. It doesn't work that way. Variation in the uF value will show up on the left side of the graph in the lower frequency range and not in the range where the phase shift is seen.

See the same responses again with one addition.



A 56uF Erse poly cap was used this time. It measures 56.64uF. It's variation in impedance is clearly seen. It is the Green line. It's phase response is shown by the Yellow line.

The light Blue line is the two 27uF values, and the Red line is the mix matched values of 39uF and 15uF (phase responses).

So as you can see the impedance of the 56uF and the two 27uF caps are nearly the same from about 2kHz and up and so is the phase response. The 39uF and 15uF values have an identical impedance to the pair of 27uF caps, but the phase shift is different.

Get it yet?

[jtwrace]

So is it safe to say then that using one really hi end cap with a lower quality cap is better as long as they match as opposed to using two hi end caps of differing values to make the same end value? 

[Jonathon Janusz]

. . . and when using a bypass cap (very low value in parallel with the necessary value), the phase shift happens further up the graph (to the right), high enough that although the shift exists, it becomes basically irrelevant?  So, when mixing high-end and low-end caps, better to do it this way to get the sound signature of the better cap in the mix while putting the potential problem areas far enough out of range that it doesn't really matter, while leaving no possible problem in the range you are using?  I think I got it. 

Danny, if it wouldn't be too much trouble, a graph showing the output from a bypassed arrangement would be nice to add to the discussion.

. . . and to clean up the OP's thread, this set of graphs/explanations would make a great sticky or "additional info" page for your web site.  The graphs make it really easy to see what's going on and make sense of it all very quickly.

Cheers!

[Danny Richie]

So is it safe to say then that using one really hi end cap with a lower quality cap is better as long as they match as opposed to using two hi end caps of differing values to make the same end value?

Even if the values are the same, if the discharge rates are different then you can still see some of this same effect.

Keep in mind that caps are energy storage devises. They store and release energy. The bigger the cap the more stored energy. This creates the smearing effect that caps cause. The smaller the cap or the faster the discharge rate the cleaner they sound.

There are a bunch of other issues that effect the sound to from the thickness of the film, type of film, type of conductive material termination of the end leads, type of end leads etc. So sometimes we can make them measure the same by the more common measurements used, but they still sound different. If only could be so easy....

[Nick77]

Danny is it safe to say that Erse and Sonicaps play well together as far discharge? That seems to be a common combo in your XO's. But the real question I would have is does Sonicap and Jantzen SuperiorZ discharge playfully?

[Danny Richie]

. . . and when using a bypass cap (very low value in parallel with the necessary value), the phase shift happens further up the graph (to the right), high enough that although the shift exists, it becomes basically irrelevant?  So, when mixing high-end and low-end caps, better to do it this way to get the sound signature of the better cap in the mix while putting the potential problem areas far enough out of range that it doesn't really matter, while leaving no possible problem in the range you are using?  I think I got it.

Yes, yes, yes...

Danny, if it wouldn't be too much trouble, a graph showing the output from a bypassed arrangement would be nice to add to the discussion.

When I read your request this morning I decided to give it a shot. Good idea! Hang on data coming.

[Danny Richie]

Danny is it safe to say that Erse and Sonicaps play well together as far discharge?

Value for value The Sonicaps are faster, and sound cleaner. Sometimes though you have to weight out their position in the circuit and decide if any performance improvement merits the cost difference. Sometimes I am using a big value in a lower frequency range and the performance difference becomes less while the price difference becomes pretty sever. So sometimes the Erse poly cap is a better choice for the application.

That seems to be a common combo in your XO's. But the real question I would have is does Sonicap and Jantzen SuperiorZ discharge playfully?

Now you are talking about a lot of differences in makeup that cause big differences in sound, and not just discharge rates.

[Danny Richie]

Okay, now we are looking at by-pass cap effects.

Here is a base line so that it can be easily compared to the other graphs.



Blue and light Blue is the 56uF cap again. Purple and Green is a 12uF cap. We will be using the 12uF cap for seeing the effect of the by-pass as is it is really hard to see the effect on the 56uF.

Here is the same graph again but I have zoomed in quite a bit on the phase response so we can see the effect.



The Yellow line shows the effect of the by-pass cap. Remember, this is pretty zoomed it. And I tried a .1uF by-pass cap, and a .22uF by-pass cap and could see little to no difference. This is a .33uF by-pass cap.

Note that the difference in phase shift is slight and only in the upper range.

I also noted yesterday, and more so on these measurements, that I had to be very careful not to hold or touch the caps when taking the measurements as the pressure and temperature from my hand can cause some effect. Pressure and temperature can also effect the uF value as well.

[HAL]

Ok, from Danny's last comment about pressure and temperature changes, that would also lend support to having an external crossover for a speaker.  The internal temperature and pressure change with speaker signal input will vary the capacitance value vs time.

Just a thought.

[Danny Richie]

Now back the original poster...

The mix matched pairs in his network are in a notch filter. The inductor and capacitor are saying that anything in a frequency range that falls between these two components will be shorted back to ground, and the resistor level determines how much of it is shorted back to ground.

So any phase shifts that the caps make are not passing on to the drivers output.

Minimizing the stored energy of the bundle using decent poly caps, and if real serious about it, by-passing it with a small value will help as well. That way the range that is intended to be passed back to ground does so quickly and leaves no residual overflow headed on to the driver. 

Another thing to keep in mind with all of this is the end use.

If the speakers were placed in a bedroom and hooked up to a receiver to play a little mood music or something then it really wouldn't matter too much if the 5.6uF value on the tweeter were made up with a 4.1uF and a 1.5uF cap value. No one would really notice the difference.

The same goes for cap brand and type. Keep in mind the application.

If you are just watching movies only and through a receiver then springing for all Sonicaps might not be the money best spent.

But if you are serious about you music and your system reflects that then a lot of these effects are not that hard to hear.

[*Scotty*]

Danny, I think your fingers were capacitively coupling to the caps when you touched them rather than seeing a heating effect or perhaps in addition to. I know that proximity effects are very important as the frequencies get higher.
Scotty

[bprochford]

The OP here,

Thanks again, Danny.  I now understand the function of the notch filter, and why it works the way it does. I also understand, because of your testing and charts, the way phase changes based on capacitor use in bundles.  All things I didn't know before.

FWIW, I'm planning on upgrading my Kestrel components based on your recommendations.  I do care about how they sound, and would like to see if I can increase the performance and value of an already awesome sounding design. I've even got some NoRez left from my N3 build and might see what I can do to fit some in the Meadowlark cabinets.

This is why I spend a lot of time in this circle, and why I'm in this hobby. The guidance and support really helps those of us who aren't engineers and designers understand what's happening with our equipment and why.

Best,
Brad