Type 'A' tweeter x-over

[Erik Mandaville]

As mentioned in 2-channel, I built a new crossover network to compare atttenuation differences between the usual autoformer and one using a fixed resistor L-pad. I'm comfortable with the results of that, but the tweeter crossover has been a little perplexing. Since it does not use a value of capacitance for a 8ohm driver using a 1st order network (which would be a little over 3uf rather than the existing *approximate* 2uf), I started to think that the autoformer was possibly causing a reflected impedance of around 16 ohms, which would be in line with the existing 2uf capacitor for a crossover at 6000Hz.

Looking at it some more, I drew a schematic of just the tweeter section, and what it resembles exactly is a 3rd order network: a series capacitor (13uf) connected to the input, followed by the autoformer (functioning as inductor? in parallel with the load), and the last element, the series capacitor (2uf) connected to the driver. The autoformer appears to be essentially a multi-tapped inductor, and tap 5 (the common point for the input capacitor and 2uf tweeter capacitor) has continuity with ground. In other words, it's possible to draw the autoformer as a conventional parallel inductor for the tweeter, since the tweeter does not use it's capacity as an attenuator in the way the squawker does.

The series connection of the 13uf and 2uf capacitors has always interested me, and remember that series connected capacitors are subtractive, unlike resistors that are additive. The calculated value of 13uf and 2uf in series = 1.7uf. Keep that value in mind.

Next, I ran calculations for a 3rd order network for an 8ohm tweeter with a crossover at 6000Hz, and the values of the two series capacitors in front of and behind the inductor were 2.37uf and 6.2uf. Those two capacitors in series also results in a value of 1.7uf.

Right or wrong, it's just a hunch that this may be more in line with what is going on. The difference in values of capacitance between what exists in the 'A' network may have to do with the adjustments of capacitance and inductance that needed to be made for the squawker, since the input cap is common to both HF and midrange drivers.

Then again, we may be looking at a second order, 12dB tweeter branch. That uses two components in the tweeter leg, a series capacitor and once again an inductor in parallel with the load. The value of capacitance in this case is something around 1.65uf (if I remember right). Close enough to be rounded up to 2uf if all you have on hand is a bin piled high with 13uf and 2uf motor run oil caps!

Dunno, just thinking about it.

Erik

[Erik Mandaville]

Good job, Mandaville!!![:S]

I meant to put this over in the Technical forum. Anyway, it's here. now.

[Erik Mandaville]

This may be at least partly why I *seemed* to hear slightly greater tweeter output from this "true" (as I referred to it, because the 'A' tweeter branch never really looked like a first order network to me) network I built. All that exists between the input and the driver in mine is a 2uf and 3uf capacitor in parallel for 5uf.

Erik

PS: I have to emphasize here that I don't think there is anything wrong in the least with the 'A,' 'AA' or any other. They are all subjectively good networks, meaning that one person will like one type better than another. I just accidentally stumbled on something I happen to like more when I was doing a test for something else. The only way I could do it was by making an entirely new network that was completely and entirely without an autoformer. For me, it was a happy accident, just like ice cream probably was, and lots of other things.

As it stands now, I'm not totally sure I prefer the 4K crossover. I detect some graininess in the treble for some reason, and maybe I'm just tired of listening to it. I may end up just sticking with the 4.5k. I'm picky about what I hear, and just want to make sure before I haul the networks out again for more surgery.

[BEC]

Erik,

 If you go back to that other thread and look at the voltage output trace of the type A crossover, you will see the tweeter slope is certainly 1st order.  Look at the output of the tweeter at 2.5 khz and again at 5 khz.  The rise for that octave is about 6 db.

 Bob

 

[Erik Mandaville]

Bob:

Thanks for taking the time to reply. I did have a look at your plot again last evening. I'm going to draw out a few schematics, including of what you were actually testing by connecting the tweeter to the input side of the 13uf capacitor. I don't have a means of doing this on the computer; I have to draw them on paper and then take a picture.

What this implies is something I referred to earlier about the autoformer and tweeter relationship. The thing that keeps coming up is that, if this isn't an 18 or 12dB/octave arrangement -- despite what it looks like on paper -- the response indicates differently, then the autoformer may be influencing the tweeter in the way it also does the squawker. If it didn't, the capacitor for the 8 ohm tweeter @ 6k would be around 3.3 uf instead of what it is.

Erik

[Deang]

If it didn't, the capacitor for the 8 ohm tweeter @ 6k would be around 3.3 uf instead of what it is.

Unless the 3dB down point is considered the "crossover point".

http://www.peavey.com/support/technotes/processors/crossover101_a.cfm

[Erik Mandaville]

This first drawing shows what a conventional high pass section looks like for a parallel first order tweeter (1).

(2) Shows a schematic diagram of what is between the input and driver in the type 'A' tweeter branch.

That they are both 6dB octave slopes is evident from the test, however they arrive at that slope by more complex routes. In the case of 1), the impedance of the driver is maintained, however, my feeling is that the same is not the same for 2) because of its parallel connection with the autoformer.

[Erik Mandaville]

3) is a diagram of what a 3rd order parallel network looks like. It is the schematic twin of what is shown for the input-to-driver diagram for the type 'A' tweeter branch shown above. The complexity is also greater than the conventional high pass (series cap only) for a first order tweeter x-over.

4) This is the schematic of the test relocation of the tweeter section moved to the input side of the 13uf cap on the Type 'A'. My concern with this connection, however, is that I believe a relationship with the autoformer has remained in place. The connection is not completely isolated from it.

[Erik Mandaville]

( forgot the flash with this one)

5) is what an 8ohm amplifier actually sees with a type 'A'. I indicate a 16 ohm tweeter because of how I think the autoformer influences it (I could be wrong). I'm going by the value of capacitance used for a 6k crossover. For an 8ohm tweeter, which is what we use, the value of capacitance needed for 6kHz is about 3.3 uf. What's actually used is in keeping with what's needed for a 16 ohm driver.

6) Shows what an amp actually sees on the midrange branch with the squawker. Electrically, both of these are extreme simplifications of the interaction, but my point is that the autoformer is used to alter the impedance characteristics of the drivers. Moreover, I think that if the tweeter branch of the network were 100% isolated from the autoformer, as it is in the conventional first order network, that there MAY be a slight increase in the perceived output from the driver. In that case, though, the tweeter would need to be treated like the 8ohm driver it is, which implies a change in the cap value.

In other words, even though the tweeter is on the top of the autoformer (tap 5) it may be down slightly from a direct and isolated connection to the input.

edit: an 8ohm amp is just used as an example. It could be a 4 ohm amp, which would also influence the power transfer factor. Note, too, that the schematic symbol for an iron core inductor (the autoformer) should have the two vertical lines to the outside. I just drew it without those.

[Erik Mandaville]

"Unless the 3dB down point is considered the "crossover point"."

This is associated with the point I'm trying to make.

[Erik Mandaville]

"Unless the 3dB down point is considered the "crossover point".

In other words, if 3dB down from a direct input is what is going on, how is that reduction being achieved? I don't see either a single series resistor or a series/parallel resistor combination L-pad on the tweeter in the 'A' network.

[Erik Mandaville]

I want to qualify all of this again who may not follow (or even care -- I respect that!) some of this, including my hand-scrawled diagrams. I'm trying to show what I see in my mind's eye.

The important thing I want to say: This issue about reduction of output between one kind of attenuator and another is purely an academic thing. I wasn't setting out to design a new network at all, just compare two that use different approaches to the reduction of volume from individual drivers. Reduction in terms of one type being a little lower down than another given the same input -- is simply that, a slight difference of attenuation. In no way shape or form does it mean one is better than the other. That may be the case on an idividual, subjective basis. Compared to the Klipsch design type 'A' that I made, and this other one I mocked up, I prefer the more recent of the two. I truly don't care which one is used; I'm just making a subjective choice between them. From what I've found, mine is slightly more efficient than the 'A,' which possibly makes it better, for me, for the type of amp I tested it with -- a low-powered single-ended 2A3. For amps with more horespower, and/or for those who listen at much louder volumes than I, it might not be the best choice.

Erik

[Erik Mandaville]

I'm ending this.

I went out yesterday to get resistors for both new squawker and tweeter L-pads, which I was going to set up to switch out together using a dual-throw, dual position switch, and couldn't find the values I needed. What was available was too far off for comfort. In any event, I wondered why in the world I would do this without even trying the amp with something that I know I could easily balance with the Lexicon. I already had one Teac 3-channel digital amp in the system that I was using for the center and surrounds, so I took off the Moondogs from the Klipschorns, and added the other Teac I have. I was remdinded again of what good amps they are, and it's surprising how using the same amp on everything can really improve the timbre matching between speakers.

There is not the slightest hint of grain or roughness that I heard yesterday, at least so far. Maybe I was listening louder with the Moondogs than I thought and something else was starting to happen.

It's done for now until I'm able to get the exact values I need for the band pass, but as it is, I think I'm so close in terms of both what I'm hearing and where I know the upper and lower limits are that I probably wouldn't be able to tell the difference. I've been listening very loud, and there is not a hint of stress or distortion of any kind.

The Lexicon really is an outstanding component, too. With the new center Heresy, I think this is the best I have had so far.

Erik

[Erik Mandaville]

I know it's not an elaborate setup with a nice big screen (that would be cool to have someday!), but it's much better than what we had in the past. The orange thingies are a new tennis ball that I cut in have to use as a temp. riser. The Heresy has made an important difference, overall. It would have upset things in the past to use just one, because they were being used as the surrounds. I've changed some things. On the crossovers, 4k Hz. it is unless something happens that sounds bad. With the Teacs, and I'm sure the Dynaco, too, no problem.

Erik

[NOSValves]

Erik,

Man you okay? You seem to be talking to yourself...............

Craig

[Deang]

!

[Erik Mandaville]

Craig:

Yeah, I'm fine. There are forum members who would rather not get bogged down in some of the unconstructive and really sort of useless elements of topics, and I have been sharing the development of this crossover for those who expressed interest.

I'm not done yet, though. I'm going to experiment with both a second and third order filter for the tweeter, and see/hear if a sharper cutoff might be better still for this lower crossover range. I generally haven't preferred higher order networks in the past, but this might be a case where it works. The rest as far as the bandpass and low pass will remain essentially the same.

Thanks for asking (wink!)

Erik