Reversed tweeter polarity

[martinbls]

Hi!

I'm in the middle of a progress here, designing a 2-way crossover for a 8" midbass and a 1" soft dome tweeter, only by listening (no measurement equipment at hands, unfortunately). I'm observing a strange phenomenon here: Some crossover designs call for a reversed polarity of the tweeter. Otherwise I get a deep notch in the x-over region and I can clearly tell which part of the signal comes from the woofer and which comes from the tweeter, as it should be. With reversed polarity of the tweeter everything falls into place, the mids are open, clear and the whole 2-way speaker sounds like one single fullrange driver.

So far, so well - but: Whenever I do this and I listen to some music for a longer time period I find something I dislike. Hard to explain, but when the polarity is reversed, despite of a great midrange and smooth blending of both drivers in the x-over region, I find the sound somehow, well, "greyish", particularly in the upper mids and the highs in general. That's strange and as said, very hard to explain, but I think the tweeter itself sounds different when connected reversed, which obviously seems to be not the case with many commercial or non-commercial designs where the tweeters are often connected reversed.

In my case it's a rather old Vifa D27TG-46-06, which sounds completely fine when connected with the same polarity as the midwoofer.

Does anyone have a clue how this could be explained?

Thanks a lot!
Martin

[5th element]

The lack of measurement equipment means you'll only be stabbing in the dark as to 'why' this could be happening. While it is true that with even order xovers, wiring one of the drivers out of phase to how it should normally be wired, should create a deep suckout, this relies heavily on the fact that you're crossover has been designed adequately in the first place. Often it takes a lot of fiddling about with asymmetrical slopes to actually get the drivers to be in phase in the first place.

If you've managed to arrive at a flat response with good phase tracking throughout the xover region without measuring anything then I would be quite surprised. With a less then perfect crossover reversing the tweeter polarity wont give you a noticeable notch whatever way you wire it, however altering the polarity will alter the frequency response, often putting lumps and bumps in different places and thus giving you a different presentation of sound.

[infinia]

Not surprised with a 8" 2 way , very difficult to get right even for the pro's. It takes a special mid-bass driver and low cross point in most cases.
publish your circuit and box alignment, nearfield applications can help some.
I'd venture to guess crossing lower with a 3rd order high pass section will give better listening.

[Nanook]

In particular if a 3rd order x-over is employed, reversing the polarity to the tweeter ccan help reduce a phase "tilt", regardless of the physical design of the enclosure (with "time aligned" offset or not).

A picture of your enclosure and the x-over circuit might help this. I suspect the "greying" of the sound may be due to this tilt, and any offset of the drivers in the horisontal plain.

You might have a look here to gain a little insight.

One of the first pieces of equipment that I had reviewed were a pair of loudspeakers that suffered from this same issue. All high quality parts were there (including Solen caps, good circuit board, good drivers and enclosures) but there was something very "odd" about the sound. They seemed very "spacey" (for lack of a better descriptor). I contacted the manufacturer, who said they usd an aysymetric cross over (3rd order on tweeter, 2nd order on mid/bass). After some thought and recalling I had read in the Loudspeaker Design Cookbook (by Lance Dickenson) that there can be phase issues with tweeters using a 3rd order x-over, I decided to wire the tweeters out of phase. Instantly the sound became excellent. These loudspeakers were a typical rectangular box, with no "time alignment" sloping baffle or anything of the sort.

[martinbls]

First of all: Thanks a lot for your replies!

Ok, here goes...

The speaker is from Reference 3a, a very early incarnation of the current model De Capo i:
Reference 3A Loudspeakers

The main driver is a hand made 8" carbon fibre midwoofer with stiff membrane and a large magnet, designed to not require any crossover at all. The tweeter in my early version is the Vifa D27TG-06-35 (I switched to the D27TG-06-46 because of a slightly softer sound), and it's crossed over just by a single 4,7uF cap and a zobel for impedance linearisation (22uF & 15R0). The enclosure is about 22l ported.

Here are some pics of my speakers. As you can see, the front baffle is slightly sloped to physically time align the drivers (at least to a certain degree).

[martinbls]

Now, what I think is that, at least in this old model, the midwoofer does have some resonances, and while the sound is incredibly open and fast, with some recordings the upper mids are a bit too excessive. That's why I decided to design a simple crossover, to "calm 'em down a bit".

Here is a measuring plot of the direct successor (MM De Capo, no x-over on woofer, single 5uF cap on tweeter), tweeter and woofer measured separately:

[martinbls]

I know, it's not exactly what you guys are used to, but the sound definitely has something to it!

I wanted to just smooth out the upper mids a bit, so I went for a single 0,47mH inductor and the 4,7uF cap for the tweeter, without the zobel. While the result is quite good and listenable with most recordings, on some recordings I felt there could be done more. So I measured impedance plots of both drivers, calculated a bit and tried serveral reasonable crossover variants, 6dB/6dB, with and without zobel for tweeter/woofer, 6dB on woofer, 12dB on tweeter and finally 12dB/12dB, with reversed polarity on the tweeter. Due to the smooth and extended response of the woofer and because I really like the sound of higher crossed midwoofers better, the crossover frequency was always chosen within the range of 3,5kHz up to 5,5kHz.

Pics of the impedance plots (measured with ARTA):

[infinia]

I'd need better specs on the woofer, the response curves w/o any data from the OLD baffle used, shows nasty peak below 1KHz followed by sharp dip. either way I don't think any reasonable 2 way Xover is the answer to that. Maybe if you like the wide open sound of the woofer by itself using a helper horn loaded tweeter crossed at 6K- ish would be one simple answer to help match directivity up there. other more complex Xover circuits would need more precision measurements.
so in effect going back to BSC inductor on the woofer and 1st order on a new HORN tweeter.

[jerome69]

The woofer looks to an Audax HM210C0 surround reversed.
http://www.audax.com/archives/HM210C...gue%201994.pdf
I recommend to try a 6dB on the woofer and 12dB on the tweeter.
Edit :
Same polarity of the drivers
Woofer L=1mH Tweeter values recommended by Vifa C=8.2u L=0.47mH.

[Nanook]

I dug out an old review of these....

Apparently the tweeter is a Phillips one, unless 3a changed it during production (the review was from 1990).

Here are a couple of recommendations that are based with general experiences and not specific to 3a:

the mid/bass: If the woofer resonates, try tightening the screws or use hex headed thread inserts and the corresponding correct screws. You can also take some time and dampen the mid/bass frame, etc. If wanting to improve the sound and control the woofer a little more, you might try stuffing a bundle of drinking straws into the port(tightly). This may or may not help, but it is a very cheap fix if it works, is easy to do and is completely reversible. You can also add a support from the rear of the cabinet to the back of the driver that puts some force on the magnet.

the tweeter: Not every tweeter can be used at a low frequency. Get an exact replacement or seek technical advice from 3a, and ask for a recomendation if the old tweeter is not available. It may look like a Vifa (and in the pictures within the old review the tweeters do look like Vifa), but check with 3a, based on your serial numbers.

Apparently there is no x-over to the mid/bass drivers and only a cap to help keep the tweeter intact.

Please do not take this as an attack: the problems you are having could well be attributed to taking something that works brilliantly due to its simplicity, and complicating it enough to ruin the work of the speaker designer. If you want to work at getting a new cross-over developed, I would start with the purchase of a Behringer DCX2496 or mini-dsp electronic crossover. Then add whatever mikes and measurement system and go from there. The true "art" of loudspeaker design is in the cross-over (I think) and is often minimized by some of us "DIY types".