Help with Super 12 crossover.. please!

Compared to the diagrams it would be necessary reversed the polarity of the tweeter.....
Aldo: This is definetely not true! Changing from 8 Ohm to 4 Ohm has nothing at all to do with polarity.

double the values for the condensers
half the value for the choke

Supposed Your new 4 Ohm tweeter is otherwise completly identical to the 8 Ohm used before - that`s all.
Oops, sorry that I misunderstood You!:eek:
I know nothing about the speaker in which this crossover is used, so I can`t tell if there is a fault in the schematic.
And even with some more information about the drivers, without knowing their individual acoustical phase and amplitude response, still not yet it`s possible to judge this correctly.

What is the reason why the polarity should be reversed, in your opinion?
Here are the tweeters

Old 8 ohm tweeter: Audax TMO25F15

New 4 ohm tweeter: ATD RDT-25 (bottom of this page)

Note: I already have some good preliminary results. If you compare the response graphs of the two tweets (ATD is a PDF file), you will see that the ATD drops off much faster than the Audax from 5K and lower. So I merely put a 4.7 uF cap in series with the ATD for a 8.5K first order cutoff (original was 3rd order 8.5K) and it seems to work very well! Now with above comments, are you saying that I should reverse the polarity?
I have to go to bed now (here it is 5 AM already) but I`ll take a closer look tommorow.
But as it looks now You are going to change the speakers construction fundamentally and I believe it becomes hard to give You good advice from far away. Also 8,5khz fc appears to be very high.
What kind of speaker is this going to be at all??
Maybe You can provide some more information about it meanwhile, particular for the "woofer" or whatever this is.
Sorry I should have given more info. I can understand your scepticism. It's the driver (and cabinet) of the Hammer Dynamics Super 12 (, which is basically a reformulated Eminence Beta 12LT...

Full specs/graph of the Beta 12LT is on this page:

It's a psuedo full-range driver (wide-range?) that is 99db and goes from 45-10K. So the tweeter is being used as a super tweeter, hence the 8.5K cutoff.

Here's the original crossover...

I have another discussion going about this little project/experiment here...
Okay Brad, as nobody else jumped in meanwhile let`s continue:

After a closer look, I can`t see any good reason why You want to change the tweeter from the original design to the ATD and I`ll tell You why You should not in my opinion:

I think one very good reason (very likely this has been the main reason) why the Audax in this project has been choosed it`s high sensivity. Obviously another reason is because it is physically small.
The ATD which You are going to implement now has much lower sensivity and is physically larger.
The manufacturers specs given according the link You gave, don`t tell the hole truth about this driver. Here a link to actual measurements from a German DIY magazine (I trust them much more than the manufactorers specs):

ATD-Ribbon actual measurements
(I need this webspace soon again and will delete this page after a while, so please save it to Your harddisk )

From the graphs You can see that the sensivity is much lower (for reference: the peak in the response of the upper graph just hit the 90dB line @ 18kHz!) than the value specified by and shown at the graphs from the manufactorer (neat designer graphs in my opinion). The on-axis sensivity of the ATD in the range from 5kHz and above actually seem to be 5-8dB under the Audax!! The off-axis response is even much worse! That`s not a little thing moreover in regards that it seems that even the Audax can`t quite hold up with the full-range Beta 12LT what concerns sensivity. You mentioned at the Decware thread, that with Your modification the speaker "sounds smoother albeit a little chesty in the mids". That`s no surprise as likely the speaker is missing the top range now and probably lacks some resolution. And I doubt that this will change much even after You`ve finished the cabinet treatments finally.
Second, on the graph on the lower left You can see the distortion characteristic of the ATD (I marked k3 in blue, full scale range is 2%!, measurements were made at only 2,83V = ~2W). The distortions (k2,k3) is overall pretty high as You can see and k3 is even fast raising below about 7kHz.
Therefore Your idea of using just a first order highpass might not be such a good idea.
I wouldn`t worry too much about the ATD`s faster cutoff below 5kHz - this might even be of benefit as the notch-filter for the Beta full-range likely was designed for compensate for some raise in amplitude in the upper mids (due to beaming/ cone breakups). Also this might be the reason why the Audax is crossed steeper by the 18dB filter (besides of the benefit of limiting cone displacement, hence lower distortion).

As I said before I can`t say if the polarity should be reversed with the first order high-pass and the ATD. This would require doing some phase or at least amplitude measurements of the hole speaker. But simply try it. If You hear a big difference than there is only one of the two possibilities which is right. If the difference is not so big without help of measurements it`s difficult to predict wether polarity better should be reversed or not.

For doing a comparison between the original design and Your modification I would suggest directly comparing the designs by leaving one speaker unmodified and placing them side by side and hearing music which is switched to mono mode (if possible)and while manually switching from one speaker to the other.
While this might sound a little strange during the first minutes You`ll get used to this after a short while. I´m convinced this will be the only way which can help You to decide wether or not to implement Your modification finally.

While I really appreciate the efforts, the tweeter I have is a completely different animal than the one you point to in the link.
I believe this is a completely different model. The magnet structure on mine is MUCH smaller and is far, far more efficient than 87db. I have it hooked up now with a 4.7 uF cap and it plays at the same level, perhaps just a hair louder, than the original tweeter. I believe it is at least 94db. Yes, the baffle/face-plate will have to be ground down, but this is no problem. The link shows the basic specs and response graph. The performance I am hearing gives me no reason to doubt those specs.

Why I want to replace the original tweet is partly to try to get better dispersion (the original makeup has a smallish sweet spot) and partly pure experimentation. Also, I like things simple, and I think there is a chance for better performance with a more simple 1st order crossover - ie only one good quality cap like an Audio Note PIO or Hoveland in the signal path.

thanks! -Brad-
hmmm...when You say Your tweeter is at the same level or even somewhat louder Your`re very propably right and they have a different magnet structure.
As the basic structure and the size of the diaphragm otherwise seem to be exactly identical there is no reason at all why the dispersion characteristic should differ from Your version.
Actually the dispersion do not seem to be superior to the Audax but is much worse as You can see for the 30° off axis on the graph I pointed in comparison to the Audax graph for that corresponding angle. The graph of the link does not show amplitude response for different angles off axis at all - maybe for some reason...

The sweet spot You mentioned may be is based in the nature of the design itself, namely due to the very high crossover frequency and beaming of the big "woofer" and maybe isn`t the fault of the tweeter!?
Regarding sweet spot a first order high-pass (besides the other drawbacks regarding distortion) might make things even worse in the crossover range.

Also as the basic design of the tweeters isn`t any different unless the magnet structure, the distortion characteristic will likely apply to Your version also to a great extend.

Me too always try to keep things as simple as possible but simple isn`t always better - You should keep this in mind.
The proper application of the basic filter structures is far more important than a better quality cap or one more cap in series. Don`t get me wrong - I don`t say quality caps don`t matter and I don`t say that the 18dB filter is better. I just say You shouldn`t disregard trying other (to say more complicated) options because You are thinking simpler is better!
As the sensivity isn`t the problem anymore, You still might try the original 3.order filter setup (with adapted values due to the changed impedance) in comparison with Your first order filter. What I said in my first post regarding the part values with the changed impedance still applies. If You want to experiment You still could play somewhat with the values of the caps (increase).

Have fun in experimentation!
To clarify my previous posts, when I say "original tweeter" I'm referring to the Audax. Anyway, I don't think you can attribute any of the characteristics of the tweeter (and graph) that you were looking at - it's a completely different tweeter, regardless of how it looks.

One thing I know, there's no disputing the difference in response graph of the ATD from the original Audax. The Audax does not drop off steeply below 5K like the ATD does. I believe this was the reason for using the original 18db filter. Because of this I don't think the 3rd order filter is needed, or will work well, with the ATD tweeter.

Now that I have both speakers running the ATD with the 4.7 uF cap, it's definitely running hotter than the original. It's extremely more detailed, but this is often the case with a (too bright) presentation. I need to give further listening before attempting to drop the level (with a resistor?) or change the xo point with different cap values. In other words, time will tell if it's just a more detailed presentation, or it's un-naturally bright. I know it's all too easy to be fooled into thinking something sounds (overall) better just because it's more detailed. First listen had a big wow mental factor that needs to be overcome - if you know what I mean here.

I think I will tweek the ATD 1st order xo as best I can and then just compare it to the original tweeter and xo - and choose the one I like best. Attempting to get a 3rd order filter right with the ATD would take more time than I'm willing to commit, if it's possible at all.

thanks for your help/comments, more later. -Brad-
ATD tweeters

Btw, given all the different tweeter models from ATD that use basically the same type of technology, I think it's safe to say that there are various differentiating factors besides just the magnet structure. Diaphram material and thickness, and trace configuration are two that come to mond. An example of this is although the circular ribbon tweeter used in my Infinitesimals looks the same as my ATD tweeters, the circular ribbon used in a pair of Genesis IM 8300's I used to own had a completely different diaphram - ie, the traces on the mylar diaphram looked very different.
I know it's all too easy to be fooled into thinking something sounds (overall) better just because it's more detailed. First listen had a big wow mental factor that needs to be overcome - if you know what I mean here.
I know exactly what You mean - and it`s very good that You`re aware of these effects!:)

As it seems there are actually much more models from ATD than I was aware, which optical appearance is identical though the technical properties are not. Sorry for the confusion I might have caused eventually!:(

I think I will tweek the ATD 1st order xo as best I can and then just compare it to the original tweeter and xo - and choose the one I like best. Attempting to get a 3rd order filter right with the ATD would take more time than I'm willing to commit, if it's possible at all.
Brad, no offense intended at all but have You ever considered buying one of this inexpensive (relatively) PC analyzers + MIC there are today? I always wonder about people experimenting with high quality and often expensive DIY speaker stuff (and who want to do their own designs - not just kit building) or modifying existing designs and who do not have such a thing. Okay...I know, the machine can not replace ones ears of course but it would be of great help finding the right direction to go or at least to avoid impractical solutions. Many times I met people who claimed that they do not need this because they can do all this by their ears alone (what I question). After examining those peoples constructions somewhat closer more often than not they turned out to be some kind of really weird and by far have not exhausted all of the potentially qualities of their parts.
Since those analyzers cost not more than a good pair of drivers, this investment is the most profitable and useful You could do for a DIY speaker hobby, in my opinion !
moved xo to 10.5K

Further listening leads me to think it's a bit bright sounding with that 4.7uF/8.5K xo . I dropped the cap to 3.8uF for a xo of ~10.5K.

More listening needed to see if it made things better or worse.


1. They're both now a 2uF Solen parlleled with a 1.8 Bennic. Do you think any harshness/grainyness can be attributed to these caps and alleviated later with better ones like Audio Note PIO's or Hovelands?

2. If I keep the 8.5K xo but drop the level with an L-pad, will the sound suffer much as a result of the added resistors?
to question 1:

I can`t comment on the caps You mentioned because I never tried them. Moreover as I said in an earlier post I think what matters most is the value of the cap and the basic filter configuration (slope, steepness, L-Pad or no L-pad). I`m convinced that wether You use a 4µF cap or a 3µ7 cap will make more difference than the type of cap You use (preassumed it`s a high quality cap). Paralleling similar value caps as you did, always is good idea. I have also experimented with different high quality caps but I found that the differences to be very small but paralleling them had a bigger effect.

to question 2:

There is no reason at all (neither in theory nor in praxis) why an L-pad should introduce signal degradation (at least not for a flat impedance planar tweeter). Regarding the value, the same applies as for the caps - the value is more important than the resistor type. But stay away from those wire-wound types in the white or grey ceramic housing (they are inductive and have high tolerance up to +/-10%). What works very fine is to use many of this smallish metal film resistors (tolerance usually:+/-1%) which are normally rather used for general purpose electronics. Though they have only very low power rating (1/2 to 1W) You can use many of them in parallel. For example 10 each 10Ohm/1W metal film resistors in parallel make a fine (and cheap) 1Ohm/10W/1% resistor. Anyway if Your speaker sounds to bright now an L-pad will result in an improvement (assumed that an L-pad is necessary actually. It can easily be that another filter configuration or another cap value might do an even better job!?).>>>>>>>
>>>>And now again we`re at the same point again about a speaker analyzer:
Though the situation in taking the right decisions with the above questions is comparable simple, You can see how many options there are already (can You imagine how many options there are in doing a 3-way speaker and doing it without analyzer - You`re lost!). Defenitely an analyzer would be very handy in helping to quickly determine which configurations are worth further investigations (to say: to hear).

About Your question of recommending an analyzer package:
since I have stopped searching and comparing specs/ features/prices after I bought my analyzer a few years ago (still highly satiesfied with the performance) it`s sure that I`m not up to date.
How about open a new thread asking about speaker analyzers - would be very interesting and I´m sure many members of this board here have one and that You`ll be given good advice about what is worth to buy?
But here just a few arbitrary links that You get an idea about what I´m talking. As far I know the cheapest of those analyzers are about $350.
(DAAS3L+ is what I use - I plan to upgrade it to the windows version at some time)
You know that parallel caps topic is interesting. I thought there was concensus about the advantage of paralleling caps. Audience the maker of Auricaps (a very highly regarded and popular cap) does not recommend it. They say that multiple paralleled caps introduce different time-constants into the signal tending to blur the sound. This is surprising given it would be in their best interest as a cap mfg'er to push paralleling since that technique sells more caps - this is actually the reason why I am skeptical about paralleling.
From the Audience site...

Do not use bypass capacitors in the signal path. A single capacitor for DC blocking/AC coupling creates a simple path with one time constant. The signal quality will be compromised if a bypass or multiple bypass capacitors are added to a signal path capacitor. Bypass capacitors were used in the past to bypass low quality film capacitors or electrolytic capacitors. The bypass was the lesser of two evils. With the advent of better quality film capacitors the need for a bypass capacitor was eliminated. Bypass capacitors create multiple signal paths with multiple time constants. These time constants are very short but they can still be heard as a smear or overall loss of focus.