my first attempt.... Usher 8945P + Peerless HDS tweet

[peterS]

Advice and comments needed

using this driver: http://www.partsexpress.com/pe/pshow...96-602&scqty=2

and this tweeter:
http://www.madisound.com/catalog/product_info.php?cPath=45_229_270_324&products_id= 1744


I do not own them yet so I am relying of Zaph's measurements

http://www.zaphaudio.com/

subsequently, the Usher has a 4db dip @1.25K and a small resonance @2.5k

the Peerless a slight dip at 6.5K

here is what I have come up with
note:
-the transfer function is electrical not acoustic
-the attenuation of the tweeter is a guess and I will fine tune it by buying several resistors
- the low pass filter has a high Q to compensate for the dip in frequency response @1.25k
-I could not successfully adjust the frequency response @6.5k so I gave up
-The box will be .75ft^2 ported at @38hz and I was planning on offsetting the tweeter 1"

-most importantly I have no idea how these will sum together

ops:

[ThomasW]

If one hasn't arrived already, I'm sure you'll have an email about this design very soon....

[peterS]

that bad?

[ThomasW]

If you haven't received an email it would appear Mr Kim or his proxies aren't lurking here as much as they have in the past.

Checkout this thread for one person's take on this design. If you want to discuss that design please so on the DIY Audio forum....

http://www.diyaudio.com/forums/showthread.php?threadid=108054

[Kevin Haskins]

I use the Peerless HDS and for my preference, I ended up shelving the top-end. It has a rising response, from about 8K-15K which I found too forward. I ended up dropping it a good 2-3db on top. Sounds very smooth now but it took a lot more crossover work (and components) and voicing than I would have preferred.

The 6K dip is too small to solve with the crossover. It diminishes in magnitude off-axis so its not anything to be concerned with.

You need measurements to do good design work so if you don't have that capability, plan on acquiring it.

[peterS]

I use the Peerless HDS and for my preference, I ended up shelving the top-end. It has a rising response, from about 8K-15K which I found too forward. I ended up dropping it a good 2-3db on top. Sounds very smooth now but it took a lot more crossover work (and components) and voicing than I would have preferred.

The 6K dip is too small to solve with the crossover. It diminishes in magnitude off-axis so its not anything to be concerned with.

You need measurements to do good design work so if you don't have that capability, plan on acquiring it.

I do
thanks

[peterS]

If you haven't received an email it would appear Mr Kim or his proxies aren't lurking here as much as they have in the past.

Checkout this thread for one person's take on this design. If you want to discuss that design please so on the DIY Audio forum....
http://www.diyaudio.com/forums/showt...hreadid=108054

ok thanks

[Kevin Haskins]

Here are the off-axis measurements, you can see the magnitude of the 6.5K dip go away. The on-axis is the yellow and all the baffle diffraction related issues are most obvious in that measurement. I don't have a perfect jig for off-axis data so there was a little bit of difference related to the slight change in measurement distance.

The crossover is around 1.6K and the Peerless has no issues @ higher SPL with that and a 4th order acoustic solution.

Image not available

[mazurek]

Kevin Haskins, can I direct you to my other post, I'm wondering what you think. I've been trying to make my design with the HDS tweeter perfect for quite a while. I'm wondering if the rising 2nd order at ~7k is what makes the tweeter sound a little hot, even off axis (or just that I am used to more rolled off off axis).

https://www.htguide.com/forum/showpost.php?p=376059&postcount=89

[peterS]

Here are the off-axis measurements, you can see the magnitude of the 6.5K dip go away. The on-axis is the yellow and all the baffle diffraction related issues are most obvious in that measurement. I don't have a perfect jig for off-axis data so there was a little bit of difference related to the slight change in measurement distance.

The crossover is around 1.6K and the Peerless has no issues @ higher SPL with that and a 4th order acoustic solution.

Image not available

Wont it still be an issue if I offset the tweeter? Do you think it would be better to not offset ?

[Kevin Haskins]

Kevin Haskins, can I direct you to my other post, I'm wondering what you think. I've been trying to make my design with the HDS tweeter perfect for quite a while. I'm wondering if the rising 2nd order at ~7k is what makes the tweeter sound a little hot, even off axis (or just that I am used to more rolled off off axis).

https://www.htguide.com/forum/showpo...9&postcount=89

Na.... assuming your distortion measurement is good its 40db down and its second order, not likely to sound bright.

Your brightness is because of the slight rising response of the Peerless. It also presents itself off-axis. It took a fair amount of effort to get something that worked for me. Take a look at the off-axis @ 1.5M. Voicing has as much to do with off-axis as it does with on-axis (probably more). If you have a dip off-axis at crossover then a long wide plateau from the tweeter response up to 10K it will sound very bright, no matter how flat the on-axis response measures.

Here is my electrical solution.

[mazurek]

Thanks for the tips and sharing your transfer function. I've been optimizing my transfer functions using spatially averaged (about +/-45 horiz, +/-15vert) measurements at about 1 meter. I've played with high frequency shaping similar to that shown in your transfer function. One big difference though is that I have a notch one my tweeter for what I assume was baffle diffraction. The diffraction notches on my tweeter and woofer could contribute to the phenomenon you were discussing (dip then plateau=bright). I'm curious if the diffraction pattern is not fully evolved at 1m, quick measurements do show some difference but by that distance everything is hazier (room). Perhaps is is just a matter of on vs. off axis equalization. Thanks for your help.

[peterS]

I think im going to have to abandon this.... that 1.25Khz dip seems impossible to rectify w/o using 4th order....
the cost of that other design's passive nears the cost of the difference in price of the scanspeak and usher
I have been able to design a XO that should improve upon the other design but Im not willing to invest the money

[augerpro]

I would listen to them before you abandon the whole project. A small dip at 1.25k is pretty small beans when compared to all the factors that can help or hurt the sound quality.

EDIT: I see you haven't bought these drivers yet. I would still move ahead with the project if it were me. Two things to consider: you could model a baffle that compensates for this dip, and if you're like me, frequencies in teh 1-2k range can be very irritating if there is any bump there. I bet you won't even notice a dip for that matter. I sure wouldn't write off all the other positive qualities of the Usher just because of a dip at frequency that makes me cringe anyway.

mazurek: here is the transfer function for the newest XO I developed for my speaker using the 810921



Pretty much like Kevin's. I also found massaging the level around the XO of 1750 Hz helped a lot too, although as I mentioned above i find that particular frequency range annoying anyway.

[mazurek]

Thanks for all the tips and transfer functions, after a while I start to psych myself out and get annoyed whenever I hear any high pitched sounds whether they are right or wrong.

[Kevin Haskins]

I'd not try to do too much with the baffle diffraction. As you go off-axis you can see most of the diffraction ripples go away. Just don't optimize around one measurement.

What I do is get on & off-axis data and then I design a crossover with the on-axis data. Sometime during the design process I swap the FR off-axis with the on-axis data and I look to see what my solution looks like modeled off-axis. Then I build one and re-measure both on & off-axis. I focus on the horizontal off-axis first and only after I have a solution that is pretty close to done do I look at the vertical off-axis.

Another tip is to work with smoothed data. A passive crossover can only give you down to around 1/6th octave adjustments so trying to adjust for narrowband notches or peaks is an exercise in frustration. Its also fair because you will find our ear does some natural smoothing. As Augerpro has said you won't miss the dips so favor a response where the large peaks are brought down and leave the dips.

You will notice moving off-axis as little as 10-15deg helps a great deal with the diffraction related stuff. I find that leaving the on-axis with a couple bumps is MUCH better than trying to get it perfectly flat at the expense of the off-axis response. If it bothers you, just listen off-axis 5-10 deg. :W

A dip @ 1.25K is probably a surround related resonance in the driver. I've not looked at the response of the Usher woofer but you could cross the Peerless HDS low and steep. You can probably go as low as 1.4-1.5K with a steep crossover. It will cost more, especially on the woofer but it may be a workable solution if the problem is ugly.