A 3 way design study

This plot I got from hificompass. normalized nearfield for the same driver. It shows a deep notch around 1kHz similar to mine.
Maybe I am not explaining myself very well, it shouldn't be a surprise that the surround resonance occurs at the same place as the nearfield nulls do as they are related to the size of the wavelength of the piston's diameter.

Here is a graph from Don Keele's nearfield paper showing the nulls. The same ones appear in your and Hificompasses measurements as expected.

Nearfield nulls.png
 
Round-2 box measurements

Now I put both the woofers in the box and took each woofer's impedance measurements. All holes in the box were sealed.
Here is the pic of the setup:
1655549259086.png


Here are the impedance plots for the top and bottom woofers in the box.
1655549365332.jpeg


Looks like the impedance peak of the top woofer is higher but both woofers have the same width for impedance peak.
The top woofer box tuning frequency is about 45.6 Hz and the bottom woofers tuning is around 46.3 Hz.
Overall, the values look close. But is there something that needs to be done about the height of the impedance peaks, to bring them approximately to same level. If so what can I do?
Does the 0.7Hz difference in tuning frequency matter?
In general are the above things worth loosing sleep over/digging deeper into? 😀
 
Have you got the free air impedance sweeps for both drivers to compare ? It isn't that unusual for drivers to have variations in parameters.

You can try to break the drivers in by running low frequency tones through them for a while to loosen up the suspensions.

Nothing there to lose sleep over. Are you going to power them separately or run them in series or parallel?
 
  • Like
Reactions: vineethkumar01
To round out the measurements, here are the impedance and nearfield frequency(baffle step not adjusted) responses of both woofers overlaid.

There was a slight variation of about 1dB difference in sensitivity between the two drivers at the moment i think. These two woofers were from different production batches and were purchased almost 1 year apart.

1655551637145.jpeg


In above plot, i offset the level of one of the woofers by about 1.3 dB or so to get the above frequency reaponses. i think there is also a difference of a few mm between the two woofers centre and the mic when i took frequency response measurements.
Impedance measurements were taken using dats v3 as usual.
 
Have you got the free air impedance sweeps for both drivers to compare ? It isn't that unusual for drivers to have variations in parameters.
I had taken the free air impedances of all 4 woofers. But i didn't take them today. I will try to take them/find the existing ones and post here later 🙂
You can try to break the drivers in by running low frequency tones through them for a while to loosen up the suspensions.
Ok. This I will try to do.. and then take measurements again.

Nothing there to lose sleep over. Are you going to power them separately or run them in series or parallel?
I am planning to have a separate dsp channel and amp for each driver. 😀
Is it a good approach?
 
If you have them available it won't do any harm 🙂
I thought the problem with series connection is that i will loose some sensitivity. With parallel connection, the minimum impedance will go around 2.5 ohms. My amps will never handle below 3 ohms.
So i thought to go with separate amps and dsp channels for each driver. I am planning to setup 8 channel dsp.
I thought this will also give the freedom to play with different configurations like these woofers in a 2.5 system in addition to the 3 way configuration along with a horn top and other top modules .. 😀
 
  • Like
Reactions: fluid
Round-3 assembling full system and making it play before taking polar measurements due to curiosity 😀

So i assembled the entire thing, paralled the woofers per cabinet, used the prototype crossovers i had developed for the foam box system and finally made it play as a 2 way system (SB audience Rosso 65CDN-T on Faitalpro LTH142 horn as the top module).

Here is me and my current system. The 6 channel amps are yet to arrive for 3/2.5 way crossover.
IMG-20220619-WA0002.jpg


I have no words.. It is so good 😀 Just happy even with the way it is playing even now.. 😀
If the way it is currently playing is any indication of how much good it can be after proper measurements, and crossover based on it, I have succeeded in this personal pursuit to create a proper stereo system that suits my tastes.. (until the next better system comes along) 😀

Ecstatic today. 😀
I will try to take proper measurements in coming days and refine the entire system. Be it crossover-wise, placement of the speakers wise, and in as many aspects as i can to improve the overall system. 🙂
 
fantastic work Vineeth!

From down under I can feel the joy and quietly delirious excitement that words cannot quite capture.

It’s good to see a thread, from conception to completion. And see you learn and grow and now enjoy.

And yet we cannot hear what you hear!
The next revolution in audio must be this, coming soon, surely.
 
  • Thank You
Reactions: vineethkumar01
But I can't see any benefit either. Unless you need extreme spl levels.

I run my 2x8" SS 21W 4R connected serial, with Hypex FA123
https://www.diyaudio.com/community/threads/avalanche-as1-modernization.321711/#post-5411266
Juha,

It very much depends on the amplifier’s voltage swing capability, in 8 or 4 ohm nominal loads.

excursion usually limits SPL at the bottom octaves, particularly in vented boxes. Voltage/power handling usually limits it in frequencies that are not displacement limited, usually above 200Hz-400Hz in most woofers, and 400-800Hz in most midranges.

One amplifier I use clips around 400/4ohm, 580W into 2 ohm. And so i partnered this with dual XXLS 12” P830845 connected in parallel; for total 3 ohm min impedance.

By using a sealed cabinet of 120L, the cone excursion meets but never exceeds the 13mm Klippel verified x-max. After baffle step correction, Net 4pi sensitivity is 93dB/2.83V, and F3 is 40L with a 12dB/octave roll off below, suiting room gain of medium to large rooms.

One could call it “extreme SPL” because the excursion limited SPL at 40Hz is about 109dB; and the rest of the system is about 113dB per speaker, anechoic, from anywhere above 50Hz.

But what it sounds like is incredible dynamic- from pianissimos (0.01W) to fortissimos (200W).

It’s like going out for middle of the theatre movie experience, except at home, with music! And it can play Ricky Lee Jones- Ghetto of my Mind, and amps never clip, the drivers never exceed x-max.
 
Last edited:
  • Like
Reactions: vineethkumar01
So i did this today to start figuring out what the system is doing currently and before i get time to do full measurements for individual drivers. 😀
IMG_20220622_194538.jpg


On the stand is minidsp umik-1. on the floor are probably 20cm thick acoustic wool and on top of it are 3 sheets of basotect melamine foam.
Mic stands at a height of 103cm from floor and aimed at the centee of the compression driver and at a ditance of 130cm from baffle surface.

Here is the on axis single speaker frequency response (with current crossover) i got with a gate time of 10ms
SYSTEM_ON_AXIS_FREQ.jpg

Here are the time domain plots
SYSTEM_ON_AXIS_time.jpg


Is it worth shelving down the entire response by about 10dB so that it becomes flatter till about 15kHz?

I will try to do polar measurements anyway during next week or so but wanted to get an idea about the kind of on axis i should aim for..

Also here is the previous prototype xps foam box based full system responses
1649911379760 (1).png


The current on axis response shape looks pretty close to the previous one above.. 😀
 
OK so this is where you can go off track a bit, I think. Are you trying to measure at 1.3m to get a feel for what you may be hearing? Or checking to see if your crossover for the XPS is applicable for this cabinet? Because I think you can be mislead. Remember that on-axis is just one aspect of overall tonal balance. It's an important part of the puzzle, but only one part. It's probably tempting to measure and see what you're getting.

What's probably more instructive is to procced and measure the polars, like you need to do for your crossover design. You know, Hor 0 out to 180 degrees in 10 degree increments, and a few vertical, at least +/- 30 degrees for the listening window, then input them in VituixCAD. and look at the output from the crossover. VituixCAD is VERY accurate in it's crossover modelling (as long as you have accurate measurements- as Erin Hardison has shown). So now if time to get that analogue mic and go dual channel for measuresments, the way recommended by Kimmo-
https://kimmosaunisto.net/Software/VituixCAD/VituixCAD_Measurement_REW.pdf

Once you have good data, you can simulate and see, to your hearts content, and since you're using DSP you can quickly prototype your crossover and listen to the changes.

My own preferences is to aim for close phase alignment at the crossover region, maintaining good sensitivity (don't knock off 10dB!), and getting a good tonality score (preference score 7+). I've found that there's trade-off that you'll have to choose from, eg. on axis vs listening window vs predicted in-room response, but real in-room response can differ significantly from the predicted, so I aim for on axis and listening window.

Remember that we can all say that we aim flat and smooth on axis, but remember that just +/-1 dB can make a BIG difference, particularly over a wide band. Like when you turn your volume control up or down 1dB- it's definitely noticeable.

So in practical sense, there's no perfect on-axis response. Well there is, but we don't have the technology to have it at the +/- 0.01dB like we can have with DACs or amps. So in the meantime, we do our best and set our levels correctly. . When working with crossovers I end up using a +/-5 dB scale... I really want to see what's going on. A 40dB scale is the difference between 0.1W and 100W, right?
 
  • Thank You
Reactions: vineethkumar01
@tktran303
Thank you.. 🙂
There are definetly good pointers for me as to what to aim for as target system frequency responses in your reply.

I am definitely going to do polar measurements using my dayton emm6 mic with loopback before finalizing the crossover on this specific build.

I took this 1.3m measurement partly out of curiosity and partly because of the following horn response related doubt that i had.

For an almost exactly same configuration of this horn (kept on top of a 30ish cm wide cabinet in the previous xps foam prototype build (freestanding?) ), These are the horizontal polar responses out to 90 degrees i got for the horn with just a 150uF capacitor in series with it.
1649611440130 (1).png

In the above plot, we can see that the horn's response droops down slightly after 4kHz eventually ending up about 10dB below the level at which it started around 4kHz.

Even in this build I will be crossing the woofer to CD+horn somewhere around 1kHz. So unless i do something to shape the overall response of the horn, i am probably going to get similar horizontal polar response for the overall system above 1-2kHz.
Hemce the doubt about whether i need to shelve down the horn response such that it looks flat from say 2kHz to 14kHz.
Also the compression drivers sensitivity is 109dB/1W/1m.
The woofers even in parallel are going to give me at the most 88dB +6dB = 94 dB/1W/1m
 
Last edited:
Isn't this 2-way just a temporary stage? A sealed bass box is the most simple thing of a 3-way.
Yes this 2 way is a temporary stage till my 6 channel amps arrive.. 🙂
But even once it arrives, the first configuration i want to experiment with is a 2.5 way where the bottom woofer is highpassed and slowly rolls off around 350Hz while the top woofer goes and meets the horn at around 1kHz.

Somehow i cant get over this compression driver.. The kind of details it pulls out from music and the kind of ear splitting levels it can reach with ease from the puny class D amp i use with it is something that i have never experienced before.. 😀

My 2 way boxes with 5inch mid and 1inch tweeter will take some time to get ready. Once they arrive, I will have a proper 3 way where both the woofers are high passed around 300-350Hz and the mid driver takes over from there. 🙂
 
  • Like
Reactions: Juhazi