RS4 - Digital 4-way XO, Long. Lam, Dayton Reference, Adcom System

Status
Not open for further replies.
The speakers are coming along very nicely. I hope the chamfering for the RS100's was enough - that baffle still looks pretty thick.

When you get to the XO development stage - check out implementing the quasi transient perfect Harsch XO. You may have the perfect set of drivers to do that. The sound realism will be much better than a symmetric Linkwitz Riley.

In a nutshell, the Harsch XO applied to a 4 way are the following electro-acoustic target functions:

Woofer: BW4 LPF (~100Hz)
Mid bass: Bessel 2 HPF (~100Hz) + BW4 LPF (~700Hz)
Mid: Bessel 2 HPF (~700Hz) + BW4 LPF (~3.5kHz)
Tweeter: Bessel 2 HPF (~3.5kHz)

All drivers positive rising leading edge polarity.

Then set delay of each driver above at half the period of the XO frequency of that driver, working from woofer up (this includes acoustic offset). Do this in PCD or Xsim first.

More info here:

http://www.diyaudio.com/forums/multi-way/277691-s-harsch-xo.html

Here is the response and phase of a Harsch XO, do this for each XO section:

494790d1437489518-filler-driver-ala-b-o-harsch-xo-plot.png


You can then get a step response that approximates a right triangle and phase is flat except for 55deg bump at XO point.


Xrk971great info thank you! With the miniDSP these types of things should be very doable to test with, will definitely be looking into that XO type more thank you =)


Sent from my iPhone using Tapatalk
 
Ok final update on the speaker build I believe as they're done!

Last night I installed all drivers and finished wiring

Wiring for subwoofer

An externally hosted image should be here but it was not working when we last tested it.


Wiring for midbass

An externally hosted image should be here but it was not working when we last tested it.


An externally hosted image should be here but it was not working when we last tested it.


Wiring for mids which are in parallel

An externally hosted image should be here but it was not working when we last tested it.


Tweeter wiring
An externally hosted image should be here but it was not working when we last tested it.


And...the final result! The sense of satisfaction is high once you finish building some speakers, what was once only in your mind manifest! I spent about an hour just sitting and staring at them, taking in every detail that at one point was just a concept and to see how similar or different it turned out once finalized.

3ee66bd22e77eba66a8de5e1d442324a.jpg


Final detail on terminal block

An externally hosted image should be here but it was not working when we last tested it.


An externally hosted image should be here but it was not working when we last tested it.


Thanks for all the interest and support everyone!


Sent from my iPhone using Tapatalk
 
Last edited:
When you get to the XO development stage - check out implementing the quasi transient perfect Harsch XO. You may have the perfect set of drivers to do that. The sound realism will be much better than a symmetric Linkwitz Riley.

I second this, even though with my DSP it didn't work well (as you can read in the last 4 pages of the Harsch thread). Don't forget to also add delays for differences in pathlength to the mic due to driver spacing and acoustic centers.

I'm now using linear phase FIR filters to get at the same transient perfect goal. With your miniDSP it should be easy to switch between profiles and you can hear the difference immediately, well worth the effort in my opinion.
 
Finally finished the speaker wires, it is a complete harness made from 8 pairs of twin conductor 16 gauge copper wire

An externally hosted image should be here but it was not working when we last tested it.


9e587044cb768e1a10eb3ab2de3b0f80.jpg


3b9e904d97dc0e39cc3e43b6e9259d6f.jpg


Been playing with the miniDSP and REW, plan on doing more with a friend this weekend, learning a lot and enjoying tweeting.

These are not real bass heavy, I don't think I'll have trouble making them flat to 30hz, not big bass monsters by any means but the bass reproduction is very tight and accurate.

They sound very good with jazz and any type of live or acoustic music. On good recordings the sound stage is so fun to experience, to close your eyes and pick out the locations of different vocals or instruments.

Hope to post some graphs soon.


Sent from my iPhone using Tapatalk
 
Ok finally made some good progress with the miniDSP and REW, really pleased with how these are sounding! My friend Brian came over today who is a miniDSP whiz and we spent an hour or so running measurements and making some adjustments.

One of the biggest improvements we made was delaying the left subwoofer 6ms which significantly reduced some in-room dips we were measuring. There is a large narrow dip at 120hz which we believe is the Allison Effect, if anyone has any suggestions on that I'm all ears.

Frankly these things are really flat above 100hz and I actually don't have any EQ'ing going on other than the subwoofer. After making some adjustments to the tweeter above 10kHz I bypassed everything and they just sound really good.

Bass is accurate, not boomy at all, flat down to 20hz essentially, I have a subsonic filter set up at 23hz with 48db/octave. They easily reach 110db, I haven't really cranked them up all the way, even at those levels the Adcom THD warning lights never come on.

Here is an in-room response graph as the speakers sit:

fetch


Thanks!
 
Some great tips on how to adjust your EQ to get the most out of your speakers is in the Two Towers thread.

Do a measurement first at 0.5m to reduce the effect of the room. In the IR button, choose "Frequency Dependent Window" and pick 1/6th octave. This basically shows you what is correctable from the speaker and use this as the basis of broad smooth EQ corrections in the PEQ section. Correct it here and when you look at the non-FDW gating, it will be quite amazingly flat.

Look at at with 5dB per major division - you are at 20dB per division and have huge variations. These drivers can do much better when EQ'd. Also, EQ with cuts and avoid using boost. Don't use any EQ more than Q=2.5.

Show us a XO plot - each driver in a different color and the summed response. That will tell you what the balance is.

It's fine to tweak on the fly in real time to get the feel of it. But the best way (and actually most time efficient way) is to measure each driver running full range with speaker and mic untouched. Then measure combinations of each driver with another. The most important is the tweeter+mid, mid+mid bass, mid bass+sub, and maybe tweeter+mid+midbass. Convert all of these to minimum phase FRD files and import into a XO simulator like PCD which has a DSP PEQ simulator.

Adjust the offset in Z in the sim to get the combined sum response from the sim to match the measured one exactly. That gives the acoustic offset. Now you know the sim program works.

Then use the PEQ and XO filters in PCD to get the curves to match the textbook slopes. In this case, I suggest the Harsch XO which requires an acoustical 2nd order Bessel high pass on the tweeter and 4th order BW low pass on the mid. Then 2nd order high pass on the mid and 4th order low pass BW on the woofer. And the woofer high pass to be 2nd order Bessel and the sub low pass to 4th order BW. Start from the sub and woofer and add a delay to the higher freq driver equal to half of the period of one cycle at the XO frequency.

If you get this right, you will have a quasi transient perfect time aligned system with great realism, and it will probably be flat within a few dB over the whole range, except for room cancellations and reflections. Like the 120hz - probably a floor bounce cancellation.

Good luck!
 
Some great tips on how to adjust your EQ to get the most out of your speakers is in the Two Towers thread.
sion - you are at 20dB per division and have huge variations. These drivers can do much better when EQ'd. Also, EQ with cuts and avoid using boost. Don't use any EQ more than Q=2.5.

Good luck!


Thanks for the advice and info, definitely will be doing more of this and focusing on flattening out the curve higher up and isolating drivers, will keep you posted.

Thanks!


Sent from my iPhone using Tapatalk
 
Took some nice pictures today, really enjoying these and still playing with tuning. They are incredible with acoustic music, sound good with rock and pop but not great, planning another project for something that will really play loud with impact.
ae652c3f3a9d01559cbf369950ca19d2.jpg
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
29895d93e3cbad338f065201c4204c71.jpg
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
b4b871633b7f7a61d64d974bb7bb8b97.jpg
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.



Sent from my iPhone using Tapatalk
 
Status
Not open for further replies.