There was no .zip file, only a .mdat file (file format of REW) with the name "mid-twt-0-60.mdat". And although it is technically possible, it did not contain any impedance measurements.
Best regards
Michael
Measurements would be much better, because they have to be imported in VituixCAD as well as the frequency responses to achieve accourate results.
Best regards
Michael
Of course, it turns out I never used REW's impedance measurement feature because I've had a Dayton DATS for years & even though I measured & saved data on multiple drivers, I never used it to develop crossovers as my projects have all been active for a decade.
Newly measured DATS impedance data in txt file uploaded.
ps -- memory at 3am was faulty: the GRS planar driver is a flat 3R, the NE123 mid a bit more complex.
Newly measured DATS impedance data in txt file uploaded.
ps -- memory at 3am was faulty: the GRS planar driver is a flat 3R, the NE123 mid a bit more complex.
How does a *.mdat file work? Is it some kind of compressed file which contains frequency response measurements?
In order to take a look at these measurements, I will need frequency response curve files, in a standard *.frd or *.txt format.
In order to take a look at these measurements, I will need frequency response curve files, in a standard *.frd or *.txt format.
It's Room EQ Wizard's native file format. Quite comprehensive -- from as single sweep, lot of data is extracted.
I'll see about exporting to other formats. In about 15-30min.
I'll see about exporting to other formats. In about 15-30min.
This is what I got for the midrange driver 0-axis
This is what I got for the tweeter 0-axis
Please check to make sure this looks like your data. ... i.e. confirm that I extracted it correctly.
This is what I got for the tweeter 0-axis
Please check to make sure this looks like your data. ... i.e. confirm that I extracted it correctly.
Since the system uses a very capable DSP, the passive crossover is not required to do much or any EQ or voicing. You can adjust that later based on measurements.
The 5.5k peak in the tweeter response looks suspicious, as others have pointed out. It could be a measurement error. There is also a 5.5 k peak in the NE123 response, and this could be a coincidence, because that is exactly where a 12 cm cone driver would probably have a peak. On the other hand, it could be part of the same measurement error.
If you have a chance to repeat the measurements, it would be nice to start them at a lower frequency... 50 Hz for the mid, and ~ 600 Hz for the GRS tweeter. It would also be nice to verify that you are using a time windowing properly so that these are truly quasi-anechoic measurements. The large oscillations from 1k - 3k in both driver measurements leads me to believe these are non-windowed scans.
But assuming these are valid measurements and assuming you can use the DSP equipment to flatten the response out, it is possible to come up with some passive crossover options.
j.
The 5.5k peak in the tweeter response looks suspicious, as others have pointed out. It could be a measurement error. There is also a 5.5 k peak in the NE123 response, and this could be a coincidence, because that is exactly where a 12 cm cone driver would probably have a peak. On the other hand, it could be part of the same measurement error.
If you have a chance to repeat the measurements, it would be nice to start them at a lower frequency... 50 Hz for the mid, and ~ 600 Hz for the GRS tweeter. It would also be nice to verify that you are using a time windowing properly so that these are truly quasi-anechoic measurements. The large oscillations from 1k - 3k in both driver measurements leads me to believe these are non-windowed scans.
But assuming these are valid measurements and assuming you can use the DSP equipment to flatten the response out, it is possible to come up with some passive crossover options.
j.
Non-windowed. Yes. In REW, time window can be applied after the measurement.
But I can redo them tomorrow morn & repost. Export to TXT format.
Thanks, Jim!
But I can redo them tomorrow morn & repost. Export to TXT format.
Thanks, Jim!
I'm very late to this. The idea of using a hybrid crossover, with the tweeter/mid passive is something I've seen before. Given my recent experiences measuring off-axis I would not do that. I'd put the passive stage between the mid and next woofer down. Of course, this means you are crossing at a lower F and require larger drivers. My reasoning is thus: Keeping the mid/tweeter in the digital domain makes it easier to time align and user LR4 filters where they are needed the most. Having higher order filters results in excellent vertical as well as horizontal coverage. This is a lot less critical in the lower octaves, and you can achieve nearly equal off axis response with lower order filters.
The trouble with that is the combined power requirements for 250~3000 Hz is far greater than 1200 Hz on up. The amps are 6x150W + 2x300W for the subs. Also, much greater complexity as the lower mid needs both low & high pass.
When the sub(s?) are introduced to the system later, it will become a 5-way, with crossover points at 70~80Hz, 250Hz, 1200Hz, 3kHz. The last one is the simplest, lowest hanging fruit to turn passive in order keep the number of amp channels to 8.
On another note, continuing the transition of these lx521 clones...
I set up the newly modified speakers in my son's place with his help last night. Overall, the sound quality & presentation isn't much changed, based on admittedly unreliable acoustic memory: the spacious dipole effect is there in spades, detail, dynamics... But there's added bass punch & depth, tho sometimes subtle. It's in the acid test of high volume modern rock/pop that the modified speakers + more powerful amp pass with flying colors. At over 100 dB peaks, the room shakes with deep bass, the impact hits in the gut, without any sign of woofer over extension or amplifier clipping. The smooth ease at lower volumes isn't quite maintained but it's not bad. I'm half surprised it's worked out this well; I had visions of a driver or amp module burning out dramatically in this test. Yeah, it was that loud.
A good sub or 2 will help the system stay more composed at such levels and provide better bass.
Funny how these speakers, designed mostly to be neutral high performance HiFi, are doing double duty as a near nightclub DJ rig. 😲 I keep my fingers crossed that all the components are robust enough to do such heavy duty well for good while.
In case I wasn't clear, I did not implement the passive crossover between tweeter & high mid that Jim kindly developed for me. Didn't have time because Dave was missing his system & wanted it back. (He lives in Vancouver & I'm on an island 50km away. Getting back and forth with heavy cargo is only practical by car & ferry and the limited winter schedule of the latter makes logistics challenging.)
But the passive will be implemented when I build & integrate a sub into the system. The 2x300W will then be bridged for 600W mono <80Hz to the sub, and the current woofers will be run on a 2x150W module.
This too is a trial -- I hope the bridged amp is capable enough for the sub, and that the 2x150W is enough for the woofers that become midwoofers. 🤞
But the passive will be implemented when I build & integrate a sub into the system. The 2x300W will then be bridged for 600W mono <80Hz to the sub, and the current woofers will be run on a 2x150W module.
This too is a trial -- I hope the bridged amp is capable enough for the sub, and that the 2x150W is enough for the woofers that become midwoofers. 🤞
A few days ago, @mikessi sent me updated frequency scans of the two drivers. I requested 3 ms time gating, because that minimizes reflection hash and given we are mostly interested in the response above 1k, a longer gating does not buy us much...
Here is my (some-what edited) response
======== =======
I think I have something workable. See if you find this acceptable.
First I will through the steps to formulate a crossover. Here are the two driver raw responses:
NE123:
GRS Tweeter:
Next I applied an idealized LR4 filter at 3k. I simulate this as if it were an active DSP filter. This is just to see what kind of slopes work best with these two drivers. I next tried 3rd order and 2nd order. I was not happy with either one... What seemed to work best is a 2nd order filter.
A passive filter which approximates this active filter would seem to work well.
The next step is to see if a passive filter can be designed which matches the above.
What I found is that a filter which exactly matches the active simulation results in too low of impedance. It gets down to 2.6 Ohm at 3k. However, with some tweaking I could get the minimum impedance up 3.4 without too much impact on the frequency response. Note that the polarity of the tweeter needs to be reversed for this crossover design.
The full 6-pack curves for this filter:
Ignore the phase plot above. Somehow you got some excess delay built into the frequency responses for the two drivers. It is an easy measurement mistake to make. Fortunately, it makes no difference to the simulation, but it does make the phase plot look crazy.
Since you have the opportunity to apply some DSP eq on this channel for the NE123+GRS tweeter, I wanted to show you what might be possible
j.
Here is my (some-what edited) response
======== =======
I think I have something workable. See if you find this acceptable.
First I will through the steps to formulate a crossover. Here are the two driver raw responses:
NE123:
GRS Tweeter:
Next I applied an idealized LR4 filter at 3k. I simulate this as if it were an active DSP filter. This is just to see what kind of slopes work best with these two drivers. I next tried 3rd order and 2nd order. I was not happy with either one... What seemed to work best is a 2nd order filter.
A passive filter which approximates this active filter would seem to work well.
The next step is to see if a passive filter can be designed which matches the above.
What I found is that a filter which exactly matches the active simulation results in too low of impedance. It gets down to 2.6 Ohm at 3k. However, with some tweaking I could get the minimum impedance up 3.4 without too much impact on the frequency response. Note that the polarity of the tweeter needs to be reversed for this crossover design.
The full 6-pack curves for this filter:
Ignore the phase plot above. Somehow you got some excess delay built into the frequency responses for the two drivers. It is an easy measurement mistake to make. Fortunately, it makes no difference to the simulation, but it does make the phase plot look crazy.
Since you have the opportunity to apply some DSP eq on this channel for the NE123+GRS tweeter, I wanted to show you what might be possible
j.
