Human 81 FR Response

Is pink noise the cause of the roll off? I would think white noise would give a flatter looking response, and I would also suspect that the speaker would be very treble deficient if the graph were accurate, but you say it sounds fine.
 
One of my earlier posts has an entire box frd file. I did that with a calibrated SPL meter at 1/3 octave per Hz tone and used the Xsim FRD tool to create a complete file with phase. You can see when it gets to about 10,000 Hz that it is rolling off pretty quick. It sounds good enough that I never gave it much thought. This concern pretty much started when I began playing with that Dayton mike and a smart phone app to see how they performed and posted that first graph. Now I'm reconciling things I have done in the past but overlooked. Huw probably put that resistor on there to keep people from frying their tweeters and then crying about it. No point in asking Huw. He doesn't even put the ohms rating on his speakers.
 
But the dive at 8 or 10k follows a consistent and significant downward trend starting under 1k. It ought to sound terrible if the first curve is remotely accurate. I wonder about the methods so I'll look at your other posts.
 
Did you measure one speaker, or the other? How did you position the microphone? What kind of room did you measure this in?
 
But the dive at 8 or 10k follows a consistent and significant downward trend starting under 1k. It ought to sound terrible if the first curve is remotely accurate. I wonder about the methods so I'll look at your other posts.

This was my thought as well. Something seems amiss
 
One speaker, pointed the mic directly at the speaker, average size popcorn vaulted ceiling living room for a 3 bedroom house. I won't know more until I get some frd files and run some modeling on Xsim.

Got the T/S on the woofer


* f(s)= 22.21 Hz
* R(e)= 3.53 Ohms
* Q(ms)= 4.099
* Q(es)= 0.3113
* Q(ts)= 0.2893
* SPL= 88.36 1W/1m
 
Last edited:
I measured and entered 7" for piston diameter. Thanks for pointing that out. Looks about 5 times too high. Fixed and posted corrected woofer t/s above. My scale was not calibrated for added mass method and was overstated on the input.
 
Last edited:
There is no point going any further with this experiment. Attached is a reliable SPL text file of the raw tweeter in the box. I used a recently calibrated Radio Shack digital SPL meter. I sent a 1/3 octave tone from 13Hz to 20,000Hz and recorded the level. I then used the Xsim FRD tool to create a 500 line FRD file. This is very liberal since the result is more like Auditory ERB response as opposed to a 1/24 octave one. Thus, the standard deviation between 1khz-20khz is 10dB at the very least.

I also deleted those woofer T/S specs that Huw doesn't have on his website. I also deleted my VAS spec because it was grossly overstated. I did leave up the tweeter specs since Huw doesn't have those on his website. If anybody would like the ZMA files for the woofer and tweeter let me know

I'm pretty sure the goal with the Model 81 is to reduce listening fatigue in the old New England fashion. If you look at the lineage of this design there was and is a reluctance to provide a lot of technical data. The sound is what it is and it suits me just fine. I use these to complement a pair Sansui SP-2500 boxes that have been heavily modified.
 

Attachments

  • Human 81 tweeterFRD.txt
    11.3 KB · Views: 30
Last edited:
All the FRD files would be great.
I've wondered for years if Huw's aluminum dome tweeter had "good"/correct performance.
There were measurements online many years ago by I think Mark.
 
Yeah, that is some good work there. I saved the page for reference. Seems like I'm getting something around the blue 45 off axis response in the second graph. I like his comments at the bottom. Pretty much explains it all...

Comments


A well built tweeter. Quite a bit of frequency response irregularity, though interestingly, the linear distortion curves look respectable, only moderately worse than the 27TDC. Very good nonlinear distortion numbers-but some of this may be artifactually low, given the drooping FR curve. Still, even corrected for flat, the nonlinear distortion numbers look very good, much better than the high Fs would suggest. If you can at least modestly effectively contour the curve to your liking, this should be a very good tweeter. Still, it seems like alot of work!

 
Last edited:
Attached are the ZMA. FRD files can be created as described above with just an SPL meter and Xsim 3D. The Xsim FRD tool is in the Xsim 3D version, which is free to download and use.

Xsim-3D
http://libinst.com/Xsim/BetaTest/

Regular Xsim
http://libinst.com/
 

Attachments

  • Human 81 Tweeter ZMA.txt
    11.6 KB · Views: 5
  • Human 81 woofer ZMA.txt
    12 KB · Views: 5
Last edited:
Oh, dear. I just realized I was measuring everything with the box laying on its side. No wonder the response is similar to the off axis graph on Mark K's page (second graph).
 
Last edited:
Looks like you would only want to use the tweeter between 1800-6000hz! I believe a Fountek cdneo 3.0 could farmed in above that.
 
Also, attached are the graphs for an off axis Model 81, lol. Smoothed to auditory ERB. 1/24 shows the crossover at 1765dB.
 

Attachments

  • Model 81 impedence.jpg
    Model 81 impedence.jpg
    66.9 KB · Views: 33
  • Model 81 45 off axis.jpg
    Model 81 45 off axis.jpg
    63.6 KB · Views: 34
Back
Top Bottom