Announcement

Collapse
No announcement yet.

Build - Polytope

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #31
    Re: Build - Polytope

    Part 6 - Tuning and Crossover Design

    There are many great tools out there for this part of speaker design. My favorites are: Speaker Workshop, Passive Crossover Design, HOLMImpulse, and the LinearTeam calculators. While fiddling with the crossover design actually started before any of the cutting, actually evaluating them didn't happen till near the end.

    The nominal volume is 2.5 ft^3, and after a bit of playing, tuning to low 20's gives a decent rolloff, with only a little bump between 48 and 100. Down 3 db at 28 and down 6 db at 23. Two ports with 1.5" diameter, 6" long.

    Click image for larger version

Name:	SpeakerWorkshop-BoxResponse.jpg
Views:	1
Size:	54.9 KB
ID:	1147323

    A quick on the LinearTeam website for the Acoustic Power calculator and Vent Calculator give these:

    Click image for larger version

Name:	LinearTeam-AcousticPower.jpg
Views:	1
Size:	90.9 KB
ID:	1147324

    Click image for larger version

Name:	LinearTeam-VentCalculator.jpg
Views:	1
Size:	100.8 KB
ID:	1147325

    Looks like the vent mach # will be ok. Would prefer a little lower, but it will do.

    With much fussing and twiddling of values in PCD and Speaker Workshop, the crossover looked like this:

    Click image for larger version

Name:	SpakerWorkshop-3WayXover.jpg
Views:	1
Size:	57.8 KB
ID:	1147326

    The LCR in there is to tame a bump in the midrange. After putting in the offsets for the three drivers, both SpeakerWorkshop and PCD were pretty close in their predictions. PCD looked like this:

    Click image for larger version

Name:	PCD-XOver Frequency Response.jpg
Views:	1
Size:	210.4 KB
ID:	1147327

    The prediction is a frequency response that is within a few db over the majority of of the spectrum. The impedance is a bit squirrly, but I'm not going to drive it with tubes so really not a problem. Just need a 4 ohm safe amp.

    Tangent - While I think tubes have a nice sound, particularly as they near their limits, I'm a bit biased against them. As a kid, when the TV failed, we would have to: unscrew the back, take out the dozen or so tubes, take them down to Radio Shack, test each one individually on the tube tester machine, buy a replacement for the failed one, bring them all back and put into TV, screw back onto TV. What a pain. Just give me solid state with 3-4 times the rated power of the tube amp and it will sound fine...

    Note that this is still a paper exercise. Speaker placement and room modes will result is much larger variances. Additionally, I've noticed that both SW and PCD seem to assume point sources and really don't cleanly predict the response for tweeters on waveguides. I've seen this in a couple of builds now.

    Comment


    • #32
      Re: Build - Polytope

      I got a chance to do a few measurements on the speakers. I tried three different amps (Old Dayton T-Amp, AudioSource AMP 100, Peavey IPR DSP 1600) and two speaker cables (12 AWG and 14 AWG) with HOLMImpulse to double check that I was measuring the speakers, not what's in between. The mic is calibrated and comes into the PC through a Xenyx 302/USB so the measurements are reasonably accurate. I settled on using the Peavey as the amp for most uses as it has sufficient overhead to cleanly drive the speakers to full volume. I'm also hoping to make use of the built in parametric EQ to further flatten out the response. Not worth doing too much on that while they are in the garage.

      The HOLM Impulse response measurements are shown below. Given the space I'm testing in, it will be a bit tough separating the room modes out and determining what areas to focus on. Any suggestions will be appreciated.

      Click image for larger version

Name:	HOLMImpulse Measured Response - Peavey.jpg
Views:	1
Size:	221.9 KB
ID:	1147489

      Tests of White noise give similar results to the frequency sweep performed in HOLMImpulse. Here are left and right speakers with solid green showing the averaged output and dashed green showing the measured/averaged response. Levels were adjusted so the two graphs were close to superimposed. Repeating those tests with pink noise gave comparable results.

      Click image for larger version

Name:	White Noise - Right - Averaged.jpg
Views:	1
Size:	66.1 KB
ID:	1147490

      Click image for larger version

Name:	White Noise - Left - Averaged.jpg
Views:	1
Size:	68.5 KB
ID:	1147491

      Pointing the mic between the speakers and averaging the spectrum for a hour or two gave pretty reasonable looking results. The first image is a variety of Beethoven, the second is a variety of STS9. The 30-60 Hz bump in the latter is pretty typical of that kind of music, and quite expected.

      Click image for larger version

Name:	LongTermSpectrum-pic3-Beethoven.jpg
Views:	1
Size:	109.8 KB
ID:	1147492

      Click image for larger version

Name:	LongTermSpectrum-pic1-STS9.jpg
Views:	1
Size:	107.2 KB
ID:	1147493

      Comment

      Working...
      X