Howdie, I added this to the DIY speaker testing thread, but thought it might be of more interest here
I lent ASR member Rick Sykora a hand with a new Purifi-based design that he's been working on. I found it had problems in a sealed box, and I made a recommendation for a tweeter to use. Sharing the design process here.
Box Type
Here I proposed the Purifi in a sealed box, but mentioned "Caveat: I still want to calculate the distortion created by the compression of the air itself in such a small box". Turns out, there's an issue using the Purifi in a sealed box.
If air is compressed too far, it's no longer a linear spring. Even a perfect speaker driver will have distortion from the non linear restoring force of the air in a small box, if pushed too hard.
Linkwitz
here estimated the % of second harmonic distortion as =0.014*SD (in cm^2)*driver displacement(in mm)/box volume (in L).
Here's the predicted 2nd harmonic distortion with the Purifi in a sealed box (which by necessity must be small), caused by just the air itself. THD will be higher. I highlighted some measurement points from ASR and compared to the ASR vented proto test results for
total HD.
The sealed distortion will be quite a bit higher than a vented because the parameters of the driver demand such a small box. You could argue about how audible this is, and in the table I show that the distortion won't be too bad when playing >300 Hz alone. OTOH, I've never tested the following theory, but I expect the mids might still distort when played simultaneously with loud bass because the air itself is distorting (Amir doesn't measure this).
The Purifi is all about playing loud with low THD, but it struggles to pull that off in a sealed box because of its unusual TS params.
Tweeter Choice
Rick had initially suggested the
SB26STWGC-4. I`ve never used the
SB26STWGC-4. On vs off axis looks great and it fits. But that top end is rough and would need at least a few extra components to EQ (mandatory IMO), and space will be tight
My understanding is that its a waveguide put on the SB26STC (which gets some love). I compared the STC distortion tests
here to the drivers below, and the ones below have much better all important higher harmonic distortion performance. The small waveguide doesn't have enough gain (5 dB) to make up the difference. It would really be nice to have distortion measurements of it.
Other good low distortion choices with nice combination of power handling and dispersion that fit, but no waveguide
- Revel uses a variant of the SB26ADC-000-4, some great distortion numbers here and here 10dB better distortion numbers 1-2k than the DXT. Great price. Troels makes it work off axis here so the lack of waveguide might not be a big deal
- An oldie but still goodie SB26STAC, MarkK tests and here
- DA25TX00-08 is promising based on raves in this and this but its faceplate is probably too large for this build
Another choice could be SB26ADC and Augerpro`s free waveguide design
here. I checked the CAD file and the 4in would definitely fit, maybe even the 5ìn (more background discussion
here and
here). The STAC might work in his waveguide as well
The Seas DXT is still a good turn key drop in with a nice balance of tradeoffs (waveguide, but more distortion than SB26). I was surprised to read this
diva complaint from a manufacturer, as the driver looks well behaved on IEC baffle.
A ring radiator would really need a waveguide to give it enough power handling to keep up with the Purifi. I took a quick look and found a couple promising turn key combinations
- DX25TG09-04 and Monocor WG-300 waveguide, tests results here show very nice directivity and great distortion results. But the waveguide is big and might not fit
- XT25TG-30 and Monocor WG-300 waveguide, tests results here also show very nice directivivity but the distortion is much higher, the DX25 combo wins
Visaton also makes a waveguide
here, though it`s also big (DIY with XT25
here wasn`t too successful). Mouser and digikey sell it. 617 found it worked best with Dayton RST28F
here, haven`t found any drop ins with ring radiators. Alternatively, one of augerpros waveguides
here might work for a ring, but it might not. His elliptical waveguides provide more options with small baffles. I looked at the wavecor tweeters
here but their distortion results aren`t very impressive
This
thread is another option to read through for candidates
SB26ADC-000-4 with or without waveguide looks like a front runner to me. Best way to predict if a waveguide is needed is to take a diffraction modeling program and model the effect of the baffle from 1 to 8 kHz, then combine with the driver IEC measurements from SB, and see if it works without a waveguide. I explain one way to do it
here. Its an old write up and there are better free tools around now, but the concepts are the same.