I suspect the TL box might be a bit much with just my table saw here!
A tablesaw works fine. The TL is made from a single width cut to length, and the lengths are not that critical (0.1in o.k). The widths need to be the same so that all of the internal walls fit in between the walls and make a good seal. Even internally, the angles of the dividers does not have to be perfect. But it is easy to do right angles well. All cuts are 90 deg. The tricky part is the baffle rebates but you have to do that with every speaker using a router and circle jig. There really is not much to making a TL.
The only way to get good polars as far as I can tell is to use a waveguide.
Problem with poor off axis isn't that you sit and listen in that direction, it is that this energy will eventually reach your ears and the off axis polars is what your are mostly listening to. Mr Bose was correct; 8/9 of all sound that reaches your ear is reflected sound. So work on those polars for best results!! This is wht serious an insightful designers here keep posting the "spinorama" charts.
//
Agreed. More measuring required to be sure. I’m only guessing at the moment. But it sounds very good. But I also feel like I can get more out of it.
What does 'good polars' mean?
Sorry to use such a vague term as "good polars". I am for one, fine with the polar response that a 3.5in full range provides as I sit at the sweetspot when listening. But what I noticed on speakers with a waveguide is that I can move about even within a 60deg cone and it sounds about the same. On a speaker with an acoustic diffraction lens like a Karlson, this is even wider - maybe 90deg included cone. On the WG300 waveguide integrated with RS28F and PTT6.5, I am getting +/-2dB within a 60deg included cone angle - and that is more than good, it is fantastic!
Here are the polar measurements (technically not on a "polar plot") - these are half angles, so included cone angle is 2x, or 60deg is the green line:
Compared to a bare RS28F-4 as measured by Dayton:
Attachments
Last edited:
Maybe try Peerless NE180W-08. It could be forced to a 4th order using 2nd order electrical at similar 3.6kHz region, but what to do about that 4kHz breakup peak? Would need a notch, so additional L and C and R - not too much more but adds phase wraps of its own. Could just try it and see what happens.
https://www.parts-express.com/pedocs/specs/264-1134--tymphany-ne180w-08-spec-sheet.pdf
https://www.parts-express.com/pedocs/specs/264-1134--tymphany-ne180w-08-spec-sheet.pdf
Dynavox LW6004HR-N might work - has smooth roll off around 4k.
Dynavox LW6004HR-N 6-1/2" Honeycomb Cone Woofer $56 is very budget priced compared to PTT6.5. Will not have the same bass extension though - could work well in a straight MLTL.
https://www.parts-express.com/pedocs/specs/295-626--dynavox-lw6004hr-n-spec-sheet.pdf
Dynavox LW6004HR-N 6-1/2" Honeycomb Cone Woofer $56 is very budget priced compared to PTT6.5. Will not have the same bass extension though - could work well in a straight MLTL.
Extremely rigid fiberglass honeycomb cone nearly eliminates cone breakup
3" copper clad aluminum voice coil wound on an aluminum former keeps excursion linear
Extremely open frame design with extensive venting for low power compression
Smooth response up to 4,000 Hz makes this driver ideal for 2-way designs
https://www.parts-express.com/pedocs/specs/295-626--dynavox-lw6004hr-n-spec-sheet.pdf
Last edited:
Yes, more like 30-60-90 though and not 45-45-90.
You can read about how many commercial speakers J.A. of Stereophile has measured that has has a right triangle response here:
Dunlavy Audio Laboratories SC-IV loudspeaker Measurements | Stereophile.com
SC-IV:
Quad ESL-63:
Here is one of my other speakers that does this - 10F/RS225 FAST:
You can read about how many commercial speakers J.A. of Stereophile has measured that has has a right triangle response here:
Dunlavy Audio Laboratories SC-IV loudspeaker Measurements | Stereophile.com
SC-IV:

Quad ESL-63:

Here is one of my other speakers that does this - 10F/RS225 FAST:

Last edited:
I received the new RST28F-4 tweeters the other day. They are a drop in fit as far as the flange and bolt pattern are concerned. However, the protective wire mesh grille (not removable) is 41.6mm dia vs the RS28F is 40.0mm dia and the WG300 aperture is exactly 40.0mm dia. As a result, the WG sits about 1mm above the dome with a small gap. I added a felt spacer to cover this gap and used the foam spacer provided from the RS28F in addition. The RST28F already has a foam gasket, but it is glued to the back for traditional front mounted on flat baffle appication.
Here is the dia of the RST28F grille:
Here is the dia of the WG300 aperture (blue putty is sealant between adapter spacer ring and WG - needs to be airtight so woofer doesn't pressurize the tweeter dome front):
Here is the felt spacer used to close the gap:
I added Noico mass loaded butyl autosound dampening sheets to absorb any vibrations in the plastic WG or 3D printed adapter:
Here is a closeup of the front of the WG300 with the ne RST28F installed:
The rest of the installation went smoothly. For what it is worth - the RST28F is a handsome and well made quality looking tweeter with a solid aluminum faceplate (not plastic), a heavy magnet, and shaped rear acoustic chamber, and a nice metal grille protector. In taking it apart to see if the grille was removable, I discovered that the diaphragm and grille and voice coil are replaceable and there is ferrofluid in the voic coil gap. Not bad for a $38 tweeter.
So how does it measure compared to the RS28F that it replaced?
Here are the measurements of the RST28F in the WG300 at 2.00Vrms and 0.5m with the same XO shown earlier. They are very similar, but RS28F seems to have a rising reach at 20kHz whereas the RST28F is falling at 15kHz. Overalll match is acceptable as you will see in the following step response:
Here is the comparision of the performance when used with the PTT6.5 woofer and same crossover - there is probably a bit smoother response in some ranges and a bit more presence at 10kHz to 12kHz. A little less "air" at the top register and this could be the result of the non ideal waveguide fit. The magenta curve is the RS28F combined data, the blue curve is the RST28F combined data:
The measured distortion levels are similar for both but it is puzzling that the the fundamental for the RST28F appears to be falling at 15kHz, and is quite a bit lower at 20kHz, but this is not apparent in the total SPL data in previous graph:
Here is the measured step response of the RST28F in the WG300 and PTT6.5 - very nice actually:
I am listening to it now and it sounds very good - near identical to the RS28F that it replaced. No adjustments to the XO needed. Some people who can hear above 15kHz may miss the few dB loss, I can't really hear that high so not an issue for me. So I think the conclusion is that the RST28F is a suitable replacement for the RS28F for this speaker design. This is good because finding RS28F's in the wild is sort of like looking for a unicorn.
Dayton Audio RST28F-4 1-1/8" Reference Series Fabric Dome Tweeter with Truncated Faceplate
Here is the dia of the RST28F grille:
Here is the dia of the WG300 aperture (blue putty is sealant between adapter spacer ring and WG - needs to be airtight so woofer doesn't pressurize the tweeter dome front):
Here is the felt spacer used to close the gap:
I added Noico mass loaded butyl autosound dampening sheets to absorb any vibrations in the plastic WG or 3D printed adapter:
Here is a closeup of the front of the WG300 with the ne RST28F installed:
The rest of the installation went smoothly. For what it is worth - the RST28F is a handsome and well made quality looking tweeter with a solid aluminum faceplate (not plastic), a heavy magnet, and shaped rear acoustic chamber, and a nice metal grille protector. In taking it apart to see if the grille was removable, I discovered that the diaphragm and grille and voice coil are replaceable and there is ferrofluid in the voic coil gap. Not bad for a $38 tweeter.
So how does it measure compared to the RS28F that it replaced?
Here are the measurements of the RST28F in the WG300 at 2.00Vrms and 0.5m with the same XO shown earlier. They are very similar, but RS28F seems to have a rising reach at 20kHz whereas the RST28F is falling at 15kHz. Overalll match is acceptable as you will see in the following step response:
Here is the comparision of the performance when used with the PTT6.5 woofer and same crossover - there is probably a bit smoother response in some ranges and a bit more presence at 10kHz to 12kHz. A little less "air" at the top register and this could be the result of the non ideal waveguide fit. The magenta curve is the RS28F combined data, the blue curve is the RST28F combined data:
The measured distortion levels are similar for both but it is puzzling that the the fundamental for the RST28F appears to be falling at 15kHz, and is quite a bit lower at 20kHz, but this is not apparent in the total SPL data in previous graph:
Here is the measured step response of the RST28F in the WG300 and PTT6.5 - very nice actually:
I am listening to it now and it sounds very good - near identical to the RS28F that it replaced. No adjustments to the XO needed. Some people who can hear above 15kHz may miss the few dB loss, I can't really hear that high so not an issue for me. So I think the conclusion is that the RST28F is a suitable replacement for the RS28F for this speaker design. This is good because finding RS28F's in the wild is sort of like looking for a unicorn.
Dayton Audio RST28F-4 1-1/8" Reference Series Fabric Dome Tweeter with Truncated Faceplate
Attachments
-
RST28F-WG300-PTT6.5-Step_Response.jpg190.4 KB · Views: 1,324
-
RST28F-WG300-PTT6.5-Distortion.jpg247.8 KB · Views: 1,826
-
RST28F-vs-RS28F-WG300-w-PTT6.5-Compare.jpg228.5 KB · Views: 1,324
-
RST28F-vs-RS28F-WG300-Compare.jpg200.3 KB · Views: 8,558
-
RS28T-WG300-Cloesup.jpg461.5 KB · Views: 1,367
-
RS28T-WG300-Noico-Wrap.jpg323.8 KB · Views: 1,337
-
RS28T-WG300-Felt-Spacer.jpg346.3 KB · Views: 1,382
-
WG300-Aperture-Dia.jpg385.1 KB · Views: 1,341
-
RS28T-Grill-Dia.jpg431.5 KB · Views: 1,355
Last edited:
A member requested an STL file of the entire TL speaker for making a CAD model. A STEP file is also included below.
Here you go...
Here you go...
Attachments
Last edited:
I am looking at the data a little more carefully from the RST28 and WG300. Despite having a very good step response, the phase measurement appears to just rise and not have a 55deg bump like a true Harsch XO. Whether or not this makes an audible difference vs the RS28F, I am not sure. I do want to let folks know about this and so they can choose their tweeter accordingly. The RS28F will measure better according to my data.
Here is measured phase for the RST28F-4 in the WG300 integrated with the PTT6.5:
Here is measured phase for the RST28F-4 in the WG300 integrated with the PTT6.5:
Attachments
An absolute goldmine of information here. Thank you.
Is there software for modelling TL speakers X? I want to have a play
Is there software for modelling TL speakers X? I want to have a play
I use Akabak - very old high level programming language. Free to download and use for non commercial use from R&D Software. Although looks like they updated it recently now there is a free version. But can’t save stuff. The new version looks much more like what ABEC3 was so not sure what they did recently.
AKABAK
The old version only ran on Win32 (XP virtual machine).
It’s very powerful but steep learning curve. There are threads in DIYA on how to use it. I spent 2 years modeling every speaker I could find on DIYA.
AKABAK
The old version only ran on Win32 (XP virtual machine).
It’s very powerful but steep learning curve. There are threads in DIYA on how to use it. I spent 2 years modeling every speaker I could find on DIYA.
A couple of questions on measuring your acoustic offsets way back in post #1:
Did you do the offset measurements with a crossover in place or with just the raw drivers? Is your typical offset measurement procedure with or without a crossover in place?
This wording baffled me for a while. Did you mean the 47.5 mm is the additional delay for the Harsch xover (Fc/2)? So total delay = acoustic offset (19 mm) plus Harsch delay (47.4 mm)?
Thanks...
I get a 19mm offset of the tweeter and waveguide based on the interferogram match of the simulated and actual combined measurement.
Did you do the offset measurements with a crossover in place or with just the raw drivers? Is your typical offset measurement procedure with or without a crossover in place?
Furthermore, since the Purifi PTT6.5 driver has such a nice flat response out to nearly 4khz, I was able to run a high XO point of about 3.6kHz, which corresponds to a time of flight delay distance of 47.5mm (1.87in).
This wording baffled me for a while. Did you mean the 47.5 mm is the additional delay for the Harsch xover (Fc/2)? So total delay = acoustic offset (19 mm) plus Harsch delay (47.4 mm)?
Thanks...
Offset measurements are always done with the raw drivers, no XO. Take data with just woofer, then tweeter, then connect both in parallel. Being careful not to physically bump or move the speaker or mic during the wiring changes. Easy if all external spadde/bullet connectors.
The acoustic offset is intrinsic to the speaker (drivers and how they are mounted). It is used to get an accurate simulation of the XO and exists already as provided by the physic arrangement. The extra circa 50mm of delay for the Harsch XO was achieved by putting the tweeter offset from the woofer axis (usually underneath on a baffle). Set the speaker so that the woofer axis is the main listening axis/height. This gives that approximate 50mm extra delay at short distances, and at farther distances, it works for a wider up-down sweetspot.
The acoustic offset is intrinsic to the speaker (drivers and how they are mounted). It is used to get an accurate simulation of the XO and exists already as provided by the physic arrangement. The extra circa 50mm of delay for the Harsch XO was achieved by putting the tweeter offset from the woofer axis (usually underneath on a baffle). Set the speaker so that the woofer axis is the main listening axis/height. This gives that approximate 50mm extra delay at short distances, and at farther distances, it works for a wider up-down sweetspot.
@xrk971, assuming the TL works well, and nor reason to doubt it will, would it be possible to turn this into a floorstander using just two legs of the TL and a taper from top to bottom and back to top again? This would almost half the depth of the cabinet.
Supplementary question, does the TL output need to be close to the Woofer or would the more normal Woofer top/TL output bottom work assuming taper and length are the same? This would result in something more like the a PMC 23 series speakers.
Supplementary question, does the TL output need to be close to the Woofer or would the more normal Woofer top/TL output bottom work assuming taper and length are the same? This would result in something more like the a PMC 23 series speakers.
Doesn't the passive crossover components "move" the drivers also ?
Running a 6db cap on a tweet, i was supprised how far back it had to be to hear the best percussive transients......
I think a cap "moves" the tweet forward, thusly needing to be slid back more.
Running a 6db cap on a tweet, i was supprised how far back it had to be to hear the best percussive transients......
I think a cap "moves" the tweet forward, thusly needing to be slid back more.
All that is accounted for in the XSim simulation as the electrical phase and impedance are all part of the simulation.
- Home
- Loudspeakers
- Multi-Way
- Simple Passive Harsch XO Using PTT6.5 and RS28F in a Waveguide