That looks better, having the waveguide a db lower or filling in the saddle in the woofer response might help a little to avoid that slight rise and dip between 1 and 5K in the combined response.
Just remember that without phase data for the CD and waveguide any crossover simulated here won't work out the same in reality. In trying to show you an example before I used a 2nd order crossover without needing to flip the phase of one side. In real life it would have had a massive dip at the crossover point.
Just remember that without phase data for the CD and waveguide any crossover simulated here won't work out the same in reality. In trying to show you an example before I used a 2nd order crossover without needing to flip the phase of one side. In real life it would have had a massive dip at the crossover point.
With the help of @fluid I started a full scale print. A first prototype to see what problems might occur.
- mounting screws are almost to high and are very close to the (inner part of the) Waveguide. Will switch to the outer two mounting holes as mentioned by fluid.
- Printer has developed some weird behavior. Going to try to solve it by increasing belt tension.
- round holes are not round. Might be caused by stl export. Investigation needed.
- Ran out of filament. Well, that was actually expected.
- Find a way to fill it.
Some drivers come with threaded studs which would mean only a washer and nut needs to be put on in the cutout.
It looks like you used gyroid infill, if so that can be filled with resin as there are gaps through the infill to let the resin flow. Just be careful as many resins get hot when they cure which could warp the plastic.
Is the horn stub three separate pieces?
The point of an OS waveguide is said to be smooth and limit the 2nd derivative (still don't understand what exactly...); smear the change in angle out evenly as opposed a diffraction slot.
The problem here is that there is a 1mm gap. And also because of the throat angle of the CD and this gap, the throat of the waveguide does not have the correct size. It feels a bit penny wise and pound foolish: have a very well defined waveguide and then accept a gap where the CD and waveguide come together.
The problem here is that there is a 1mm gap. And also because of the throat angle of the CD and this gap, the throat of the waveguide does not have the correct size. It feels a bit penny wise and pound foolish: have a very well defined waveguide and then accept a gap where the CD and waveguide come together.
Leave some tolerance on the CD mount bolt holes so that the throat can be aligned. My commercial waveguides have some allowing align/misalign.
Mabat did some tests with the exit angle mismatch and the effect wasn't much in simulation at least (in ath thread somewhere). Angle is customizable with ATH script anyway. I think the conclusion was that all that matters is the exit wavefront of the driver and while a waveguide could be made to help fix/match poor CD it is very hard to do as we don't have perfect models for the driver. Just use best/good driver, which you know you have if measurements match simulations closely (enough).
For the gap due to the foam, people mention to fill it with some putty, parhaps play'doh or something I don't know. Mark100 has posted he doesn't bother these days as it doesn't seem to matter with his MEH builds, square throat.
If you have time to experiment with these please post your results and thoughts as this is something common to all. Thanks!
Mabat did some tests with the exit angle mismatch and the effect wasn't much in simulation at least (in ath thread somewhere). Angle is customizable with ATH script anyway. I think the conclusion was that all that matters is the exit wavefront of the driver and while a waveguide could be made to help fix/match poor CD it is very hard to do as we don't have perfect models for the driver. Just use best/good driver, which you know you have if measurements match simulations closely (enough).
For the gap due to the foam, people mention to fill it with some putty, parhaps play'doh or something I don't know. Mark100 has posted he doesn't bother these days as it doesn't seem to matter with his MEH builds, square throat.
If you have time to experiment with these please post your results and thoughts as this is something common to all. Thanks!
Last edited:
Hey,
Very interesting thread. I’ve been around the block building speakers for personal use. Perhaps an open baffle design may be of interest? If so google Dick Olsher.
He offers some diy designs that you may find interesting.
Regards
Btw,
The Calpamos is an intriguing design. I’m a big fan of humble homemade hifi. I especially like the capacitor evaluations.
Last edited:
I used plasticine putty to fill a cavity between an Eminence screw-on adapter and a JBL waveguide. It works perfectly fine. The effect of a small mismatch of exit angles and throat is not a big deal.
https://www.diyaudio.com/community/...-design-the-easy-way-ath4.338806/post-6475246
https://www.diyaudio.com/community/...-design-the-easy-way-ath4.338806/post-6728508
https://www.diyaudio.com/community/...-design-the-easy-way-ath4.338806/post-6475246
https://www.diyaudio.com/community/...-design-the-easy-way-ath4.338806/post-6728508
I'm slowly moving forward. Recently I've been trouble shooting the 3d printer (belts too loose, excentric bolts to tight, underextrusion,...). At the moment I'm doing autopsy on some blisters on top. Curiously, a more experienced 3d printing collegue on a different brand printer has them also...
In the mean time I'm reading up on measuring and DI.
The foam is 2mm thick and I think it won't compress more than 0.5-1.0mm overall, even though by applying localized pressure you can easily squash it.
I can compensate the throat radius for that. Angle stays the same. I'll oversize the mounting holes a bit, but the printer is quite good at printing the right dimensions. The foam is positioned much less precise.
Edit: One should not put PLA in the sun. Even in the Dutch spring they deform easily in slight pressure.
Last edited:
It looks like you are too close to the bed if the blisters are from the bottom. I would not worry too much about the transition. A perfect seal is much more important.
Edit: I re-read your post. If the blisters are from the top layer, I would increase the number of solid top layers, the infill looks also a bit too sparse.
Edit: I re-read your post. If the blisters are from the top layer, I would increase the number of solid top layers, the infill looks also a bit too sparse.
^^ yeah looks like the foam is not well centered. Perhaps it doesn't matter because the company is probably well capable of precise construction but still didn't apply it here. You could cut foam donut from around the throat and reposition it better. Not sure how tight the glue is, perhaps try small cut on the outer edge first as it is not that critical there if it gets ripped/ruined.
Orange phase plug tells it is a FaitalPro driver? I think my hf10ak has the foam bit of as well, shame on them.
Orange phase plug tells it is a FaitalPro driver? I think my hf10ak has the foam bit of as well, shame on them.
Last edited:
The same driver can play lower on a deep horn than a shallow one. My own tests with a 1" show about 150Hz difference when comparing a short horn to
a longer one (similar to the Celestion 90*40 you linked earlier).
i.e. the 20cm deep Celestion horn will support a slightly lower crossover than most PA horns.
The basic idea is fine, but not that particular horn, IMO. The 2380 has hard angles (a slot in the throat) to scatter the output, to allow a 2" driver to have wide horizontal coverage, to cover a large audience.
Since that is not your application, a horn that does not have those hard angles will be better. I note that is what you are going with in your 3D printing.
Why not revisit that idea by using a large horn (or a found object) for the mouth of your project - and then you only have to print the throat, and glue them together. The attached image is a speaker I was working on, which used an Ikea bowl as the horn. I've seen multiple other found objects that would work similarly.
I found that the afp file as generated by ath is very easy. And as I still don't like Fusion, I reverted to something I know better. I wrote a matlab/octave script to generate afp file based on the ROSSE function and some of @fluids ideas on how to make the waveguide. It's not as neat and handy as I would like it to be, but it's a start. Max print angle, minimum flange radius and minimum thickness can be entered. Flange radius is changed to match the maximum print angle.
Picking up this project after it has been collecting dust for some time as I've been busy trying to keep the energy bills from exploding...
I switched back to mabat's approach and printed the throat. I tried to outsmart mabat by adding an alignment feature, but completely overlooked the print support needed and the consecutive less than optimal fit.
Some say I'm a perfectionist and my pile of test print parts, but no finished part might be considered evidence. This time I said to myself that I was going to finish it, perfect or not. We'll see
I switched back to mabat's approach and printed the throat. I tried to outsmart mabat by adding an alignment feature, but completely overlooked the print support needed and the consecutive less than optimal fit.
Some say I'm a perfectionist and my pile of test print parts, but no finished part might be considered evidence. This time I said to myself that I was going to finish it, perfect or not. We'll see
