Looks can be deceiving...
If it really does anything, that remains the question...
Then again, I think most would have had a hard time believing that a 6mm cone extension would do anything useful... (me too if I hadn't seen measurements online) All I hoped for is for it not to be detrimental, and was real happy to find it could actually bring something useful.
I don't believe in this above just jet, but it would be useful to see the trend it brings to the FR response. It will probably need huge mesh resolution and as a result be a pain to sim. And that will be my start in Abec?
If it really does anything, that remains the question...
Then again, I think most would have had a hard time believing that a 6mm cone extension would do anything useful... (me too if I hadn't seen measurements online) All I hoped for is for it not to be detrimental, and was real happy to find it could actually bring something useful.
I don't believe in this above just jet, but it would be useful to see the trend it brings to the FR response. It will probably need huge mesh resolution and as a result be a pain to sim. And that will be my start in Abec?
Attachments
Last edited:
Did you see the effect at 4 KHZ of that measly 6mm chamfer in front of the baffle?
This is another 6mm on top of that, creating a steeper angle.
I wouldn't want to bet on it, without testing...
But first I have to see if I could fabricate it as an add-on:
An not make it look out of place...
It is at 4 KHz and above where I want it to have an influence.
This is another 6mm on top of that, creating a steeper angle.
I wouldn't want to bet on it, without testing...
But first I have to see if I could fabricate it as an add-on:
An not make it look out of place...
It is at 4 KHz and above where I want it to have an influence.
Attachments
Last edited:
Uploaded as a printable object online it is about 10 euro a piece .
For that money it should better work wonders! No doubt testing it could be done with easier and cheaper with filament printers, but they won't have the resolution and versatility of SLS printers or Multi Jet Fusion.
But for that kind of money I could make entirely new front baffles.
Luckily the weather seems to improve next weekend so I'll be too busy to think about silly projects like these. Let's first see what we've got with the new filters before going wild again. That doesn't mean I couldn't bring my PC to it's knees with some ABEC crunching .
OK then, one more view:
To show the differences in dimension.
.For that money it should better work wonders! No doubt testing it could be done with easier and cheaper with filament printers, but they won't have the resolution and versatility of SLS printers or Multi Jet Fusion.
But for that kind of money I could make entirely new front baffles.
Luckily the weather seems to improve next weekend so I'll be too busy to think about silly projects like these
. Let's first see what we've got with the new filters before going wild again. That doesn't mean I couldn't bring my PC to it's knees with some ABEC crunching .OK then, one more view:
To show the differences in dimension.
Attachments
Last edited:
That does look different and less likely to cause a problem
I don't think it will have to be huge but if the faces of it can be simplified to fit with the 6mm chamfer that will make things a lot easier and will still give virtually the same result as if it was fully rounded. If you can model that then I can simulate that for you fairly easily. To get effects higher in frequency then a phase shield like a dome tweeter uses would probably be needed, again the effects of that can be simulated quite easily by just hanging a flat disc or dome in front of the model without having to worry about the reality of making it until the idea is shown to have benefit.It will probably need huge mesh resolution and as a result be a pain to sim. And that will be my start in Abec?
If it is just done on a single driver basis then it will only be on it's knees for a few minutes at a timeThat doesn't mean I couldn't bring my PC to it's knees with some ABEC crunching
I'll think about how I can simplify the shape, as that is indeed answering if such a thing would be useful...
The ABEC frequency curves show greater stability in movements up/down or from different distances with the filters in place, so I'm very pleased to see that.
All in all, that does seem like a worthy upgrade I can look forward to.
Above 4K this tweak is only meant to limit vertical dispersion (bring off axis down in the vertical direction). I don't see how we could improve the dispersion of the top end, as a full range driver will have gone to non pistonic behavior there anyway. And it isn't that bad at angles of interest to cover the listening area. I did think of a circular phase plug like compression driver has, you know, similar to what Geddes developed quite a while ago. But that would be a real long shot (lol).
The ABEC frequency curves show greater stability in movements up/down or from different distances with the filters in place, so I'm very pleased to see that.
All in all, that does seem like a worthy upgrade I can look forward to.
Above 4K this tweak is only meant to limit vertical dispersion (bring off axis down in the vertical direction). I don't see how we could improve the dispersion of the top end, as a full range driver will have gone to non pistonic behavior there anyway. And it isn't that bad at angles of interest to cover the listening area. I did think of a circular phase plug like compression driver has, you know, similar to what Geddes developed quite a while ago. But that would be a real long shot (lol).
Here are the simulation results for wesayso's lip idea. I think it has some merit. It has made the vertical more similar to the horizontal removing the big bulge at 5K which may have been partly responsible for the ceiling splash horns seen in the vituix sims. It has made the Horizontal slightly worse off axis above 5K.
Comparison included below to make seeing the effect easier.
Comparison included below to make seeing the effect easier.
Attachments
Thanks for running the sim, fluid. Kind of cool to see it working as expected.
A bit sad to see the influence on the horizontal dispersion from this tweak, though it isn't worlds apart.
The trend in off axis response in vertical direction works for almost all of the off axis energy around 4K (a reduction even in the widest of angles), so it might be worth it to be able to reduce the floor/ceiling spray. At 5K and above it does not seem to hold this up though.
I guess I'll have to find some CNC power if I want to create a baffle like that. That would be next to impossible for me to make/create manually with simple tools.
Some optimization might be possible, I guess I really do need to learn to do this myself with ABEC..
A bit sad to see the influence on the horizontal dispersion from this tweak, though it isn't worlds apart.
The trend in off axis response in vertical direction works for almost all of the off axis energy around 4K (a reduction even in the widest of angles), so it might be worth it to be able to reduce the floor/ceiling spray. At 5K and above it does not seem to hold this up though.
I guess I'll have to find some CNC power if I want to create a baffle like that. That would be next to impossible for me to make/create manually with simple tools.
Some optimization might be possible, I guess I really do need to learn to do this myself with ABEC.
.
Last edited:
I don't think it is the lip that is causing the change in horizontal directivity, I think it is the small step that is in this simulation but not the other. When I edited your CAD file I made the chamfer flush to the surround, in this one that you made (that matches your own baffle) the step is there.
If you want to test that then I can rerun the sim if you send me a new quartered step file with the surround flush to the chamfer.
I think you would take to ABEC fairly easily, having someone to point you in the right direction makes it much more achievable
If you want to test that then I can rerun the sim if you send me a new quartered step file with the surround flush to the chamfer.
I think you would take to ABEC fairly easily, having someone to point you in the right direction makes it much more achievable
That makes sense... all little details will influence the results after all.
It is great to see it confirmed how little tweaks like these can go a long way.
I'll send you a similar model, I might have to tweak it a little more to confirm to what I've got. Or maybe I do need to learn how to do it myself.
It is great to see it confirmed how little tweaks like these can go a long way.
I'll send you a similar model, I might have to tweak it a little more to confirm to what I've got. Or maybe I do need to learn how to do it myself
.This is quite a simple one to make changes on and it doesn't take a huge amount of time to simulate, the last one took 58 minutes due to being a denser mesh that the first ones.
This would be a good starting introduction as you already understand the model so explaining what is happening in the script would be easier. You decide if or when you want to dip your toes in the water
This would be a good starting introduction as you already understand the model so explaining what is happening in the script would be easier. You decide if or when you want to dip your toes in the water
Finally some good weather over here, so I could pick up some work in the garage again. The bad thing is that the extensive break had me going nuts with extra simulations, so I made up a few extra filters, as mentioned earlier in this thread. A few bypass filters that go in parallel with only one driver in a string of 5. So instead of finally closing up the first array I was actually making more filters...
The 4 extra filters (2 per array), below the first set of filters for the second array:
and their impedance measurements against an 8 ohm resistor:
(that's all 4 of them + the resistor, good enough quality control I'd say)
The 4 extra filters (2 per array), below the first set of filters for the second array:
and their impedance measurements against an 8 ohm resistor:
(that's all 4 of them + the resistor, good enough quality control I'd say)
Attachments
Last edited:
Question for fluid,
I read a lot about the meshing of objects and ABEC can read stl.
Depending on a couple of variables I can let Inventor decide where the triangles should go. I won't have total control, but it seems smart enough not to divide the larger area's into smal triangles.
Here's a more standard way to mesh as stl:
mesh size set to ~5mm edge length, the deviation (another variable) determines the number of slices to use for more complex shapes.
So even if I set the edge length to something like 50mm:
Due to the surface deviation allowed, it will divide large flat faces into big pieces, and only use smaller ones to be true to the original shape.
This deviation setting is limited though. it varies from 0.003595 to 0.3595.
There are two other variables left that also determine the end outcome, but I'd have to guess what it is they do.
Would this help? I see a lot less triangles in the second example, except where it counts to stay true to the original shape.
Edit: just played with the "normal deviation variable, this one allows to create even simpler shapes decreasing the triangle count further.
All of them are sliders, though with limited choices.
I read a lot about the meshing of objects and ABEC can read stl.
Depending on a couple of variables I can let Inventor decide where the triangles should go. I won't have total control, but it seems smart enough not to divide the larger area's into smal triangles.
Here's a more standard way to mesh as stl:
mesh size set to ~5mm edge length, the deviation (another variable) determines the number of slices to use for more complex shapes.
So even if I set the edge length to something like 50mm:
Due to the surface deviation allowed, it will divide large flat faces into big pieces, and only use smaller ones to be true to the original shape.
This deviation setting is limited though. it varies from 0.003595 to 0.3595.
There are two other variables left that also determine the end outcome, but I'd have to guess what it is they do.
Would this help? I see a lot less triangles in the second example, except where it counts to stay true to the original shape.
Edit: just played with the "normal deviation variable, this one allows to create even simpler shapes decreasing the triangle count further.
All of them are sliders, though with limited choices.
Attachments
Last edited:
The first one will work the best but there is way too many triangles for the surround to deal with the curvature, for a one off you could let that run and it would give good results but it would take a while.
The third one is about right for the surround but the main panels have quite large triangles.
The edge length needs to be kept below 20mm and ideally 10mm for the front baffle in order to have valid results up to 12 to 15K. Beyond the front it matters much less. Sometimes you can mesh it in two different densities and save them separately and only call specific faces at high or low resolution. That only works for square corners or rectangular surfaces of which you have none where it might work.
The third one is about right for the surround but the main panels have quite large triangles.
The edge length needs to be kept below 20mm and ideally 10mm for the front baffle in order to have valid results up to 12 to 15K. Beyond the front it matters much less. Sometimes you can mesh it in two different densities and save them separately and only call specific faces at high or low resolution. That only works for square corners or rectangular surfaces of which you have none where it might work.
That image is exactly what I do by splitting the bodies with the y and x planes then hiding the quadrants I don't need.
You can also model the driver separately, export the mesh from ABEC with all symmetry and use that as a template to cut the baffle or chamfer so the vertices that will be used in ABEC match those on the solid model.
In fusion I create a surface patch to cover the surround or chamfer and then use that to extrude a cut or split a body apart. By converting mesh to Brep solid tools can be used on the mesh points.
Then when meshed those vertices are kept. Inventor could probably do the same.
The new one will
, I'm still using an ancient version that lacks a lot of surface tools (but had some other important features for me, that were crippled in later versions when I was working on my Apfelbeck project). I have/use the new one at work and it features basically everything that Fusion has.
Just trying think of ways to keep prep to a minimum
. I am used to model in solids, so having to think in surfaces is somewhat new to me
. Fusion (to me) is like being in your favorite store, but someone rearranged all the products! One thing that gmesh does that I suspect an stl file will not do is tag the contiguous surfaces from the step file separately. So to define the driver you might select tags 1, 2 and 3, without tags you might have to have the sections as separate STL files.
That mesh look a lot better, high element count but not ridiculous for no reason.
I feel the same about Inventor, why did they hide the sauce in the pet food aisle
Rhino seems to be the best for surfacing, but that feels like being blindfolded in a new supermarket
That mesh look a lot better, high element count but not ridiculous for no reason.
I feel the same about Inventor, why did they hide the sauce in the pet food aisle
Rhino seems to be the best for surfacing, but that feels like being blindfolded in a new supermarket