The planar falls off higher in the scale. It doesn't go high enough to get the absolute full spectrum. The ribbon plays better in the 5-20khz range. The planar also becomes higher erratic higher up and I do not believe I could filter all of the down to an accurate response
Right, and the holes in that jig on the left and right are for the microphone to be in position with the mid and the woofer respectively.
I have considered this. Just taking it at one point directly where my ear will be. I am just doing it "by the book". My build is pretty far off from normal though I'm not sure where "by the book" ends and "this needs a bespoke method" begins
I see, so I am sort of fighting geometry here.
What really matters is 0 degree and 15 degree. I won't go too far outside of that. I took all of the measurements just to be thorough. I may measure again with 0, 5, 10, and 15 degree so I cna be more precise within my listening window
Yes, this I understand. So are they just getting further and further out of phase? I would think if they were, that the off axis would get a larger and larger drop. Similar to what is happening between the mid and the woofer.
The mid to the tweeter is far more erratic and is not following this expected pattern
After some thought - looking at your measurement rig - I'm not sure if you are measuring the right things. You said they will be used on a desk - but your measurements do not include desk bounce (I don't see it in the photos, infact you raise the speaker to eliminate floor bounce). I think desk bounce will make a measurable contribution to the FR.
- Your speaker is built.
- You are measuring in baffle - not simulating.
- You are building for yourself.
I will do this on Monday.
I am interested to see how the low end plays out considering the port is refiring into the wall. With the XO of the woofer so high I feel some of the higher frequency noise of the port is going to come off the wall and blend or cancel out other wavelengths of the mid.
The idea was to create something that didn't have to live only on this desk, in this situation. Many people actually want to purchase the plans for this when I am done. I may make two XOs. One for generalities and one for my desk specifically.
I will see what measurements look like on my desk
Well yes, to a very small degree. You have created a cool looking design, but sorry, I just think a 3-way is needlessly complex for a desktop system, especially when there are plenty of drivers available to cover the range as a 2-way.
Best of success with your adventure.
Yes, exactly, now you understand what I am going for.
"Unnecessarily Complex" is the goal.
This is for my office. I run the fabrication shop of a highly regarded engineering consulting firm. This is my space to "show off". I definitely did not need a push push sub that doubles as a coffee table. I definitely do not need it to hit to 16 Hz. It did not need to have a clear top. But wow is it super cool to listen to and look at.
And I do not need a 3 way on my desk man will it be cool when I do.
I think hifi isn't all about practicality. If it were, we would all just listen to 2 way studio monitors.
Attachments
I have the means and ability to make this........ We create enough fires in the shop on accident so I don't think we need to be causing them on purpose lol
Possibly, Yes. I am not clear on the terms - they seem interchangable, although I doubt it. Lobbing may be more correct to what I suspect is happening here.
I think it is lobing. I do not think I fully understand lobing. Wouldn't any and all drivers lobe no matter how close or far they are? Exempting coaxials of course. Do they just lobe so far away from the XO point that I don't notice? Are these lobing so close to the XO point because they are, in fact, so close together?
I would think they would lobe less the closer they are. I can visually imagine lobing in my third eye but I'm just not understanding why I wouldn't always have this issue if that's what it is.
On a separate note, I have been messing with this XO inbetween today's activities. I really got nowhere. I came back to the same XO over and over. I pushed the XO lower to no avail because of this dip in the tweeter response messing everything up.
Pushing it higher makes the lobing worse (we are going to stick to calling it lobing until someone figures out if its something else).
And then....... I just took an inductor and stuffed it here. It puts an odd kink into the lower end of the filter but it makes the XO so steep that the lobing is pretty much rememdied.
I will take some measurements inside of the actually office tomorrow. We shall see what difference it makes being in the actual room.
I would think they would lobe less the closer they are. I can visually imagine lobing in my third eye but I'm just not understanding why I wouldn't always have this issue if that's what it is.
On a separate note, I have been messing with this XO inbetween today's activities. I really got nowhere. I came back to the same XO over and over. I pushed the XO lower to no avail because of this dip in the tweeter response messing everything up.
Pushing it higher makes the lobing worse (we are going to stick to calling it lobing until someone figures out if its something else).
And then....... I just took an inductor and stuffed it here. It puts an odd kink into the lower end of the filter but it makes the XO so steep that the lobing is pretty much rememdied.
I will take some measurements inside of the actually office tomorrow. We shall see what difference it makes being in the actual room.
Not in an appreciable way if the driver spacing is closer than 1/4 wavelength at the cross point or if the crossover effective slope is very high with sufficiently small spacing. Joseph Infinite Slope is one approach to this, but there's still much more response variation vertically (see below and note JA's positive comment about it even though it's not perfect).
But yes, most normal speakers do experience crossover lobing and poorer response off axis for the axis where the drivers are spaced apart. That's one of the points I was getting at earlier with the vertical response plots. The vertical plot is half the range of the horizontal, but the vertical plot is normally much more chaotic in general once you get into the range where the high cross point is influencing things (other than for co-ax's, as you noted).
https://www.stereophile.com/content/measuring-loudspeakers-part-three-page-7
. . .
https://www.stereophile.com/content/joseph-audio-rm7si-loudspeaker-signature-measurements-part-2
"In the vertical plane (fig.7), suckouts develop in the crossover region more than 15 degrees above or below the tweeter axis. This is quite a window, presumably due to the minimal overlap between the drive-units. The speaker will be relatively unfussy when it comes to stand height."
Last edited:
So here's the horizontal
This used to be absolutely chaotic before I dug into it with filter components. I do not like the jump at 5500hz but I cannot fix it without ruining other aspects. I'd much rather have a dip than a bump. I know from car audio that bumps are what make systems sound harsh. Better to have drivers not quite crossover so there's a dip than have them cross into each other have them combine into a harsh tone at a specified frequency.
Yes, but even a single element can lobe, depending on it's size. This is the basis behind antenna directivity which was studied long before it was of interest with speakers, so there are plenty of interesting things to find.
Interesting..... now you've given me another research rabbit hole to go down.
In respect to speaker drivers this is mostly based on size though?
We usually look at it in terms of two points with a distance between them. The reality does include their sizes and can be a little different. I typically measure polars and calculate power after crossing to make finer adjustments than the simulation allows, but I'm only saying this now to make a point.
This is the source of the problem. Vertically offest drivers have lobes in the vertical plane, Horizontally offset drivers have lobes in the horizontal plane.
There is a good visual representation of what happens in this article
https://www.prosoundtraining.com/2013/01/25/tweeter-placement-two-way-loudspeakers/
With a tweeter and mid it is generally impossible to have the crossover frequency be low enough for this not to present as an issue
A very steep crossover sometimes helps in one axis, but creates bigger problems in the other axis or off axis. Since you have both vertical and horizontal displacement - I return back to my "measure it on the desk where it will be, from the position of your head" and design as best you can for that. And if 15 degrees off-axis is the most linear - then just point them straight out. They don't need to point at your ears. If there will be a wall beside and to the back - include in your measurements. This is a "custom" speaker, for a specific use-case. Don't torture yourself too much 🙂