Ah I see, M is a brace for D. I thought M was a brace for L / G. In that case I will explore leaving it in or maybe deleting it and adding extra bracing behind D, as it seems like F-G-H is pretty well braced by L / N / I / J / K. Thanks.
Guys,
just something that is puzzling me,
it is easier to do the bracing per panel in a discreet manner
instead of connecting one panel to the other , with the little braces in the original plan
as there is very small area reinforcing the flex parts
is it ?
i am not a structural engineer in any manner
i wish some one with access to FEM software can sim the keystone panels
and see what is the optimum way of doing it,
im not saying in any manner that the design is not right , i mean it was built dozens of times
as shown over here plus the people who built and are using it but they not post anything here.
so there are many keystones around the world.
so , it will feasible to do the bracing like the below pic ?
( i removed the baffle and the center panels so to have a better view )
just something that is puzzling me,
it is easier to do the bracing per panel in a discreet manner
instead of connecting one panel to the other , with the little braces in the original plan
as there is very small area reinforcing the flex parts
is it ?
i am not a structural engineer in any manner
i wish some one with access to FEM software can sim the keystone panels
and see what is the optimum way of doing it,
im not saying in any manner that the design is not right , i mean it was built dozens of times
as shown over here plus the people who built and are using it but they not post anything here.
so there are many keystones around the world.
so , it will feasible to do the bracing like the below pic ?
( i removed the baffle and the center panels so to have a better view )
Large surface area is not required for bracing.
It would be easier to use longer, heavier braces to do what the smaller, lighter braces "K" and "J" already do well.
As drawn, the bracing would not brace part "H", the horn back.
If the long braces were angled, they would brace both part "H" and "B", the cabinet back.
A single Keystone uses two 4'x8' sheets of plywood almost completely, longer braces replacing "K" and "J" would require purchasing additional material.
Hey Art,
I just drawn the red "braces" but i know all of the panels must be braced.
So you mean that doing long pieces of wood , will use more ply than the plan braces if I understand correct.
So let's say level 1 of difficulty from 1 to 5 , 5 being the most difficult to build, the long discreet bracing will work if adding them to all the internal and external panels.
I also read that you was going to publish the b-low keystone that had a lower tuning but somehow you did not like the testing phase and decided to cancel the project.
Is it really hard to add more horn length to lower the tuning 8 to 10 Hz or is the keystone shaped exit who do not cooperate if you lower the cabinet tuning.
I just drawn the red "braces" but i know all of the panels must be braced.
So you mean that doing long pieces of wood , will use more ply than the plan braces if I understand correct.
So let's say level 1 of difficulty from 1 to 5 , 5 being the most difficult to build, the long discreet bracing will work if adding them to all the internal and external panels.
I also read that you was going to publish the b-low keystone that had a lower tuning but somehow you did not like the testing phase and decided to cancel the project.
Is it really hard to add more horn length to lower the tuning 8 to 10 Hz or is the keystone shaped exit who do not cooperate if you lower the cabinet tuning.
Yes, longer pieces use more material, increasing cost and weight.
Discreet means careful or intentionally unobtrusive.
Discrete means distinct or unconnected.
The short braces in the plan could be considered "discrete" units.
Long braces would not change the build difficulty level much at all.
I have no way of knowing what three discrete channels in the horn path created by long braces on the back of the cabinet would do to it's response.
I built six of the taller, longer path length Keystones expecting the low corner (cone excursion minima) would drop from ~37Hz to 31Hz (the low "B" note of a five string bass is 30.87 Hz). The low corner did not drop.
The exit size had to be reduced compared to the standard Keystone to match it's low corner, it did not "B-Low" 😉 .
You can read about it in post #1135.
https://www.diyaudio.com/community/threads/keystone-sub-using-18-15-12-inch-speakers.185588/page-57
There was no difficulty adding horn path length by increasing the cabinet height by 3", but I was not expecting to lower the tuning by more than 6Hz.
As it turned out, there was no exit placement or shape that did what I expected from the response.
Using the BC18TBW100-4 rather than the BC18SW115-4 may be responsible for unexpected response difference.
Leoandstrom's recent measurements using BC18TBW100-4 in post 1694 show a higher low corner than my BC18SW115-4 Keystones had.
At any rate, Hornresp simulations can't properly model the influence of the exit shape and position, and I'm not interested in learning Akabak, which might (or might not) be able to account for those details.
You seem interested, maybe you could figure it out !
Art
Last edited:
Nope, no other processing other than LPF/HPF. There was actually two vans behind the cabinet in the previous post, along with a small shed out of frame to the right. All the structures was within 6-10 meters, so I decided to take another measurement in a better location:
And here's the result:
Was really hoping that the 50hz rolloff would be due to the nearby structures... The cabs are built based on the plans in post #487, with only (very) minor changes because of the difference in wood thickness (3/4" vs 18mm).
Does anyone else have measurements of a 18TBW100-4 in a Keystone?
I just ReSimmed both drivers against the Keystone HR Screen
77 Volts
i got the below
i dont see too much deviation
have you checked your filters again ?
maybe run a 2 volt sweep with no filters and see
if you see any change on the low end response ?
other than that.. maybe a build "typo" on the angles or position of the panel inside ?
maybe an air leak ? or something that is not sealed completely ?
77 Volts
i got the below
i dont see too much deviation
have you checked your filters again ?
maybe run a 2 volt sweep with no filters and see
if you see any change on the low end response ?
other than that.. maybe a build "typo" on the angles or position of the panel inside ?
maybe an air leak ? or something that is not sealed completely ?
I don't recall any 18TBW100-4 measurements other than yours, and mine in the "Tall" version, which also seem to share the feature of a higher rolloff point.
What frequency is your excursion/impedance minima at?
There have only been a few other measurements, but with different lower BL drivers that also roll off from around 50Hz.
Speetpeet made a 3d printed Keystone, his 18SW115-4 response looks quite different than any others, ~-12dB at 30Hz from 50Hz, with response dropping above 65Hz.
It would be interesting to see side by side comparisons between the drivers in the same location.
Art
The drivers are long gone so I can't test, but I don't recall my old 18TBW100-4 loaded Keystones having a low corner that high.
From the looks of this graph I would assume 35hz?
I followed the guide here and did the measurement in the same location as in my most recent post. I don't have a picture of it, but the setup was basically: Laptop --> Scarlett 2i2 --> circuit board with 100ohm resistor and messy soldering --> Speakon. Does the data look "correct"?
First of all, thanks for taking the time to sim the drivers! Yeah, I've gone through the whole rack multiple times now and can't find anything weird. The only processing taking place really is the LPF/HPF in the PA2. Tried two different laptops as well as an external audio interface with no luck. I've also double-checked the Windows settings to make sure no "audio enhancements" are enabled. The cabs are definitely air-tight as well... I've gone through the plans once and couldn't find any major deviations, but I might just have to do it again more thoroughly.
How exactly do I perform a "2 volt sweep", and why 2 volts? I'm still kind of new to the world of audio and have lots to learn haha. What frequency range would you consider safe when doing such a sweep without any filters?
Yes, that looks like a 35Hz impedance minima. Neither your impedance sweep or frequency response measurements suggest any cabinet build defects.
Maxolini did not share the inputs he chose for his Hornresp sim, but neither response matches either of our measured responses using those drivers.
Your measured response is -4dB@40Hz compared to 50Hz, the sim -2dB.
My measured response is -0dB 40Hz to +2 @60Hz, the sim shows a dip in that region.
2 volts into a nominal 4 ohms load is one watt, 2.83v in to 8 ohm load is one watt, 4v into 16 ohms is one watt etc.
Sensitivity is often quoted as 1watt/1 meter, though to avoid near field measurement problems, subs may be measured at longer distance.
I'd recommend at least 2 meters for the Keystone sub due to it's height.
At two meters, there is a -6dB level drop from one meter, 4v would be required for a 1w/1m equivalent. At 10 meters, 20 volts.
In an outdoor test environment higher voltage levels reduce the influence of wind and other ambient noise pollution.
With no filters, a short sweep test of using 20 volts at any frequency would be completely safe for the 18TBW100-4.
That said, a BW24 30Hz filter should only reduce response -3dB at 30Hz, no effect at 40Hz.
Art
For the sake of science I ordered a single 18SW115-4 to see how it compares to the 18TBW100-4 side by side. The measurements below was taken at the same location as in post #1708 with the microphone two meters away. Each trace is the average of four runs, all of which were performed at 4v (measured while plaing a 60hz sine wave). Earlier measurements were done using a UMIK-1, so this time I tried using a ECM8000 along with an audio interace. No processing or EQ was applied, and LPF/HPF was disabled in the DSP.
Here is the REW .mdat for anyone curious: https://drive.google.com/file/d/1-47E9zjXvIH1C43zbU0DgOyzRZG7LVto/view?usp=sharing
The measurements are rather dissapointing, seeing that the low corner is pretty much unchanged. I can't come to any other conclusion than that my cabinets deviate too much from the original plans...
Although, I did manage to get the low corner down from to 40hz at the expense of ~3-4 dB by covering a big part of the exit. Not sure if it's something worth pursuing though.
The thing that bothers me the most is that I was really careful not to deviate too much from the original plans when creating the CAD. The known deviations are due to differences in plywood thickness, which is a mm or so. Slots were cut out for all the ribs so the margin for error during assembly was minimal.
oh, and I almost forgot: the 18SW115-4 was "broken in" for an hour or two at quite high drive levels before measuring. Not sure how much of a difference it makes, but thought I'd mention it!
Here is the REW .mdat for anyone curious: https://drive.google.com/file/d/1-47E9zjXvIH1C43zbU0DgOyzRZG7LVto/view?usp=sharing
The measurements are rather dissapointing, seeing that the low corner is pretty much unchanged. I can't come to any other conclusion than that my cabinets deviate too much from the original plans...
Although, I did manage to get the low corner down from to 40hz at the expense of ~3-4 dB by covering a big part of the exit. Not sure if it's something worth pursuing though.
The thing that bothers me the most is that I was really careful not to deviate too much from the original plans when creating the CAD. The known deviations are due to differences in plywood thickness, which is a mm or so. Slots were cut out for all the ribs so the margin for error during assembly was minimal.
oh, and I almost forgot: the 18SW115-4 was "broken in" for an hour or two at quite high drive levels before measuring. Not sure how much of a difference it makes, but thought I'd mention it!
Remember - the Hornresp TH model does not include the impact of box losses. The LF corner will therefore always be shallower and lower than predicted by the Hornresp model.
A more accurate way to determine if the build matches the sim is measure the impedance curve and compare it to the one predicted by the sim. If the peaks and dips are at the same frequencies, then the sim is a good match. Furthermore, the heights of the measured peaks (after the first one) compared to the heights of the ones in the sim is basically a visual indicator of the effect of box losses.
You could also export the sim's FR from Hornresp and import it into the program you're using to perform the measurements (REW?) to do a more direct comparison between what was predicted and what you measured.
A more accurate way to determine if the build matches the sim is measure the impedance curve and compare it to the one predicted by the sim. If the peaks and dips are at the same frequencies, then the sim is a good match. Furthermore, the heights of the measured peaks (after the first one) compared to the heights of the ones in the sim is basically a visual indicator of the effect of box losses.
You could also export the sim's FR from Hornresp and import it into the program you're using to perform the measurements (REW?) to do a more direct comparison between what was predicted and what you measured.
Thanks for posting the different responses!
Your measured relative difference between the 18SW115-4 and 18TBW100-4 from 40 to 50Hz is about 2 dB.
Your measured relative difference between the 18SW115-4 at 40 and 50 is ~2.5dB, mine is 0dB, a similar difference.
Your tests were further from any buildings than mine, that could account for at least part of the difference.
I used a B&K 4004 mic for many of the Keystone tests, an RTA-420 for others.
The RTA-420 response was -2dB at 30 Hz and -1dB dB at 50 Hz from the B&K 4004.
Your mics could easily vary by a few dB from mine.
A fair amount of testing of ECM-8000 mics have shown response at 40Hz to be more than the difference between our measurements.
I think the measurement mics and cabinet location relative to buildings could easily account for the differences noted, even if the drivers and cabinet construction were identical.
The Hornresp predictions have never matched the measured response, the effect of the Keystone exit can only be approximated.
Leosandstrom's measured frequency response deviations from mine were what he was concerned about.
Art
Hello there.
Some friends want to build a keystone (for now, another one in the future), but it's our first PA speaker (I have some experience but with home hifi speakers)
It will be used with all types of music, including DNB/DUB with sustained low shine waves
We have some doubts, I suppose many of them are from beginners...
1- Can we build it with straight cuts, and fill the gaps with glue or silicone? In the place where we buy the boards, they can only make straight cuts.
2- Should we use white wood glue for the joint and silicone to seal the joints?
3- What type of screws are recommended, width and length, and what distance between them should there be approximately?
4- Do you see this model as suitable for a first project or would you recommend another one?
Thank you very much.
Some friends want to build a keystone (for now, another one in the future), but it's our first PA speaker (I have some experience but with home hifi speakers)
It will be used with all types of music, including DNB/DUB with sustained low shine waves
We have some doubts, I suppose many of them are from beginners...
1- Can we build it with straight cuts, and fill the gaps with glue or silicone? In the place where we buy the boards, they can only make straight cuts.
2- Should we use white wood glue for the joint and silicone to seal the joints?
3- What type of screws are recommended, width and length, and what distance between them should there be approximately?
4- Do you see this model as suitable for a first project or would you recommend another one?
Thank you very much.
You can, but gaps filled with silicone are not very strong.
Use yellow glue like Titebond III.
Mix with sawdust for joint sealer.
1 5/8" fine thread drywall screws work OK, 4-5" approximate spacing, ~2" from corners. Depending on the type of plywood, countersinking and pilot holes may be required.
I would not recommend building the Keystone without the correct angles.
Can't think of any other tapped horn of similar size and response that does not also use angles, and most are more difficult to build than the Keystone.
Art
I used straight cuts for my panels when I built mine. Around the smallest panel of the horn (top panel of the horn) there is a lot of pressure building up, I dont remember if I used regular wood glue or polyurethane glue but it broke. I didn't use screws, only small nails from my nail gun. Screws+glue would probably be fine. Should be ok as a first project, just prepare accordingly and think about the steps beforehand, and/or proceed slowly. Good luck!
