First time DIY, Beyma 8" Coaxials

Either zip them, or rename them to TXT files, both work for attachments.
 
Either zip them, or rename them to TXT files, both work for attachments.
Thanks, I renamed to .mdat.txt but both times the attachment never appeared, strange it's worked before.

Not a problem, I think I'll setup a public folder on my server for all this stuff.
 
most of these had some sort of octave smoothing on, are you basing the "speaker" correction on this?
 
just looking at the correction filter, you showed

1664922430760.png


and (I think) said the W correction was based on NF which looks like a fairly typical rolloff

1664922477491.png


I don't see how you get from one to the other

fwiw the mdat is v hard to follow because it contains a mix of measurements + intermediate stuff

what you might say is a traditional approach does work well... take measurements (not just on axis) and as quasi anechoic as you can, decide how to combine them to design purposes (listening window, power response, some combination or whatever) and then design filters against that, listen, repeat to fade.
 
just looking at the correction filter, you showed

View attachment 1762052

and (I think) said the W correction was based on NF which looks like a fairly typical rolloff

View attachment 1762053

I don't see how you get from one to the other
The filter is a combination of the woofer correction, filters 4,5,&6 and tweeter 1,2&3. The 24dBLR crossover was done in the Crown XLS1502 power amp so after the processors PEQ. I then added filter 7 an additional correction from the measurements I got from the combined filers and crossover.

I did the woofer correction with two PEQs instead of a shelving filter because I didn't want to elevate <120Hz too much. I guess I could have used a shelf to level the rolloff and a steep HPF
to protect the driver, I guess that would have shown clearer intent.

fwiw the mdat is v hard to follow because it contains a mix of measurements + intermediate stuff
Sorry, I thought I'd picked out the relevant measurements so you could see what I started with and the progression as the crossover and each PEQ was added.

What would be better to share?

what you might say is a traditional approach does work well... take measurements (not just on axis) and as quasi anechoic as you can,
Is there some canonical guide to this?

I used these resources to get me started:
http://audio.claub.net/tutorials/FR measurement using ARTA.pdf


decide how to combine them to design purposes (listening window, power response, some combination or whatever)
Listening window is about dispersion, angles from the axis of the driver?

From here:

They talk about
The five frequency response measurements that are averaged for the Listening Window are: on-axis, 15 degrees left and right off-axis, 15 degrees up and down off-axis.
In your opinion is 15 degrees enough for a coaxial or should I ignore on axis and average 15 and 30 degrees, to avoid over correcting the on axis response?

I just took one 15 degree off axis in both planes measurement that looked typical and worked from that, that's the tweeter measurement that's in the mdat I posted. I have the 15 degree incremental measurements I posed earlier I can re-work from those or do them again.



Power response is totally unknown to me, I found this:

but it doesn't help me understand what I should be considering.

The only possibly related thing I'm doing is EQing down mid woofer response (filters 4&5) to achieve a flatter response down to 120Hz but I'm not sure that the same thing at all.


and then design filters against that, listen, repeat to fade.
That's what my measurements were, I thought...

I appreciate any help but I'm a bit confused now.
 
This looks interesting, just dropping a link here so I don't forget about it.

 
Car passed it's MOT without serious work required :)

So I ordered three more drivers, I can build LCR & SL SR and get a real situation for these drivers. There were three in stock at Blue Aran which was the best price so they are on their way. :cool: I also ordered parts to make proper tweeter protection and 10dB L-Pads.

If it turns out there are slight differences with the older eBay drivers I have, those two can be surround and the new three for LCR.
 
Is there some canonical guide to this?
not really, https://kimmosaunisto.net/Software/VituixCAD/VituixCAD_Measurement_REW.pdf gives a good view on how to measure but leaves you to your own devices on how to design the XO. You just have to read a bunch of threads basically and look at other designs.

Listening window is about dispersion, angles from the axis of the driver?
perhaps a picture helps, this is my CD 0-45 degrees at ~7.5 degree steps. The highlighted line is a vector average of those (a quick example of what you might use as a listening window measurement). Power response takes that further and is basically the total power radiated by the speaker

the Q is which one do you use as the target for your speaker? or do you look at all of them and consider them in total? or do you combine them with different weights in some way? There's no right answer to this but it's definitely the case that better results should be achieved by not just using a single measurement (or even just a few close to axial response)

1665004179233.png
 
Sorry, I thought I'd picked out the relevant measurements so you could see what I started with and the progression as the crossover and each PEQ was added.
it's the naming of the measurements that is a bit confusing along with the not so obvious EQ references

e.g. what does the XLS suffix mean? multiple measurements called "add peq" or a 1.5k xo without knowing what the xo is

I was only really looking for the raw measurements anyway to see what you were basing it on
 
looking at the T measurement itself, I'd look into the measurement environment. It appears there are some significant reflections early on which don't seem to be close enough to be the driver or the box

1665042180151.png


1665042265939.png


for comparison, the measurement I used (which was done in less than ideal conditions due to the time of year)

1665042397025.png
 
comparing those responses, the 2.5k peak on yours is really large

1665042597779.png
 
@mattkhan

Thanks for the input, really appreciated. :cool:

I'm going to re-measure, I may be able to try in the garden today, as long as I don't miss the DPD guy with my new drivers! Our garden is really small and complicated but ought to be a little better than the living room, if not brother in law has a huge empty patch of lawn with nothing around it.

I'm ok with crossover design, it's just basic electronics, I studied that at college, I can remember enough. :cool: The things I'm not up with yet are the realities and practicalities of measurement and cabinet building, learning that very slowly. I think I'm assuming I know more than I actually do...

With the labelling, I should really use the description field more, so I can be less terse... Sorry, next time I post some measurements I'll explain better, in advance.

XLS is to remind me I used a Crown XLS 1502 amp, Woofer on CH1 Tweeter on Ch2.
The crossover (in the Crown XLS) can only be 24dB Linkwitz Riley, which is why I didn't specify it. I tried a bunch of Xos 1.5k, 2k, 2.5k and 3k and the 1.5k sounded best and was smoothest. It can do staggered Xos but only at crude 0.5k intervals so I didn't try that. I'm not sure about committing to using Crown XLS amps for crossover anyway, I'm just using it because it's here and allows me to listen to a few things without building crossovers.

I have no idea what those reflections could be, there really wasn't anything I can think of that could cause that timing. I'll look closer at my next set of measurements, I'm really only used to looking at room reflections, not sure what I'm doing yet, that close in.

I posted a pic of the setup I used for measuring here.
#45
I'm as far from the walls as I think I can be, the room has heavy curtains and 350mm to 500mm bass traps in 6 corners, I expected most problems from the back wall as that's only treated with curtains.

As to the giant 2.5k peak, h'mm. I'll measure the other driver I have and the new ones will be here before lunch so see how they compare. Encouraging that yours is lower, maybe it's down to my poor woodworking, unfortunate co-incidence of resonance between the horn and my box? I will add some damping, more bracing and compare.

Thanks
 
Didn't get time for measurements today, I did get as far as unpacking the new drivers, checking them with DATS, freeair sweeps and played a bit of pink noise and some sweeps through them to hear if they were ok.

22-10-06 16-03-34 6502.jpg


I've overlaid all the DATS freeair sweeps to compare the drivers.

CX8-AllWoofers-Freeair.PNG

CX8-AllTweeters-Freeair.PNG


The only real discrepancy is the eBay driver I haven't done anything with yet, it's the green one in both graphs. Might be nothing, don't know how repeatable drivers usually are in this test.

Hoping for some weather and a gap in work to do some outdoor measurements tomorrow.
 
The only real discrepancy is the eBay driver I haven't done anything with yet, it's the green one in both graphs.
The cyan one looks different too as far as I can see

I posted a pic of the setup I used for measuring here.
The massive mic holder looks bad for this purpose, the ideal is a really long pole exactly the diameter of the mic housing (never worked out how to make such a thing myself)
 
The cyan one looks different too as far as I can see
Yeah, I ran several sweeps on each driver and there seemed to be small variations each time so hopefully it's not too significant.

Edit: Cyan is one of the new ones.
The massive mic holder looks bad for this purpose, the ideal is a really long pole exactly the diameter of the mic housing (never worked out how to make such a thing myself)
Ah, thanks, that I can change easily.

The umik holder would be better, it could screw to the end of the boom. Otherwise a bit of 22mm drainpipe might fit over the umik, if I slit it so it can expand and a bit of foam inside to wedge it over the end of the boom, gaffer tape to suit.

If it works I'll post a pic, there has to be some use for the offcut bits I have left over from the bathroom. :D
 
Overflow and drain pipe offcuts didn't fit but maybe this?

22-10-07 10-27-35 6504.jpg

22-10-07 10-27-02 6503.jpg


15mm pipe lagging, have to wedge it right up to the back of the umik to stop it sagging but so far it's been there for 20 mins and no sag.

Otherwise I can glue a tripod thread to the back of the umik and screw it directly into the end of the boom.

Going to try this first though, tapered the front face of the lagging to make that less frontal area.
 
Too windy outside for measurements today :(

I learned a few things in between prep for a project on Monday. I can use the DATS hardware to sweep the driver for measurements, it works fine as a USB audio device in Windows, not in MacOS for some reason. I installed Windows 10 bootcamp on my old spare Mac laptop, won't run modern MacOS but it will run Windows 10 with a bit of futzing about and that's now filling up with HT related software and stuff. :cool:

REW is just horrible on Windows, asio drivers and a lot of silliness.

So I am now ready to do portable measurements, middle of the local cricket ground seems favourite. :D


I tried the pipe lagging umik thing and I can't see that it makes any difference. I took a bunch of sweeps and they all look the same, these are a couple for comparison.

MicClip.png



What does make a difference is moving the sofa around so it's right under the mic, that removes a small wobble but as it's the earliest one I can see properly I think that makes it important.

Like this, the orange and red have the sofa in the better position.

MicClipSofaPosition.png

The mdat has a bunch of these measurements, I did three of each in case there was some extraneous noise in one of them.

Lag is the umik in the pipe lagging
bigclip is the horrible mic stand clip.
sofa 1.5m is the sofa in it's usual spot like this.

22-10-07 16-50-59 6506.jpg


sofa 1m is the sofa moved forwards like this.

22-10-07 16-50-29 6505.jpg

mdat, actually now a public folder for this project.

I don't see any difference with the mic clip, I'll take some outdoor measurements as VituixCad requirements and see where that gets me.
 
the 1.5m is definitely worse

you can see some difference with the clip but that gets swamped by everything else going on so agree it's not really any benefit in this setting
 
to highlight a couple of points about the 1m measurement

firstly, it seems inverted
secondly, once the speaker itself dies down then you can see a lot still going on such that there is no obvious first reflection. This is good in some respects (something must be damping down the first reflection) but not in others (there's too much signal bouncing around making some degree of mess so your freq resp is noisy

1665213517025.png



for comparison, taken roughly in my yard (which isn't massive and has big hard brick walls nearby but no ceiling cos it's outside and the speaker was a reasonable distance from the floor)

1665213681230.png


sharper positive peak + a more obvious 1st reflection

I think this driver is inherently difficult to measure mind you, bound to get a load of diffraction mess given it's built into the design
 
FR comparison

1665213814020.png


yours is bouncing up and down throughout the range as a result of those reflections & you can't really window them out either
 
to highlight a couple of points about the 1m measurement

firstly, it seems inverted
I'll double check the wiring, I just checked and the amp I'm using doesn't inherently invert so I don't know why it would be inverted.

That said, the first peak is an upswing, so why do you say it's inverted?

I've just been reading the REW guide on this and they don't say anything about using the largest peak rather than the first peak, what am I missing here?

I looked at some other sweeps I did to test with the DATS hardware and they look very similar but with a little more symmetry around the first swing.

DatsSweep.png



secondly, once the speaker itself dies down then you can see a lot still going on such that there is no obvious first reflection. This is good in some respects (something must be damping down the first reflection)
There are thick three layered curtains on the back and side walls, ceiling is bare though.

I tried covering the ceiling fan with a soft throw but that didn't seem to be the cause.


but not in others (there's too much signal bouncing around making some degree of mess
There is a bookcase near the back wall, maybe there's enough diffraction from that to create the mess?


so your freq resp is noisy

View attachment 1763693


for comparison, taken roughly in my yard (which isn't massive and has big hard brick walls nearby but no ceiling cos it's outside and the speaker was a reasonable distance from the floor)

I think I'm starting to see what you are looking for in the traces, initially I thought they all looked the same.



View attachment 1763694

sharper positive peak + a more obvious 1st reflection

Ok, sun is out, not enough wind to go sailing, wife is making soup, time to try some outdoor measurements.

I think this driver is inherently difficult to measure mind you, bound to get a load of diffraction mess given it's built into the design
Very diplomatic, thank you.
 
first peak is an upswing, so why do you say it's inverted?
Seems not is, i.e. it looks like it might be. I don't recall seeing such behaviour (extremely strong extremely early reflection) before, depending on your appetite for risk/tweaking you might try reseating the diaphragm on the CD to see if it changes. This could also be worth trying on the driver that behaves differently to the others on the impedance sweep.
 
Seems not is, i.e. it looks like it might be. I don't recall seeing such behaviour (extremely strong extremely early reflection) before,
I haven't seen much of anything before and it looks pretty wonky to me too.
depending on your appetite for risk/tweaking you might try reseating the diaphragm on the CD to see if it changes.
No problem, I've done a lot of miniature camera work, microscopic electronics and lens rebuilding, I have suitable tools, I'm happy to have a go.

This could also be worth trying on the driver that behaves differently to the others on the impedance sweep.
I'm going to try a few of the other drivers in the box too, that should tell me if it's the box or the drivers it seems it's not the environment.

I managed to do a couple of sweeps in the garden today, not easy working there, our lawn is very uneven. I think the sweep looks better and I got first reflection at 8ms, 1.9m off the ground. Doing a full spin is going to be difficult, I need a better mounting system to be able to do that.

22-10-08 12-58-02 6508.jpg


22-10-08 12-58-15 6509.jpg


I think I need to rig something with my biggest camera tripod, it's a heavy box to put on a camera head though.

Ouside1.9m.png


Cleaner after the initial mess but initial mess is just as messy, I guess this means I can work on the box and the driver inside and only go back out when I've got a cleaner initial impulse.

Ouside1stattempt.mdat in the same folder.

Edit wind picked up again and that stopped me getting any more measurements, although I think they are not useful yet.
 

The latest video from AVForums

TV Buying Guide - Which TV Is Best For You?
Subscribe to our YouTube channel
Back
Top Bottom