Hi Chris, yeah it's a fun project in a different direction than all my previous work.
I've started getting the straight-line corner setup sounding pretty good. Interestingly, best sounding EQ has come from flattening one center section running alone, and then using that EQ set for the entire array.
Just got the CBT version running too, with amplitude shading.
It also sounds better using the single section EQ set from the straight corner tuning. Same drivers, makes some sense at least....
Different animals than my PA point source boxes for sure, ....working with speakers so room interdependent.
But I'm already thinking they may become my low volume go-to speakers. Mid range really is pretty.
I also have trouble getting good sounding low volume out of my point sources...nothing likes running at minus 40-50 dB huh??
But when it comes to big SPL with sparkling transients, ripping dynamics, and bass with balls, I already know it's back to my PA boxes. I mean, they ARE the reason I forsake home gear
I've been dropping hints in various posts that I might be interested in updating the 22-channel amplifier design that I used for my line array prototypes, but so far there hasn't been any response. Is this something that would interest enough people to make a group effort worthwhile?
Each board would have 3 ADAU1701 chips for crossovers and delay for 24 channels, and there would be a total of 10 SSM3582 chips for 20 channels of amplification, each at about 20W. There are two additional DAC outputs for driving a pair of subwoofer amplifiers. The delay, shading and crossovers would be controlled via a cell phone app using WiFi (MQTT protocol). See this article for more details: 10. Case study #3: A line array with DSP – Audiodevelopers Reborn.
Attachments
Last edited:
Well, you've inspired me to get back to my arrays - DIY Column speaker for PA use
Going for an RCF Evox killer, with 16x drivers per side and a curved array.
I'll give your EQ method (EQ one block, apply to all) a go. For now I've just been applying delays to make a curved array, and then EQing the result.
FWIW, I've got an MA5002VZ here that seems happy at -50dB - I'm using it for the lounge stereo. I'm pretty sure I could launch the cones across the room - talk about headroom...
Chris
For me, it would depend. I don't want to build a board but I could easily be tempted to buy one if the price were right. I'm not negotiating, just being honest. I've gone some distance down a different path; already have half the amps I would need for electronic curvature and I'm using PC for DSP. In fact, I would be interested in an 8 channel amp board or two, no DSP, balanced inputs.
I'm also not sure I want to do electronic curvature; with a straight floor to ceiling array it shouldn't be needed. Looking to this thread and perhaps an update to yours for confirmation. But your technology is ideal for shorter arrays, like I might use for surrounds and ambient.
Looking at chip specs, I see the AD family you are using is in the 90 db S/N range while the more expensive solutions I'm moving towards are 10-15 db better. I'm not sure that matters though given the background noise level in my home ( it did matter when I was using a high efficiency CD) and your other features (like wireless) are really attractive.
A firm requirement is for FIR, does your board do that?
Hi Mark, I'd have to take a look. It's been two or three years since I made those measurements. I don't have access to that space anymore. I'll take a look over the weekend and see what's happening with the phase. I was also pretty new to REW at the time and was using Holm before that. I would expect relatively flat phase once the delay is removed. That's what I've seen with the array, both in the old and new space.
On tapering versus shading: Shading could be done simply by using power resistors, no need for multiple channels of amplification. Tapering could also be done using passive components but gets a little more complex.
Last edited:
Neil, what an absolutely gorgeous design. Love the story behind the wood.
How much would the fully done up board and app cost?
Hi all, sorry to have disappeared for a bit....real life needed a little tending to, having been somewhat obsessed with building the boxes haha.
I haven't had any time to make further measurements or play with various shading or delay strategies, but I thought I'd at least share some subjective listening comparisons.
I've gone back and forth between the CBT and straight line versions (as well as comparing to my 4-ways).
Listening has been one speaker only, stereo summed to mono.
The straight line has been in a corner, the CBT out in the room with amplitude shading.
Both sound very nice and close to each other throughout the midrange, which makes sense given the same TC9 drivers. Very nice midrange really..
The straight line corner loading of course helps the low end, vs the CBT.
I could maybe live with the straight line in the corner without a sub, but I'd have to have a sub with the CBT. Either that, or have two lines of drivers like the Parts express CBT 36, to get stronger lows along with tighter HF spacing.
I think the biggest take away so far is in comparison to my 4-ways. As pleasing as the CBT and corner line sound as I go back and forth playing with tuning by ear, as soon as i switch to a 4-way, I realize just how much is missing both on the low end, as well as the high end. It's a bit disillusioning really....
I knew to expect less low end and plan to tie the lines to subs, but i didn't expect to hear much difference in the highs.
Maybe it's just having all the highs come from one point (bms4594he on xt1464)..I dunno....
Anyway, I still have have lots of listening/measuring/experimenting to do before I trust these early impressions. Just wanted to check in and give a quick status report...thx for listening.
I haven't had any time to make further measurements or play with various shading or delay strategies, but I thought I'd at least share some subjective listening comparisons.
I've gone back and forth between the CBT and straight line versions (as well as comparing to my 4-ways).
Listening has been one speaker only, stereo summed to mono.
The straight line has been in a corner, the CBT out in the room with amplitude shading.
Both sound very nice and close to each other throughout the midrange, which makes sense given the same TC9 drivers. Very nice midrange really..
The straight line corner loading of course helps the low end, vs the CBT.
I could maybe live with the straight line in the corner without a sub, but I'd have to have a sub with the CBT. Either that, or have two lines of drivers like the Parts express CBT 36, to get stronger lows along with tighter HF spacing.
I think the biggest take away so far is in comparison to my 4-ways. As pleasing as the CBT and corner line sound as I go back and forth playing with tuning by ear, as soon as i switch to a 4-way, I realize just how much is missing both on the low end, as well as the high end. It's a bit disillusioning really....
I knew to expect less low end and plan to tie the lines to subs, but i didn't expect to hear much difference in the highs.
Maybe it's just having all the highs come from one point (bms4594he on xt1464)..I dunno....
Anyway, I still have have lots of listening/measuring/experimenting to do before I trust these early impressions. Just wanted to check in and give a quick status report...thx for listening.
Hi and thanks ra7,
Yep, I get that shading could be done via resistors...I guess it's just too easy otherwise, since I have the amp channels to spare...
So far, on the corner version, I've only played with a little arc-creation delay, timing the sections to arrive together at listening spot. It didn't sound as good as left alone.
I think such delay will probably only have potential benefit out freestanding.
It would be very interesting to see a measurement in listening position with mic in exactly same position (i.e. not moved) and a sweep from both the 4 way and the straight line.
//
I often wondered during the last couple of years of observing and participating in these line array threads, that if the TC9 does so well, what does that say about the need for $$$ beryllium tweeters and compression drivers. I know Geddes believes that midrange is of primary importance and deprecates above 13.2 khz. I know my own hearing is already tailing off by 16 khz. TC9 users equalize to target curve up almost 20 khz, but beaming starts way earlier with a 3.5" driver so I wonder if that is it.
Are you listening on axis? If more than 10 degrees off, you may be sitting outside the beam; not totally but enough for the top half octave to be down more than 6 db from target.
It will be interesting to see what acccounts for your subjective preference but that might not be possible until you have a pair of fully tuned arrays with integrated subs.
Hi nc535, I completely agree it will probably not be possible to know what accounts for my subjective preference until full tuning on a pair with subs is done.
Until then, it's just a bit of intermingled learning and speculation.
My listening has been both on and off axis, even from other rooms.
The HF impressions are the same everywhere.
Your HF hearing sounds somewhat better than my 66 year old ears...I think my ears crap out at around 14kHz.
I'm currently believing the difference I'm hearing between the lines and the 4-way is about time, not frequency.
I'm thinking it is about multiple arrivals from about 1150Hz up...where 1/4 wave length center-to-center is exceeded with the TC9's.
The 4-way sounds like it has a higher energy, pin-point kind of sound, compared to a lower energy rounded off kind of sound from the lines.
And I think it comes off as a tonal shift.....basically from about 1-2kHz up to where I quit hearing.
Or so I'm currently speculating....
I made wings for the corner line today, to reduce any side-wall bounce. Measurements are showing it helps. I used foamular board because it was easy to try....need to get the more absorbent corning stuff..
Not arguing here, just discussing, while I wait for my own LA parts to arrive.
Multiple times of arrival gets down to the difference between an ideal line source and the discrete equivalent. That difference decreases with distance from the source so your report that your here the same delta from another room argues against that point.
Perceiving it as a tonal shift points in a different direction - to some issue with the power response. We hear sum of direct and reflected and a tonal shift where direct sound is flat or nearly so implies reflected sound has a different spectrum, e.g. reflections from off axis lack highs).
So again, I postulate beaming HF compared to relatively flat HF from a constant directivity waveguide.(or perhaps better, beaming starting at much higher frequency).
This was a big concern with me during design so I spent a fair amount of time looking at smaller drivers. None of them measured or sounded better in test boxes than the SBC65 I eventually selected. The really small ones were too hard to mount, next size up needed woofer support, various 1.5" and 2" drivers had measurement defects.
Multiple times of arrival gets down to the difference between an ideal line source and the discrete equivalent. That difference decreases with distance from the source so your report that your here the same delta from another room argues against that point.
Perceiving it as a tonal shift points in a different direction - to some issue with the power response. We hear sum of direct and reflected and a tonal shift where direct sound is flat or nearly so implies reflected sound has a different spectrum, e.g. reflections from off axis lack highs).
So again, I postulate beaming HF compared to relatively flat HF from a constant directivity waveguide.(or perhaps better, beaming starting at much higher frequency).
This was a big concern with me during design so I spent a fair amount of time looking at smaller drivers. None of them measured or sounded better in test boxes than the SBC65 I eventually selected. The really small ones were too hard to mount, next size up needed woofer support, various 1.5" and 2" drivers had measurement defects.
I see a lot of conflicting info been thrown around, lets first see what happens out in the room.
Mark, I know you don't want to "EQ the room". But you have to keep in mind what's happening here. There will be differences between an array free standing outside and one in a room corner. It is part of the design concept that it uses the floor and ceiling so you simply can't use the outside information to gather enough data to EQ the line for inside use.
Secondly, I see the old quarter wave length argument being used. Think about it, please.
If two sources are used at a frequency where they are within a quarter wavelength CTC at it's origin position, they will act as one source of sound from there on. That's the lesson, right?
If you have an array of drivers, what happens there? First we look at two drivers, super small ones. They are neatly within a quarter wavelength at say 10 KHz. Get another one in that row to make a larger array. Are both outer drivers in that array still within a quarter wavelength at 10 KHz? Even if they are within a quarter wavelength from that center driver, they won't be within that same distance over the entire length of the array, right? So it will never act as one point source as is. Even with the smallest of drivers used.
So that's not a good rule of thumb for an array to determine the boundary where combing starts, is it? You can't use point source theory on a line.
Does that mean size does not matter? Sure it does, but even a line with the smallest of drivers is still going to exhibit combing, largely depending on the actual distance from the array. Smaller drivers do have their advantages though. I could say lots more about that, but first let this sink in. Think about what happens to your quarter wave rule. If you expect a cylindrical wave front you'll be disappointed.
For the floor to ceiling array I would advise to measure at listening distance out in the room. Seeing your corner there I would advise to use many measurement points around the (desired) listening position(s) and average them. I'd also advise to start using DRC or something very similar with at least a frequency dependent window function available. How else will you learn what truly happens out in your room?
I see a lot of vertical ridges in that photo, parallel to the array. I must have warned you about that, i always do. Every single ridge like that is going to turn up in your measurements. That's why I strongly suggest using the multiple measurements to average, to filter out the placement dependent results from those ridges.
I also see a slanted ceiling, you've probably mentioned it before, I don't remember. But it is a discontinuity for the "floor to ceiling" concept. How much that influences things we will have to see, I think it could still work, but it just might be suitable for some type of shading on the top end.
I do believe we must first come up with a game plan, before we start criticizing the array as they sound right now. I agree with nc535 that the balance between direct and indirect sound is not going to be ideal here from what I see so far.
I do believe we've had lots of lessons by Dr. Geddes about horns and discontinuities within that horn. Think of your corner walls as the inside of that horn. Does it show discontinuities? Do they matter?
The answers you seek are not solved by measurements taken from a joining room. They could be made more clear by first making a plan of how to get the best out of what you have to deal with.
In this case I'd get that array out of this corner. The corner simply doesn't support the idea of horn loading as it will create more havoc than it solves.
I'd bring it into the room with the baffle about half a meter from the back wall and play with toe in while collecting measurement data.
I'd use damping on that short wall up to the doorway on the right, but I know you weren't too keen on solutions like that.
My own decision to go for a free standing array were based on a corner much like yours. I simply could not see that corner being beneficial to use for that purpose.
I'm real sorry to be this blunt, I just look at it from an engineering point of view. Nothing personal here, no offence ment.
Mark, I know you don't want to "EQ the room". But you have to keep in mind what's happening here. There will be differences between an array free standing outside and one in a room corner. It is part of the design concept that it uses the floor and ceiling so you simply can't use the outside information to gather enough data to EQ the line for inside use.
Secondly, I see the old quarter wave length argument being used. Think about it, please.
If two sources are used at a frequency where they are within a quarter wavelength CTC at it's origin position, they will act as one source of sound from there on. That's the lesson, right?
If you have an array of drivers, what happens there? First we look at two drivers, super small ones. They are neatly within a quarter wavelength at say 10 KHz. Get another one in that row to make a larger array. Are both outer drivers in that array still within a quarter wavelength at 10 KHz? Even if they are within a quarter wavelength from that center driver, they won't be within that same distance over the entire length of the array, right? So it will never act as one point source as is. Even with the smallest of drivers used.
So that's not a good rule of thumb for an array to determine the boundary where combing starts, is it? You can't use point source theory on a line.
Does that mean size does not matter? Sure it does, but even a line with the smallest of drivers is still going to exhibit combing, largely depending on the actual distance from the array. Smaller drivers do have their advantages though. I could say lots more about that, but first let this sink in. Think about what happens to your quarter wave rule. If you expect a cylindrical wave front you'll be disappointed.
For the floor to ceiling array I would advise to measure at listening distance out in the room. Seeing your corner there I would advise to use many measurement points around the (desired) listening position(s) and average them. I'd also advise to start using DRC or something very similar with at least a frequency dependent window function available. How else will you learn what truly happens out in your room?
I see a lot of vertical ridges in that photo, parallel to the array. I must have warned you about that, i always do. Every single ridge like that is going to turn up in your measurements. That's why I strongly suggest using the multiple measurements to average, to filter out the placement dependent results from those ridges.
I also see a slanted ceiling, you've probably mentioned it before, I don't remember. But it is a discontinuity for the "floor to ceiling" concept. How much that influences things we will have to see, I think it could still work, but it just might be suitable for some type of shading on the top end.
I do believe we must first come up with a game plan, before we start criticizing the array as they sound right now. I agree with nc535 that the balance between direct and indirect sound is not going to be ideal here from what I see so far.
I do believe we've had lots of lessons by Dr. Geddes about horns and discontinuities within that horn. Think of your corner walls as the inside of that horn. Does it show discontinuities? Do they matter?
The answers you seek are not solved by measurements taken from a joining room. They could be made more clear by first making a plan of how to get the best out of what you have to deal with.
In this case I'd get that array out of this corner. The corner simply doesn't support the idea of horn loading as it will create more havoc than it solves.
I'd bring it into the room with the baffle about half a meter from the back wall and play with toe in while collecting measurement data.
I'd use damping on that short wall up to the doorway on the right, but I know you weren't too keen on solutions like that.
My own decision to go for a free standing array were based on a corner much like yours. I simply could not see that corner being beneficial to use for that purpose.
I'm real sorry to be this blunt, I just look at it from an engineering point of view. Nothing personal here, no offence ment.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.