TC6WD (PE buyout) Line Array - Lessons Learned

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
First of all, thanks to everyone on the diyaudio forum, and the creators of my "inspiration" projects below, for all the helpful information you've shared here. I'm also indebted to James R. Griffin for the paper Design Guidelines for Practical Near Field Line Arrays, as well as published works by Don Keele and Floyd Toole.

Some impressions and lessons I learned from building floor-ceiling corner line arrays of 40* 2" TC9WD drivers:

Design inspiration

Roger Russell "IDS 25" (full range array of 25* 3.5" TC9FG)
Two Towers (full range array of 25* 3.5" TC9FC)
Kuze 3201 (shorter full range array of 24* 2" TB drivers)
Murphy Corner Line Array 24 * 3.5" ND90-8

Project details

Drivers

Drivers are Vifa/Peerless/Tymphany TC6WD02-04 and can be purchased in crates of 80 for $2.98/unit.

They are pretty sensitive through the midrange, but quite low in the treble, so they require a lot of EQ. Since line arrays require EQ anyway, this isn't a huge deal. Importantly, they have a lot of Xmax, strong motors and lightweight cones.

Each line has 40 drivers divided between 5 cabinets. Each cabinet module has 8 drivers wired in series. Cabinets are wired in parallel. At the time, I wasn't aware of any theoretical issue with series wiring drivers in the same cab, nor have I experienced any practical problem so far. Since the drivers are 3.2 ohms, each cabinet totals 25.6 ohms and when the 5 cabinets are paralleled, system impedance is 5.2 ohms. Going the other way would put each cabinet at 0.4 ohms for a total system impedance of 2 ohms, which is pushing it.

Cabinets

I wanted to mount these in the corners since I have very little space, and also because I would prefer to reduce the imaging effect of side and front wall reflections. MCLA design was a starting point, from which I added a bend into each "wing", making the shape closer to a 1/4 cylinder.

To maximize the internal volume while maintaining rigidity and minimizing spacing from the front/side walls, I used very thin (~1/4") birch ply walls, braced in between each driver. The wings are thicker - 5/8" plywood - and the drivers are mounted in a double layer of 1/4" ply. The other reason I selected such thin wood was to employ laser cutting for the ribs, spine, and mounting baffle. The driver magnets are being hugged by the spine, so the mass of the drivers dampens the lightweight cabinets.

EQ and Crossover

A laptop computer running Linux provides the prototype DSP. An 8-output $20 USB soundcard allows for up to 4-way crossovers. Analog or digital inputs can be processed through the computer with minimal latency. For better quality DACs, I have used an Asus Xonar PCI-E card in a desktop computer.

Using JACK and the Calf Plugins pack LR2, LR4, or LR8 crossovers can be implemented as well as EQ for each output. This is a great system for tweaking since you can adjust the crossovers and EQs in realtime. Measurements are performed with Qloud, adjust, measure again. I can (but usually don't) generate FIR filters in Windows using Rephase (manual lin phase EQ / xover) and/or IK Arc 2 (automatic room correction) and run them in JACK using Jconvolver.

I have a MiniDSP 2x4 HD on the way which will be able to implement all of this (for 4 outputs only) in a much more compact and efficient setup.

Listening impressions

When I was half done wiring up the cabinets, I decided to set one up for a mono impression. I was *extremely* disappointed. The sound was incredibly dry, with basically 0 ASW (apparent source width). Without EQ, the tonality was downright horrid, which I expected, but not to that degree. After measurement and EQ, the tone was decent, but the lack of ASW was quite troubling. This points to an almost complete lack of room interaction - there are no floor, ceiling, side or front wall reflections to broaden to sound. You simply get a very direct sound.

A few months later, I decided that it would (at least) take up less space to put the remaining drivers into their cabinets and get them mounted in the corners. I was pleased to find that the same qualities which made the speaker sound so lifeless in mono provided for *exceptional* stereo imaging - that is, if the source material had appropriate stereo information.

The good

Where the system really excels is in true stereo recordings. Obviously, classical music is the obvious candidate here - I have probably found the most enjoyment from listening to completely unprocessed recordings posted on Gearslutz, where location recording engineers tend to post microphone comparison clips. XY (coincident 90 degree pair) recordings have a strong mono presentation and limited width. Spaced pairs have a Left-Right dominant soundstage with a recessed center. Equivalent stereo techniques sound simply REAL, that is, where mics are placed 10-30cm apart and angled outward. The ability of the system to accurately present true stereo information is in my experience (which includes mixing, mastering, and film studios) just unparalleled.

Any recording which has a substantial mix of true stereo elements will have the ability to transport you into the recorded environment. Think 90s rock recordings which had dominant drum overhead pairs or stereo room mics - Steve Albini, etc. Compression and EQ sounds like.. compression and EQ. Any deviation from reality is quite obvious and perceived as an effect. One exception to this would be stereo reverbs, which sound more realistic than on a typical system due to the high degree of envelopment with which the system presents out-of-phase information.

Midrange distortion is absurdly low. Between 150Hz and 5kHz you won't find even the second or third harmonics hitting 0.1% - higher order harmonics are down at 0.01%. This contributes to a transparent, effortless, natural, non-fatiguing sound (depending on the source recording of course!)

I would definitely repeat the "modular" concept though as it makes moving them much easier, and for versatility in installations (line length can vary depending on ceiling height, listening distance required..)

The not-so-good, not-so-bad

Mid-side oriented material (a lot of modern pop music, hip hop is mastered this way - the mid (stereo sum) and side (stereo difference) tracks are EQ'd and compressed independently) takes on a surreal quality, with a dry and highly localized center channel, but outrageously wide side channel (which really does seem to be coming from the sides - not frontally located at all). This is pretty fun to demo, since it just sounds totally unlike any speaker that has substantial room interaction.

Bass performance, as one might expect from 2" drivers in a small cabinet, wasn't usable below 100Hz, though a substantial improvement over a single driver. At a reasonable SPL (louder than typical listening volume) distortion exceeded 1% only below 150Hz. This was low enough to cross to non-coincident woofers, placed on the front wall a foot or two inward of the corners.

The bad

Pan-potted mixes (mono tracks mixed in varying degrees between left and right channels) maintains the troubling "dry" sound with eerily precise imaging and 0 envelopment. Without stereo reverbs, this kind of mix sounds really quite bad here, since it relies on the playback room to liven up the dry recorded sounds.

Running the line array floor-ceiling (5 modules) actually made things worse in my room than running only 3 modules (from the floor to just below standing height). In an ideal room (simulation), the line would sound the same at any height. In a real room, there's a strong asymmetry top-bottom... ie. there's a bunch of furniture and diffusing objects in the lower half of the room, but practically none at the ceiling. This led to an increased level of reverberant energy, even some nasty flutter echo, up near the ceiling in my room. Subjectively, this colored the midrange in a nasty way (flutter echo), and pulled the image upward, away from ear level. There was some material that this sounded kind of good for, but I didn't want the system introducing false height cues. At my listening distance and crossover frequency, the line performs just as well in the lower ranges even at 3 modules high. Lower frequencies couple to the ceiling despite the gap.

Finally, the main problem is the comb filtering above 1 wavelength center-to-center spacing. Based on the "inspiration" designs linked to above, I had formed the impression that the comb filtering mainly affected the high frequency rolloff, and could be compensated for with EQ. This has absolutely not been my impression.

No matter how much I adjusted the high frequency EQ - flat voicing, straight slope, rolloff curve - it just sounded "off". The best voicing I could come up with was a straight line from 20Hz-20kHz, 10dB down at the high end. (Note: I'm referring to the measured response, not what the EQ curve looked like).

I wasn't noticing any particular deviation in the frequency response as I moved around the room, so I didn't initially suspect comb filtering. I re-read James Griffin's NFLA paper and it suggested that the issues I was perceiving, in the frequency range I was perceiving them, were related to the center-to-center spacing imposed by the 2" drivers. At 2.25", the one-wavelength rule corresponds to a frequency of 6kHz. Above this frequency a "blurry" and "dull" sound was exhibited that EQ couldn't fix.

What I would do differently

First of all, I would design as a 3-way system from the get-go. Having crossed the lines LR4 at 4.5kHz to a single ring-radiator tweeter (placed just outside of the lines, still in the room corners), the treble problems completely disappeared while still maintaining the benefits of insane imaging and vanishing midrange distortion. I'm curious to try a tweeter line but can't find any suitable (1/2" with low Fs) drivers except these terribly inefficient Daytons.

Printing and assembling all the laser cut pieces was very time consuming. I would opt for a simpler, (if thicker) cabinet next time. Maybe nothing but a curved baffle that can be sealed to the walls.

Magnet-mounting the drivers via the laser-cut spine greatly complicated assembly. There is not a lot of "give" to the plywood, and certainly none on the magnets, so any increase in magnet size (there's a pretty loose tolerance given) meant hand-filing the spine to accommodate the extra size. I would skip this next time as well, unless I can think of a nice compliant material that wouldn't bunch up or tear as I squeezed the drivers in...

One of the greatest time wasters was preparing the wires - cutting, stripping, twisting and tinning. I found that you can buy short tinned copper wire "jumpers" on eBay for next to nothing, so I will be using these next time.

I might consider using larger drivers such as the TC9 next time, but that would force my crossover a bit lower. From my experience, it is worth staying far away from the 1-wavelength combing frequency with the crossover. If I were using the Dayton 1/2" drivers linked above, a 3kHz crossover may be feasible, or even less, depending on distortion. For a single tweeter (which obviously has some integration issues with a line), a 1 1/8" driver such as the RS28 could work.

To deal with the increased reverberant energy near the ceiling, I would plan for a sitting-height-only array. When listening distance or low frequency response dictates a longer line, I would consider frequency tapering the line somewhat so that only low frequencies were emitted by the highest module(s). Room treatment would also help a lot - just a mix of absorbers and diffusors on the sidewalls.

Finally, a large diffusor around ear level placed on the front wall in between the columns would almost certainly improve the sound. I plan to install some wood 2x2 QRDs which I built a few years ago - will report back.

Conclusion

Certainly a ton of work, but a worthwhile project. Since domestic line arrays are so rare in the real world, I really just wanted to hear them, and I'm glad I did. For natural, immersive stereo reproduction, the corner line arrays can't be beat!

But, make sure that you take the center-to-center spacing comb filtering seriously. I wonder if the predominantly older guys who have designed say the IDS-25 and MCLA are relying on the measurements, rather than their hearing, in the HF (which of course, indicate a correct treble response with enough EQ). Just trying to figure out why they have decided that this is a "non-issue" despite the theory, and contradicting my own observations.

Another explanation which is often offered is that the increased directivity from a 3.5" driver would mitigate the combing effect, but the frequency at which the driver starts beaming is proportional to its size - and so is the comb frequency. I mean, the TC6 starts beaming at 6kHz but it certainly didn't avoid combing issues above that frequency. For me, the line arrays just unusable past that frequency, which is exactly what the theory suggests.
 

Attachments

  • tc6 frequency and impedance response.png
    tc6 frequency and impedance response.png
    71.6 KB · Views: 1,106
  • fullheight.jpg
    fullheight.jpg
    138.3 KB · Views: 459
  • angle.jpg
    angle.jpg
    74.8 KB · Views: 385
  • 3-modules-plus-ot19.jpg
    3-modules-plus-ot19.jpg
    42.9 KB · Views: 432
  • single-module-with-bc25.jpg
    single-module-with-bc25.jpg
    75.7 KB · Views: 432
  • lasercutting.jpg
    lasercutting.jpg
    85.2 KB · Views: 291
  • inside-cab.jpg
    inside-cab.jpg
    145.7 KB · Views: 1,078
  • ribs.jpg
    ribs.jpg
    98.6 KB · Views: 1,093
  • construction-2.jpg
    construction-2.jpg
    130.8 KB · Views: 1,099
  • construction.jpg
    construction.jpg
    115.7 KB · Views: 1,107

ra7

Member
Joined 2009
Paid Member
Hi,

Thanks for describing your journey here. I too have a very similar corner array built out of TC9s:
http://www.diyaudio.com/forums/multi-way/284371-corner-floor-ceiling-line-array-using-vifa-tc9.html

I like your idea of a curved cabinet. My cabinets are triangular in shape and I tried to make the front baffle as small as possible to get the drivers close to the corner.

In terms of EQ, you mention several options at your disposal. Can you show a FR measurement at the listening position? Maybe 1/12th octave smoothed? If you are familiar with REW, it has an excellent new feature called frequency dependent windowing that really allows us to see what we hear. Basically, it has a long window at low frequencies and short window at high frequencies. Try it out.

What is your listening distance? Comb filtering is a problem only at short distances. I tried to investigate this phenomenon and you can see my measurements and conclusions here (posts 127-130):
http://www.diyaudio.com/forums/mult...line-array-using-vifa-tc9-13.html#post4584879

Take a close look at post 128. Comb filtering moves higher up in frequency the farther you sit from the arrays. CTC is obviously important, and in an ideal world we would have infinitely small drivers with 20-20 kHz bandwidth. But even with the 3.5" drivers, if you sit far enough, comb filtering can be avoided. In your case, this distance would be even smaller. I recommend making some measurements and seeing how comb filtering changes with distance. Typically, it moves up one octave for every doubling in distance. Can you try them in a bigger space?

I'm also interested in learning more about your experiments with a tweeter. Can you post measurements?

EQ is extremely crucial to get these arrays to sound right. I use DRC.
DRC: Digital Room Correction

This is a fantastic piece of software to do auto correction. It is a bit arcane, but there is a thread here in the full range forum that can help you get started.

Don't give up on these just yet. In my opinion, this is the ticket to home hi-fi. Small footprint, low distortion, avoidance of floor/ceiling reflections. These are important issues in sound reproduction in the home.
 
Last edited:
Thank you dynomike for the excellent writeup! Nice work too.

Having not heard tall line arrays, I have been questioning the treble quality because of comb filtering. Kind of nice that you heard it, but still a bit disappointing, because the concept is charming. Your notions about imaging vs. recording techniques was an eye-opener. Your room looks quite small and perhaps listening distance is sort of too small for LAs. I will follow this thread!
 
Last edited:
Welcome to the club!

I built a pair of 3-way line arrays 3 years ago for my garage, the idea was to keep the sound off the floor/walls/ceiling.

I went with 12 five inch woofers crossing to 21 three inch mids at 350Hz then the mids crossed to 48 10mm (0.4") dome tweeters at 6KHz. The center to center spacing on the mids is 3.3" and the tweeters at 1.3". Run the EQ at a boost of +7dB at 15KHz and it is very close.

After wiring in 81 drivers per speaker or 162 drivers total, I can attest to the time waster portion of your experience. On paper, I should have some combing at around 4,300Hz to 6KHz and over 10KHz... They are not perfect but the way they work overcomes those issues since they are in a garage with cement floor--something has to give.

I use subwoofers with the arrays but the total of 24 five inch woofers will give good response down to around 70Hz in the sealed box. Punch a hundred or more watts into them and you'll fell the chest hit which is fun at BBQs.

If I was to do it again, what would I change? Well, I'd increase their size from just over 6 feet tall to almost 8 feet, use 25 of the Vifa 3.5" full ranges crossing to 72 of the Dayton 5/8" dome tweeters at 4,500 Hz. Set the crossover to the subs at 100 to 150Hz depending on use (block party would be 150Hz) and that would do it.

Many people love those line arrays in the garage, when they ask if I can "help build a pair" I decline! They are a great way to learn about the dark side of audio, it will increase your wood working abilities and very educational to get them to work properly. But only once...
 
Could you post an IR of the line array doing the full range duty?
Judging the original response of the driver I'm not completely convinced comb filtering is to blame for what you are hearing in the higher frequencies. It's an easy target to point the finger at, but with a line array you get the sound of the driver that is used on steroids. IIR EQ alone is not going to fix that.
There has to be a reason why the driver does not have smooth FR above 6 KHz by itself. Seeing an IR and STEP response of a single driver would help too.
 
Thanks all, for the replies and advice so far! Sorry I haven't posted measurements; I wrote the OP on the bus, and I'm still out of town.

Could you post an IR of the line array doing the full range duty?
Judging the original response of the driver I'm not completely convinced comb filtering is to blame for what you are hearing in the higher frequencies. It's an easy target to point the finger at, but with a line array you get the sound of the driver that is used on steroids. IIR EQ alone is not going to fix that.
There has to be a reason why the driver does not have smooth FR above 6 KHz by itself. Seeing an IR and STEP response of a single driver would help too.

This is a very good point.. I did EQ a single driver before and thought the results were good, but it was a while ago. I will try this and post back.

I am pretty afraid of that 200+ page thread of yours wesayso.. but I'm gonna have to dig through it to find the gold huh? I did check out the posts you had linked in the OP and saw a lot of good stuff about DRC and windowing for EQ.
 
Thanks all, for the replies and advice so far! Sorry I haven't posted measurements; I wrote the OP on the bus, and I'm still out of town.

This is a very good point.. I did EQ a single driver before and thought the results were good, but it was a while ago. I will try this and post back.

I am pretty afraid of that 200+ page thread of yours wesayso.. but I'm gonna have to dig through it to find the gold huh? I did check out the posts you had linked in the OP and saw a lot of good stuff about DRC and windowing for EQ.

It could be worth a shot to try a FIR correction. If the TC6 is reasonably well behaved it would be worth it. Sorry about the length of the thread :D. I did try to put up some useful links in the first post to make it accessible (as you noticed). One cool thing to try would be runing the APL_TDA demo to see your timing at the listening position. I'm not sure it will run on Linux though.
 

ra7

Member
Joined 2009
Paid Member
The TC6 looks fine to me. It has a strange shelf near 3 kHz, but it is well-behaved off-axis, so I presume there is no anomaly. I think it is your listening distance. How far away is your listening position? From my measurements, you need to sit at least about 1.5 to 2 times the array length away from the array. The smaller drivers in your array could make this distance smaller. Raw measurements at the listening position would help to see what you are trying to EQ.
 
Could you post an IR of the line array doing the full range duty?
Judging the original response of the driver I'm not completely convinced comb filtering is to blame for what you are hearing in the higher frequencies. It's an easy target to point the finger at, but with a line array you get the sound of the driver that is used on steroids. IIR EQ alone is not going to fix that.
There has to be a reason why the driver does not have smooth FR above 6 KHz by itself. Seeing an IR and STEP response of a single driver would help too.

This is a good point.

Here's a few observations:

I have a bunch of the Gento 2" drivers. I didn't think they were anything special until I heard them in Bill Waslo's Synergy horn. In his Synergy, they sound great.

Inspired by that, I built a line array using the Gentos. I put it up on the shelf after just one hour of listening. The treble is just "off," I can't put my finger on why but it sounds wrong. It reminds me of the sound of paper cone tweeters, like they used to use in the 80s. They don't sound right. Probably the damping, or lack thereof? The reason you don't hear this problem in Bill's Synergy horns is because he wisely crosses them over long before that's a problem.

I have tried three of the Tymphany TC drivers. I've tried the 3.5" that Wesayso uses, I've tried the 2" that's the subject of this thread, I've tried the 3" too.

All of them sound different. IIRC, the 3.5" and the 3" were smoother than the 2".

I've generally found that small aluminum coned drivers work better as full range array drivers than paper coned drivers. Paper coned drivers are great if you're trying to maximize efficiency, because they generally have a higher FS and efficiency. Aluminum coned drivers are less sensitive, but generally better damped.

I have a hunch that the issues you're hearing with the array are due to poor damping. Basically the response may measure flat, but if you look at a waterfall plot I'll bet you'll see a ton of resonances above 5000hz. BTW, a lot of compression drivers suffer from this too, it's one of the reasons I mostly use domes for treble these days.


If you want the best of both worlds check out the SB Acoustics 2.5". It has an aluminum cone with a paper dustcap.
 
I have tried three of the Tymphany TC drivers. I've tried the 3.5" that Wesayso uses, I've tried the 2" that's the subject of this thread, I've tried the 3" too.

...

All of them sound different. IIRC, the 3.5" and the 3" were smoother than the 2".
If you want the best of both worlds check out the SB Acoustics 2.5". It has an aluminum cone with a paper dustcap.

I'll check out the SB!

It could most certainly be my chosen drivers, though I have reason to believe it isn't, since they sounded quite good as singles.

I have tried most of the TC series at this point: TC6WD,FC00,FC02,FD00,FD02,TC7FD00,TC8FD00,TC9FD18,TC9FD00, and the aluminum Peerless 2".

I did not like the aluminum Peerless, I found it had fairly high distortion above "whisper quiet" volume. Most of the TC drivers I like quite well, the TC9FD18 being the clear champion for full-range and "bang for buck".

My favorite all around is the TC6FD00. It has incredibly little midrange distortion. The FC (neo version) is similar; subjectively a bit better control in the treble? The TC6xx02 have noticeably better low-midrange performance at a given SPL, but lose the smooth HF response - haven't tried EQ'ing these to match though. TC6xx02 is also great bang for the buck in terms of its midrange performance... just not as good full range as the 00.

The TC7 did not sound great to me, the rising treble response is a bit harsh. I prefer the TC6 always. TC8 isn't as good as the TC9FD18, although the models with larger motors do have lower distortion in the low-mids at SPL.

So I considered the options in the TC series quite carefully, and found that the TC6 (any variant) and the TC9FD18 were the most appealing to me. The TC9 sounds starts beaming quite low, so subjectively I just prefer the more even illumination of the smaller TC6 for "mid-tweeter" duties. The TC9 is looking like a better option if a tweeter line is thrown into the mix, and can be crossed that much lower.

I ended up with the TC6WD simply because of price. It does have pretty decent midrange performance and high Xmax for the price / size. And with (substantial) EQ, the TC6WD isn't that far off from the TC6Fx00 in treble response either. But I've always wondered what the arrays would sound like with the "proper" TC6Fx0x (at 2-3 times the cost)...

Since I have all these drivers on hand, I should definitely do some comparative measurements and post them here... especially if it helps get to the bottom of this "why does the treble from my line arrays sound bad" dilemma!
 
I recommend making some measurements and seeing how comb filtering changes with distance. Typically, it moves up one octave for every doubling in distance. Can you try them in a bigger space?

I'm also interested in learning more about your experiments with a tweeter. Can you post measurements?

EQ is extremely crucial to get these arrays to sound right. ...

Hey, thanks for the links and insights from your own line arrays! This is all great feedback from what seems like a pretty small group of people who have built them :)

I do see the HF smoothing trend in your measurements. I would say yes, the treble from my 2" lines is wackier in close measurements (1 meter, 1 foot, 1 inch) than at listening distance (2.5 meters). They are about that same distance apart, its not a very long wall. I'm more interested in subjectively they sound OK at that greater distance.. easy enough to find out. I should definitely try them in another (bigger) room.

At this point it is a pressing question whether
- the drivers individually or maybe even as groups (because of the series wiring) are exhibiting poor impulse response / CSD / damping
- the close listening distance is exaggerating the comb filter effect
- major improvements could be made with DRC / FIR correction

Look forward to troubleshooting this and seeing if they can't be made to behave full range, or at least what flaws in the system can be corrected.
 
Edge simulation at 3m distance - Line array vs single 50mm driver, mic height varying.

The funny thing that makes me wonder is that many hifi'ists say they can tell a tweeter from other or even a cable change. I guess it should be easy to hear the difference in treble of an array vs. single driver.

Is it so that we hear what we want to hear and vice versa?
 

Attachments

  • linearray edge vs single-vert.jpg
    linearray edge vs single-vert.jpg
    210.5 KB · Views: 507
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.