Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Wesayso]

After an email conversation with our host Jim I was encouraged to post a little something of my Speaker project on this forum.

I'll add a little foreword: This is just my journey to learn how we perceive music trough speakers, trying to improve the experience, in other words: Just my humble opinion. I'm still learning, still experimenting. It's a fun and fulfilling hobby. I'll share my journey below:

Let me first tell a bit of background. I've been an avid music lover for as long as I can remember. I remember messing about with speakers at an early age to get me into the music. I've build some speakers, but basically without knowing what I was doing. Way later in life I decided to learn a bit more about the art of reproduction. I figured to start this project in my car. I did not have any clue how well I would do, as a car isn't the most ideal place to listen to music. The idea did have some advantages though, as in my car, I was usually driving alone, commuting from home to work and vice versa. The car I owned at the time had a very bad setup in it that was distorting and basically sounding awful.
I spend a little time researching my options and started working on that project. The goal? Getting to enjoy my music on that daily commute.
I've documented that project on DiyMobileAudio.com, it can be found here: http://www.diymobileaudio.com/forum/build-logs-project-install-gallery/86635-1982-porsche-911-first-build-simple-system.html

This project, taking my time to learn, turned out very well for me. For the very first time I actually got to listen to stereo the way it was meant to be heard. It was very engaging and drawing me into the music. It also made me sit (very often) in a dark garage listening in a car while standing still  . I had to admit to myself, my car was actually sounding a lot better than my living room stereo.

I decided that had to change, so plans were made for a set of home speakers. But I had a lot of things to consider. I remembered battling real hard with finding crossover points that worked for all of the songs I liked. As I had an active setup I was always fiddling with that.

Due to several reasons I decided to try and chase a phase linear system. It wasn't long until I found JRiver that could help me achieve my goals. I downloaded the 30 day trail and started to repeat some of the experiments I had done in my car with my home setup. In short, after those 30 days I couldn't stand being without JRiver and bought it. It would still be a while before I actually started building my dream home setup.

After considering many options, like Linkwitz's open baffle speakers, horn setups and Keff's dual concentric I opted for full range speakers. Remembering the battle I had with crossover points I figured: why not go without? My car setup also learned me I wanted bass response down low!
No full range speaker would do that all by itself. So how about line arrays? Line arrays made of several full range drivers, all playing the same thing.

JRiver would be able to give me all the help I needed to provide the EQ a line array like that needs. Now before you start: yes, I was aware of comb filter problems an array of drivers would exhibit. After reading papers from people like James Griffin (http://www.audioroundtable.com/misc/nflawp.pdf), Roger Russell, but also the papers on arrays by his successor, Dave Smith at McIntosch I decided to take a leap of faith. From my car experiments (in a car you deal with many comb filter problems due to the many reflections) I figured it had to be possible to make the arrays work.

Based on a lot of reading and research I got a crazy idea how to build the speakers.
I started messing about with my favorite CAD program to make a design:


The idea? Stacking layers of birch ply on top of each other to create the shape I want. I wanted it smooth on the outside, avoid diffraction etc, and scattered on the inside to try and limit the standing waves. Many cross braces to make a stiff enclosure and all kinds of damping to avoid box resonances. Ambitious right?

The total design I cooked up looked like this:


Two tall towers, standing 2.25 meter high. (excuse my metric standards )

[Wesayso]

Next problem of coarse: how are you going to make all those layers for the stacked construction?
The original plan: CNC, laser cutting or water jet cutting. A pretty sound idea right? After looking up what services were available (and already aware of a water jet cutting firm nearby) my mind was set on having it all done for me. But something happened to upset these plans. Due to the world getting in a financial crisis I lost my job. I had been an ICT system administrator and helpdesk but the firm I worked at had difficulties. They decided to outsource all ICT services and drop the ICT department (with me being the only full time member).
We were still very much in that recession so I figured it might take me a while to find new employment. Especially since my knowledge of ICT was self taught, I didn't have any papers to prove my skills. My original education was in mechanical design.

I figured I would have more time than money available and decided to take the opportunity to start building my speakers while hunting for new employment. I had already bought the speakers needed and even a Behringer DEQ 2496 to have 2 routes of processing available. One with the PC + JRiver and the other with the Behringer.

The speakers? Vifa's TC9 FD18-08.

(3D model made of that speaker to aid in the design phase)

The best plan I came up with: use aluminum templates and cut all of the needed pieces out of Birch Ply using a router in a table.

Here you see the templates and a few parts of the stack done... I used a router bit with bearing to follow the aluminum template. That way I could repeat this over and over again. I'll say to you I had no real clue of the amount of work I had forced upon myself. But it was good therapy to get my mind off the setback of being unemployed after 24+ years of working day in and out. It kept me sane and positive and most of all: busy.

The idea came together and this is actually the short version (lol)

Gluing the stack by clamping them with threaded rods, that would become part of the enclosure.

With the future baffle as a guide it looked like this:

[Wesayso]

Going fast forward I rinsed and repeated


The original plan included a coat of epoxy and fiberglass matt on the outside of the enclosures. But after sanding down the towers and seeing all that wood I wanted to preserve that wood look.


Though the color had to change. Between sanding I made some samples and decided to go with a dark stain and several coats of PU floor paint.
I thought they looked lovely:


I coated the inside and out (though obviously no stain on the inside) to prevent moisture problems. But keeping this condensed version short, the inevitable happened... I had parked the enclosures inside to start working on the baffles.



One night we were sitting in our living room and heard a loud noise. We couldn't figure out what that was, but the next day it became obvious...
Painfully obvious to be honest... 

[Wesayso]

Painful as it was, I couldn't do much about this... winter had set (I started in the spring!) and decided to think about my options and let this be for a while. I got many ideas from others, trying to help, nearly all of the concentrated on removing the rods. Funny enough, that was the last thing I wanted to do. After a long winter (at least it felt that way to me), still being without a new career, I picked up the project again.

The idea: cut up the speakers in 25 pieces and cover it with epoxy + fiberglass matt. Close to the original plan (the plan I had before I wanted to preserve the wood look) but with a twist. The cuts would separate the towers into smaller pieces. Foam would cover the slots so the wood could expand and contract without upsetting the outer epoxy shell.

I'll make it quick...

Putting the saw into the stack:


Using expanding foam (PU glue) to seal them:


And covering it with Epoxy plus matt:


This was a recipe that worked. It made me able to finalize the speakers. Now I could go on and on about the baffles. Even those were cut and polished in my garage. Again using a simple router and also the cause of many head aches ...


But in the end I did finish the speakers,




and that's where the JRiver story starts....

I'll continue another time, I do hope you guys enjoy this journey.
Believe it or not, I tried to keep it short!
More to come.

For those that cannot wait, there's a whole lot more background following this link:
http://www.diyaudio.com/forums/full-range/242171-making-two-towers-25-driver-full-range-line-array.html

[Wesayso]

With the speakers finished I let the arrays run for a couple of weeks with some fixed PEQ setting, basically gambling what the arrays did. No measurements taken at that time.

Both left and right got the same curve, a boost on the low end, below 85 Hz, a cut in the middle section between 85 Hz and 3 KHz and some boost above that. I chose the curve based on published graphs from similar line arrays. It was fun to finally get sound, the arrays did impress from day one. But it still was far from ideal. Good enough for casual listening and running in the drivers. All along this period I was working with impedance plots to solve some minor baffle vibrations.

Once the baffle issues were under control I started measuring the arrays with REW, using JRiver's loop back feature. The WDM driver wasn't available yet. With the measurements at hand I let REW do it's auto EQ and put the PEQ settings into JRiver's Parametric Equalizer.

About 17 to 18 PEQ cuts/boost per side. This improvement was noticeable. Very entertaining and balanced sound and good imaging. Yet I came to the conclusion I needed to add a damping on the left wall (behind the curtain you see in the second picture of the finished arrays).

I made a huge panel using wool felt and fiberglass insulation (7 cm thickness) and mounted the panel behind the curtain out of sight. I left a gap of about 2 cm behind the panel. The right side looked way cleaner in the IR impulse with the first big reflection at ~6.5 to 7 ms.

Repeating the measurements and PEQ's gave a very rewarding stage improvement. Both speakers disappeared into the stage. While before adding the damping panel I could still point out the left speaker.

I've done a lot of chasing of reflections and after several tries found the cause of the 6,5 to 7 ms reflection. It was coming from behind the microphone, the wall behind the couch. Obviously I wanted to get rid of that one too! But the wall being in plain sight gave me something to think about. I had an old Led Zeppelin poster hanging there for years, with a Plexiglas cover. Time to negotiate with the boss, my girl. She agreed I could make a damping panel mimicking the old poster.

The old poster:

Lot's of battle scars from old age, but with nostalgic value for me

I sourced a similar picture on the net of high enough quality and redid the lettering to match my old poster, though it is a bit different. The old poster was dating back to the 80's somewhere, I got it as a gift long ago. It featured half of the Houses of the Holy Vinyl cover with lettering above. The picture I found was of the whole cover. This is how that one turned out:

I even thought about doing it in the newest "remastered" color scheme.

The fabric I used to print the poster is Kona® Cotton ordered from Spoonflower. I got that idea from another line array builder and several posts on, let me think, I think it was HTforum. I had found a thread with tests of several materials and Kona was the most transparent.
(reference link to fabric material: http://www.spoonflower.com/spoonflower_fabrics#view_fabric)

The finished panel behind the couch:


Whenever I use the stereo I put pillows between the damping panel and the couch. A later addition to my setup features ambient channels firing to the sides between the couch and back wall. They are behind the cushion of the couch itself.

In addition to this poster/panel I made another huge panel behind the second curtain on the couch side of the room. It is symmetrical with the other side.

Though I believe it's best to solve as much as you can the old fashioned way (like with damping/diffusion panels) my goal here was to get maximum absorption of first reflections with minimal intrusion in the living room. Only one panel is in view, the Poster.

Still, I had to promise my girl not to add any more panels . So this was the best I could do (or get away with). For that reason I only used absorption and figured I could do the rest with processing. My IR's were getting pretty clean after adding just these 3 panels. But two of them (out of sight) were huge.

So how did that sound? This was a way different sound than before. While I gained some clarity, I felt I also lost what I would describe as a sense of envelopment. Before adding the last two panels the sound was more engaging with big orchestral music. Now I'm not an avid classical music listener, I mainly use the orchestral tracks of movie soundtracks. (my bad )
So while this change was beneficial, but for some tracks the massive impact I had before, being engulfed in the music, was reduced. Should I have done diffusion instead of damping? My goal was to clean up the first 20ms of the measured impulse at the listening position. Trying to get close to the reflection free zone as used in Studio environments. Obviously I'm not able to get to that level in a living room, but I went with maximum absorption. I'll add a little spoiler: the plan along the road was to experiment with a Haas Kicker, simulated by ambient speakers.
A Haas Kicker is a late reflection, but still within the Haas limit. Usually between 15 ms and 30 ms behind the main pulse. That plan explains my choice for absorption instead of diffusion. My room is just too small to achieve that amount of delay. But JRiver can do it!

[blgentry]

This is a nice writeup!  Fun to read about your experiences.  I have a few questions:

How are you amplifying these?  I.E., now many amps, connected to how many speakers per amp channel?
Are you using any passive crossover at all?  Active?  I guess not since they are all the same drivers.  You mentioned active, so you probably mean active EQ/convolution in MC right?

I was expecting that you'd be crossing these over at 150 Hz or higher (high pass) because of the driver size and the very high resonant frequency.  But it seems you are running them full range and adding bass boost.  Are you concerned about mechanical damage to the drivers due to over excursion around and below resonance?  Do they have adequate bass for your listening levels?  How loud have you listened to these?

I guess that's actually a bunch of questions. 

Brian.

[Wesayso]

Hi Brian, thanks... I'm just starting, more to come.

As amplifier for the arrays I use an old, but well maintained Pioneer A757 Mark II, 100 watt into 8 ohm. I'm saving up for a more modern and powerful amplifier.

The separate towers, with each 25 full range speakers are wired 5 in series and that 5x parallel. The impedance curve looks like a single 8 ohm unit. No crossovers, PEQ plus convolution is used. I do have an impedance flattening circuit hooked up, targeted at the FS bump. I'll show some graphs later on.

As said, no crossovers and bass boost till about 30 Hz, gradually falling off below that. The Sd area of one array is about the same as a 15" driver, one could look at these speakers as a 15" driver with 2.6 mm x-max. X-mech will be higher than that. To reach x-max at 30 Hz I'd need about 240 watt input.

They do have plenty of output down low, but in all honesty, running them as Home Theatre they would be better off with sub support for the first octave.

In music I haven't found a track I couldn't play yet. A song like Lullaby from A Perfect Circle does give them a proper workout.

I'm at the limit of my amp before I reach x-max. One thing that is important is the DAC. Because of the large boost you need the proper headroom to stay well above the noise floor. A 24 bit DAC becomes important. I let JRiver take care of leveling between songs. I run loop back measurements on each part of the chain to see where I'm at.

An example measurement, measured at the listening position showing REW's THD plot:


My usual listening levels vary between 85 dB to 88 dB average. I always keep a RadioShack SPL meter nearby. My measurements are all SPL calibrated with the RadioShack meter. Don't forget, arrays drop about 3 dB per doubling distance instead of the usual 6 dB you get with point source speakers. I won't claim to get the 3 dB drop per doubling distance over the complete frequency spectrum but it does come quite close. Sadly my room doesn't provide the room to thoroughly test that, but it has been done by another Array builder on DIYaudio.com.
The bass is very even throughout the listening area. Spreading the bass over a large area (floor to ceiling) seems to work as a distributed sub system.

Look at this plot, several measurements with both speakers playing taken along the listening area:


This plot was made before I added the 2 extra damping panels. Due to both speakers playing you see most of the cancelations (between left and right speaker) happening between 200 to 2 KHz.
The bass however is very stable throughout the entire area as evident in the above plot. The average of all these measurements was very close to the target set at the sweet spot (at that time). I don't aim for flat response at the listening spot. I aim for a gradually declining curve from 25 Hz to ~17 KHz. This measurement does not show my preferred target curve. This measurement was made to check the balance over a wide area only.

I'll get to the convolution part soon. Then on to ambient speakers etc. etc.

[Wesayso]

After living with IIR EQ for a while I noticed a couple of things. I had used REW's auto EQ feature and it had a couple of strong corrections around the 80 to 200 Hz area. At times it didn't sound quite natural to me. I went for a fix by removing all PEQ's with a Q of 5 or higher while preserving the overall shape of the graphs. This did improve the sound. In hindsight, I was EQ-ing the area below 200 Hz (that is mostly dominated by the room) using an un-gated response. That is more likely the real reason for the unnatural sound I got at the time. Replacing the high Q PEQ's for more gradual setting did help though.

I was ready to try the next step, convolution. For this adventure I used the free DRC-FIR. After reading up on the basic principals and some step by step guides of other convolution programs, like Audiolense and Acourate, I was ready to take the plunge.

I made new measurements at the listening position and noticed I needed to provide some help to get the raw FR within DRC's scope. As said, the line arrays need quite a big boost at the lower end while the area around 200 Hz is there in abundance.
I decided to do some pré EQ, with broad low Q "strokes" before feeding the IR to DRC-FIR.

This pré EQ looks like this when viewed in REW:


Here's what happened with each step, first the raw measurement:

(from the level of the reflection at ~6.5 ms I can see this measurement was made prior to hanging up my damping poster)

Throw on top the pré EQ:

This shows the (P)EQ used to flatten the FR didn't exactly improve the IR.

After feeding the above IR (with pré EQ) to DRC-FIR the result looked like this:


These were some of the very first experiments with convolution. The difference in sound was drastic. Convolution really did change the sound drastically. That doesn't mean it was spot on, but an improvement on one side, while missing out on other things.
From this point on I started to hunt for improvement. DRC-FIR uses a frequency dependent window to apply it's convolution tricks. I started to experiment with these window settings to get a grip how that changed the sound. There are so many things you can change, I won't even try to name them all here. I've often played with settings and DRC's predictions to see the effect before applying it for real. Just to learn the new tool I was playing with.

I did say I could tell the damping panel behind the listening position wasn't in place yet. Let's look at a more current IR:

It's clear to see the improvements from all of that experimenting. Part of the shape of the IR (the downward peak) has to do with the target I have chosen. But the main difference is the improvement in all of the swings after the main pulse in the early tries vs a much cleaner pulse I have now. Some of it due to adding the damping panels, some of it due to using different settings and changing the frequency dependent window.
I also started using JRiver's second bank of PEQ's to do some additional EQ after the convolution step. Primarily to balance left and right FR but also to smooth out some over correction from DRC-FIR. (for that I look at early waterfall plots, meaning: with short rise time)

Here's the left vs right impulse:

As always, taken at the listening position. See how close both pulses match? That's because the Left and Right channel balance matches:

Left and Right FR, 1/12 smoothing within REW. At ~70 Hz I have a room mode on the left array, I compensate slightly with the right channel. Below 40Hz the left side needs less boost to get these SPL levels so there I dial back the right side and let the left channel handle the lower notes.
As that one Array (the left) is placed in a true corner it makes it easier to get higher SPL levels compared to the right side.

[Wesayso]

Let's move this thread a bit more towards JRiver now. The above story tells (in a really condensed way) how I used JRiver convolution and PEQ to get my FR curves right. There's a lot more to that story, most, if not all of it in detail in that huge thread I linked above. But I need to get a few things off my chest first. I do believe the IR tells us a whole lot about the sound we perceive. But you need to dissect that impulse to find it.
I can't look at an impulse and say: "yeah that will sound great" by just glancing at it. We need a bit more than that. Let me explain a view parts of it...

I mentioned my quest to get a phase linear correction. After all, it was one of the reasons to use full range line arrays. All of that quest was handled by slowly creating my own templates for DRC, tailored to my specific setup. I went on and on, trying to learn and read all graphs within REW to get a handle on how the speakers and room interact. I've said it before and I'll say it again. In my opinion the Frequency Dependent window chosen is a large part of how it sounds. I tried to find the shortest window I could get away with, while still improving the frequency curves. While it may seem I use excessive force to get the curves to look like this, It is actually done with relative short windows. All of the more "normal graphs just look better if you correct the first 5 or 6 cycles of the sound. On the top end my window is slightly longer. Because I use arrays, the high frequency sounds arrive first from the middle (ear height speakers) and a bit later from the top and bottom ones. By increasing the number of cycles I got more focus at those higher frequencies. For conventional speakers I'd say the 5 to 6 cycles would be enough though. The biggest advantage is: the sound is also pleasant at off axis or outside the sweet spot positions.

It's very easy to use convolution and get great looking FR curves (with brute force). It's a lot harder to find the setting that give you great sound as well as great looking curves. We need to look at a lot more than just the IR (which is merely a representation of part of the frequency spectrum, highly dominated by the higher frequencies) and FR curves. Look at early waterfall plots, it will show you what signal your ears hear first!

(example of an early waterfall plot, again at the listening position)
Settings used in close up:


Look at spectrogram plots, they will show you the speaker and room interaction. Gate it by playing with the Time Range. And definitely look at the filtered IR plots, use those filters and see/learn what influence the room has at different frequencies. Compare it to how a true Dirac pulse looks.

At one point I checked my specific setup and quest for phase linearity with the program: APL_TDA. I just used the Demo version to see what that showed me.

A combined stereo measurement showed I got excellent phase tracking and I was exited (and scared) to try this program to see what my (multiple) experiments got me. I'll show the combined response as measured with REW (again at the listening spot) filtered with REW's Frequency Dependent Window:

Phase tracking the FR plot. Some changes were made, to make the phase a little closer to a "minimum phase band pass", in listening tests that one clearly won over the more "linear phase". I can't seem to find that graph right now. No doubt it is somewhere in that big thread .


After running the APL_TDA demo (with some minor startup problems getting it to work with JRiver's WDM driver) I got this out of it:


Not bad right? It shows the sound arriving at the ears is almost instantaneously, the higher frequencies arrive at the same time as the lower ones, AT the listening spot. It also shows what the room is doing, after that first wave front. All of those wiggles at the bottom are reflections. After all, this is no studio. In fact, a fully functional living room with only 3 big damping panels.

Here's a wavelet from REW, taken that very same day:


Which one of these graphs makes it easier to see the actual timing?

(sorry guys, I do notice I'm making this thread almost as long as that huge thread on DIYMA, it's difficult to get everything I deem important in here. This is a study over a period of at least one year I'm trying to put into words. I hope I'm not boring anyone... you can always stop reading I guess (lol). I can't promise to make the next part of this story any shorter than the above. More to come, crazy right? I'll be back soon!)

[Wesayso]

Let's talk about the effects on sound all these changes had. As said my efforts were to get early reflections down as much as possible. Starting with the damping panel beside the left array and continuing with a damping poster (and another side wall panel to the side of the listening position).

The first panel, beside the speaker got me a wide stage where the speakers disappeared into the sound field. It worked very well with big orchestral pieces. Even though there were reflections from the wall behind the listening position I liked it very much.

After my first convolution experiments the sound got different. It was excellent from a pin point image perspective. But it also took away a bit of the overpowering dynamic sound. Adding the damping panel behind the listening position further enhanced that.

I guess you could say I started missing some of that envelopment, some ambience if you will. I started snooping around in JRiver Effects in the DSP environment. I tried the variety of "Environment" settings and remember liking the "Recording Studio" setting for some songs. But it didn't work all that well on other songs I tried. I started playing with the Surround Field settings. Up to Medium enhancement there was something that happened that I liked. My imaging stayed pretty much in the same position but the sides really came alive. Back ground vocals got more body.
But it was both a blessing and a curse. The phantom center became a bit distant and dark. Not by itself, but compared to the lively sounding sides. I enjoyed this for a long time, but I was always searching for the right amount to really please me.

In a later stage I started using the free Voxengo mid/side processor MSED, to dial back the Medium enhancement a bit by boosting the center signal. But never really got a result that worked on all music. Some songs were great, better that great, magical even while others felt unbalanced. Very recording dependent.

I kind of compared it to the Stereophile review of the BSG qøl Signal Completion Stage: http://www.stereophile.com/content/bsg-q248l-signal-completion-stage#MI5JxQGYzvgMsE1B.97

It does work on similar principles, old ones actually. The Blumlein shuffler. I have used it for a long, long time. Every once in a while comparing it to the sound without the effect applied. On most music I liked the result better with that effect in place. But never got to the point that I could get a good balance throughout the entire stage, left to center to right. This didn't pose a big problem, but it made me do several experiments.

At some point I got into a discussion about the tonal balance in stereo listening on the DIYaudio forums. A thread was started to dive into this topic and I actively participated. The thread: http://www.diyaudio.com/forums/multi-way/277519-fixing-stereo-phantom-center.html

I tried the suggested shuffler, what looked like a series of very early reflections and even though I could detect a change in tonal balance, this shuffler had the effect of gluing the sound to my speaker positions. Something I never had before. In other words: the speakers didn't disappear anymore. So I gave up on that idea until someone proposed the "S-curve" mid/side EQ. Immediately I was all over that, with the JRiver effect "Surround Field" in the back of my mind.

With JRiver, something like this is easy to setup. Especially with a few plugins. I setup something close to this:

This is how I started. Here I have EQ in between the mid chain of the mid/side processing. One Voxengo MSED is setup to encode the mid/side signal, the other is setup to decode it back to stereo. Not long after that I began to EQ the side chain too. I was very glad to see JRiver introduced mid/side processing into it's current version. Proof that JRiver is always evolving and bringing us more tools. Though props to Voxengo too, for actually supplying us with a range of useful plugins, even free ones!

At first I based my EQ on the S-curve found in these two documents:
http://www.sengpielaudio.com/FrequenzabhHoerereignisrichtung.pdf
http://www.sengpielaudio.com/FrequenzabhUebersprechen.pdf

This immediately got me better results than all I had done previously with JRiver's "Surround Field" effect and the Voxengo MSED vst plugin. I used the setup as pictured, except for the introduction of a second EQ for the sides.
I started with the "S-curve" and slowly but surely edited that curve (both mid and side EQ) to get my own preferred balance.
In my thread you can find many more details, at some point I also dissected what the JRiver Surround Field effect was doing.

Feeding JRiver a left, mid and right signal got the above resulting wave when run trough the surround field effect.

I had started a thread, asking about this JRiver "Surround Field" effect, here on this forum: http://yabb.jriver.com/interact/index.php?;topic=95679.0
Running a couple of tests like the picture above finally gave me the answers I was looking for.

So what was I compensating for with all of this processing? I had done the best I could to get a clean pulse at the listening position. Making sure the left and right signals were as close to each other as possible. All these deviations from that pure stereo concept point at the "problem" we actually listen to our speakers with two ears. That is the main reason (for me) to like a bit more bass below ~1500 Hz in the side (pure left and right) signal and a bit more top end in the phantom center. More on this later on.... This "S-curve" had me pleased for a mighty long time.
Now I hear some thinking: wouldn't this be compensated for in the recording? Well, not if they listen to headphones. And I do believe a near field setup as is often in use in studios is behaving quite different from a home setup where we sit at about 3 meter distance or more.

(finally some real JRiver talk )

[Wesayso]

Okay, time for the next experiment! Adding ambience....

Ever since my CAR audio days I've been interested in what a Haas Kicker could do in a home audio setup. After spending a lot of time getting rid of reflections I'm actually going to (re)introduce some controlled reflections from the back. In my CAR audio days I had done a lot of reading on this subject. A car is a pretty small place. Yet by introducing a delayed, band passed and attenuated (L-R) and (R-L) signal from the back speakers you could fool the listener to believe he was listening in a much bigger space.
I lacked the processing to get this in my car, but swore I'd try it in my home setup. With a add-on card for my Asus Xonar Essence ST(*) I would have the extra channels to play with this kind of stuff. JRiver would have to do the rest.
(*) = The mains are still fed trough optical out of the Asus ST to my Musical Fidelity M1 DAC, but some quick tests confirmed I could use the added channels to feed this ambient back channels.

Some interesting vies on Haas Kickers can be found here: https://www.gearslutz.com/board/studio-building-acoustics/708997-who-needs-haas-kicker.html. It gets many conflicting views in the studio world. Yet the Kicker often gets praised to enhance the listening pleasure. Time to test it myself, right?

I made 2 small speakers, with the Scan Speak 10F 8424G00 as the driver. These would go somewhere behind the couch to find a place that works. They looked like this:

I figured I could get away with using speakers this small due to the band pass from ~200 Hz to about 3500 Hz. They wouldn't need to do a lot of high SPL, just get me some diffused sound from the back. I acquired a sister amplifier to my trusty old Pioneer and started the experiments.

The speakers got a nice coat of paint, slippery and hard to photograph:


The first experiment was to aim the speakers... As I wanted to mimic reflections coming from the back, like in a larger room I aimed them Up, Sideways and toward each other. The sideway aiming worked best. As the speakers are behind the cushions of the couch, the sound bounced from the side wall on one side and a desk on the other side back toward the listener. Some FIR filtering was added to get the desired EQ curve,
delay set in JRiver and even the signal fed to the speakers is made with JRiver's mix channels option within the PEQ banks.

Here's how it ended up, though I play with setting from time to time. It will give an idea on how to do it:


So what do we see? Getting a (L-R) signal and an (R-L) signal for left and right, with a bit of center mixed in at a lower level.
The center get's a different delay compared to the surround left and surround right. This was determined again in many many listening sessions.
You also see some specific EQ, even more in the second bank of PEQ settings, plus the use of a FIR filter in convolution.

What does this do? If we look at the filtered IR tab in REW, of the mains playing in stereo we see this:

A clean peak, the big drop in level of early reflections and a steady drop after that.

With the Haas kicker in place it looks like this:

The peak is the same, a reflection free (or reduced) zone of about 20 ms and then the Haas Kicker shows up.

By themselves the output of the Haas Kicker looks like this:


The big question: how does it sound? In my humble opinion: a more 3D like stage. More engaging and involving. Remember I said that all of the convolution and damping panels took away that specific part? With this ambient add-on it was back. But this time without a negative influence on imaging. It actually worked positive, reinforcing the stage. At no point you actually hear sound coming from the back. But the sound in front of you becomes deeper, more life like.

Instead of relying on me, the "designer" of this setup, here's a review done by a friend. He mixes his own music in a small home studio and I can count on him being as critical as I am about music. Here's what he had to say:

Hi Ronald. Here is my report


I will make apologies in advance for my English but I give it my best shot!

----He explained it all to me. The whole building process of his now grownup 7 ft twins. It is impressive, the way they gaze at me with each holding a long row of little speakers. I do not understand and comprehend everything Ronald told me about his project in a very enthusiastic way but I feel a lot of respect and admiration. Blood sweat and tears were shed to accomplish what is standing in front of me. “This is it, they produce sound and I think I did a good job” Well,….. a good job is a big understatement…because I have listened and they certainly produce sound. To call it just sound is way too simple to describe what is entering your ears. Tone colors is a better way to put it. Any color you like is there. Anyway, I was asked to give my honest and frank opinion about his tall and rank self-built stereo speaker set.

So, how do I do that? I asked myself. I don’t want it to become a lyrical piece of prose so people will doubt my honesty…..but I am lyrical. And that is because I am a critical listener, picking up details and can feel music to the bone. But anyway…at this my findings, while having a coffee on the ‘hotspot’ at Ronalds place.


Opeth, Windowpane from the album Damnation. I immediately feel like I’m on very right spot in the concert hall. The ultimate place to be, everybody is always looking for, I am experiencing now. Every instrument is clearly in place and it is very extraordinary, I can almost feel the space in between. It is a wide sound with room for everything that is going on. There is prominent clarity but the bass and deeper tones are perfectly present without domination. Hearing and feeling them in a very pleasant way as if sound and air cooperate to make it just right. This is how it should be. Like putting on 3D glasses. The musical landscape gets an extra dimension. It is warm and deep and I can’t help it so say….”Holy macaroni” (which is an old Dutch saying meaning: ….I’m flabbergasted) This is some hearing experience!!


Something completely different. He puts on the Sugababes. Hole in the head. It is not music I prefer but that is not the reason why I am sitting on the hotspot. So babes, let’s go J. I’m surprised that my ears tell me something different from my prejudgment. The extra dimension these speakers offer makes it very nice to listen to. While listening I made an image of a rectangle on my piece of paper. This rectangle is where this dominant kind of pumping song has its body. This was meant to be I guess…for the crowd….in your face…. But don’t get me wrong…everything is still on place and has its own distinctive position with clarity and warmth. And then there is out of the box… there is this nice tingling Ping-Pong panning effect which makes you happy… No other way to put it. It is hard to explain what makes these speakers so special but I think it has to do with a very nice wide stereo image without losing the natural reproduction. A sort of…the real thing!


David Bowie, Ashes to Ashes. I will confine myself to the striking details now. That I am very enthusiastic about the speakers needs no further explanation. I now notice how many tracks have been used for vocals in this song. There is a upper stream and a undertow with vocals and bizarre effects. These speakers apparently do have the distinctive character to reveal these details.


Blackfield – Blackfield. The clarity of the rhythm guitar is striking. Not that I couldn’t hear it on my own speaker set but over here, on the hotspot it comes more forward.


Blackfield – Hello. In a way a little shocking to find out about the honesty of these speakers. The high frequencies that are floating above it all in this song are too dominant in my opinion, it is really hurting a bit. Too bad….but a honest message. Didn’t notice this before, at least not as strong as now, not in a way it disturbed me like now.

Overall….

Like I said, I’m a devoted listener but not a technician. Don’t know about charts and measurements and why it should have this or that value. I know what I hear and I know it sounds great. Like I mentioned, I’m a critical listener but….. It was a revelation.

Ronald, for me, for what it is worth, you accomplished a great result. Thanks for being a little part of it!

Copied as I got it, no editing. Later on he told me it felt like being in the right spot at a good concert, like North Sea Jazz. That doesn't happen all that often, here you get to sit in that right spot whenever you want .