This should be interesting [DEQX by two]

[Ken Pachkowsky]

Ken,

Have you been able focus on getting DEQX to emulate exactly what the Westlake active crossover does as far as real voltage drives are concerned? I will have my second unit within the next week and have the ability to go 4-way.

There appears to be a notch at 4.5Khertz and at 6.5khertz? They maybe crossover point related.

Ian

As I have not used the crossover abilities of the Deqx I have no idea how it will compare although from what I have read and heard it should do very well.

As far as the 4.5k and 6.5k notches I agree as the crossover point is 4.5k.

I wish I had a way of measuring the voltage drives of each channel on the hrx. I simply don't have the equipment or the know-how.

Ken

[Mr. Widget]

Bingo.

So I guess I am surprised that you haven't seen these suck outs before... maybe it is because you use Smaart to capture more room and less speaker?

Edit: Oh - I see...now I understand - that's the theoretical response.

No, it isn't theoretical... it is real, but it is only flat at the mic capsule. Since DEQX "heard" all of the bounced information along with the direct information it corrected the speaker based on faulty (not particularly useful) data.

That is why for it to really work, it is necessary to use the DEQX correction technology in an anechoic chamber... NHT does this with their system that they sell with built in DEQX. Basically if you can get a true anechoic measurement of your speaker, the DEQX technology can make it flat to within a fraction of a dB... the next best thing and only practical solution would be to go outdoors...

But then it wouldn't be good when you placed the speaker back in the room?

The point is that no speaker is even remotely flat... here is a real anechoic measurement by JBL of the 4345... if they had used a DEQX in there the result would be that this specific 4345 that was measured would come out dead flat. Even the comb filter ringing that occurs around the 10KHz crossover point can be eliminated by using a super steep crossover slope. The room will then mess that flat response up, but that will always be the case as Steve said... DEQX has a room correction feature too... for room correction you take your "flat" responding speaker and take a series of in room measurements... DEQX will then use it's 10 band parametric EQ to reduce the effects of the room.


The bottom line is that it really does what is advertised... unfortunately with most speakers that we like it doesn't work out too well... it seems to me that Kim Ryrie and the gang designed the system for a two-way mini monitor and subwoofer... for this type of system, in room measurements can be quite easily done and the overall improvement can be remarkable. For me since my speakers weigh over 300lbs each, I find the digital crossover, preamp, equalizer, and DAC functions awesome and better than anything else in this price range by a wide margin and I simply don't use the correction features.


Widget

[Shane Shuster]

The room will still introduce problems when the speakers are brought back in, but at least the room will not influence the measurements.

Having said all that, I've wondered what the results would be if all DEQX measurements were taken with the mic in the listening chair. Perfect amplitude, phase and arrival time in that spot might be great, although the next seat over could be awful.

Then it might not be like a Tact. With a Tact they recommend you move the speakers close to a wall to get bass boost and a better impulse response. You put the mic at the listening position and it pulses the speakers to get a measurement. It then "fixes" level balance, time alignment, and frequency response depending on what you draw in for a target curve.

[Mr. Widget]

Then it might not be like a Tact. With a Tact they recommend you move the speakers close to a wall to get bass boost and a better impulse response. You put the mic at the listening position and it pulses the speakers to get a measurement. It then "fixes" level balance, time alignment, and frequency response depending on what you draw in for a target curve.

All measurement techniques whether using MLS, FFT or whatever process respond essentially the same way to speaker/room issues... placing the speaker near a wall and the mic at the listening position sounds pretty screwy... especially if you don't normally listen to the speakers up against the wall.

Have you used the Tact system? I know that unlike DEQX, they make a wide variety of pieces... and they are generally well regarded.


Widget

[boputnam]

So I guess I am surprised that you haven't seen these suck outs before... maybe it is because you use Smaart to capture more room and less speaker?

In the home application, surely. As Steve posited some posts back, I too measure at and optimize for the listening position. Anything else seems pointless. I don't sit at 1m distance; I sit at closer to 4-5m, so there is where I measure the response.

Also, as per the multitude of pics of my 4345 roost, there is a nice carpet that may mitigate floor bounce.

Regardless, never in my personal experience have I seen a bounce that extreme. Where it possible, I would like to know the coherence of it - it may not be so very high. And in that case, I wouldn't toil over it too much anyway...

[Mr. Widget]

As Steve posited some posts back, I too measure at and optimize for the listening position. Anything else seems pointless. I don't sit at 1m distance; I sit at closer to 4-5m, so there is where I measure the response.

Well... you're wrong.

I should have dragged CLIO along when I was over last time.... It is best to get a nearfield (more speaker, less room) measurement to set the L-pads, amps, perhaps a bit of EQ... and then move to the listening position to tweak it a bit. You really need to start out with a "speaker" measurement so that you can correct for it's own unique issues before you try to fix the sound of the room... measuring at a great distance you get too much erroneous data... much of what the mic "sees"... we don't hear, and in many cases the room masks what we are "seeing" on the screen but are hearing. This is one of the reasons that Zilch's RTA plots look smooth but the speakers don't sound that way. (Zilch, sorry to use you as a negative example... it was the first thing that came to mind.)

Also, as per the multitude of pics of my 4345 roost, there is a nice carpet that may mitigate floor bounce.

Ken had a pile of pillows on the floor between the speaker and mic. I have been able to get quite a bit of that floor bounce removed by placing a 1' high pile of Sonex sheets... that does help quite a bit. A high nap carpet probably won't affect frequencies lower than maybe 1000Hz.


Widget

[Zilch]

This is one of the reasons that Zilch's RTA plots look smooth but the speakers don't sound that way. (Zilch, sorry to use you as a negative example... it was the first thing that came to mind.)

Oh, POO, Widget.

I've also posted some of the most unsmooth RTA plots ever seen in these forums.

What you see is what I get....

I ignore the LF, typically, 'cause I know there's little to be done about it.

And it's WRONG, anyway, unless measured nearfield or groundplane.

[Or in-box, maybe, which I haven't tried yet.... ]

[Mr. Widget]

I ignore the LF, typically, 'cause I know there's little to be done about it.

And it's WRONG, anyway, unless measured nearfield or groundplane.

Exactly.... except it isn't just the lows that can be wrong... without time windowing the highs are always in question too...

You've read the "good book" ...you know what I mean. I was just pointing out that you, I, and probably everyone who has used an RTA in a smallish space and has EQ'd their speakers to look flat... are not really listening to flat speakers.


Widget

[Shane Shuster]

All measurement techniques whether using MLS, FFT or whatever process respond essentially the same way to speaker/room issues... placing the speaker near a wall and the mic at the listening position sounds pretty screwy... especially if you don't normally listen to the speakers up against the wall. Have you used the Tact system? I know that unlike DEQX, they make a wide variety of pieces... and they are generally well regarded.
Widget

I own a Tact system. It uses FFT. The idea behind it is that you put the speakers closer to the wall for better coupling and then the room correction cuts the bass back. Less woofer excursion and more amplifier headroom. If you move the speakers or change the toe-in you have to run the measurement process again. It works great on some speakers and less on others. I liked some pulled out in a normal position, others needed the wall.

I haven't tried measuring close like you describe. I guess I always figured if the listening position is where I want it to be flat, then thats where I should measure. I will try measuring 18" out or so this weekend and see if it is better.

Depending on which speaker I was measuring, I have the same or worse frequency dip that Ken has in roughly the same spot. It depended on the woofers height.

I do have a couple of questions for anyone. I somewhat know how to read a impulse response graph. But what does a really great system's graph look like. The first blip should die down right away and stay flat, right?
Any further blips are reflections or driver delays? Theoretically perfect would be a sharp line blip then nothing?

The 2nd question is using a FFT, what is considered a good at the listening position frequency response. (20-20k) Plus or minus 6db is the best I could get in my small uneq'd, untreated room.

Thanks.

[Ian Mackenzie]

Shane,

Reading and interpreting these graphs as as much a science as it is an art and the same applies to the measurement techniques.

The graphs help confirm what you might be aiming for with a design but they should never be used as a yard stick of how it sounds.

Usually the early reflections 3ms < are what influences the tactile transient response and they relate to the baffle surface and diffraction effects around the driver. There should be rapidly damped ringing after the initial pulse. Small glitches immediately after the pulse may suggest early reflections and multiple glitches may mean time alignment issues in a multiway system. Waveguides are in vogue as they can help minimise early reflections in small to medium size systems.

John Dunlavey formerly of Duntech did a lot of research on transient impulse testing and you may find a few references on the www if you do a search. In short its a complex topic and there is as much agreement as there is disagreement on what matters and what is important in speaker design.

I am not aware of a loudspeaker than can product a perfect square wave.

The thing is many modern systems like the Duntech and Vanderstein systems claim transient perfect behaviours but they just don't have the same snappy transient capability or realism of a JBL or a TAD or Altec system which oddly enough may not exhibit well controlled early reflections or time alignment. Why? Dymanic power compression.

Just trust your ears.

Bo or Mr Widget may be able to assist with your last question.

My suggestion is to attempt to control worst of the bumps in response by careful selection of the listening position and careful selection of room furnishings.

[Shane Shuster]

Thanks Ian. Thats about what I was guessing.
It seemed like most webpages and books like The Master Handbook of Acoustics assumed the reader has already measured lots of rooms or is a sound professional.

[Mr. Widget]

I own a Tact system. It uses FFT. The idea behind it is that you put the speakers closer to the wall for better coupling and then the room correction cuts the bass back. Less woofer excursion and more amplifier headroom. If you move the speakers or change the toe-in you have to run the measurement process again. It works great on some speakers and less on others. I liked some pulled out in a normal position, others needed the wall..

The problem with sticking the speakers up against the wall is two-fold. First, while it is true that you will get better coupling and therefore "more" bass, it isn't linear... you will not get an even increase in bass... this is even more the case with corner placement. Of course some people want more bass and aren't terribly concerned about the quality of that bass or simply never paid enough attention to notice that they have created a quality of bass that isn't particularly natural sounding... They don't mind having certain notes accentuated and others muted... that's fine. Others are dealing with domestic aesthetic concerns and these weigh more heavily than getting the finest reproduction possible... I can understand that. I am guessing that you are not in either of those camps and are after the best possible sound based on your comments.

Secondly, for most speakers, placing them near the walls interferes quite dramatically with sound stage depth and image specificity.

I haven't tried measuring close like you describe. I guess I always figured if the listening position is where I want it to be flat, then thats where I should measure. I will try measuring 18" out or so this weekend and see if it is better.

I have no practical experience with any of the TACT equipment, but if you can first measure just your speaker and correct any frequency or phase anomalies that are inherent in it and then make additional adjustments based on it's actual in room placement I am sure you will have better results. If you can drag the speakers outdoors to do the initial measurements and corrections you would be better off especially considering that your room is small and not controlled.

I do have a couple of questions for anyone. I somewhat know how to read a impulse response graph. But what does a really great system's graph look like. The first blip should die down right away and stay flat, right?
Any further blips are reflections or driver delays? Theoretically perfect would be a sharp line blip then nothing?

I pretty much agree with what Ian said... a clean straight line is the ideal, it doesn't exist, and systems that approach that ideal are not necessarily the ideal speakers. It is simply more complicated that we are able to measure at this time.

The 2nd question is using a FFT, what is considered a good at the listening position frequency response. (20-20k) Plus or minus 6db is the best I could get in my small uneq'd, untreated room.

That is is the realm of acceptability. You can probably do a little better... at some point your room will take over and there is nothing you can do.


Widget

[boputnam]

Well... you're wrong.

I should have dragged CLIO along when I was over last time.... It is best to get a nearfield (more speaker, less room) measurement to set the L-pads, amps, perhaps a bit of EQ... and then move to the listening position to tweak it a bit. You really need to start out with a "speaker" measurement so that you can correct for it's own unique issues before you try to fix the sound of the room...

That's exactly what we did, except used Smaart. The L-pads were and are set to "0", best as can be interpreted from the decal on the foilcal. We made some minor adjustments near-field.

Then we moved to the listening position and fine-tuned things.

But, what you are suggesting really requires two EQ's - one for attaining a flat-as-possible speaker response and leaving it set that way (with my road system I use the KT DN9848 loudspeaker processor with parametric and time-align the mains and subs); the other EQ is to mitigate the room affects (for this I use the KT DN 370 graphic). At home, I feel I get very good results using just the DN370 and optimizing for the listening position. It keeps uneccesary filters out of the signal path.

[Mr. Widget]

That's exactly what we did, except..

That's sorta what we did. I would have pulled both speakers out from the wall to flatten them, and then placed them where they sound best in the room... in this case Susan determined where they sound best so that tosses a bit of a wrench in the works, but what the hell the client is happy so all is good, right?

I don't use two sets of EQs... I save the settings from nearfield and then tweak them... comparing the tweaks to the pure nearfirld settings until I (the client, in this case) am happy... it takes about an hour to get close and then a lifetime to complete.


Widget

[Ian Mackenzie]

The paper speed, scale and division make a big difference to how the graphs appear. Smoothing and 1/3 octave pink noise can make a graph look very presentable for marketing. But on the other hand how audible is every wiggle and glitch?

I tend to shoot for overall balance in the near field with scale so all the drivers have balanced levels as a priority. If there are specific issues with the near field graph I look for the cause. ie room or driver or crossover. Then determine if those issues are likely to impact at the listening position. I think there can be an over emphaisis to obtain as super smooth response without due attention to mitigating other causes of less than optimium subjective performance.

By way of example a Wilson system can be measured in room and show all sorts of response defects. But I have yet to see anyone complain about them in that they sound bad because the review graph is not perfect and tweek the response with 1/3 active eq.