The No.1 Website for Pro Audio
 Search This Thread  Search This Forum  Search Reviews  Search Gear Database  Search Gear for sale  Search Gearslutz Go Advanced
My Listening Room Studio Monitors
Old 17th August 2015
  #511
Lives for gear
 
jim1961's Avatar
 

Quote:
Originally Posted by Audionut11 View Post
I'm not sure that's the best way to think about it, but it's to early in the morning to expand further.



To be fair Jim, it's only an issue because you have less then when you were optimising for magnitude. That green line is remarkably flat.

Have you taken your ears for a test run?

With a better understanding of phase matching speakers, it shouldn't be to hard to work on magnitude while maintaining phase.

Can you share the mdat?
Yesterdays mdats

BTW, I dont mind you being critical. In fact, I invite criticism. Thats how things get better
Attached Files
File Type: mdat 08 16 15 REW 256k LL RR 90.mdat (5.66 MB, 119 views)
Old 18th August 2015
  #512
Lives for gear
 
Bjorn Omholt's Avatar
 

Interesting. Based on the few researches on audibility of GD, there are no indications that GD is audible at low frequencies. My guess is that you're hearing other changes.
Old 18th August 2015
  #513
Lives for gear
 
jim1961's Avatar
 

Quote:
Originally Posted by Bjorn Omholt View Post
Interesting. Based on the few researches on audibility of GD, there are no indications that GD is audible at low frequencies. My guess is that you're hearing other changes.
Perhaps. But let me illustrate something. The two plots are ballpark similar in magnitude, somewhat different decay rates, but have very different GD's.

The blue one sounds much better.

There is no question these sound different. But I agree there is some debate as to what it is I am actually hearing.

Edit:The Decay graphs is 1/1 63hz filtered.
Attached Thumbnails
My Listening Room-gd-comp.jpg   My Listening Room-fr-comp.jpg   My Listening Room-decay-comp.jpg  

Last edited by jim1961; 21st August 2015 at 02:50 PM..
Old 18th August 2015
  #514
Lives for gear
 
Bjorn Omholt's Avatar
 

One obvious problem is that the frequency responses are clearly different. Even a difference of 0.1 dB is audible if it's broad enough and it's much larger here.
Old 18th August 2015
  #515
Lives for gear
 
jim1961's Avatar
 

Quote:
Originally Posted by Bjorn Omholt View Post
One obvious problem is that the frequency responses are clearly different. Even a difference of 0.1 dB is audible if it's broad enough and it's much larger here.
Yes. But the GD and Decay results vary more than do the FR's. But in an absolute sense, your right.

I dont think it likely you could align two FR's within +/- 0.1 db that had wide variances in GD. So we just have to derive conclusions based on biased data.
Old 19th August 2015
  #516
Gear Head
 

Something that may interest you.

You loose 4.1dB at 21hz, but that looks be be a dirty big room mode anyway (prime for EQ). There's also still excess energy in the region of EQ, but it's far smoother in the time domain and should translate to listening.

I've added some boost to compensate for the negative EQ, so you should double check signal levels for clipping, if you feel like trying it. If you want even smoother, you can apply more negative gain in the first EQ slot, but (obviously) you loose more magnitude.

I've found REW to translate fairly accurately between expectation and final results, but I obviously can't confirm with further measurements, especially any delay problems with the filters themselves.

You should be able to load this straight into the mini-dsp.

edit: It doesn't translate as smoothly on the left channel, so you'll probably want to find some compromise based on further measurements. Best would probably be to equalize each sub separately.
Attached Thumbnails
My Listening Room-jim_fr.jpg   My Listening Room-jim_160.jpg   My Listening Room-jim_100.jpg   My Listening Room-jim_300.jpg  

Last edited by Audionut11; 19th August 2015 at 05:15 PM.. Reason: I found a dictionary.
Old 19th August 2015
  #517
Lives for gear
 
jim1961's Avatar
 

https://dl.dropboxusercontent.com/u/...tting%2090.xml

Here is my current filter set (for 2.1 adv)
Old 20th August 2015
  #518
Gear Head
 

Quote:
Originally Posted by jim1961 View Post
https://dl.dropboxusercontent.com/u/...tting%2090.xml

Here is my current filter set (for 2.1 adv)
Heh, so maybe it's not a room mode.

I think I recall the minidsp applying some internal gain reduction to cover off any boost, but it's probably better practice (methodology) to apply some negative gain on the inputs to cover off boost, rather then the outputs. You've probably checked the voltage levels, but a voltage meter won't show clipped waveforms if the minidsp is clipping internally.
Old 20th August 2015
  #519
Lives for gear
 
jim1961's Avatar
 

Quote:
Originally Posted by Audionut11 View Post
Heh, so maybe it's not a room mode.

I think I recall the minidsp applying some internal gain reduction to cover off any boost, but it's probably better practice (methodology) to apply some negative gain on the inputs to cover off boost, rather then the outputs. You've probably checked the voltage levels, but a voltage meter won't show clipped waveforms if the minidsp is clipping internally.
As you can see, ive gone to some lengths to make sure the miniDSP doesnt clip. My philosophy was to turn down the inputs a bit (-1.2db), and subtract output level enough to nullify any added gains in the PEQ area. In other words, if the maximum PEQ gain was +4db, id turn down the output to -5db. And so on. I make up the difference by turning up the plate amps input levels.

What I have also done, using a sine wave generator and a Fluke digital meter, is look at the bass frequencies and make sure every main and sub reach their maximum rated output voltage at the same time.
Old 20th August 2015
  #520
Gear Head
 

(0) - (1.2) = (-1.2) + (10.6) = (9.4) - (11) = (-1.6).

The end result is below +/-0 dBfs, it's the bit in red that could be an issue. The minidsp doesn't have internal headroom to cover off the 9.4 dB positive gain (I think it applies internal gain to cover off boost, but why chance it), you need to apply that headroom manually by adjusting the input gain.

(0) - (11) = (-11) + (10.6) = (-0.4). No peaks above 0dBfs.

Last edited by Audionut11; 20th August 2015 at 03:12 AM..
Old 20th August 2015
  #521
Lives for gear
 
jim1961's Avatar
 

Quote:
Originally Posted by Audionut11 View Post
(0) - (1.2) = (-1.2) + (10.6) = (9.4) - (11) = (-1.6).

The end result is below +/-0 dBfs, it's the bit in red that could be an issue. The minidsp doesn't have internal headroom to cover off the 9.4 dB positive gain (I think it applies internal gain to cover off boost, but why chance it), you need to apply that headroom manually by adjusting the input gain.

(0) - (11) = (-11) + (10.6) = (-0.4). No peaks above 0dBfs.
If what your proposing was the case, wouldnt it be reflective in the distortion graphs where the boost is being applied the most (62/63hz)? And at any volume level (since I am pretty sure neither the analog inputs or outputs are being taxed)?

If this were an analog device, I am certain you would be right. But the part you put in RED is in the digital domain. I wouldnt think they would allow +/- 16db PEQ magnitude variables if it meant distortion before the output stage. I could be wrong though.


But, you do have a good point. I might give it a go tomorrow. As you say, why chance it?

Last edited by jim1961; 20th August 2015 at 03:39 AM..
Old 20th August 2015
  #522
Gear Head
 

In the top example, you rely on the device to apply the necessary gain to ensure the device does not exceed 0 dBfs. Even in the digital domain, signals cannot exceed 0 dBfs, and you currently have the potential to apply 9.4 dB above 0 dBfs.

I would call it good gain structure. Extremely important with analog equipment, still relevant today when downmixing (and bass managing) channels.

The top way probably works fine, but you did invite criticism.
Old 20th August 2015
  #523
Lives for gear
 
jim1961's Avatar
 

Quote:
Originally Posted by Audionut11 View Post
In the top example, you rely on the device to apply the necessary gain to ensure the device does not exceed 0 dBfs. Even in the digital domain, signals cannot exceed 0 dBfs, and you currently have the potential to apply 9.4 dB above 0 dBfs.

I would call it good gain structure. Extremely important with analog equipment, still relevant today when downmixing (and bass managing) channels.

The top way probably works fine, but you did invite criticism.
https://dl.dropboxusercontent.com/u/...ting%2090B.xml

How about this?
Old 20th August 2015
  #524
Gear Head
 

Quote:
Originally Posted by jim1961 View Post
Yeah that's probably more optimal. Your EQ requirements and the limits of the minidsp means we're (just me actually ) probably splitting the worlds thinnest hair.

Anyway, here are two files. The first tweaks the gain settings a touch since you were leaving a dB or 2 of maximum output on the table.

https://dl.dropboxusercontent.com/u/...gain_tweak.xml

The second uses PEQ to ensure the peak signal remains as close to 0 dBfs as possible. Beware that this setting leaves some extra signal at the low end on channel 3 due to no remaining PEQ slots.

https://dl.dropboxusercontent.com/u/...gain_tweak.xml


If you're inclined, I'd be happy to look at unequalised mdats of each sub separately, and the mains, and see if I can find a more optimal EQ strategy (maybe I can't, but the offer is there). I'd really be happy just to see what you're doing with the funky EQ on channel 4 (and channel 3 to a lesser extent). All of that boost is non-conductive to a device that only outputs 0.9v maximum (edit: or did you get the balanced unit?). As you know, you can't actually apply boost with the minidsp, since that would drive the signal above 0 dBfs, so you're effectively reducing the remaining frequencies by 10.6 dB with the boost you're applying at those specific frequencies, which means, if my maths is correct, those remaining frequencies are output at around 0.26v. dB chart voltage power table conversion

Did I mention something about a thin hair being split.
Old 20th August 2015
  #525
Lives for gear
 
jim1961's Avatar
 

Lets start with what may or may not be obvious. You know I am using the 2.1 ADV plugin on the miniDSP (non-balanced)(2V input version). In my case, referring to the drawing, sub is the main sub, LS = Lmain, and RS = Rmain as far as the DSP is concerned.

Ch 1 = main sub (250W, 2x10" ported)(6 cu ft box)
Ch 2 (not used)
Ch 3 = LS (75w, 10" sealed)(1.2 cu ft box, stuffed)
Ch 4 = RS (75w, 10" sealed)(1.0 cu ft box, stuffed)

In the FR chart, one graph (blue) is Lmain, SL, and main sub, the other (purple) is without SL (left sub). This should explain why ch 3 has the EQ it does. A similar situation exists for the right channel and Ch 4.

Ch 3 or LS is there to fill in the dip centered around 59hz for the left channel. Ch 4 does a similar duty for the right channel.
Attached Thumbnails
My Listening Room-grid-big-4.jpg   My Listening Room-no-subl.jpg  

Last edited by jim1961; 20th August 2015 at 04:06 PM..
Old 21st August 2015
  #526
Gear Head
 

Thanks for the extra detail and your patience Jim.

Quote:
Originally Posted by jim1961 View Post
Ch 3 or LS is there to fill in the dip centered around 59hz for the left channel. Ch 4 does a similar duty for the right channel.
That EQ is working well, pretty well behaved in the time domain.

I think at this stage the horse has long lost feeling to all the beatings I'm giving it. Hope you enjoy your weekend.
Old 21st August 2015
  #527
Lives for gear
 
jim1961's Avatar
 

Quote:
Originally Posted by Audionut11 View Post
Thanks for the extra detail and your patience Jim.



That EQ is working well, pretty well behaved in the time domain.

I think at this stage the horse has long lost feeling to all the beatings I'm giving it. Hope you enjoy your weekend.
Surviving a few beatdowns is the only way I will know I am going in the right direction

I thought that about things over a year ago, but keep finding ways to improve things.
Old 17th September 2015
  #528
Gear Head
 

Hey Jim,

Your recent discussion over at AVS got me looking at some measurements, and I found some things in one of your measurements I'd like to discuss.
It looks to me like your kicker, and perhaps mainly the tail of the kicker, has excess magnitude around 11Khz.

Two different spectro graphs showing pretty much the same thing. Then looking at a sliced ETC overlay, the excess magnitude appears to be only in the tail. As you know I like to just kill everything, so with no first hand experience I'm keen to hear your thoughts.

The reflections around 11-18ms are somewhat interesting too. Granted they are 25dB down., but still, they produce a clear output above what I guess could otherwise be considered as the noise floor.

edit: The file of interest: 08 16 15 REW 256k LL RR 90.mdat
Attached Thumbnails
My Listening Room-spectro.jpg   My Listening Room-spectro1.jpg   My Listening Room-etc.jpg  
Old 17th September 2015
  #529
Lives for gear
 
jim1961's Avatar
 

Quote:
Originally Posted by Audionut11 View Post
Hey Jim,

Your recent discussion over at AVS got me looking at some measurements, and I found some things in one of your measurements I'd like to discuss.
It looks to me like your kicker, and perhaps mainly the tail of the kicker, has excess magnitude around 11Khz.

Two different spectro graphs showing pretty much the same thing. Then looking at a sliced ETC overlay, the excess magnitude appears to be only in the tail. As you know I like to just kill everything, so with no first hand experience I'm keen to hear your thoughts.

The reflections around 11-18ms are somewhat interesting too. Granted they are 25dB down., but still, they produce a clear output above what I guess could otherwise be considered as the noise floor.

edit: The file of interest: 08 16 15 REW 256k LL RR 90.mdat
Funny you should bring this up, just what I was working on.

Ill be posting new mdats on my thread soon, but these represent where I am now.

Basically, ive managed to extend downward the bandwidth and flatten the response of the terminator. To favor the 1.5k - 6k region instead of the above 8K .

BTW, why do you use a 500ms window when looking at the first 60ms of it? I usually scale the window to the view. Is there a right and wrong way to do this?
Attached Thumbnails
My Listening Room-sp2222.jpg   My Listening Room-1k3ksp.jpg  

Last edited by jim1961; 17th September 2015 at 06:01 AM..
Old 17th September 2015
  #530
Lives for gear
 
jim1961's Avatar
 

Got the gamma wrong. This one is the same as your earlier pic.
Attached Thumbnails
My Listening Room-correct-gamma.jpg  
Old 17th September 2015
  #531
Gear Head
 

Quote:
Originally Posted by jim1961 View Post
Basically, ive managed to extend downward the bandwidth and flatten the response of the terminator. To favor the 1.5k - 6k region instead of the above 8K.
Looks lots better. Is there an easy way to widen the Q of those peaks?


Quote:
Originally Posted by jim1961 View Post
BTW, why do you use a 500ms window when looking at the first 60ms of it? I usually scale the window to the view. Is there a right and wrong way to do this?
The time range just determines how much data past T=0 is generated. Accuracy may increase very slightly if I only generated data till 60ms (I guess), but otherwise that's just where the setting was previously. Based on the REW help, it makes sense to scale the time range with the graph axis. I'll ask over on the REW forum and see if there's something I'm missing.

I scaled the time range as such, because I like to zoom in on the region of interest.
The FR scaling was mainly just to try and show the data of interest, plus the settings panel, without the settings panel overlaying the data of interest. If we agree on some basic settings, we should remove the settings panel, allowing better viewing of the data.
Old 17th September 2015
  #532
I'm not sure why, but I've spent the last 2 hours reading this thread from start to finish; your determination/passion is admirable. While I only understand about 20% of the information being discussed, I am very impressed with the frequency response of your room, and hope I can make large improvements to whatever space I decide to turn into a mixing room. If I was nearby I would ask to rent your listening position to mix from my laptop!

Anyway just wanted to chime in and say good job, and I hope you get your room as musically enjoyable as possible!

Chris
Old 17th September 2015
  #533
Lives for gear
 
jim1961's Avatar
 

Here are the latest mdats.

Here are three spectrographs of RR at 10ms, 30ms and 100ms time windows. I am open to opinion to which best describes what is actually heard. The scale range being 12db. Showing to -12db is useful because this is the detection threshold for the Haas Effect.

A good link to a post where the Haas Effect and termination are discussed. https://www.gearslutz.com/board/7014534-post11.html

On the ETC, what I want to point out is the 24.5ms reflection is at 120 degrees, and the 27ms ones are at 60 degrees, where 0 degrees is directly forward and 180 directly behind.

I am deviating a bit from customary theory here, favoring directional specificity over a smooth tail.
Attached Thumbnails
My Listening Room-etc-rr.jpg   My Listening Room-100ms-spec.jpg   My Listening Room-30ms-spec.jpg   My Listening Room-10ms-spec.jpg  
Attached Files

Last edited by jim1961; 17th September 2015 at 05:46 PM..
Old 17th September 2015
  #534
Lives for gear
 
jim1961's Avatar
 

Here are the spectrographs with a 19db range. I chose this range because according to the Haas detection threshold, -19db is the detection threshold for 25ms.

So what I am trying to do is graph what reflections are detectable via frequency.
Attached Thumbnails
My Listening Room-r-19-spec.jpg   My Listening Room-l-19-spec.jpg  

Last edited by jim1961; 17th September 2015 at 05:03 PM..
Old 17th September 2015
  #535
Lives for gear
 
jim1961's Avatar
 

For reference

For reference.

Last edited by jim1961; 17th September 2015 at 04:41 PM..
Old 17th September 2015
  #536
Lives for gear
 
jim1961's Avatar
 

So in summation, I can describe my reflections in these two ways.

1) Detection threshold (-19db) = 830hz to 20k (+/-)
2) Haas Effect threshold (-12db) = 1K to 15K (+/-)

I am looking at this because a reflection should emulate the same tonal characteristics as the direct in regards to FR or linearity. In other words, you want your reflection(s) content to be broadband.

Now, if I were to take this literally, I might conclude that if I am going to have a 25ms delayed reflection, it should go down to 24hz (my system F3). We can augment that some right off the bat because directionality only exists down to 80-100hz. So I can revise this a bit to 100hz-20K. I dont know if this is the right number or not, but one thing I am sure of it I am not going to get there in my room. The main obstacle is because I am having to bounce the termination reflection twice (off two walls) to get the desired delay of 25ms. I have found that its exponentially more difficult to preserve the magnitude of the lower frequency bounces when two surfaces are required. The surface area of each reflection point has to get larger and larger as you descend in frequency. This is far easier when your only doing one bounce.

At any rate, the above plus some room cancellation (???), has me only down to around 1K in terms of usable reflection returns. Still, an improvement over where I was before, 1.5K. But more importantly, the response is flatter over it 1K-20k range. Before the high end (10K+) was favored (2-4db louder). Now the 1.5k - 8k range actually is favored by 1-2db.
Old 17th September 2015
  #537
Lives for gear
 
jim1961's Avatar
 

Quote:
Originally Posted by Obe1 View Post
I'm not sure why, but I've spent the last 2 hours reading this thread from start to finish; your determination/passion is admirable. While I only understand about 20% of the information being discussed, I am very impressed with the frequency response of your room, and hope I can make large improvements to whatever space I decide to turn into a mixing room. If I was nearby I would ask to rent your listening position to mix from my laptop!

Anyway just wanted to chime in and say good job, and I hope you get your room as musically enjoyable as possible!

Chris
Thanks for tuning in!

This thread has become a place to discuss things, but also a room diary so to speak. Its helpful sometimes to go back and see where I was and what I was thinking in the past.
Old 18th September 2015
  #538
Gear Head
 

Quote:
Originally Posted by jim1961 View Post
Now, if I were to take this literally, I might conclude that if I am going to have a 25ms delayed reflection, it should go down to 24hz (my system F3). We can augment that some right off the bat because directionality only exists down to 80-100hz. So I can revise this a bit to 100hz-20K. I dont know if this is the right number or not, but one thing I am sure of it I am not going to get there in my room.
We could probably conclude that it only needs to reflect down to the schroeder frequency. And then if I understand correctly, there is a transition region between wave theory and ray theory, so it probably only needs to reflect down to some frequency above the schroeder frequency. Maybe the boundry of reflection zone.

Which raises another question, how much benefit does generating an ISD termination in a small room have? <--That question is off-topic.

Your latest changes have removed almost all of the phase wrap at 11K, and you can see much more smoothness in this area in the GD plot.
The "tail" is broader in the time domain.
The region between 450hz and 1.3Khz is much smoother too. This is very noticeable in the spectrogram, both with and without normalising to peak of each frequency.

There's probably some excess energy around 11.6K, which corresponds to the sole remaining phase wrap in the upper frequencies.

Thanks for the other links and detail. I'll need some time to assimilate. The measurements suggest the listening experience went up yet another notch Have a nice weekend.
Old 18th September 2015
  #539
Gear Head
 

Quote:
Originally Posted by jim1961 View Post
Here are three spectrographs of RR at 10ms, 30ms and 100ms time windows. I am open to opinion to which best describes what is actually heard.
Using a smaller window allows the reflections themselves to be more easily distinguished in the time domain. Using a larger window allows the effect of the frequency response and decay to be clearly visible.

"The width of the window that is moved along the impulse response to generate the spectrogram is set by the Window control. The corresponding frequency resolution is shown next to the window setting."

Simplified REW Setup and Use (USB Mic & HDMI Connection) Including Measurement Techniques and How To Interpret Graphs - Page 415 - AVS Forum | Home Theater Discussions And Reviews

So if we imagine a 10ms window moving through your impulse response, it is looking at small chunks of the time domain. 0-10ms, 10-20ms, 20-30ms etc. It can more accurately pinpoint the kicker (in the time domain), since it is only visible in one of the captures (20-30ms).

So for instance, in the 10ms graph, we can clearly see where the reflections are in the time domain. Using a 100ms graph, increases the frequency resolution at the cost of time domain resolution. We can no longer see exactly where in the time domain the reflection occurred, however, we can see with higher resolution the frequency response of the reflection.

My comment regarding graph specifics was more focused on us being able to post graphs we can both understand without having the settings panel getting in the way. For instance, there are clear differences between 10ms and 100ms window settings, and we could easily settle on those two settings without having to describe to each other what window setting was used. I used 12dB scale range, since this clearly showed your reflection minus what can otherwise be described in this example as noise. I wanted reflection detail, without all of the other reflections clouding the data.
Old 18th September 2015
  #540
Lives for gear
 
jim1961's Avatar
 

Here are some rough 256K mdat's in an attempt to lessen some of those higher frequency reflection spikes.

Looking at Toole's charts, I am inclined to conclude the low threshold for reflection spaciousness or utility maybe in the 500-1K range, and the upper end in the 6-7K area.

An additional reason to not be too concerned keeping things level beyond 7K. Seemingly no reason for spikes or elevated output there.
Attached Thumbnails
My Listening Room-toole-reflection-fr.png  
Attached Files

Last edited by jim1961; 18th September 2015 at 06:03 PM..
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Similar Threads
Thread
Thread Starter / Forum
Replies
soundbarnfool / Mastering forum
21
Chaotic / Studio building / acoustics
21
Gerax / Live Sound
8
Sound Chaser / Remote Possibilities in Acoustic Music and Location Recording
2
wildplum / Remote Possibilities in Acoustic Music and Location Recording
3

Forum Jump
Forum Jump