Simplified REW Setup and Use (USB Mic & HDMI Connection) Including Measurement Techniques and How To Interpret Graphs - Page 253 - AVS Forum
First ... 251  252  253 254  255  ... Last
Audio Theory, Setup, and Chat > Simplified REW Setup and Use (USB Mic & HDMI Connection) Including Measurement Techniques and How To Interpret Graphs
Audionut11's Avatar Audionut11 09:53 PM 12-23-2013
Quote:
Originally Posted by AustinJerry View Post

My recommendation is that the measurement level be 40dB above whatever the poster is using as the noise floor (typically 45dB or 50dB).  IOW, a measurement level of 85 or 90dB.

Agreed.

The top graph only looks better because you stuck to a consistent scale. IMHO the graph should be scaled to the peak of the signal down to -40dB with a minimum peak signal level of 90dB.
This way the waterfall will behave in the same manner as the spectrogram, and will show the data of interest only.

markus767's Avatar markus767 01:24 AM 12-24-2013
Quote:
Originally Posted by AustinJerry View Post

Comments?

I'm not sure the term "noise floor" is fully understood. From Wikipedia: "In signal theory, the noise floor is the measure of the signal created from the sum of all the noise sources and unwanted signals within a measurement system, where noise is defined as any signal other than the one being monitored."

Example animation:


kbarnes701's Avatar kbarnes701 04:22 AM 12-24-2013
Quote:
Originally Posted by AustinJerry View Post
 

OK, as promised, here is an analysis of the effect of measurement levels on Waterfall graphs.  Or, as Markus might say, watch the iceberg emerge from the water...

 

 

 

Comments?

 

Full agreement from me Jerry on your suggestion. I do my waterfall readings at about 90dB anyway as you know. I think we absolutely need a standard - say 40dB above the noise floor - for the purposes of comparison and I think all members should be directed to adopt that standard if they want feedback on their graphs.

 

Here is a similar experiment I did - here there is only a few dB difference between the measurement levels, but look how the second waterfall is 'cosmeticised' by that small difference. At a casual glance one may not even notice the level difference (if the graphs were not posted side by side) so this shows the importance of a proper standard being agreed and followed.

 

 


Audionut11's Avatar Audionut11 04:50 AM 12-24-2013
Quote:
Originally Posted by kbarnes701 View Post



Do you think you can hear the area circled in red? What about the area circled in green?
kbarnes701's Avatar kbarnes701 05:50 AM 12-24-2013
Quote:
Originally Posted by Audionut11 View Post
 
Quote:
Originally Posted by kbarnes701 View Post


Do you think you can hear the area circled in red? What about the area circled in green?

 

Good question. Difficult to say. Probably not for the area circled in red. It's pretty difficult to hear bass that low anyway, due to the way our hearing works (eg look how quiet Audyssey's 75dB bass is on the test tones!) and when it is 30dB down, well I imagine it is pretty difficult to hear the original sound let alone any reflected sound that is so much quieter. 

 

The area circled in green is more problematic. I have stopped trying to improve the red-circled area as I have come to the view expressed above and also accept that in my room it is pretty much impossible to go any further with treatments - I just don't have the space to use sufficiently thick gubbins to have any further effect that low down. I also do not want to 'chase graphs' as an exercise in itself.  But the area in green is around 50Hz which is much more audible and problems are beginning to develop only 15dB down. I do believe that this can be heard - although the sound is currently of a standard where it isn't bothering me unduly. HST, I would still want to try to improve this area, but it is so far defying all attempts. I have doubled the gubbins in many of my panels to 8 inches for example, hoping it might be beneficial but without much impact at that frequency. I have since installed 5 new panels but have not yet measured as I am relocating my Height speakers after Xmas and wanted to run Audyssey and measure again once that is done. If it makes any impact on the green-circled area I will be delighted (I will post the results either way of course). If it doesn't then I think I am going to have to live with it. Considering what a very poor space I started with, I have already achieved far, far more than I ever thought would be possible and I accept that at some point I will reach a limitation that I cannot overcome. Maybe I already have... we shall see.


Audionut11's Avatar Audionut11 06:45 AM 12-24-2013
Quote:
Originally Posted by kbarnes701 View Post

Probably not for the area circled in red. It's pretty difficult to hear bass that low anyway, due to the way our hearing works

Our hearing is excellent at filtering out noise, which is exactly what that area (in red) is.

Have another look at the graph markus767 posted above.
Quote:
Originally Posted by kbarnes701 View Post

The area circled in green is more problematic..................But the area in green is around 50Hz which is much more audible and problems are beginning to develop only 15dB down. I do believe that this can be heard

Agreed.
Quote:
Originally Posted by kbarnes701 View Post

I also do not want to 'chase graphs' as an exercise in itself.

Which is why you probably shouldn't be concerning yourself with graphs that show unwanted detail. Consider these 2 graphs below, top 1 scaled to -40dB, bottom 1 scaled to remove the noise. Note: The time range has been changed in the bottom graph also to further help show the area of interest.






The top 1 looks like crap (for the sake of the discussion) and might concern the listener that there are bigger problems then otherwise exist. The areas circled in red are room modes with have been dampened. The 2 right most areas in red are 22dB below the primary signal and do not effect response until 105ms after the primary signal. They can be considered noise. The blue areas are noise.

The bottom graph filters all of this noise out. Doesn't it look better! It's much more representative of what I can hear wink.gif I can be very happy with my system response smile.gif


Since sooner or later I am going to want to strive for more, there are 3 areas I can deal with.

Sound insulating the room (not easy), to reduce ambient noise (lower the noise floor), and increase dynamic range.
Further dampening the existing room modes (not easy)
Or deal with the decay at VLF. The area circled in red on the left hand side of the graph above, and better represented in the graph below.



Again, not easy.

A 4th way to better increase the listening experience would be a better sub.

TLDR; Graphs scaled to a fixed value will vary rarely focus on the detail of interest, and include far to much noise in the analysis .
Having a defined upper limit also wastes valuable resolution in the provided graph. It wastes space at the top. Graphs should always be scaled to 1-2dB above the maximum source signal.
I personally think a -40dB bottom graph limit is fine as it limits noise interference in the graphs, and is probably about the best fixed value to help keep things reasonably consistent for newcomers.
kbarnes701's Avatar kbarnes701 07:35 AM 12-24-2013
Quote:
Originally Posted by Audionut11 View Post
 
 


Again, not easy.

A 4th way to better increase the listening experience would be a better sub.

TLDR; Graphs scaled to a fixed value will vary rarely focus on the detail of interest, and include far to much noise in the analysis .
Having a defined upper limit also wastes valuable resolution in the provided graph. It wastes space at the top. Graphs should always be scaled to 1-2dB above the maximum source signal.
I personally think a -40dB bottom graph limit is fine as it limits noise interference in the graphs, and is probably about the best fixed value to help keep things reasonably consistent for newcomers.

 

I found your comments interesting, but I have never before seen a waterfall that even vaguely resembles the one above, neither in this thread or elsewhere, so I am struggling to take on board what you say it represents.


Audionut11's Avatar Audionut11 07:46 AM 12-24-2013
Quote:
Originally Posted by kbarnes701 View Post

I found your comments interesting, but I have never before seen a waterfall that even vaguely resembles the one above, neither in this thread or elsewhere, so I am struggling to take on board what you say it represents.

Minus the tiny little spikes, from 43hz-200hz the decay is -24db @ 100ms. From 25hz-33hz, the decay is only -10dB @ 100ms. We want the decay to be even over the entire FR and through the time domain. This way the room effects all frequencies evenly. The quoted graph focuses in on the excessive room energy in the lower frequencies that otherwise is very difficult to see on graphs scaled as per the guide for instance.

The area between 25hz-33hz is only 10dB below the primary signal @ 100ms, and 14dB louder then 43hz-200hz.
AustinJerry's Avatar AustinJerry 08:26 AM 12-24-2013

Audionut11 uses different setting in the Controls for his waterfalls, which explain why the graphs look somewhat different than the ones we are used to.  Here are two waterfalls, one using Audionutt11's settings, and one using the settings recommended in the Guide.  I am not saying that the Guide's recommendations are "correct", only that we should be consistent in out presentations.  The Guide states, "These recommended settings are taken from a series of articles published here, titled Bass Integration Guide Parts 1-3, by Paul Spencer. A highly recommended read."

 

 


Audionut11's Avatar Audionut11 08:36 AM 12-24-2013
Quote:
Originally Posted by AustinJerry View Post

Audionut11 uses different setting in the Controls for his waterfalls,

Defined to focus in on the areas of interest. Here is the same measurement at guide settings.


cadett's Avatar cadett 10:26 AM 12-24-2013
I prefer the visual representation of the graph Audionut11 posted. It gives more resolution & detail of what is taking place. JMO
AustinJerry's Avatar AustinJerry 10:31 AM 12-24-2013

OK, testing for understanding from everyone, here are the proposed "standards" for publishing Waterfall graphs:

 

- Using the RTA procedure described in the guide, determine the noise floor for your listening environment (typically in the 40-50dB range).

- Measure at a level at least 40dB above the measured noise floor (90dB is a good target level)

- Set the Waterfall graph scales as follows:

  . Horizontal scale should be 15-300Hz (10Hz if you are showing off, or your name is Keith)

  . Vertical scale lower limit should be the measured noise floor

  . Vertical scale upper limit should be immediately above the upper limit of the measurement (say 95db if the measurement is to 90dB)

- Settings should be as follows, with the Time Range parameter adjusted to show fully decayed resonances at the lower limit of the graph:

 

 

- Graph color should enhance clarity when viewed in the thread.

 

Open for discussion.


AustinJerry's Avatar AustinJerry 10:32 AM 12-24-2013
Quote:
Originally Posted by cadett View Post

I prefer the visual representation of the graph Audionut11 posted. It gives more resolution & detail of what is taking place. JMO

 

Which one?


Audionut11's Avatar Audionut11 11:27 AM 12-24-2013
Quote:
Originally Posted by AustinJerry View Post

Open for discussion.

Measure at a level 85dB or more.
Vertical scale lower limit should be -40dB from the peak level.
Vertical scale upper limit should be just high enough to see the peak level.

I prefer total slices to be 100, showing higher resolution.

This produces graphs as such.






Also note the last graph with the recommended slice setting.


IMO, it's much easier to see the dips in the time domain with the higher resolution from the higher slice setting.
AustinJerry's Avatar AustinJerry 11:36 AM 12-24-2013
Thanks for the feedback. Of the settings, the number of slices alters the overall graph the least, and I agree, it can make the variations a little easier to see.
Audionut11's Avatar Audionut11 11:51 AM 12-24-2013
Quote:
Originally Posted by AustinJerry View Post

Thanks for the feedback.

No problem smile.gif

Vertical lower limit of -40dB with a time window of 300ms isn't ideal for all measurements either, but I think it's the best compromise for keeping things reasonably consistent for new comers, while still limiting the graph the useful detail.

Of course where someone has posted the waterfall as per the guide (however that's updated in the future), they can then discuss it further with members to narrow their graphs down even further to help focus on areas of interest, and eliminate noise that might be effecting their expectations.

Dismissing the different graph properties of my measurements above, you can see how different settings target different areas of interest (hopefully leading to further discussion of the pros/cons of each area).
cfraser's Avatar cfraser 01:18 PM 12-24-2013
Quote:
Originally Posted by Audionut11 View Post

Quote:
Originally Posted by AustinJerry View Post

Thanks for the feedback.

No problem smile.gif

Vertical lower limit of -40dB with a time window of 300ms isn't ideal for all measurements either

I think 300ms is way too short. Based on what I've seen. So do the guys who wrote the paper Sanjay referred to the other day. I am not sure that 300ms is even an adequate time for higher frequencies, never mind bass, usually longer decay times are recommended, but IDK. I realise "40dB down" is already relaxing the standard a bit...

"Standard:
Resonances from 35Hz-300Hz should not extend beyond 350ms before decaying into the noise floor or reaching a level of -40dB. Below 35Hz this standard is relaxed to 450ms"
kbarnes701's Avatar kbarnes701 02:32 PM 12-24-2013
Quote:
Originally Posted by AustinJerry View Post
 

Audionut11 uses different setting in the Controls for his waterfalls, which explain why the graphs look somewhat different than the ones we are used to.  Here are two waterfalls, one using Audionutt11's settings, and one using the settings recommended in the Guide.  I am not saying that the Guide's recommendations are "correct", only that we should be consistent in out presentations.  The Guide states, "These recommended settings are taken from a series of articles published here, titled Bass Integration Guide Parts 1-3, by Paul Spencer. A highly recommended read."

 

 

 

The only thing I would say is that of all the hundreds of waterfalls I have seen, all over the Internet, most resemble those described in the Guide and Hi-fi Zine. I have never once seen one that resembles Audionut's until today. This makes me wonder who is out of step - Audionut or everyone else? Of course, the latter is possible...


kbarnes701's Avatar kbarnes701 02:42 PM 12-24-2013
Quote:
Originally Posted by AustinJerry View Post
 

OK, testing for understanding from everyone, here are the proposed "standards" for publishing Waterfall graphs:

 

- Using the RTA procedure described in the guide, determine the noise floor for your listening environment (typically in the 40-50dB range).

- Measure at a level at least 40dB above the measured noise floor (90dB is a good target level)

- Set the Waterfall graph scales as follows:

  . Horizontal scale should be 15-300Hz (10Hz if you are showing off, or your name is Keith)

  . Vertical scale lower limit should be the measured noise floor

  . Vertical scale upper limit should be immediately above the upper limit of the measurement (say 95db if the measurement is to 90dB)

- Settings should be as follows, with the Time Range parameter adjusted to show fully decayed resonances at the lower limit of the graph:

 

 

- Graph color should enhance clarity when viewed in the thread.

 

Open for discussion.

 

I think this is exactly what I already do. I will check...

 

Yes, my noise floor is 45dB. I have my lower limit set to 45dB.

 

I am measuring at about 95dB or even a little more... that is 40dB above the noise floor.

 

My horizontal scale is 15-300Hz (for publication - for private dual Submersive gloating it is 10-300Hz :) )\

 

My vertical scale upper limit is about 7dB above my peak measurement, so I guess that is close enough.

 

Time range is 450ms,showing full decay as requested,

 

This results in the following waterfall:

 

 

So, Jerry, isn’t this how we have always done it?  I am sure I only ever followed the Guide...

 

 

 

EDIT: here is the above graph but using 100 slices:

 

 

 

 

 

It does look easier to read, but essentially has the same info.


Audionut11's Avatar Audionut11 03:25 PM 12-24-2013
Quote:
Originally Posted by cfraser View Post

"Standard:
Resonances from 35Hz-300Hz should not extend beyond 350ms before decaying into the noise floor or reaching a level of -40dB. Below 35Hz this standard is relaxed to 450ms"

This applies to room modes only afaik. Room modes starting at the peak of the frequency of interest and extending through to the noise floor.

Room modes that have been dampened such as in this example,



Are no longer having the same effect. Here they have only become noticeable at -22dB below the source signal and 105ms after the source signal. In the same manner that lossy codecs remove information that is not audible (due to the brains excellent ability to ignore noise), the room modes have been reduced in amplitude, and extended far from the source signal in the time domain that they can be considered noise. Noise filtered out by the brain.
Quote:
Originally Posted by kbarnes701 View Post

The only thing I would say is that of all the hundreds of waterfalls I have seen, all over the Internet, most resemble those described in the Guide and Hi-fi Zine. I have never once seen one that resembles Audionut's until today. This makes me wonder who is out of step - Audionut or everyone else? Of course, the latter is possible...

There is no "out of step". The graph I posted was specific to my setup showing a specific area of interest. It has nothing to do with, "start adjusting graph parameters like this"!
Quote:
Originally Posted by Audionut11 View Post

The bottom graph filters all of this noise out. Doesn't it look better! It's much more representative of what I can hear wink.gif I can be very happy with my system response smile.gif

Does the above quote not carry the implication that I intended? I then went on to say,
Quote:
Originally Posted by Audionut11 View Post

Since sooner or later I am going to want to strive for more, there are 3 areas I can deal with

One of which was the very low frequency decay which I highlighted using the waterfall, with the parameters required to do so.
Quote:
Originally Posted by kbarnes701 View Post

It does look easier to read, but essentially has the same info.

A waterfall graph extended out to 300ms with a total of 30 slices will show a line of resolution (detail) every 10ms. The same graph with 100 slices will show a line of resolution every 3ms.
Extend the graph out to 600ms and with 30 slices you have a line of resolution every 20ms vs every 6ms.

1080p video displayed on a 1080p 55" TV is easier on the eyes then 320p video. But essentially has the same info wink.gif
rhed's Avatar rhed 10:50 AM 12-25-2013
Quote:
Originally Posted by AustinJerry View Post

The REW Guide has been updated to Version 3.0.  Changes in this version are:








3.0




Sept 12, 2013




-          Added link (p. 3) to guide on how to use REW with the Apple MacBook Pro, authored by forum participant J_P_A.





Special thanks to J_P_A for taking time to create this guide addendum, which should be a welcome addition for anyone using an Apple laptop.



Where do I find this guide for the Mac? I might use my wife's Mac Pro for now with the USB mic I got from cross spectrum labs. Her Mac book has a HDMI output. Thanks
AustinJerry's Avatar AustinJerry 11:14 AM 12-25-2013
As the post says, open the guide, scroll down to page 3, and you will see an embedded link that will take you there.
FLViking2011's Avatar FLViking2011 05:41 PM 12-25-2013
I just received my umm-6 mic and installed rew. I noticed rew automatically detected the mic and asked for the calibration file. Is this mic now an auto setup one like the umik-1 where one does not have to do the spl measurement or is that step still necessary? Also, I have currently ordered the following mini stereo to RCA cable: http://www.amazon.com/gp/aw/d/B002K7ULJQ/ref=mp_s_a_1_fkmr1_2?qid=1388018048&sr=8-2-fkmr1&pi=AC_SX110_SY165_QL70
Does this go straight from the laptop headphone jack to the left and right RCA inputs on the front of the receiver? Or will I need a y-splitter in addition to this cable?

Thanks,
artur9's Avatar artur9 09:14 PM 12-25-2013
I got a sub amp to drive my passive subs. That extends the bass quite a bit further than just using some channels on the multi-channel amp. That's good.

I am hoping someone can give me some pointers on what may be causing the dip(s) between 40 and 70 Hz. This is using just the front two passive subs. For comparison, the FR of the 3rd sub is included. I know that the hump near 30Hz is a room mode I have to address at some point. Are the dips strictly a sub positioning thing?

This really is a dance of two steps forward and one step back. My bass is extended but not quite as smooth as previously and the ringing according to the waterfall is somewhat worse. I'm thinking of putting up about 9 feet of 3.5" acoustic panel on the rear wall (behind MLP) to address that. That is, 6-7 1.33'x4'x3.5" DIY panels.



As always TIA.
AustinJerry's Avatar AustinJerry 09:27 PM 12-25-2013
Quote:
Originally Posted by FLViking2011 View Post

I just received my umm-6 mic and installed rew. I noticed rew automatically detected the mic and asked for the calibration file. Is this mic now an auto setup one like the umik-1 where one does not have to do the spl measurement or is that step still necessary? Also, I have currently ordered the following mini stereo to RCA cable: http://www.amazon.com/gp/aw/d/B002K7ULJQ/ref=mp_s_a_1_fkmr1_2?qid=1388018048&sr=8-2-fkmr1&pi=AC_SX110_SY165_QL70
Does this go straight from the laptop headphone jack to the left and right RCA inputs on the front of the receiver? Or will I need a y-splitter in addition to this cable?

Thanks,

Whether the UMM-6 needs to be calibrated depends on your calibration file. Open the file in a text editor to see if the first line in the file starts with the word "sensitivity". If yes, then the calibration file is automatically setting the mic level in REW. If not, you will need to calibrate the mic using an external SPL.

As for the cable you have, when you plug that into the headphone jack on the laptop, REW will output a mono signal to either the left or right RCA jack. On the AVR side, you will need a Y-splitter cable to take the mono signal to both the left and right inputs on the AVR in order to measure left+right channels, and center channel.
AustinJerry's Avatar AustinJerry 09:34 PM 12-25-2013
Quote:
Originally Posted by artur9 View Post

I got a sub amp to drive my passive subs. That extends the bass quite a bit further than just using some channels on the multi-channel amp. That's good.

I am hoping someone can give me some pointers on what may be causing the dip(s) between 40 and 70 Hz. This is using just the front two passive subs. For comparison, the FR of the 3rd sub is included. I know that the hump near 30Hz is a room mode I have to address at some point. Are the dips strictly a sub positioning thing?

This really is a dance of two steps forward and one step back. My bass is extended but not quite as smooth as previously and the ringing according to the waterfall is somewhat worse. I'm thinking of putting up about 9 feet of 3.5" acoustic panel on the rear wall (behind MLP) to address that. That is, 6-7 1.33'x4'x3.5" DIY panels.



As always TIA.

Wow, the baseline is WAY better. If all you did was change how you power the subs, and you didn't move anything, I just can't understand why the measurements are so different.
Audionut11's Avatar Audionut11 09:47 PM 12-25-2013
Quote:
Originally Posted by artur9 View Post

I am hoping someone can give me some pointers on what may be causing the dip(s) between 40 and 70 Hz. This is using just the front two passive subs. For comparison, the FR of the 3rd sub is included. I know that the hump near 30Hz is a room mode I have to address at some point. Are the dips strictly a sub positioning thing?

The dips/humps are strictly a room thing (throw more pink fluffy at it :P ), which can be adjusted via sub placement.

The first thing I would try is moving the subs around and measuring, rinse and repeat. Someone posted a link to sub placement tips for multiple subs in the last page or 2 (sorry no time to check which link it is personally).
artur9's Avatar artur9 10:17 PM 12-25-2013
Quote:
Originally Posted by AustinJerry View Post

Wow, the baseline is WAY better. If all you did was change how you power the subs, and you didn't move anything, I just can't understand why the measurements are so different.

I didn't move anything but how the 2 passives are connected is quite a bit different. I am using the sub amp's crossover instead of the pre pro (but see below). I like the sound of my LR speakers in bypass better than how they sound if they are digitized for doing crossover handling. But I do want deeper bass than my bookshelves offer (darn cello music). I'm trying to get the best of both worlds by using the sub amp's crossover.

I was thinking the sub amp's crossover might be causing the dip but the front subs by themselves have always had that dip according to old measurements. Maybe the front subs have never been contributing as much as they should have.

Since I have some time now I want to address that if I can without coating the room in fiberglass.
artur9's Avatar artur9 10:22 PM 12-25-2013
Quote:
Originally Posted by Audionut11 View Post

The dips/humps are strictly a room thing (throw more pink fluffy at it :P ), which can be adjusted via sub placement.

The first thing I would try is moving the subs around and measuring, rinse and repeat. Someone posted a link to sub placement tips for multiple subs in the last page or 2 (sorry no time to check which link it is personally).

I did try moving one of the subs around sort of within a 5ft circle. I don't have any spare speaker wire at the moment to do more than that but there was no real improvement in that area of concern. Placement options are limited, as usual, as my room is a strange shape.

Nice thing about the passives is that they don't weigh very much.
AustinJerry's Avatar AustinJerry 10:23 PM 12-25-2013
In my experience, positioning addresses frequency response issues, while treatments address room resonance issues. Best to explore options to flatten the response before you invest in a ton of treatments. JMO.
Tags: Dayton , Dayton Audio , Room Equilizer Wizard Rew
First ... 251  252  253 254  255  ... Last

Up
Mobile  Desktop