ProSoundWeb Community

Please login or register.

Login with username, password and session length
Advanced search  

Pages: 1 2 [All]   Go Down

Author Topic: Smaart RTA 1dB/Octave HF Roll Off?  (Read 1309 times)

Art Welter

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1187
Smaart RTA 1dB/Octave HF Roll Off?
« on: October 12, 2017, 11:44:19 am »

In the outdoor test measurements below the microphone is on the ground 2 meters distant from the cabinet face also on the ground, the “tapped horn” mouth  exit center approximately 50 centimeters off  the ground, 11 meters from a house.

The Smaart Magnitude Response has been equalized for flat response from 30-100 Hz, the same signal measured using the RTA results in about a 3 dB drop from 30-100 Hz, with a less steep lower and upper roll off. The roll-offs are consistent over time, passing traffic and airplanes were mainly visible only above or below the pass-band.

The RTA’s comparative rolloff in an outdoor environment has me curious to the cause. Anyone familiar with the explanation?

Cheers,
Art
Logged

Frank Koenig

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 393
Re: Smaart RTA 1dB/Octave HF Roll Off?
« Reply #1 on: October 12, 2017, 02:15:37 pm »

Hi Art,

Sticking my neck out here, but at 2m measuring distance it seems you might be flirting with the near field. I wonder what happens if you go out to, say, 6m?

Have you looked at the impulse response to see hints of any inadvertent reflections (thinking of that house)? Unfortunately, Smaart does not give us direct control (or knowledge) of the measurement window it uses for the transfer function. This might be be easier to sort out using ARTA, or the like.

I think we agree that in the anechoic far field the RTA (with pink excitation) should agree with the (smoothed) magnitude of the transfer function. I look forward to the explanation.

--Frank

Logged
Yes, it is a giant stereo system!

Mark Wilkinson

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 556
Re: Smaart RTA 1dB/Octave HF Roll Off?
« Reply #2 on: October 12, 2017, 05:18:55 pm »

Hi Art, interesting...........what's signal source?  And X-overs in play ?
Logged

Art Welter

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1187
Re: Smaart RTA 1dB/Octave HF Roll Off?
« Reply #3 on: October 15, 2017, 12:10:24 pm »

Hi Art,

1)Sticking my neck out here, but at 2m measuring distance it seems you might be flirting with the near field. I wonder what happens if you go out to, say, 6m?

2)Have you looked at the impulse response to see hints of any inadvertent reflections (thinking of that house)? Unfortunately, Smaart does not give us direct control (or knowledge) of the measurement window it uses for the transfer function. This might be be easier to sort out using ARTA, or the like.

3)I think we agree that in the anechoic far field the RTA (with pink excitation) should agree with the (smoothed) magnitude of the transfer function. I look forward to the explanation.

--Frank
1) My sub testing has shown the influence of reflected sound becomes worse as the distance is increased, and though 1 meter is too close, 2 meters is sufficiently "far field".

2)Not specifically, but coherency was 100% in the pass band. The house should  influence both RTA and magnitude response- in NM I found a convenient location out the shop door had something like +4 dB at 40 Hz compared to getting 25 feet away from the nearest building measuring at 2 meters.

3)Me too.
Logged

Art Welter

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1187
Re: Smaart RTA 1dB/Octave HF Roll Off?
« Reply #4 on: October 15, 2017, 12:20:38 pm »

Hi Art, interesting...........what's signal source?  And X-overs in play ?
Signal source was Smaart pink noise. Processing was a DriverackPA.
Without processing, it would be hard to notice the 3 dB drop in the raw response with all the usual peaks and dips, but since the sub's output was equalized flat, it became obvious.
Logged

Mark Wilkinson

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 556
Re: Smaart RTA 1dB/Octave HF Roll Off?
« Reply #5 on: October 16, 2017, 10:28:26 am »

Signal source was Smaart pink noise. Processing was a DriverackPA.
Without processing, it would be hard to notice the 3 dB drop in the raw response with all the usual peaks and dips, but since the sub's output was equalized flat, it became obvious.

Hi Art, if transfer's reference channel is straight from the soundcard (ie loopback), I don't have a clue.

But if you're getting reference from a mixer, maybe you have a mixer filter in play you don't mean to ? 

I screw that up every now and then.......
It's probably too simple to even mention to you, but since it could cause what you're seeing, I though i should....

Logged

Chris Tsanjoures

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 65
Re: Smaart RTA 1dB/Octave HF Roll Off?
« Reply #6 on: October 16, 2017, 11:30:50 am »

I'd be curious what the measurement looks like using Smaart's modern TF algorithm (try using the demo of v8). The v6 TF is effectively 24 FPPO. The under-the-hood mechanics of the TF engine from v7 forward uses a far more refined scheme of sample rate decimation and various FFT sizes to produce a measurement with different time and frequency resolution in the various frequency ranges - called MTW or Multi Time Window. MTW is effectively 80 FPPO.

You may also find adjusting the TF engine settings from Polar to Complex will give you a slightly different result depending on what you are looking for/what you are looking to ignore. In v6 I believe this is called Complex and Vector - though forgive me, it's been quite some time since I've opened that version.

Logged
I am employed by Rational Acoustics. Besides being a Product Manager for the Smaart Platform, I operate as a fully deployable Smaart Ninja (aka Instructor/System Guy).

Art Welter

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1187
Re: Smaart RTA 1dB/Octave HF Roll Off?
« Reply #7 on: October 16, 2017, 12:58:35 pm »

Hi Art, if transfer's reference channel is straight from the soundcard (ie loopback), I don't have a clue.

But if you're getting reference from a mixer, maybe you have a mixer filter in play you don't mean to ? 

I screw that up every now and then.......
It's probably too simple to even mention to you, but since it could cause what you're seeing, I though i should....
Mark,

Good point, but after I realized I could just use some simple adapters and a phantom PSU, stopped using an outboard mixer, so no mixer filters.
Logged

Art Welter

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1187
Re: Smaart RTA 1dB/Octave HF Roll Off?
« Reply #8 on: October 16, 2017, 01:31:14 pm »

I'd be curious what the measurement looks like using Smaart's modern TF algorithm (try using the demo of v8). The v6 TF is effectively 24 FPPO. The under-the-hood mechanics of the TF engine from v7 forward uses a far more refined scheme of sample rate decimation and various FFT sizes to produce a measurement with different time and frequency resolution in the various frequency ranges - called MTW or Multi Time Window. MTW is effectively 80 FPPO.

You may also find adjusting the TF engine settings from Polar to Complex will give you a slightly different result depending on what you are looking for/what you are looking to ignore. In v6 I believe this is called Complex and Vector - though forgive me, it's been quite some time since I've opened that version.
Chris,

Don't find any adjustment screen called "TF engine settings" or "Polar" or  "Complex". Was using the 32K FFT Size, which claims a 1.5 Hz frequency resolution.
What would  a higher effective FPPO rate have to do with the difference I displayed between the RTA and Magnitude response?

Having "upgraded" to OSX 10.6.8, the last available for my MacBook processor has downgraded the performance of Smaart considerably, using "Hide" will crash it.
Nothing higher than Smaart V6 will run on my laptop, so can't do the comparison you are curious about- you'll have to satisfy your curiosity yourself ;^).

Cheers,
Art



Logged

Chris Tsanjoures

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 65
Re: Smaart RTA 1dB/Octave HF Roll Off?
« Reply #9 on: October 16, 2017, 02:39:04 pm »

Chris,

Don't find any adjustment screen called "TF engine settings" or "Polar" or  "Complex". Was using the 32K FFT Size, which claims a 1.5 Hz frequency resolution.
What would  a higher effective FPPO rate have to do with the difference I displayed between the RTA and Magnitude response?

Having "upgraded" to OSX 10.6.8, the last available for my MacBook processor has downgraded the performance of Smaart considerably, using "Hide" will crash it.
Nothing higher than Smaart V6 will run on my laptop, so can't do the comparison you are curious about- you'll have to satisfy your curiosity yourself ;^).

Cheers,
Art

re: engine settings, this is in v8. Polar averaging is useful for getting a response that 'looks' more like it 'sounds' and since the RTA is most closely related to human hearing, I would think that polar averaging would make the two match pretty well. This is a moot point however, as you can't run v8 cause your computer is at 10.6, and I can't run v6 cause my computer is at 10.12!


With a higher frequency resolution, comes more data points. In the case of MTW, we have 1Hz resolution up to 140 Hz, and 1/48th octave resolution from 60Hz up - which is double the frequency resolution of v6. It could be possible there is information outside the window of v6 TF that is not showing, causing the de-correlation of the RTA to the TF.
Logged
I am employed by Rational Acoustics. Besides being a Product Manager for the Smaart Platform, I operate as a fully deployable Smaart Ninja (aka Instructor/System Guy).

Frank Koenig

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 393
Re: Smaart RTA 1dB/Octave HF Roll Off?
« Reply #10 on: October 16, 2017, 03:46:18 pm »

In the case of MTW, we have 1Hz resolution up to 140 Hz, and 1/48th octave resolution from 60Hz up - which is double the frequency resolution of v6.

2 Hz resolution requires a .5 s measurement window which will include reflections off of boundaries within ~ 280 ft. of the measurement location (565 ft out and back at 1130 ft/s). For practical purposes this is an unwindowed measurement. Reflecting objects that subtend a small solid angle (small and far away), however, will have relatively little effect as the sound power they contribute falls off as the fourth power of their distance (inverse square law twice, like radar). This still does not explain the difference between the RTA (constant fractional-bandwidth power spectrum) and the smoothed magnitude.

I think the first step is to determine if this is a strictly acoustic phenomenon, or some system or measurement issue. As might already have been suggested (and you likely already did), take all processing out of the system and measure the RTA at the speaker terminals or the input of the known-flat power amplifier. Also perform a "loop-back" on the spectrum analyzer with the measurement and reference inputs tied together. If that checks out, insert the electroacoustic part and see what you get. If it turns out to be acoustic, the next thing I'd try is to alter the physical arrangement, such as the measurement distance and angle. Angle SHOULD have relatively little effect as the source is nearly omnidirectional, but we're probing assumptions here.

Varying the type of averaging and smoothing makes sense, too. Ideally, the smoothing of the magnitude should match, to the extent possible, the bandwidth of the RTA filters.

Next time I'm out in a field with a woofer and my measurement rig I'll try to reproduce this. It may be a while. With interest.

--Frank
Logged
Yes, it is a giant stereo system!

Art Welter

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1187
Re: Smaart RTA 1dB/Octave HF Roll Off?
« Reply #11 on: October 16, 2017, 05:33:17 pm »

I think the first step is to determine if this is a strictly acoustic phenomenon, or some system or measurement issue. As might already have been suggested (and you likely already did), take all processing out of the system and measure the RTA at the speaker terminals or the input of the known-flat power amplifier. Also perform a "loop-back" on the spectrum analyzer with the measurement and reference inputs tied together. If that checks out, insert the electroacoustic part and see what you get.
Ding Ding Ding-

We have a winner, Frank K!

The Smaart RTA function is flat when set to any of the portions of "octave".
The RTA function when set to "Log" or "Lin", rolls off the top end, on "Log" about 30 dB from 20Hz to 20kHz, or 4 dB from 30 Hz to 100 Hz as measured...
This does not occur on the Magnitude response screen, flat response looks flat  regardless of whatever you choose, although "Lin" is useless for anything below 200 Hz.

So the interesting by-line is when looking at harmonic distortion of sine waves on the RTA, the harmonics will appear to have less level (inaccurate) when looking at Log function compared to the octave functions. A 30 Hz fundamental with the third harmonic at -20 would have 10% distortion, but using the Log function would read about -24, only 6.31% distortion.

Smaart has made me a (sometimes) inadvertent "liar" when reporting speaker distortion ;^).
Well, at least I learned something new about a 10 year old software ::).

I'll be using REW for my future distortion testing, but it won't run on my laptop at all...

Cheers,
Art
« Last Edit: October 16, 2017, 05:42:43 pm by Art Welter »
Logged

Frank Koenig

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 393
Re: Smaart RTA 1dB/Octave HF Roll Off?
« Reply #12 on: October 16, 2017, 06:17:33 pm »

The RTA function when set to "Log" or "Lin", rolls off the top end, on "Log" about 30 dB from 20Hz to 20kHz, or 4 dB from 30 Hz to 100 Hz as measured...
This does not occur on the Magnitude response screen, flat response looks flat  regardless of whatever you choose, although "Lin" is useless for anything below 200 Hz.

Interesting. I don't get it. The RTA's filters should have constant fractional bandwidth (1/3 octave, etc.) and should measure flat on pink noise irrespective of log or linear scaling on either the amplitude or frequeny axes. On white noise they should show a rise with frequency of 3 dB/octave (or sqrt(2)/octave with a linear amplitude scale) as the amount of energy each filter swallows is proportional to its width in Hz, which is proportional to its center frequency. Or is Smaart just Stuupid in this instance?

--Frank
Logged
Yes, it is a giant stereo system!

Art Welter

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1187
Re: Smaart RTA 1dB/Octave HF Roll Off?
« Reply #13 on: October 16, 2017, 07:21:20 pm »

Interesting. I don't get it. The RTA's filters should have constant fractional bandwidth (1/3 octave, etc.) and should measure flat on pink noise irrespective of log or linear scaling on either the amplitude or frequeny axes. On white noise they should show a rise with frequency of 3 dB/octave (or sqrt(2)/octave with a linear amplitude scale) as the amount of energy each filter swallows is proportional to its width in Hz, which is proportional to its center frequency. Or is Smaart just Stuupid in this instance?

--Frank
It appears Smaart V6 running on OSX 10.6.8 is Stuupid in this instance. Whether this is the way it originally was written, or a OS incompatibility problem, I can't tell, not worth loading an older OS that won't work with today's interwebs back on the old MacBook.

Anybody with different versions of Smaart should be able to quickly tell if this glitch is history, or still happening.
Logged

Stelios Mac

  • Full Member
  • ***
  • Offline Offline
  • Posts: 132
Re: Smaart RTA 1dB/Octave HF Roll Off?
« Reply #14 on: October 17, 2017, 01:37:44 am »

but it won't run on my laptop at all...
Offtipic but, Smaart is really lightweight software (so is REW), they should be able to run flawlessly on any macbook, no matter it's OS.
Was the last OSX update you did a fresh install or an upgrade? If the latter, you really should do a fresh install, it'll perform like a brand new device. It just takes some time to backup & copy all of your data, plus reinstall all software. IMHO It's worth it though.
Logged

Art Welter

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1187
Re: Smaart RTA 1dB/Octave HF Roll Off?
« Reply #15 on: October 17, 2017, 11:33:59 am »

Offtipic but, Smaart is really lightweight software (so is REW), they should be able to run flawlessly on any macbook, no matter it's OS.
Was the last OSX update you did a fresh install or an upgrade? If the latter, you really should do a fresh install, it'll perform like a brand new device. It just takes some time to backup & copy all of your data, plus reinstall all software. IMHO It's worth it though.
The last "upgrade" was a fresh install on a replacement of the MacBook the Ex smashed. I purchased a refurbished unit specifically so I could run Smaart V6, and it does not run "flawlessly", although it did with the OS that was current when I bought it 10 years ago.
REW won't run on the OSX 10.6.8 in the MacBook, the last OS version possible for it's hardware .

Logged

Mark Wilkinson

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 556
Re: Smaart RTA 1dB/Octave HF Roll Off?
« Reply #16 on: October 17, 2017, 12:28:25 pm »

The last "upgrade" was a fresh install on a replacement of the MacBook the Ex smashed. I purchased a refurbished unit specifically so I could run Smaart V6, and it does not run "flawlessly", although it did with the OS that was current when I bought it 10 years ago.
REW won't run on the OSX 10.6.8 in the MacBook, the last OS version possible for it's hardware .

REW at least, has prior versions going way back http://www.roomeqwizard.com/changehistory.html

If you're still interested in getting to the bottom of what you saw in Smaart, I'd get whatever version of REW running,
 and start from ground up with a soundcard calibration.

The more I think about your original post, the clearer it seems that the reference pink is rolling off.
I kinda dismiss acoustic rolloff because the rolloff is too smooth, and besides, if it were acoustic I'd think transfer would show it as well.
I'm also a little doubtful the rolloff is due to measuring ...windowing or averaging etc,..... but I sure as hell defer to Chris or Frank here...

I left thinking something is rolling off the pink in your setup.....whether it's the soundcard, your adapters....anything that's in line with what you're using as your reference feed.
The REW swept-sine  soundcard calibration process could pinpoint the source of rolloff precisely, as you could calibrate the soundcard alone, then add pieces of the reference chain one by one.
Soundcard cal is a cool feature I wished Smaart had....
And hey, if there's no rolloff with REW all the way through the reference feed, you know Smaart's pink has stink !

Logged

Art Welter

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1187
Re: Smaart RTA 1dB/Octave HF Roll Off?
« Reply #17 on: October 17, 2017, 04:23:12 pm »

REW at least, has prior versions going way back http://www.roomeqwizard.com/changehistory.html

If you're still interested in getting to the bottom of what you saw in Smaart, I'd get whatever version of REW running,
 and start from ground up with a soundcard calibration.
Mark,

Guess you missed that my laptop is OS6.8, REW starts at OS7, and the problem in the OP was sorted, it's a Smaart RTA glitch, see post #11.

Cheers,
Art
Logged

brian maddox

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1571
Re: Smaart RTA 1dB/Octave HF Roll Off?
« Reply #18 on: October 17, 2017, 10:04:41 pm »

Ding Ding Ding-

We have a winner, Frank K!

The Smaart RTA function is flat when set to any of the portions of "octave".
The RTA function when set to "Log" or "Lin", rolls off the top end, on "Log" about 30 dB from 20Hz to 20kHz, or 4 dB from 30 Hz to 100 Hz as measured...
This does not occur on the Magnitude response screen, flat response looks flat  regardless of whatever you choose, although "Lin" is useless for anything below 200 Hz.

So the interesting by-line is when looking at harmonic distortion of sine waves on the RTA, the harmonics will appear to have less level (inaccurate) when looking at Log function compared to the octave functions. A 30 Hz fundamental with the third harmonic at -20 would have 10% distortion, but using the Log function would read about -24, only 6.31% distortion.

Smaart has made me a (sometimes) inadvertent "liar" when reporting speaker distortion ;^).
Well, at least I learned something new about a 10 year old software ::).

I'll be using REW for my future distortion testing, but it won't run on my laptop at all...

Cheers,
Art

I guess this goes back to that First Rule Of Troubleshooting. 

First, test the test device.

If i had a nickel for the number of times THAT one has bit me...
Logged
brian maddox
11brian.maddox11@gmail.com

'...do not trifle with affairs of dragons...

       ....for you are crunchy, and taste good with ketchup...'

Art Welter

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1187
Re: Smaart RTA 1dB/Octave HF Roll Off?
« Reply #19 on: October 18, 2017, 02:36:53 am »

If i had a nickel for the number of times THAT one has bit me...
We would both be so effing rich the queen would do whatever she wanted.
Test results, +/-positive...;^)
Logged
Pages: 1 2 [All]   Go Up
 


Page created in 0.035 seconds with 16 queries.