HOLMImpulse: Measuring Frequency & Impulse Response

askbojesen said:


Your soundcard is out of phase with itself. Check of the "Invert recorded signal" under device settings. At the impulse response it is very clear that the impulse is going down instead of up.


Dr.EM said:
Ah yes, that's much better! Thought the phase was a bit odd :eek:
Is there a way to compensate for interchannel latency? I could not find it in the various screen captures on the web site. This would be much needed if people are going to be using lower cost sound cards.
 
Windows 2000

Moondog55 said:
Just checked the download page, what a shame i am still using W2000 Pro; due to now wanting to make Mr gates any richer than I have too, not to mention multiple license issues not transferring to the upgrade

Hi, I just changed the text to: Windows (2000, XP & Vista)
There is no reason that it should not work on w2000, I did not know anybody was using that in 2009 ;)
 
Latency - No problem

soongsc said:
Is there a way to compensate for interchannel latency? I could not find it in the various screen captures on the web site. This would be much needed if people are going to be using lower cost sound cards.

Latency is no problem - I filter away the latency - I can deal with an arbitrary large latency, lets say up to e.g. 1000 ms. HOLMImpulse auto detects the received impulse-peak.
 
WM61A capsules

Salas said:
On what assembly have you measured the Panasonic capsules? Were they inset on the end of some tube? They normally gain 4dB up to 20kHz if there is a lip around their edge due to diffraction.

Attached picture shows the mounting in an alu-tube (Stuffed with damping materiel)

as I seen in

http://www.diyaudio.com/forums/showthread.php?postid=1841507#post1841507

I also get 4-5dB gain at 20kHz (The base level at the graphs is at -4dB) , but it drops again at 20kHz and upwards
 

Attachments

  • image0122.jpg
    image0122.jpg
    83.1 KB · Views: 2,070
Re: Latency - No problem

askbojesen said:


Latency is no problem - I filter away the latency - I can deal with an arbitrary large latency, lets say up to e.g. 1000 ms. HOLMImpulse auto detects the received impulse-peak.
I think the best way would be to just run a test to find out whether it's a problem or not. I have the right card to test this.
 
diyAudio Member
Joined 2007
Hi, I just changed the text to: Windows (2000, XP & Vista) There is no reason that it should not work on w2000, I did not know anybody was using that in 2009


Thank you,

Blame poverty; spending all of my money on drivers not operating systems, and it actually has no disadvantage except I can't play any of the new games, which saves me more money
 
diyAudio Chief Moderator
Joined 2002
Paid Member
Re: WM61A capsules

askbojesen said:


Attached picture shows the mounting in an alu-tube (Stuffed with damping materiel)

as I seen in

http://www.diyaudio.com/forums/showthread.php?postid=1841507#post1841507

I also get 4-5dB gain at 20kHz (The base level at the graphs is at -4dB) , but it drops again at 20kHz and upwards

Totally normal. I just wanted to compare some ballpark data for 20k to 40k, and your construction follows one or two of my mics.
If the capsule is made to protrude, the diffraction is roughly half, so its the rise, maybe better for high frequency impulse measurement. But if it drops on the floor, its history. See pic. On the left protruding, on the right inset.
 

Attachments

  • nosy.jpg
    nosy.jpg
    88 KB · Views: 2,069
Log-file

Ardee said:
I have installed the software but when I try to run it I get a message saying it has encountered a problem and needs to close. I'm using XPpro/SP3. Is there any diagnostic information I can post?

When the window has closed then
attach the following file:

C:\Documents and Settings\All Users\Application Data\HOLM Acoustics\HOLMImpulse\HOLMLogging.txt

(Or similar path for C:\Documents and Settings\All Users)
 
Thanks for the reply askbojesen,

When the window has closed then
attach the following file:

C:\Documents and Settings\All Users\Application Data\HOLM Acoustics\HOLMImpulse\HOLMLogging.txt

(Or similar path for C:\Documents and Settings\All Users)

There is no Holm entry in the Application Data folder and I have searched the c: drive for the file without result.
 
Administrator
Joined 2004
Paid Member
Re: Windows 2000

askbojesen said:
There is no reason that it should not work on w2000, I did not know anybody was using that in 2009 ;)

Not to worry Moondog - I just gave it up last year! Was sorry to see it go, but kept a copy. It's like XP but without the "Fisher Price" interface.

Askbojesen - thanks for posting this software. I'm most eager to try it.
 
askbojesen said:
Not sure that I have read the latest (I am a AES member, but not subscriber), so sent me a link to the article.
I have tested with digital loopback and electric loopback to ensure I have no artifacts. But I might have forgotten something - I have an open mind :)
I think Earl is referring to this one:
http://pcfarina.eng.unipr.it/Public/Papers/238-NordicSound2007.pdf
(Chapter 5)

And of course, big thanks for your software (I can only imagine the pains of windows GUI programming, I never made it beyond the commandline).

- Klaus (BTW, my primary working system still runs unter W2000.... never change a running system... and I have to use some programs that don't run under XP, let alone Vista. Some even require real-mode DOS!)
 
Farina revisited

Thank you for the links.

With ref to

http://pcfarina.eng.unipr.it/Public/Papers/238-NordicSound2007.pdf
http://www.aes.org/e-lib/browse.cfm?elib=14106

The concerns about the log sine sweep are:
  • pre-ringing at low frequency before the arrival of the direct sound pulse
    My problem here is not as big as in the article since I sweep from 10 Hz per default instead of starting at 20Hz
  • sensitivity to abrupt pulsive noises during the measurement
    No really solution to this problem! Do not slam the door when measuring
  • skewing of the measured impulse response when the playback and recording digital clocks were mismatched
    If the skewing is deterministic I solve this with a phase calibration (DAC-ADC calibration)
  • cancellation of the high frequencies in the late part of the tail when performing synchronous averaging
    I do not see this problem (Samplerate converters + Nyquist limit in the DAC-ADC chain is a much worse problem)
  • time-smearing of the impulse response when amplitude-based pre-equalization of the test signal was employed
    I make phase synchronized, and absolute gain independent average in frequency domain avoiding this effect

The ideas presented are:
  • post-filtering of the time-reversal-mirror inverse filter for avoiding pre-ringing
    Yes, HOLMImpulse does this
  • "exact" deconvolution by division in frequency domain with regularization
    Yes, HOLMImpulse does this
  • development of equalizing filters to be convolved with the test signal for pre- or post equalization.
    I do not see a reason to do this
  • counter-skewing of the measured impulse response
    when the playback and recording digital clocks aremismatched

    Yes, with a DAC-ADC calibration
  • employing running-time cross-correlation for performing proper synchronous averaging without cancellation effects
    I have a different solution mentioned above

A little demo: When we have digital clipping we have "nice" odd numbered harmonic distortions, that are seen before the impulse.
An externally hosted image should be here but it was not working when we last tested it.
 
It looks like you have all things adressed, then. And I agree, most of the potential gotchas are rather irrelevant for the way you handle things.

Your SW is running in the background while I'm typing this (no problems so far with my P4-system under XP, W2K not checked yet) and I'm trying to get familiar with it, and then will report begging for additional option/features ;)

One thing found so far: the plots don't use the full area of the pane for some reason, although they scale with window size. Probably because of my 125% scaling of fonts -- many programs have problems with this, because noone seems to use this and thus it is seldom checked for.


Anyway, big cheers for giving us this tool FOR FREE! Very much appreciated! I'll try to raise interest in the german forums.

- Klaus
 
Re: New Releases (Binary code changes)

askbojesen said:
No new features/bugfixes except:
  • The SSE2 is now disabled per default (Not all computers have SSE2)
    (This might help the users: BDP and Ardee)
  • A specific version with SSE2-support
  • 64 bit support
See here

Mr Bojesen

That you very much, your cooperation and support are very appreciated and this IS the piece of software that I have been waiting for!!

Much appreciated!