HOLMImpulse: Measurements in practice

The intention of this thread is to get practical feedback for my application: HOLMImpulse

HOLMImpulse is Non-Commercial freeware with full functionality (No limitations or trial periods) running on Windows XP/Vista

Download HOLMImpulse Here


Other diyaudio thread for theoretical discussions
I started a thread, which went rather theoretical:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=144984
If you have theoretical questions/features/bugs/doubts then please refer to that thread.

The main goals of the application
  • Measure Impulse-response
  • Measure Frequency-response (Gain and Phase)
  • Easy and fast to use (Simply select your soundcard and click measure)
  • Work with any soundcard
  • Use the latest measurements techniques without complicating the interface
I would like your response
  • I know that there is no manual yet, I'm working on it...
  • Please post your measurements in this thread, it would make me very happy to some measurements and it would help me improve the tool.
Why I'm doing this?
When I started building my DIY-speakers I could not find any freeware satisfying my needs.
I believe that whatever you can measure you must measure instead of fabulating...
  • Does the grill make any reflections?
  • Is the relative phase correct?
  • Is the relative gain correct?
  • Is the tweeter broken?
  • How does my carpet or sofa table affect my impulse response?
  • ...
Above questions can be answered by HOLMImpulse even without a calibrated microphone, since you are often only interested relative measurements (Like above)
I have attached a measurement of my girlfriend 's speakers (Rega EL8), where the manufacture had the mid/woofer out of phase in the one speaker causing the dip at 4 kHz
(It's of course resolved now...)
 

Attachments

  • screenshot004.png
    screenshot004.png
    46.7 KB · Views: 7,952
Hello!

An externally hosted image should be here but it was not working when we last tested it.


I also use ARTA and the measurements look the same.
The handling of holmimpulse ist nice. I like it more than ARTA.
Realtime gating is cool.

acoustic phase is right with the crossovers....

To measure THD in my room is a problem, because of vibrating wood panels or dvb-s receiver....

But I think you have made your main goals!

Litte things like there is a big clear (grey) area right from the Graphs are not "perfect". But the function is the important thing.

best regards
Richard
 
Time window / Gating

Thank you for the feedback :up:

schauki said:
An externally hosted image should be here but it was not working when we last tested it.

I also use ARTA and the measurements look the same.
The handling of holmimpulse ist nice. I like it more than ARTA.
Realtime gating is cool.
Yes, the Realtime gating is learning you what reflections do to the response. Why don't you use a time window / gating on your measurement then?
acoustic phase is right with the crossovers....
What do you mean by this... ?
Litte things like there is a big clear (grey) area right from the Graphs are not "perfect". But the function is the important thing.
I cannot find this big clear (grey) area - please post a screenshot

My attached an image, which shows the same impulse-response, but with a different gating. (Notice the floor reflection with 100 cm extra path)

A new version 1.1.5.2 today makes it easier to save nice png-image-files: File > "Save Frequency- & Impulse-Response
http://www.holmacoustics.com/downloads/HOLMImpulse/ChangeLog.txt
 

Attachments

  • holmimpulse.png
    holmimpulse.png
    32 KB · Views: 2,335
Re: Time window / Gating

askbojesen said:
Yes, the Realtime gating is learning you what reflections do to the response. Why don't you use a time window / gating on your measurement then?
To see the LF too.

An nice feature ARTA has is that the area wich is not a valid result marked yellow.
Measurement "dummys" will be happy if they put a short gate an have a flat amplitude...

What do you mean by this... ?
If you check the "phase" box, you see the "phase".
And its right. ARTA doesn't do it so good.
I have Filters 24dB/oktav LR an they will shown as 360° phase drift in the plot.
ARTA makes many phase shifts, perhaps its a handling fault by myself.
But HOML makes it right.

I cannot find this big clear (grey) area - please post a screenshot
Will post it soon.

A new version 1.1.5.2 today makes it easier to save nice png-image-files: File > "Save Frequency- & Impulse-Response
http://www.holmacoustics.com/downloads/HOLMImpulse/ChangeLog.txt [/B]
Download it soon.
The png export is nice but I ever do Screenshots and put it into "Gimp", and load it on my webspace.
So I can make it in the right size for forum (800x600 is good) and direct link it, and all users can see it well. Not everybody has an high resolution Screen.

best regards

P.S. sorry for my bad (horrible) english... :eek:
 
Hey Schauki, nice to see you here!

Don't forget to check(bookmark out the main HOLMI thread, too.

(@Ask, you see my "campaign" is yielding the first fruits ;))

>> I cannot find this big clear (grey) area - please post a screenshot
I had mentioned this also (in the other thread), seems to be related to screen resolution and/or a user specified font size...
(see attachment)

- Klaus
 

Attachments

  • holmi-screen.gif
    holmi-screen.gif
    91.8 KB · Views: 5,225
Hello KSTR! - Servus Klaus.

You have the screen with the grey area - good, less "work" for me.

Your capaign is a little too late in the german forum - for me, because I'm reading here too. Bu I don't write much because of my english.
And I've installed the programm 5 min later I've read the thread.

My german is not much better than english... (remeber "magst du deine Fragen mal auf *deutsch* wiederholen" - "do you want to repeat your questions in german) I wrote it in german... austrian german...

Perhaps I should write danish??

best regards
and mfg

Richard
 
>> Perhaps I should write danish??

Haha, Smørrebrød anyone?


>> Bu I don't write much because of my english.

The more you read and write on this forum the better your english will get, believe me.

And anyway, form follows function... this applies to conversations in foreign languages, too.

cu, Klaus
 
Time window

panomaniac said:
So what about the time window? Using automatic it usually isn't right for me, so I slide it until the frequency response window looks "right."
Any tips about using the time window?
  1. Include/exclude the reflections. The problem is that the first reflection is often from the floor apprx. 100 cm extra path. 100cm corresponds to a wave length at 344 Hz, so that You cannot really trust the frequency response below 344 Hz, since no whole wave have yet played.
  2. Measure in a Anechoic chamber (Or outside in the garden...)
    [/list=1]

    The Autodetect feature in HOLMImpulse (Applied after each measurement) tries to detect the first reflection and cut this of with a time window. (Autodetect can be reapplied form the options window)

    Tips? Post your measurement here perhaps with autodetected window and with our own selected window.
 
Administrator
Joined 2004
Paid Member
Well with the latest version (v1.1.5.2) it seems to be better. Also - oops! - I found that the time window was "locked." Unlocking it in the measuremant window options helped. It tends to pick about 100Hz as the low point, instead of ~300Hz as before.

But I think that recalibrating my sound card would be a good idea. Interested to know what others have done with calibration.

My current setup is:
USB>M-Audio Duo>Spdif>DCX2496>amp>speaker

I calibrated the M-Audio card with line out patched back to line in. Very flat. But I think I need to go thru the DCX (all filters off- delays on) and into the mic input of the M-Audio (padded) to get a really good cal. Or do delays make any difference?

What do you think?
 
Bug - Fixed - Misc DPI - Grey areas

Hi Schauki an Klaus
KSTR said:
I had mentioned this also (in the other thread), seems to be related to screen resolution and/or a user specified font size...
attachment.php

(see attachment)

I tried a 120 dpi screen resolution and got the same problem as you. I beleive it is fixed now:

http://www.holmacoustics.com/downloads/HOLMImpulse/ChangeLog.txt

Version 1.1.5.4 (2009-06-13)
Bugfixes:
* DPI resolution other than 96 in Windows setting
* Improved scaling when using large fonts in Windows setting

Have a nice weekend!
I won't be here more this weekend
 
DCX2496 LP/HP/BP

panomaniac said:
My current setup is:
USB>M-Audio Duo>Spdif>DCX2496>amp>speaker
I calibrated the M-Audio card with line out patched back to line in. Very flat. But I think I need to go thru the DCX (all filters off- delays on) and into the mic input of the M-Audio (padded) to get a really good cal. Or do delays make any difference?
What do you think?

It could be very interesting to see the response of the outputs of the DCX2496 with your x-overs:

USB>M-Audio>Spdif>DCX2496->Analog->M-Audio

I mean you can direct verify the LP/HP/BP of the DCX2496 :)
It is interesting to see the phase/impulse of the LP/HP/BP.
(Then you know what phase is from the DCX2496 and what is from your speaker)

If you want to calibrate (also for speaker mesurements) then I believe you should set DCX2496 in allpass mode and measure.

Waiting in excitement for you x-over response images :yummy:
 
Administrator
Joined 2004
Paid Member
I had a quick try at this - but got wierd results. LP filter was OK, HP very strange.

How should I post here? Just graphs - or can I export something and put it here for you?

Also, if I do a loop back calibration of the soundcard, then run a measurement of the sound card, shouldn't the measurement show flat? Shouldn't the calibration file cancel out any FR or phase errors? I don't see this happening in my measurements.
 
panomaniac said:
I had a quick try at this - but got wierd results. LP filter was OK, HP very strange.
How should I post here? Just graphs - or can I export something and put it here for you?

I think frequency- & Impulse-response graphs: File > "Save Frequency- & Impulse-Response
You have zoomed so we can see the impulse-response (The time window is not important when not making acoustic measurements, except, that it should be long enough)
Plot both gain and phase (not THD)

Remember then to give the measurement slots some names before saving the image. Legends will be written to the file.png when saved.

If something is really weird, then you can sent the measurement file (file.hlm) to support@holmacoustics.com, so I can load the file myself.
 
Administrator
Joined 2004
Paid Member
Here are 3 measurements for you.
Signal path = USB>M-Audio Duo>Spdif>DCX-M-Audio>USB

Calibrated on analog loop, but digital thru the DCX looks almost identical. tiny phase shift in the low end between analog and digital.

Here are some sample filters from a stock DCX2496. All from same channel (2). All at 1KHz.

First order below.
 

Attachments

  • dcx-butt6-1k.png
    dcx-butt6-1k.png
    33.5 KB · Views: 2,381
Impulse peak Time = 0 detect (bug)

Thanks for the measurements :up:

I see that Impulse peak time = 0 detect does not work for these IIR filters (I mostly tested on raw drivers or linear phase FIR)

From Measurement options you can adjust the time offset manually.

Two changes I'll implement to improve above measurements:

  1. I'll change the time = 0 strategy to a simple peak detect.
  2. It would also be appropriate to disable gain normalization
    (This is already a feature request)
    [/list=1]

    I'll be back :D