Interesting David! Having only thought about this issue for a few minutes, I can't think of any good reason why there should be any DC offset for a reasonable design. Have you figured out where this is coming from? It looks, at a glance to be constant across the time interval, is that what you're seeing?
I doubt that it is the supply voltage to the mic, charging some cap with a very long time constant, or could it be? Seems odd since you obtained measurements with and without offset with the same hardware/software. Perhaps one time you took the measurement shortly after power up and some large DC blocking cap had not charged, the other time it had been on for a while with things stabilized.
It would be reasonable IMO, for the software to remove any DC offset if we knew for sure that it was constant across the measurement interval.
I have noted on several occasions that some people using MLS measurement systems seem to be measuring an unexpected rise in the low end and this might be the reason.
Last edited:
Hi Pete,
It is a long term DC offset. I had been successfully measuring for a number of days and then more recently the measurements contain significant DC and have the abberant bass response. I don't think it is capacitor charging because it is still there after, say, an hour of the computer being on. It is possible that the cap is leaking DC from the mic polarizing voltage, but I'm a little surprised that it was good previously and now presents a problem.
My impression is that DACs have some inherent DC offset and can drift. I can't imagine that the cheap soundcard in a PC is especially good in that regard. As I recall at KEF in the early 80's, with professional HP gear, we worried enough about the issue to do the alternating impulse correction.
I wish there was a way to null it out.
David
It is a long term DC offset. I had been successfully measuring for a number of days and then more recently the measurements contain significant DC and have the abberant bass response. I don't think it is capacitor charging because it is still there after, say, an hour of the computer being on. It is possible that the cap is leaking DC from the mic polarizing voltage, but I'm a little surprised that it was good previously and now presents a problem.
My impression is that DACs have some inherent DC offset and can drift. I can't imagine that the cheap soundcard in a PC is especially good in that regard. As I recall at KEF in the early 80's, with professional HP gear, we worried enough about the issue to do the alternating impulse correction.
I wish there was a way to null it out.
David
I suppose that it would depend on the type of converter. It seems that there is plenty of time, say from -20 to -10 ms to take many samples, average them, and subtract it. Does it look to you though, as if the error shifts to a smaller amount when the pulse is present? It is hard to tell from the small plot.
Hi Speaker Dave,
If you can post the impulse response file (in ASCII text), I think I could easily get rid of the DC offset for you and send it back for you to compare it with the one you have that does not have a DC offset. That would at least tell you if the problem at low frequencies is due to DC offset.
Regards
Peter
If you can post the impulse response file (in ASCII text), I think I could easily get rid of the DC offset for you and send it back for you to compare it with the one you have that does not have a DC offset. That would at least tell you if the problem at low frequencies is due to DC offset.
Regards
Peter
I'm pretty sure (actually very sure) that the problem at LF is DC offset. You can easily see the offset in the impulse below the response curve. The curves are of the same speaker on two different days.
As (other) Dave surmises the windowing creates the character of the LF rise. With an infinite time window the presence of DC would give a non-zero DC value on the curve. Windowing simulates an impulse resonse with a relatively long non-zero value. This is what brings the error up to audio frequencies. Remember that the FFT of a square pulse is a sinx/x response with energy down to DC... that is what we are seeing on top of the normal frequency response of the system (also why the periodic nulls at low frequencies).
The question is whether there could be an automatic software means of removing it. As I mentioned, at KEF we used alternating pulses and let the software subtract positive pulses from negative pulses. Subtracting the pulses would double their strength. Since the DC offset was constant (relativley constant over the time duration of the measurements) its subtraction would cause total cancelation.
Does the software developer still monitor this thread?
Davdi S.
For a person with professional interests he has been uncharacteristically generous thus far. I enjoyed using his software, even though I have SoundEasy (not as user-friendly). Their commercial product seems to have progressed way, way beyond what HOLMimpulse offers DSPreLab.
Check the videos. Amazing. But not cheap.
askbojesen has not posted since Jan 21st, I hope all is okay.
HOLM has made measuring very easy for me but I would love to request some enhancements.
We need to be able to do off axis responses allowing overlays. It seems I can only do 3 different measurements.
ARTA can do as many as I want, saving the last one and overlaying the next one. Its just as easy now.
HOLM has made measuring very easy for me but I would love to request some enhancements.
We need to be able to do off axis responses allowing overlays. It seems I can only do 3 different measurements.
ARTA can do as many as I want, saving the last one and overlaying the next one. Its just as easy now.
Hi,
My one request ( necessary, before I eventually post the rest ) is that this program stop being free & become a commercial offering .
At the very least , it should become shareware ( with ongoing development tied to contributions; be they, $$$ or intellectual ) . Think ARTA .
Without money trading hands / this program is bound to suffer neglet, & then whither & die ( when it won't work due to a release of some new operating system )
Really, we don't even know if AskB still has a job with Holm Acoustics ( or if he's employed anywhere for that matter ) . Think Recession !
I don't mean to offend anyone here / but under the circumstances how can people ask for enhancements or even advice ( not knowing his, &/or the companies financial situation ) ?
<> cheers EarlK
My one request ( necessary, before I eventually post the rest ) is that this program stop being free & become a commercial offering .
At the very least , it should become shareware ( with ongoing development tied to contributions; be they, $$$ or intellectual ) . Think ARTA .
Without money trading hands / this program is bound to suffer neglet, & then whither & die ( when it won't work due to a release of some new operating system )
Really, we don't even know if AskB still has a job with Holm Acoustics ( or if he's employed anywhere for that matter ) . Think Recession !
I don't mean to offend anyone here / but under the circumstances how can people ask for enhancements or even advice ( not knowing his, &/or the companies financial situation ) ?
<> cheers EarlK
Well Earl K, you may be correct in your speculation. However, let me speculate.
Why would Ask be interested in monitoring this thread when there is so much off-topic chest beating and ******* contests. Personally, I would find that very disrespectful for someone who is doing us a favor - a very big favor.
Ask is a rare individual. Let us hope he has not been turned off by some of the behavior around here.
Why would Ask be interested in monitoring this thread when there is so much off-topic chest beating and ******* contests. Personally, I would find that very disrespectful for someone who is doing us a favor - a very big favor.
Ask is a rare individual. Let us hope he has not been turned off by some of the behavior around here.
You know, I don't think that this thread has been too bad and I don;t get the impression that Ask was all that offended by anyone. A lot of us praised him. No, I'm inclined to think that he is gone for some reason. But I also agree that the software needed to stop being free. Free is a sure way to kill it.
Hi Guys,
First, I don't know Ask and have never met him in person. But I know who he is. However, judging from a recent DIY microphone event in Denmark, hosted by Holm Acoustics, it appears that he is fine. He properly just tight up in other matters. I have linked to some pictures, in a danish forum, from that event. As far as I know, Ask is the guy with the ponytail.
Nerds.DK
Nerds.DK
I know the guy who organized this and he told me it was a great day.
Best regards,
mkc
First, I don't know Ask and have never met him in person. But I know who he is. However, judging from a recent DIY microphone event in Denmark, hosted by Holm Acoustics, it appears that he is fine. He properly just tight up in other matters. I have linked to some pictures, in a danish forum, from that event. As far as I know, Ask is the guy with the ponytail.
Nerds.DK
Nerds.DK
I know the guy who organized this and he told me it was a great day.
Best regards,
mkc
Its already to a development point that it just works so the current version will always be free and very useful.
ARTA isnt really shareware. Its Demo version is 100% free. Its another great measurement package. I just wish the setup (sound card calibration was a lot easier). Other then that ARTA is ahead of HOLM is many other aspects.
I have no problem with HOLM professional version become available if...
- If it has CSD, Wavelet type graphs
- If it allows for n Measurements to be overlayed on one screen
- If it allow us to EXPANDED the measurement windows to a full desktop screen, HOLM is so damn tiny on my big monitors
If they added a crossover sim package then it would match something like SoundEasy in functionality.
It resizes to whatever I want on mine. And it's easy to zoom in on a portion. I rarely run it full screen, don't need ot.