That rather sounds something is clipping or the microphone can't do such low frequencies.
At those frequencies also other things like traffic, walking around etc can sometimes be an issue.
At those frequencies also other things like traffic, walking around etc can sometimes be an issue.
No, nothing is clipping and the microphone is calibrated down to 10 Hz.
As I said to my previous post, I know that this is due to the very low frequencies, because if I put a passive high pass filter at 20 Hz at the microphone amplifier output, all the readings are exactly the same.
I have also monitor the output of the microphone amplifier with an oscilloscope and I saw very very low frequency signals (in the range of 1 to 2 Hz), probably due to electronic noises and not acoustical noises.
And I was thinking for a good solution for this in the software. Maybe a High Pass filter at 3 to 5 Hz?
As I said to my previous post, I know that this is due to the very low frequencies, because if I put a passive high pass filter at 20 Hz at the microphone amplifier output, all the readings are exactly the same.
I have also monitor the output of the microphone amplifier with an oscilloscope and I saw very very low frequency signals (in the range of 1 to 2 Hz), probably due to electronic noises and not acoustical noises.
And I was thinking for a good solution for this in the software. Maybe a High Pass filter at 3 to 5 Hz?
Member
Joined 2003
It really appears that you're fussing over noise floor stuff here that's looks to be far below the gate 1/1 oct resolution. The error you see in the frequency range you care about is a fraction of a dB. Take repeated ungated measurement and overlay them to see the reality of the situation. Multiple averages and louder signal will help with SNR. If you really want to fix the slope of the response down to 0Hz, you can post-process the measurement with Hilbert-Bode Transform.
Already being said, but a sine wave always gives a much higher SNR, NOT periodic noise.
There is also still the option to average sine wave sweeps as well.
Btw 20 averages doesn't make much sense.
Noise reduction goes with a 10log. So 3dB maximum can be expected with an average of 2. 6dB with 4 and so on.
So it's either having 16 averages or 32 averages. (unless you care about 1.5dB more)
There is also still the option to average sine wave sweeps as well.
Btw 20 averages doesn't make much sense.
Noise reduction goes with a 10log. So 3dB maximum can be expected with an average of 2. 6dB with 4 and so on.
So it's either having 16 averages or 32 averages. (unless you care about 1.5dB more)
ARTA 1.9.5 just released.
https://www.artalabs.hr/news.htm
https://www.artalabs.hr/download.htm
- Multitone generation algorithm improved.
- THD+N frequency weighting switch updated.
- Improved estimation of Directivity Index. Now user chooses the type of spatial directivity pattern used for DI estimation: cylindrical or axe-symetric spherical.
- Spherical pattern is used as defined in standard IEC 60268-21
- Directivity index can be estimated for free full space and half-space radiation.
https://www.artalabs.hr/news.htm
https://www.artalabs.hr/download.htm
Last edited:
Thanks Heinrich. Wishing you, your loved ones, and all of your faithful followers here, a wonderful Holiday Season full of love, laughter, health, joy, merriment, and fond memories to be made. Jay
In LIMP (W7, versions 1.9.4 and 1.9.5), impedance curve shows a strange fall around 12 kHz. Downgrading to version 1.8, the artifact disappears.
Tried with two different soundcards (M-audio Mobilepre and UCA 202) and changing every measurement paramenter, without success. I do not know if new method for calibration starting in 1.9.4 may be involved. Any suggestion?
Tried with two different soundcards (M-audio Mobilepre and UCA 202) and changing every measurement paramenter, without success. I do not know if new method for calibration starting in 1.9.4 may be involved. Any suggestion?
Which OS do you use? I am fiddling with linux audio where sometimes strange things happen.
What is your sample rate?
What is your sample rate?
Ok, a different building site. Anyway there is a chance that the problem is buried in your system, i.e. outside ARTA.
Member
Joined 2003
Problem appears sample rate related. 48/4 = 12. I recommend using asio driver to avoid sample rate problems, with WDM driver you have to make sure windows sound setting matches configuration in ARTA.
I thought the same but the artifact remains the same both with ASIO and WDM.
Interestingly, the impedance curve is perfect when downgrading to ARTA 1.8.
Thanks for your inputs.
I will explain what happens in new LIMP.
Three figures will help for explanation.
1) Impedamce measured with EMU Tracker pre without calibration. This curve is good reference one
2) Impedamce measured with UCA 222 pre without calibration. This curve shows significant errors in all frequency range
3) Impedamce measured with UCA pre with calibration, Again we have error. The error can be from various reasons (nonlinearity, noise, bad antialiasing filter, low current capacity of output op amps).
In version 1.8 we allowed that calibration help slightly but in most cases results are not reliable. Now we emphasize error to warn users not to us this kind of interface (as you may notice I do not recommend Behringer interfaces).
Ivo
Thanks, third picture is exactly what happens in my system. Not only with UCA, but also with M-Audio Mobilepre.
Tried Presonus Firebox and problem solved.
Tried Presonus Firebox and problem solved.
Anyone here tried the Steinberg Ur12? I'm trying to find a good interface for my system and was planing to go with the Focusrite Scarlett Solo 3, but the Steinberg is nearly half the price and $ is tight at the moment.
Member
Joined 2003
For 2 channel measurement, go with UR22. The second input on the UR12 is high Z only which may be problematic.
For cheap interface, Behringer UMC202 is about as cheap as I’d recommend. Stay away from UCA.
Check your local classifieds, often you can find a decent lightly used interface for low cost.
For cheap interface, Behringer UMC202 is about as cheap as I’d recommend. Stay away from UCA.
Check your local classifieds, often you can find a decent lightly used interface for low cost.
I currently live in Brazil and the import taxes here are crazy, making the prices unaffordable sometimes... Usually the import taxes are around 100%, but last week I was looking at the price for something on amazon and the import price along with the shipping was just a hair under 130% of the price of the product.
The UR22 cost here is about 15% more than the Scarlett Solo 3, and the Behringer UMC202 is about the same price.
I already have an ECM8000 mic that I would like to use to take measurements in REW, OpenSpundMeter or Smaart on my old imac.
Member
Joined 2003
I wouldn't use Scarlet Solo either, for the same reason as UR12. Go for Scarlet 2i2 if you want a Focusrite.
You may be able to get something functional from Solo or UR12, it may require some different resistor values to build the jig, since one input impedance will be 20-60kohm, while the other will be 1.5Mohm. Better to have 2 matched input channels for a measurement system that is constantly comparing one input channel to the other.
You may be able to get something functional from Solo or UR12, it may require some different resistor values to build the jig, since one input impedance will be 20-60kohm, while the other will be 1.5Mohm. Better to have 2 matched input channels for a measurement system that is constantly comparing one input channel to the other.
Last edited: