REW first scan - mystery

[Sokel]

I highly doubt that.

View attachment 437640

All the graphs are labelled "L+R" indicating you swept both speakers together.

The two graphs at the bottom (blue and orange) show comb filtering of the upper frequencies. This suggests that the polarity was inverted in one of the speakers. A faulty microphone does not produce comb filtering. If you look very carefully, you can see that in the green graph, the tweeter dropped in level. Interestingly, the comb filtering appears at the same frequencies upwards. This makes me think that you disconnected the cables to the tweeter, took the green measurement, and reconnected them with the wrong polarity. Then you took the two measurements at the bottom.

These are not the same measurements, even if you staggered the level. Read the REW Help thread, and make sure you are taking measurements properly. Then take another set of measurements and post them.

These can't show different levels, we're talking 250dB difference here, these must just be separated with REW's function for easy view.
What's odd at these are not the obvious high freq stuff but the low frq ones that should be immune and consistent if that's all L+R .
We see 30dB dips at red and blue disappearing at green and orange.

Strange stuff.

 

[Keith_W]

These can't show different levels, we're talking 250dB difference here, these must just be separated with REW's function for easy view.
What's odd at these are not the obvious high freq stuff but the low frq ones that should be immune and consistent if that's all L+R .
We see 30dB dips at red and blue disappearing at green and orange.

Strange stuff.

I said that he deliberately staggered the level. Regardless, I have a hard time believing that the measurements were all taken at the same time with no changes in between. As I pointed out - faulty mics don't produce comb filtering.

 

[kemmler3D]

I said that he deliberately staggered the level. Regardless, I have a hard time believing that the measurements were all taken at the same time with no changes in between. As I pointed out - faulty mics don't produce comb filtering.

Since they suspect the mic is malfunctioning, if you assume they actually took the same sweep 4 times, it looks to me like the output of the mic is varying randomly every second or two... Did they disconnect the tweeter, or did it just randomly drop the gain for a second?

Not sure as to potential cause, but if OP really did just take 4 similar measurements and didn't make any changes, what else would explain it?

 

[Salt]

Purely optical the FR looks like filter cutoff of higher frequencies as DACs do above 20 kHz, but here at 5 and 8 kHz.
Only an optical aspect.

 

[physicsplace]

Perhaps a problem with the calibration of the microphone or an acoustic problem in the room. Have you tried measuring at 0º to compare? Checking with a test signal and a multimeter can help, but before that, I would do a scan again by changing the orientation and the location of the microphone. Curious to see your results! ️

Mini Militia App Lock​

 

[Scrivs]

The response posted:
View attachment 436736

Looks oddly similar to that posted in this discussion in the REW support forum on AV Nirvana:

View attachment 436738

Which, as I posted earlier, was caused by the DAC muting the output when one channel has no signal. Output to both L+R simultaneously and disconnect the DAC output for that channel that you are not measuring.

Have you tried what Sam Adams suggested? (Check your REW preferences brosef).

 

[SashaR]

I am EXTREMELY embarrassed by my erroneous and vaguely worded statement that ALL sweeps were made in the SAME circumstances! This statement only concerned graphs with recording the level of silence in the room - which, in my opinion, demonstrate a malfunction of the microphone (REW constantly told me when checking the level, which I had to do before starting each swipe, that the level was not sufficient - and the drawing with colored sweeps demonstrates exactly those cases when REW did not have time to perform 5-second swipe without microphone glitches).

During all the measurements, no changes were made to the equipment or wires - except for changing the sound level using the DAC. The microphone was in a horizontal position. The mic and the DAC was declared as EXCLusive.
About the possible reason for the "filtering" of the high frequencies in the 4th figure (blue and orange): I made these scans trying to filter out the distortion created by the room, for which I moved the microphone very close to the speakers (keeping the measured SPL by adjusting the DAC). The attached scans were taken with a microphone that I have already returned to Amazon. The scan named "-15 dB" was captured at normal listening spot, the scan named "near field" was captured when the mic was on my work chair point (very close to the speakers).

@Sokel @Keith_W @kemmler3D @Salt @physicsplace @Scrivs

Thank you all very much!!!

 

[SashaR]

Since they suspect the mic is malfunctioning, if you assume they actually took the same sweep 4 times, it looks to me like the output of the mic is varying randomly every second or two... Did they disconnect the tweeter, or did it just randomly drop the gain for a second?

Not sure as to potential cause, but if OP really did just take 4 similar measurements and didn't make any changes, what else would explain it?

See, please, the SPL logs here - they were taken in the absolute silence of late night.

 

[Keith_W]

Please give your graphs proper names so that we know what we are looking at. You say that the earlier measurements which showed comb filtering was because the mic was placed extremely nearfield to the speaker. I can believe that this will produce comb filtering between drivers. So why the heck did you name it "L+R"?!?!?!?!? Why would you place the mic close to one speaker and then sweep them both together??!?!?

Anyway let's look at your new curves.

All these curves show that your UMIK-1 is working as expected. The only difference is SPL and the shape of the curve, probably because you were measuring from a different position. In fact, I KNOW that you were measuring from a different position because the pattern of reflections in the energy-time curves is different between all of them.

I can tell that you have a subwoofer, even though you did not provide that information. How do I know? Because these two curves (both labelled "L+R Mar 13" from file "D90SE -15dB 48kHz") have the exact same frequency response:

But the step responses are different:

You can see that the subwoofer is delayed in the green curve and the speaker can be clearly seen as a separate impulse.

These 4 graphs are all different:

All of them deviate from the measurements as shown before. So, if your mic is faulty, these are the curves that demonstrate deviations from all the other measurements. Did the sweep sound the same with all the impulses? The only curve that suggests a microphone glitch to me is the blue curve. All the others look as if you are sweeping part of the frequency response only, for example green probably has the subwoofer turned off? But let us look a bit closer at the pink and red curves to show you what I mean.

This is the pink curve ("L+R Mar 13" from "-25dB near field AND bad mic") compared to the blue curve from the same file "-25dB near field AND bad mic"). You can see that the two measurements are exactly the same up to 5kHz, when the pink curve drops in level.

This is the red curve from the same file ("-25dB near field AND bad mic"). You can see that it matches the blue curve exactly except for below 3kHz.

If you sum the purple curves and the red curve together (result = green curve), it replicates the blue curve exactly, except for a dip between 3kHz and 4kHz. The dip can be explained from a difference in phase because of different mic positions between the measurements (which I can easily see on the Energy-Time Curve). This looks as if you are measuring the high-pass filter and low-pass filter of the speaker independently.

Anyway I am finding this exercise quite frustrating because of the poorly labelled graphs, the missing information, the fact that all the measurements were taken from different positions, and so on.

If you suspect your microphone is faulty, please do this: take 20 measurements from the exact same position, with your mic on a tripod. DO NOT make any changes in position, REW configuration, volume level, or anything else. Do not touch your mic. NO CHANGES between measurements. Set it up, take 20 measurements, and listen to the sweep every time to make sure each sweep sounds the same. If you do not see any microphone glitches, take another 20 measurements. Keep repeating and discarding measurements until you see a microphone glitch. Then make an .MDAT, keeping 5 "normal" measurements with all the glitched measurements, and upload that to ASR.

 

[SashaR]

... until you see a microphone glitch. Then make an .MDAT, keeping 5 "normal" measurements with all the glitched measurements, and upload that to ASR.

Thank you so much for the careful analysis of the measurements I sent!

In order to believe that the microphone is malfunctioning, it is enough to analyze the SPL Logs that I made for myself after I realized that the constant repetitions of corrupted scans are caused by the instability of the microphone's SPL measurement - which causes a decrease in the level in certain areas of the scans - and registering the SPL level is (in my opinion) the most obvious way to detect the instability of the microphone works in registering the SPL level - see them here, please!

Please forgive me for not specifying the equipment I used! EACH series of scans was performed under circumstances that were UNCHANGED during the execution of a particular group of scans. The names of the scans I sent contain everything that describes the circumstances under which the scans were performed. All scans were performed with a microphone attached to the end of a long horizontal and perfectly straight wooden stick.
Each individual scan was performed in a room with no one in it. If the name of a series of scans contains "L+R" it means that the microphone is directed horizontally and is in the vertical plane, which is in the MIDDLE BETWEEN the speakers VELLA 407 with the declared diapason from 36 Hz to ... [today these speakers declared as from 30 Hz to 50 kHz]. Without any need and just in case, I repeat: I did not make any changes to the equipment, except for changing the level using the DAC (TOPPING D90SE). The amp used is TOPPING LA90.

Quite by chance, the measurements I have already sent you fully comply with your instructions to send measurements that were taken under exactly the same circumstances and contain both successful and erroneous scans. But because it is impossible to send more than five scans in one set, I have thrown out some of the scans. So I will send them now and I am very grateful to you for agreeing to analyze them too.
Note: Scans labeled "-15 dB" are taken from the generally accepted listening position - the point where the listening chair is usually located. Scans labeled "near field" are taken from a point that is very close to the speakers, where my work chair is located. As a rule, now, without room correction, it is at this point that the stereo effect is better heard ... according to my subjective impression. I will be very interested to know whether this subjective impression is confirmed by the analysis of scans performed by specialists and experts?!? Thank you all for your help with all this issue and with the room correction challenge!

 

[Keith_W]

If these are your new measurements, how come they overlay exactly on top of the old measurements, like this?

I checked the shape of the noise floor to see if they are the same measurement (waterfall graph). They are exactly the same.

On the left is your "old" measurement, and on the right is the "new" one. Instead of taking new measurements, you simply re-uploaded your old measurements. There is nothing new to see here.

Why did I bother to download exactly the same measurements that you posted before (5 different files), open them all up, apply smoothing to 22 separate graphs, and then find out they are exactly the same as your old measurements?!?!?!? STOP WASTING OUR TIME!!!!

I am going to start yelling at you because I am so annoyed and I need to get through your thick head.

1. LISTEN TO THE SWEEP TO MAKE SURE IT IS A MICROPHONE PROBLEM, AND NOT A DAC OUTPUT PROBLEM.
2. TAKE NEW MEASUREMENTS, STOP RE-UPLOADING OLD MEASUREMENTS WHICH WE HAVE ALREADY LOOKED AT. AGAIN: 20 MEASUREMENTS, AND DISCARD ALL THE NON-FAULTY ONES. KEEP REPEATING UNTIL YOU SEE A DEFINITE MICROPHONE GLITCH.

And with that, I am out. I'm not going to look at any more of your measurements. Look at it yourself.

 

[SashaR]

I'm not going to look at any more of your measurements.

I am very sorry that I failed to write clearly that 1) I am uploading exactly what you asked me to, and that 2) I did everything exactly as you told me to do. In addition to the additions to the scans sent earlier, I also sent the scans that were made at a different level - and they also have the same strange changes in high frequencies. I am very sorry that I did not write for the fourth time that I have already returned the faulty microphone.

It is a pity that you did not read my advice to use the logs with the results of recording the SPL of absolute silence - and therefore these logs are not subject to any possible malfunctions of my audio equipment (and if they are subject - then only to errors of my PC). These logs clearly demonstrate a malfunction of the UMIK-1:

REW first scan - mystery

Q. What test noise loudness is needed for REW measurement? There is no fixed loudness requirement for REW measurements. As long as SNR is sufficient, you can measure at 10dB SPL or 140.

audiosciencereview.com

Thanks in advance for your efforts, dear friend!

 

[Sokel]

I am very sorry that I failed to write clearly that 1) I am uploading exactly what you asked me to, and that 2) I did everything exactly as you told me to do. In addition to the additions to the scans sent earlier, I also sent the scans that were made at a different level - and they also have the same strange changes in high frequencies. I am very sorry that I did not write for the fourth time that I have already returned the faulty microphone.

It is a pity that you did not read my advice to use the logs with the results of recording the SPL of absolute silence - and therefore these logs are not subject to any possible malfunctions of my audio equipment (and if they are subject - then only to errors of my PC). These logs clearly demonstrate a malfunction of the UMIK-1:

REW first scan - mystery

Q. What test noise loudness is needed for REW measurement? There is no fixed loudness requirement for REW measurements. As long as SNR is sufficient, you can measure at 10dB SPL or 140.

audiosciencereview.com

Thanks in advance for your efforts, dear friend!

Logs are indeed strange.
That's how it's normal with silence, until I clapped three times at the 1 min. mark at an increasing fashion in level:



There's definitively something wrong, the reason though is unclear.
A new mic would definitely help for comparison.

 

[SashaR]

Logs are indeed strange.

Yes, silence logs also prove a microphone malfunction.
But, fortunately, I managed to take several scans BEFORE the microphone became completely unusable.

I sent scans that are probably not spoiled by the microphone. One set of scans is from where people put the listening chair, it is called "-15 dB" and is divided into two ZIP files.
The second and third sets of scans are made close to the speakers, almost in the plane of the speakers and their names contain "-20 dB" and "-25 dB" and in high frequencies these are strange scans. (In the set that is "-25 dB" there are 4 or 5 scans spoiled by the microphone).
I do not know what to do with the information I have collected - I can't even build a PEQ: a) my EQ has only 10 bands, and b) REW does not allow me to use shelves.
I will be very grateful to all of you for advice and help!

 

[SashaR]

I ask experts and specialists to say: whether the PEQ that I asked REW to generate using an underestimated target level (in my attempt to make it as easy as possible for REW to create the most flat response), is flawed in any sense?
Thank you in advance for your help!

 

[SashaR]

Is there a way to force the REW to ignore a specific part of the registered response? Or, alternatively, change manually a part of response?

Thank you for the help!

 

[Salt]

In the impulse window you can restrict what ever you want.

 

[SashaR]

Does anyone know how to build a PEQ for UAPP using REW ? The 10 bands that REW can be limited to allow you to choose either classic Q or symmetrical Q - but the ToneBoosters EQ of UAPP only has Analog Bell and Digital Bell. What does this correspond to?
What in version 3 of ToneBoosters (which is ported into UAPP) corresponds to REW terms?

Thank you all in advance!

 

[SashaR]

How can I download an SPL file from AutoEQ.app site? Where else can I download measurement files for headphones and, if possible, speakers?
How do I upload a target to REW and where can I download them?

 

[staticV3]

How can I download an SPL file from AutoEQ.app site?

All of AutoEQ's frequency response files and many target curves are hosted on https://github.com/jaakkopasanen/AutoEq

Additionally, you can grab curves from squig.link sites using the Developer tools built into Chromium-based browsers (the Network tab).

If you have a graph that contains the curve you're interested in, then you can convert it to .txt or .csv using https://usyless.uk/trace/ or https://automeris.io/

Targets can be loaded into REW in either the settings or the EQ panel.
REW calls them HouseCurves.

Here are some speaker targets:

A collection of speaker target responses in .csv/.txt format

In order to use REW's automatic EQ feature, I started collecting and converting every speaker target that I could find into the necessary .txt format. In this thread I want to share my work so far, as well as ask you guys for target responses that I may have missed. Comparison: So...

www.audiosciencereview.com