Inconsistent measured volume levels

Dmanstarr

New Member
Thread Starter
Joined
May 29, 2023
Messages
43
I can do a series of measurements in a row and the relative volumes are correct.

A little time later with everything at the exact same volume and the mic in the exact same place the recorded FR graph is several decibels lower or higher randomly. Help!
 

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
7,212
If you are using a receiver do you have any kind of dynamic EQ or dynamic volume mode enabled?
 

Dmanstarr

New Member
Thread Starter
Joined
May 29, 2023
Messages
43
This is measuring with a umik with a windows laptop in a car. There is nothing of the sort and I am certain nothing is changing as I said. This has been going on for the past couple years now with updates.

I’m thinking it could be a software setup issue. Perhaps the laptop is not configured right either in REW or elsewhere. I can’t recall having this problem with my desktop.
 

Dmanstarr

New Member
Thread Starter
Joined
May 29, 2023
Messages
43
This is measuring with a umik with a windows laptop in a car. There is nothing of the sort and I am certain nothing is changing as I said. This has been going on for the past couple years now with updates.

I’m thinking it could be a software setup issue. Perhaps the laptop is not configured right either in REW or elsewhere. I can’t recall having this problem with my desktop.
It almost always happens when I close out REW. When I go back to take more measurements even with everything the same, all the new ones are at a different level. FR is the same as expected, but the whole graph shows as equally quieter or louder for all measurements when they should be *very* close in volume to the earlier set.
 

Dmanstarr

New Member
Thread Starter
Joined
May 29, 2023
Messages
43
It’s a pink noise track on a usb stick. The playback is exactly the same, I am certain.
 

Dmanstarr

New Member
Thread Starter
Joined
May 29, 2023
Messages
43
I don’t have any saved that are a good example. I deleted my last measurements that displayed at a lower level as I didn’t feel like doing them all over since relative levels is what I was trying to get.

I will be doing more soon, so when it happens, I will save the files to post here.
 

Dmanstarr

New Member
Thread Starter
Joined
May 29, 2023
Messages
43
I don’t have any saved that are a good example. I deleted my last measurements that displayed at a lower level as I didn’t feel like doing them all over since relative levels is what I was trying to get.

I will be doing more soon, so when it happens, I will save the files to post here.
I can assure you the frequency response matches and it’s just publishing the result at a lower level. As if the recording level changes by itself when I restart the program. It may even happen without the restart, but I am not certain.
 

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
7,212
If I can see a couple of measurement files that show the normal and reduced levels that will provide more info.
 

Dmanstarr

New Member
Thread Starter
Joined
May 29, 2023
Messages
43
*deleted* I saw my mistake. Hopefully that's all it has been this whole time.
 

Dmanstarr

New Member
Thread Starter
Joined
May 29, 2023
Messages
43
Ok. Now it happened. I opened a saved set of measurements from earlier today. My playback levels have been the same all day. No filters to signal, same mic position. It's not a perfect example, because one measurement is with the grill off, but that is only causing a rise above 600hz. The lower freqs should be very close to the same volume, but it's a good number of decibels lower. I haven't offset any of them (which was my mistake above, I offset and forgot about it).
 

Attachments

  • L MB 6_ grill off.mdat
    59.4 KB · Views: 4
  • L MB 6_ grill on.mdat
    59.4 KB · Views: 4

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
7,212
Input volume is the same for both and the SPL offset (derived from input volume and the cal file sensitivity) is the same for both, so doesn't look like anything has changed on the input path. Looks to be a 5 dB shift between the two. Interesting that the levels are very similar at high frequencies, curious if the grill effect is somehow exactly opposite to the shift that is occurring.

1685434424107.png
 

Dmanstarr

New Member
Thread Starter
Joined
May 29, 2023
Messages
43
When I say “grill” in this case, it is an entire door trim panel and close measured 8-10” from dustcap of a pure midbass driver with car door open, outside and no filters.

I’ve had similar happen dozens (maybe hundreds) of times where it is the same shift all the way down the line. This was just the first example that occurred.

*I say “same,” I mean equal across the spectrum. I don’t *think* it’s always 5db, but I haven’t tuned for a while, so if I saw any pattern to this before, I don’t recall.

It’s very frustrating because I can’t trust my levels across different sessions.

My feeling here is like when you take a car to a mechanic, and then you can’t get the random problem to occur to show him/her.
 
Last edited:

Dmanstarr

New Member
Thread Starter
Joined
May 29, 2023
Messages
43
Input volume is the same for both and the SPL offset (derived from input volume and the cal file sensitivity) is the same for both, so doesn't look like anything has changed on the input path. Looks to be a 5 dB shift between the two. Interesting that the levels are very similar at high frequencies, curious if the grill effect is somehow exactly opposite to the shift that is occurring.

View attachment 62021
That said, the *only* thing that I changed here is the grill. Volume of the vehicle head unit, at the processor and amp gains stayed exactly the same all day.

It’s possible it’s the dsp as the software is buggy, but it’s would be surprising if I’ve never noticed it randomly changing volume over hundreds of hours tuning.
 

Dmanstarr

New Member
Thread Starter
Joined
May 29, 2023
Messages
43
Regardless, this *will* happen again when I have changed literally nothing at all producing an equal shift in the same measurement position across the board. When it does, I will once again be baffled at what triggered the behavior, but in any any case, I will sent you the mdats.

Next time I will take multiple measurements to confirm that the recorded volume shift has changed across the board from where it had remained previously. If nothing else to prove to myself I’m not losing my mind.
 

Dmanstarr

New Member
Thread Starter
Joined
May 29, 2023
Messages
43
Regardless, this *will* happen again when I have changed literally nothing at all producing an equal shift in the same measurement position across the board. When it does, I will once again be baffled at what triggered the behavior, but in any any case, I will sent you the mdats.

Next time I will take multiple measurements to confirm that the recorded volume shift has changed across the board from where it had remained previously. If nothing else to prove to myself I’m not losing my mind.
I also concede that the above example could have been an error on my part that I missed. I hope so. It’s just happened so many times where I was certain all settings where the same. Most often when it’s the next day and I open a previous set to continue and my levels are many db different from before with all settings and mic position the same.
 

sam_adams

Member
Joined
May 23, 2017
Messages
285
The info on the measurements indicate that these are RTA measurements and not sweeps. The .cal file used was the ninety degree .cal file. More than likely what you are seeing is a change in the background noise since most of the difference is below 1 KHz.

Try pointing the mic directly at the driver and use the zero degree .cal file while doing a 20 Hz - 20 KHz sweep. Use a 256k or 512k sweep at a high enough level to get the best signal-to-noise ratio. Make sure that your input and output sample rates are the same.

One other thing to consider. The door essentially acts as the enclosure for the speaker. If you measure with the interior panel on, the door acts to a certain extent as a sealed enclosure. If the panel is off, the driver is essentially in free air. Both these conditions will change the measured response whichever way you measure them.
 
Last edited:

Dmanstarr

New Member
Thread Starter
Joined
May 29, 2023
Messages
43
First of all, most people use RTA for various reasons fo car audio applications. I use sweeps for my home stereo. That close measurement above was just to verify that I had done a good treatment on the door. The files below are RTA. Mice position and all volume and gains identical. No filters. It seems like you don't know what is causing this or don't believe it's a software issue (REW or other software), but it is. If nothing else, I'm not the only person I've seen who has had this issue. See attached.

Right mR 2 was measured days after "RmR msr" because I am measuring a tweeter and needed to see if REW is showing different levels as it has done to me dozens if not hundreds of times. All the other drivers are changed by the same level if measured the same way they were before. Thanks for your help.
 

Attachments

  • R mR msr.mdat
    64.6 KB · Views: 4
  • R mR 2.mdat
    60.2 KB · Views: 3

Dmanstarr

New Member
Thread Starter
Joined
May 29, 2023
Messages
43
First of all, most people use RTA for various reasons fo car audio applications. I use sweeps for my home stereo. That close measurement above was just to verify that I had done a good treatment on the door. The files below are RTA. Mice position and all volume and gains identical. No filters. It seems like you don't know what is causing this or don't believe it's a software issue (REW or other software), but it is. If nothing else, I'm not the only person I've seen who has had this issue. See attached.

Right mR 2 was measured days after "RmR msr" because I am measuring a tweeter and needed to see if REW is showing different levels as it has done to me dozens if not hundreds of times. All the other drivers are changed by the same level if measured the same way they were before. Thanks for your help.
Also, I know the mic cal is different for each.
 

Dmanstarr

New Member
Thread Starter
Joined
May 29, 2023
Messages
43
For the record, when I restarted REW it now measures at the correct volume relative to the original measurement. So for now I'll just have to check the REW recorded volume offset via such a comparison test frequently, and restart if there is such a discrepancy. That's probably what I figured out to do last year when I was tuning last. I don't mean to seem ungrateful for this free software. I am. Just wanted you to be aware of the issue at least, even if I have to live with it, as it seems I do for now.
 

Attachments

  • R mR 3.mdat
    60.2 KB · Views: 2

sam_adams

Member
Joined
May 23, 2017
Messages
285
First of all, most people use RTA for various reasons fo car audio applications. I use sweeps for my home stereo. That close measurement above was just to verify that I had done a good treatment on the door. The files below are RTA. Mice position and all volume and gains identical. No filters. It seems like you don't know what is causing this or don't believe it's a software issue (REW or other software), but it is. If nothing else, I'm not the only person I've seen who has had this issue. See attached.

Right mR 2 was measured days after "RmR msr" because I am measuring a tweeter and needed to see if REW is showing different levels as it has done to me dozens if not hundreds of times. All the other drivers are changed by the same level if measured the same way they were before. Thanks for your help.

Your measurements have different output levels, FFT lengths, and different averaging.

529.png


531.png


If you are going to measure a single driver the best approach is to measure on axis. If you want consistent measurements when measuring single drivers, make a jig that allows you to exactly position the mic in relation to the driver you're testing so the the mic is positioned in the same place every time. Pointing the mic to the ceiling of your shop is going to measure the response of the space in addition to the driver.

It’s a pink noise track on a usb stick.

Try measuring with pink noise generated from REW. You can use the Signal Generator to generate various types of pink noise that you can either input to the system directly or save to a file for playback. Using REW to generate a file will provide a known source that can be easily reproduced.

If you believe that there is a bug in REW, update to the latest early access release from the first post of this discussion. Make your measurements and then generate a diagnostic file (Help > Generate diagnostic file). Submit the .mdat and diagnostic file in the early access discussion so @John Mulcahy can examine it to determine if there is an issue.
 

Dmanstarr

New Member
Thread Starter
Joined
May 29, 2023
Messages
43
None of that is in anyway relevant to what is on the graphs. And you can’t see the output levels because REW was not outputting anything.

However, the data offset in the info you sent above shows a 3db offset between the two. I did not add any data offset! REW did that in it’s own.

*I suspect that is the problem*

If you look at the third mdat I sent, I bet it will show a data offset that matches the first file I sent, because when I restarted REW, it went back to the matching recorded volume in the graph.
 
Top Bottom