Home Theater Forum and Systems banner

1 - 20 of 32 Posts

·
Registered
Joined
·
16 Posts
Discussion Starter #1
I'm using the ubiquitous Behringer UCA-202 as the pre-in & pre-out for my REW set-up (a RS digital meter [new version] is my mic, although I may get the Dayton mic at some point).

My problem I have is that the UCA-202 is reporting high SPLs at bass frequencies, rolling off slowly from 80 dB down to just under 30 dB at 300 Hz (and extending farther). Obviously it's impacting my measurements - my speakers (no sub) have an anechoic -3 dB point at 27 Hz; and have an audible response at 20 Hz (can't remember but it's about 15 dB down) - yet I can't get an accurate measurement because of the curve the Behringer is imposing.

I did a comparison of the whole setup measuring my speakers (Infinity Renaissance 90; great speakers) versus just the UCA-202 running without even being plugged into the mic - and you can see those curves below.

I haven't taken comprehensive measurements yet - it would be nice to have a measurement of room noise (the majority of the sound being my Thinkpad's fan - which is already pretty quiet) for comparison, and high frequency measurements of the UCA-202 alone as well. However, given that I still have the same problem with no mic, I skipped that for now.

I can see a similar bump in the RTA with no mic, so I'm not sure that it has to do with the output of the UCA-202. Again, more testing may be required to entirely eliminate that possibility. Like I said, I'm in the preliminary troubleshooting stage and have yet to perform much in the way of time-consuming serious variable control.

So has anyone else had this problem? Might it be a calibration issue? I think I have done some tests at a higher SPL and it seemed to be less of a problem. Other than the SPL calibration and the test level, the only other thing I can imagine being wrong is that my UCA-202 is defective. Anyone have any ideas?

Thanks,
Ben
 

Attachments

·
Premium Member
Joined
·
1,400 Posts
Ben said:
,,,,However, given that I still have the same problem with no mic,,,,,
> Yep, pretty strange behaviour indeed .

> You should make a simple looped-back ( output to input ) recording ( calibration ) of the UCA 202 to see what's up with it .

:sn:
 

·
Premium Member
Joined
·
1,562 Posts
...Anyone have any ideas?
There is not much info here but…
My impression is that this does not look like a soundcard issue. It looks like an issue with the calibration of the REW SPL meter.

The fan noise you show is at about 70 dB when it should be much lower (maybe 45 dB if the mic is not very close to the fan)!

The rise in the low freq is partially due to the soundcard calibration, but appears mainly due to the calibration curve loaded for the RS SPL meter. The LF for the RS meter has to be boosted a lot due to the C weighting. This makes the noise floor approach the 75 dB recommended measuring level at the very low frequencies even when calibrated correctly. If calibration of the REW SPL meter is not correct the result could look like your measurement.

If my assumptions are correct I think you will find that a proper calibration of the REW SPL meter to agree with the RS meter will solve your problem.

[You are correct that using higher measurement SPL levels to boost the signal further over the noise floor is directionally correct, but not normally necessary or recommended. When setup correctly the RS meter can be used down maybe to 25 Hz or so using the 75 Hz SPL measuring level with without significant error. If you want high accuracy at the low end you need a real measurement mic and better yet, one with a custom calibration curve for it.]
 

·
Registered
Joined
·
16 Posts
Discussion Starter #4
There is not much info here but…
My impression is that this does not look like a soundcard issue. It looks like an issue with the calibration of the REW SPL meter.

The fan noise you show is at about 70 dB when it should be much lower (maybe 45 dB if the mic is not very close to the fan)!

The rise in the low freq is partially due to the soundcard calibration, but appears mainly due to the calibration curve loaded for the RS SPL meter. The LF for the RS meter has to be boosted a lot due to the C weighting. This makes the noise floor approach the 75 dB recommended measuring level at the very low frequencies even when calibrated correctly. If calibration of the REW SPL meter is not correct the result could look like your measurement.

If my assumptions are correct I think you will find that a proper calibration of the REW SPL meter to agree with the RS meter will solve your problem.

[You are correct that using higher measurement SPL levels to boost the signal further over the noise floor is directionally correct, but not normally necessary or recommended. When setup correctly the RS meter can be used down maybe to 25 Hz or so using the 75 Hz SPL measuring level with without significant error. If you want high accuracy at the low end you need a real measurement mic and better yet, one with a custom calibration curve for it.]

No no no, that's not the fan noise in purple; that's my speakers. The ambient room noise is on the order of 50 dB broadband (I have not measured it in REW yet).

I do agree that a calibration issue seems likely - but I am in fact calibrating the REW SPL meter to agree with the RS meter. In fact, I have tried several different techniques for calibrating the RS SPL meter, because the line output is very sensitive to the range setting on the RS meter. If I calibrate at any level (I have done 75 dB up to 90 dB) with the RS meter set to the appropriate range, the input level into the UCA-202 is very high and within a few dB of clipping (I am told by REW that I have one or two dB of headroom over the level I calibrated at). I noticed how large the signal was, so to avoid clipping I matched the level on the RS SPL meter with the REW SPL, and then tried setting the RS meter to a higher range before I click "calibrate". This brought the level down to a more realistic amount of headroom with accurate REW SPL readings, but I still had the same low bass reading problem.

At some point I was able to get the erroneous reading low enough to see the 15-25 Hz levels better, but I have so far been unable to repeat that. I believe it involved setting the test level from the -12 dB or whatever the default is all the way to -40 dB and then turning up the volume on my pre-amp to reach the testing level. However, even with those readings the erroneous low bass curve is still visible, just at a lower level. So far I have been unable to reproduce those readings, but I will have more time after work today to do so.

As for the C-weighting - it was my understanding that the RS digital meter's microphone line output (not its readings, of course) is unweighted - is that wrong? As such I have left the C-weighting box in preferences unchecked but I do use the RS digital meter's correction file. I do think that the graph looks like a C-weighted inverse correction curve, though....

I will try a loopback measurement tonight among my other testing.
 

·
Premium Member
Joined
·
1,400 Posts
Ben said:
,,,As for the C-weighting - it was my understanding that the RS digital meter's microphone line output (not its readings, of course) is unweighted - is that wrong? As such I have left the C-weighting box in preferences unchecked but I do use the RS digital meter's correction file. I do think that the graph looks like a C-weighted inverse correction curve, though....
> Hmmm, everyone around here ( that I've taken notice of ) checks off the C-Weighting box within the preferences window > mic tab .

> Most also use the RS meter correction file ( though not all ) .

:sn:
 

·
Registered
Joined
·
16 Posts
Discussion Starter #6
> Hmmm, everyone around here ( that I've taken notice of ) checks off the C-Weighting box within the preferences window > mic tab .

> Most also use the RS meter correction file ( though not all ) .

:sn:
Hmm, maybe I've gotten it backwards.

Also, I should add that when I started I didn't use the RS meter correction file. The one I posted is with it, but the one without has the same problem. Obviously there's still the relatively small differences between with and without the RS correction file - a few decibels added on the low end - but the difference is nowhere near the problem I'm having.
 

·
Premium Member
Joined
·
1,562 Posts
Since the above plot was indeed your speaker reading and was around 75 dB I see no strong evidence that there really is a problem.
I mentioned that the required LF boost from that soundcard and RS SPL Meter together will make it appear there is high SPL output at very low freq when there really is none. That is a limitation of that choice of equipment. Do you really need to see what is going on below 25 Hz? If not, scale the chart from 25 Hz - 20 kHz and work from there.

Of course, particulary since I have never actually used that setup, there is the possibility there is a problem that is not apparent to me. If you are still having problems and looking for more help, it would help (at least me) to see your soundcard calibration and you mic calibration curves. You just need to check those boxes.
 

·
Registered
Joined
·
16 Posts
Discussion Starter #8
Okay, I have done some more troubleshooting today.

First, I ran the sound card calibration - that only brought minor changes, but then I realized I had a problem in calibrating - it was impossible to bring the input and output levels close to each other without turning the OS volume almost to zero (1 out of 50). That can't be good since that's digital volume... But I tried both anyway.


I tested a loopback after calibrating the UCA-202, which measured flat of course.

I'm pretty tired right now and haven't had the willpower to do a proper design of experiment so far - and as I've been playing around, I've noticed a ton of different behaviors that I haven't been able to pin down yet. So for now, I'm going to focus on the couple of things I did carefully measure.

First, I set everything up, calibrating the SPL level to the meter. This was using no calibration curve or weighting - The C-weighting would only increase my results below 10 Hz (and no higher) and the RS meter compensation would also increase the low frequency readings I got. I first had this problem without using either the weighting or the compensation, so to add them in where they would only mask my problem seems as if it would needlessly complicate the measurements.

I calibrated the SPL at 75 dB with the meter set at the 80 dB range. I can't remember the output dB FS level, but importantly it was the same throughout all of the first experiment's tests - and I believe I did not change it between calibrating the SPL and taking the measurements.

So I took my first measurement with the RS meter set at the 80 dB range - it was near clipping with the signal. Then, without changing anything else, I set the RS meter to the 100 dB range and measured again. I then changed it again to the 110 dB range (I threw out the 120 dB range numbers as I started to run into other non-linearities (presumably from the mic input into the UCA-202 from the RS SPL meter being so low.

The first graph below shows this experiment. The green line is the 80 dB range, purple is the 100 dB range and red is the 110 dB range on the meter. You can see that the output decreased (expected as the RS meter is giving less gain as you change the range) - but conversely, the erroneous sub-20 Hz response increased! It's not like it was a noise floor that stayed the same - obviously there is some interaction going on here between the meter and the sound card that is causing this. I then aligned the responses up, and you can see the increase in noise with each step better in the second plot. Again, note in the first plot that the noise actually increased as the level decreased - not just in relative terms, but in absolute terms.


Then, the second test I did: I changed the dB FS of the output level to change the level of the signal leaving the UCA-202, but adjusted the volume of the amp to reach the same SPL overall. In this manner the input signal was the same level, but the output level of the DAC was different.

In the last plot you can see this - the erroneous low bass signal was larger with the louder output level but same input level.

I'm falling asleep as I write this so I'll try to give it an analysis tomorrow - but let me know what you think.
 

Attachments

·
Registered
Joined
·
16 Posts
Discussion Starter #9
Since the above plot was indeed your speaker reading and was around 75 dB I see no strong evidence that there really is a problem.
I mentioned that the required LF boost from that soundcard and RS SPL Meter together will make it appear there is high SPL output at very low freq when there really is none. That is a limitation of that choice of equipment. Do you really need to see what is going on below 25 Hz? If not, scale the chart from 25 Hz - 20 kHz and work from there.

Of course, particulary since I have never actually used that setup, there is the possibility there is a problem that is not apparent to me. If you are still having problems and looking for more help, it would help (at least me) to see your soundcard calibration and you mic calibration curves. You just need to check those boxes.
You may be right about this... I was thinking - could the noise floor with the mic be high enough in the low bass that the boost from the mic calibration curve is only bringing it up into visibility? That wouldn't explain the level changing I have experienced, however.

I'm going to bed now but I'll pick it up again tomorrow. Thanks for the help - I would attach the sound card cal file but it's too big (and it only has a minor effect on the lowest bass and high treble); the RS meter cal file is attached but it's just the standard one from here for the newer digital meter.
 

Attachments

·
Premium Member
Joined
·
1,400 Posts
Ben said:
First, I ran the sound card calibration - that only brought minor changes, but then I realized I had a problem in calibrating - it was impossible to bring the input and output levels close to each other without turning the OS volume almost to zero (1 out of 50). That can't be good since that's digital volume... But I tried both anyway.
> I don't follow . What level controls are you talking about ?

> Read the beginning of this thread on what it takes to calibrate a UCA202 properly with Win7 .

Some Highlights ;

> ie ; Make sure Win7 is setup for 2 channel recording ;


> And make sure the Monitor Switch on the UCA202 is in the "OFF" position .

:sn:
 

·
Registered
Joined
·
16 Posts
Discussion Starter #11
Problem is, the control input/output volume mixer/volume check boxes are grayed out for me with the UCA-202, and I can't adjust the input volume or output volume through REW. The only control I have is the OS volume.
 

·
Premium Member
Joined
·
1,400 Posts
I run XP so my experience will be different in some aspects from yours, but ( for my UCA-222 ) :

> Within REWs preferences window, if I choose "USB Audio CODEC" as the Audio Device & then choose "Speaker" as the Output, those windows ( Wave Volume & Output Volume ) are made active/accessible ( if I choose to check-off the appropriate box above them ) .

> I can never get control of the "Input Volume" from within REW, no matter what my choices are within Audio Device & Audio Input drop-down menus . Input volume is always maxed at "1" ( which has never been a problem when using the RadioShack SPL meter ) .

:sn:
 

·
Premium Member
Joined
·
1,562 Posts
Okay, I have done some more troubleshooting today.

First, I ran the sound card calibration - that only brought minor changes, but then I realized I had a problem in calibrating - it was impossible to bring the input and output levels close to each other without turning the OS volume almost to zero (1 out of 50). That can't be good since that's digital volume... But I tried both anyway.

I tested a loopback after calibrating the UCA-202, which measured flat of course.
I first want clarify that that you indeed have a calibration problem. I did not consider the very large scale of the LF boost and indeed the curve would show an elevated noise floor not a smooth line. Sorry for the confusion.

The flat line you got with the loopback measurement rules out everything except the something related to attaching the RS meter. The volume should not be near zero so that is another indicator of a problem.

Is something amiss in your wiring of the RS meter? Is the adaptor you use to get the signal out of the meter a mono adaptor as needed, or is it a stereo adaptor? Check everything again to be sure it is correct.
 

·
Premium Member
Joined
·
1,400 Posts
> Here are two sound card calibrations ( loopbacks ) for my Behringer UCA222 .
> One has unwanted feedback entering into the loopback, while the other doesn't .




> Red is with the "Monitoring" switch engaged ( the physical switch is found on the UCA222/UCA202 ).
> Pink is with the "Monitoring" switch set to the "off" position .

> Asking REW to EQ ( auto-correct ) the response curve of that Red trace ( when that EQ is added to the Auto-EQ curve of the "C" weighting, & the EQ within the RadioShack calibration file ) will surely give a huge boost to the bottom end response, below say 20hz .

> My bets are now on Ben having done all his test traces with a junk SC calibration ( such as this ) loaded .

:sn:
 

·
Registered
Joined
·
16 Posts
Discussion Starter #16
I first want clarify that that you indeed have a calibration problem. I did not consider the very large scale of the LF boost and indeed the curve would show an elevated noise floor not a smooth line. Sorry for the confusion.

The flat line you got with the loopback measurement rules out everything except the something related to attaching the RS meter. The volume should not be near zero so that is another indicator of a problem.

Is something amiss in your wiring of the RS meter? Is the adaptor you use to get the signal out of the meter a mono adaptor as needed, or is it a stereo adaptor? Check everything again to be sure it is correct.
Well the near-zero Windows volume that I'm setting is for the soundcard calibration. I'm confused at what is happening though, as when I adjust the Windows volume down it is the right channel input that decreases - the output stays the same as reported by REW. Now, I didn't expect the Windows volume to affect anything, actually, as I thought that REW had exclusive control over the UCA-202 - but for it to change the input level but not the output level with a loopback connection is very strange. I will take screenshots of this tonight if necessary.

I'm just using an RCA interconnect to the right channel input on the UCA-202 (from either the RS meter or the right channel RCA line-output for a loopback connection), and I have the right channel selected as the input within the REW properties.

I am using both left and right outputs on the UCA-202, however - and I just realized that since I only ran the calibration with the loop-back connection through the right channel, the left channel output would be uncompensated for. However, the level between the two has remained the same and the same input is being used, so if I understand it correctly the only error from the left channel output should be from the difference in performance between the two channels (i.e. not much). Obviously if there was a significant level difference between them I'd hear it.

By the way, thank you so much for your help troubleshooting this so far. Obviously I have a lot to learn about using REW - it is not as straightforward as it first seems to account for everything.
 

·
Registered
Joined
·
16 Posts
Discussion Starter #17
> Here are two sound card calibrations ( loopbacks ) for my Behringer UCA222 .
> One has unwanted feedback entering into the loopback, while the other doesn't .




> Red is with the "Monitoring" switch engaged ( the physical switch is found on the UCA222/UCA202 ).
> Pink is with the "Monitoring" switch set to the "off" position .

> Asking REW to EQ ( auto-correct ) the response curve of that Red trace ( when that EQ is added to the Auto-EQ curve of the "C" weighting, & the EQ within the RadioShack calibration file ) will surely give a huge boost to the bottom end response, below say 20hz .

> My bets are now on Ben having done all his test traces with a junk SC calibration ( such as this ) loaded .

:sn:
That may well be it - but I did my first measurements (including the first plot I posted here) without any calibration whatsoever on the sound card or meter. It does, however, look just about like the opposite of the curve I have...

I checked the monitor switch very first thing before I started making measurements, but maybe I made a mistake (Well, I'm sure I made a mistake somewhere, just not necessarily here). I'll check it again.

But if I did have the monitor switch on and no calibration done, shouldn't I see a big decrease in the bass response?
 

·
Premium Member
Joined
·
1,400 Posts
Ben said:
I checked the monitor switch very first thing before I started making measurements, but maybe I made a mistake (Well, I'm sure I made a mistake somewhere, just not necessarily here). I'll check it again.

But if I did have the monitor switch on and no calibration done, shouldn't I see a big decrease in the bass response?
> Your observation of barely being able to send out any signal before overload occurs (ie; oscillation ) is quite consistent with a feedback loop being in place .

> "Feedback is feedback" no matter where it originates.
> Additional signal feedback has been known to have been added from within the computers own audio control panel ( this is called the "Thru" check-off box ) .
> The first areas to go from feedback into oscillation, will be those with the most EQ boost applied ( in your case, LF ) .
> This oscillation will also skew the over-all static snap-shot of the FR curve ( since above a certain thresh-hold it becomes self-generating ) .


> The amount of "fake" bass being reported on the screen will be proportional to the amount of feedback applied by the UCA202s' monitor switch ( FYI, I assume a nominal monitor mix of 50/50, with the recirculated signal rising above 50% when the circuit begins to oscillate ) . Add the boost EQ from the mics calibration file ( which is too heavy-handed IMHO ) to the EQ from the "C" Weighting compensation circuit and one gets a huge LF lift ( & not all of it is warranted IME ) .

> You'll eventually see a decrease in "apparent" LF response, once you get everything set up correctly .

:sn:
 

·
Registered
Joined
·
16 Posts
Discussion Starter #19
> Your observation of barely being able to send out any signal before overload occurs (ie; oscillation ) is quite consistent with a feedback loop being in place .

> "Feedback is feedback" no matter where it originates.
> Additional signal feedback has been known to have been added from within the computers own audio control panel ( this is called the "Thru" check-off box ) .
> The first areas to go from feedback into oscillation, will be those with the most EQ boost applied ( in your case, LF ) .
> This oscillation will also skew the over-all static snap-shot of the FR curve ( since above a certain thresh-hold it becomes self-generating ) .


> The amount of "fake" bass being reported on the screen will be proportional to the amount of feedback applied by the UCA202s' monitor switch ( FYI, I assume a nominal monitor mix of 50/50, with the recirculated signal rising above 50% when the circuit begins to oscillate ) . Add the boost EQ from the mics calibration file ( which is too heavy-handed IMHO ) to the EQ from the "C" Weighting compensation circuit and one gets a huge LF lift ( & not all of it is warranted IME ) .

> You'll eventually see a decrease in "apparent" LF response, once you get everything set up correctly .

:sn:
Well, I checked. I was right - the monitor switch has been off the whole time. And remember, I had this problem from the very start with no EQ applied whatsoever.

I'll give it another go tonight.
 

·
Registered
Joined
·
16 Posts
Discussion Starter #20
Okay, I'm fresh off some new tests.

The first plot has me repeating some tests I previously did, this time taking careful note of what corrections have and haven't been made. These were all performed with the self-created calibration curve for the UCA-202 and no correction curve for the RS meter.

This is what I measured:

Pink line (the straight downward sloped one) - the UCA-202 alone, no input

Green line (straight downward sloped with noise) - the UCA-202 with the RS meter, measuring ambient noise (speaker pre-amp turned off)

The three lines that track close together - the magenta/pink one that is just a little below the others is with only right channel output from the UCA-202 and running both speakers via the mono switch on my preamp. I guess my preamp doesn't compensate completely for the relative loss in gain. The other two are with both outputs and with the monitor switch on (blue) and off (purple). Clearly the monitor switch doesn't make any difference, anyway.

The second plot shows the sound card frequency response below the calibrated test. Obviously there's no massive boost - about 10 dB max and only about 1 dB at 10 Hz.

Also, I have another question - does it even matter what calibrations you have done when you actually take the measurement, since you can add (or eliminate) them at will from the plotted measurement? Obviously the input/output levels matter, but that's different.

Speaking of, selecting "microphone" from the input menu for the UCA-202 did in fact work, allowing me to set the input level. I had no trouble setting it at an appropriate level and getting a reasonable measurement range once I did that - although I had to set the input level as low as 0.100 to get it.
 

Attachments

1 - 20 of 32 Posts
Top