Popping in Sub Cal Pink Noise on Mac Book Pro

**geekwithfamily** · **09-03-06, 09:06 PM**

When I play the Sub Cal. 30-80 Hz pink noise I get a clicking or popping every half second in the output signal. It only occurs when the soundcard is set to 44.1 kHz. If I change it to 48 kHz the clicking goes away. I can't hear any clicking when I play a full bandwidth pink noise signal.

I'm on a Mac Book Pro running REW under Mac OS X (the laptop is new so I haven't had a chance to buy or setup Parallels and Windows) so the only way I can connect to my RS meter is if I have 44.1 set. Any other MBP users experiencing the same thing?

I'm pretty sure the clicks will ruin any measurements I will take.

**brucek** · **09-03-06, 09:32 PM**

Did you also set the Soundcard Sample Rate pull-down in REW to 44.1Khz to match the soundcards rate in your computer?

**geekwithfamily** · **09-03-06, 10:48 PM**

yes, the soundcard sample rate in REW is the setting I'm talking about. 48 no pops, 44 pops. Also happens on my Mac Mini (PowerPC).

**brucek** · **09-04-06, 06:59 AM**

Interesting. I will ask John (REW author) to comment on this...........

**JohnM** · **09-04-06, 08:03 AM**

Strange one. The pink noise sequence is generated in blocks of 8k samples (185ms at 44.1k) and only the filter coefficients change when switching between 44.1k and 48k, so there isn't anything happening in the code at 0.5s intervals.

If the clicks only occur when generating the sub cal and not when running a sweep then you can go ahead and make measurements anyway. If you look at the captured sweep (in the Oscilloscope graph page) and there are no discontinuities or gaps in the signal then all should be OK.

**geekwithfamily** · **09-04-06, 09:10 AM**

I spotted a related thread about another user with the same problem on a Powerbook. Guess I'll have to go the Windows route.

Would the clicks screw up the soundcard calibration/characterzation? I am connecting a double male mini-plug between the line in and out on the Macbook and keep getting errors. The impulse response is crazy like a feedback loop but I don't have monitoring on.

**JohnM** · **09-04-06, 06:29 PM**

One other possibility for the clicks/pops is interference from a wireless card, try disabling the wireless connection if it is enabled.

When you say double male mini-plug is that stereo-to-stereo? Needs to be.

**geekwithfamily** · **09-04-06, 09:07 PM**

Wireless didn't make a difference.

Here's a pic of my first graph of my sub cal:

Does this seem normal? Are the clicks in there? I'm measuring a Hsu VTF-2 and Vandersteen 1c fronts. Room is 13.5 width x 7.5 height x 30.5 length (in feet). DIY bass traps in front corners.

I'll post soon about the other little glitches in the Java/Mac interface I've found workarounds for.

**brucek** · **09-04-06, 10:00 PM**

Change the graph from Linear to Logarithmic (keeping the horizontal scale at 20hz-200Hz) and change the vertical scale to 45dB-105dB and repost it.

brucek

**geekwithfamily** · **09-04-06, 10:50 PM**

Graph is updated above. Got confused with the lin/log toggle. It is smoothed to 1/3 octave.

**geekwithfamily** · **09-04-06, 11:01 PM**

Some gotchas on my Macbook Pro with REW:
1. Graph numbers don't redraw when you change the scale or graph type. Resizing the REW window corrects the display.
2. Sound input is lost. Fix by choosing "Choose Device…" and then choosing Mac OS X Core Audio.
3. Sound output is half muted in auto measurements. I think this happens when I click on the Choose output pulldown list. Can be fixed by manually playing one of the pink noise test tones.
4. Crashing from changing tasks too quickly. Happened a couple times immediately after an Auto measure by clicking the Set Target Level button, have to force quit REW.
5. Unable to calibrate soundcard. I want to try this one again because I hadn't figured out the input problem above the last time I tried.

**JohnM** · **09-05-06, 02:36 AM**

Thanks Kyle, good feedback.

Your measurement looks fine, though would need to see the impulse response to be sure.

The graph axis redraw is something I've tried adding a workaround for in the next release, not clear how/why it happens on the Mac, must be some subtlety of the rendering engine.

Sound input and output should be easier with the next version, which can work with the OS defaults without requiring any device selection.

Interesting about the task change issue. From other reports the crash happens inside Apple's JVM, not sure what provokes that and whether I can find a workaround, but I'll look into it.

**geekwithfamily** · **09-05-06, 07:11 AM**

Any idea why I get the impulse response error every time I do an auto measure? The error reads something like: "The center of the impulse response wasn't where REW expected it."

The measurement above is with the mains on. Can't remember if I got the error on a sub only measure (with mains off).

**JohnM** · **09-05-06, 08:00 AM**

Might be because the measurements are being corrupted - easiest way to tell would be to use the Save Measured Data Set option to generate an mdat file of a measurement, zip it up and email it to me.

**geekwithfamily** · **09-05-06, 10:36 AM**

The impulse screen. The error did say something about corruption. I'll email you the mdat.

**geekwithfamily** · **09-05-06, 10:42 AM**

My mdat is attached, John.

**JohnM** · **09-05-06, 12:55 PM**

That mdat looks to be from a bad measurement - the measurement you posted earlier looks much better, which could mean that the pops and clicks you noticed are also occurring during measurement and sometimes causing corruption. If you look at the oscilloscope page after a measurement you may see obvious spikes or discontinuities in the captured sweep, which would confirm the problem.

**geekwithfamily** · **09-05-06, 01:04 PM**

My Osci graph attached

**JohnM** · **09-05-06, 02:11 PM**

Bit odd, there seems to be a huge buffer delay somewhere, the captured response is a long way after the stimulus. The pops might be indicative of some kind of buffering problem occurring between REW, the JVM and the soundcard. That might also be connected to the problem of crashes if you do something else using the soundcard (like setting target level) a very short time after a measurement sweep. REW requests replay and record buffers of 64k samples each, but the sound subsystem can ignore any requests and choose its own buffer sizes, if a very small replay buffer is assigned then the buffer can underflow causing clicks or pops when it runs out of data, but that plot suggests there is a large buffer somewhere which doesn't start emptying until it has accumulated a big chunk of data.

Are there any Mac controls that relate to soundcard buffering?

**JohnM** · **09-05-06, 02:26 PM**

For comparison, here are the scope and impulse plots for a sweep just now using my PC's speakers. Note the very short delay between the sweep end and the captured response end, that is how it would more typically look.

**geekwithfamily** · **09-06-06, 12:33 PM**

So yesterday I installed Parallels and ran REW under XP. What a better experience! It's great having control over the volume of the input and output from within REW. I still could not calibrate the soundcard, it has a feedback loop I can't get rid of even though I mute the line in. Parallels supports the Realtek XP AC'97 drivers, maybe I'm missing something in the mixer control panel. Outputting and miccing the measurements works fine.

I only get an error free measurement on every third or fourth try. Most measurements give the Impulse and corrupt data error, though the frequency plot seems normal. Could the error be from the bad soundcard cal (even though I clear it out under the soundcard menu data is still filled in on the soundcard tab and graph).

The good news is the clicks are gone (I'm using 48kHz). Every few measurements I'll get a soft click before the sweep kicks in.

Unfortunately, I didn't save any measurements from the session last night. I'll try again tonight.

**brucek** · **09-06-06, 12:42 PM**

Quote:

I still could not calibrate the soundcard, it has a feedback loop I can't get rid of even though I mute the line in.

The feedback loop is likely internal to the card, so muting line-in won't do much. You must have an internal monitor switched on somewhere.... Did you check both recording and playback mixers and ensure there is no soundcard app running like Sound Effects Manager for RealTek.

brucek

**geekwithfamily** · **09-07-06, 07:20 AM**

I was able to calibrate the soundcard last night after upgrading to the latest Parallels beta. Guess they changed how they deal with the sound drivers. I'll see if this also fixed the impulse corruption.

**geekwithfamily** · **09-09-06, 09:20 AM**

Had another measurement session last night. The graph is measuring my sub and mains mixed (both receive a full range signals: mains have inline high pass filter and the sub has its own low pass crossover dial). I'm not even sure if I need a BFD, the exaggerated level between 20 and 38 Hz will pump up action movie explosions and I mostly listen to rock music which I think only goes as low as the mid 50s Hz. Are the less than 5 dB peaks really that objectionable?

I still get some pops at the beginning of the measurement but all the data corruption errors are gone and my impulse and oscilloscope graphs look normal.

**brucek** · **09-09-06, 09:57 AM**

Quote:

I'm not even sure if I need a BFD,

It is quite a good response.

I would say the two spots I marked would be considered about 10dB after you equalized to the red line, which still provides an acceptable house curve..

I suspect it would be somewhat better if equalized..

20060908_hsu.jpg

brucek

Advertisement