Curves never the same

[jpv]

I am new at this and I am not doing something right. I have read the help and followed it as close as I could. My question is If I make two or more measurments they are never the some twice. I would think they would be mirror images of each other. I don't change anything I just press the measure button again and its different. I am measuring my PC speakers for now just to get this dowm befor I start testing my system. Where am I going wrong.
John

[brucek]

Where am I going wrong.

The measures should be a carbon copy every time.

The only time I find the measures different is when my computer decides it wants to perform some other task during the sweep. The result is an interference in the sound of the sweep. It isn't smooth and continuous, as if it had gone and performed another task and then came back to the sweep. I have come to recognize this problem and whenever I hear it I simply cancel the measure and try again.

This only happens on my old challenged computer. My fast computer doesn't have this problem.

Do you think this is your problem?

brucek

[jpv]

I think figured it out. On my sound card, Realtek HD audio, I had to slide the playback volume all the way down to zero and the record volume at 50%. There is no button to press for full duplex. I think I was getting feedback thru the mic while doing the sweep. They look much more alike now.
My next step is to measure my subs first correct? What should a good waterfall chart look like?

[DrWho]

If you haven't done it already, I would calibrate your soundcard first...just to make sure you got rid of all the feedback.

[Wayne A. Pflughaupt]

My laptop isn’t the fastest in the world, but it’s certainly no dinosaur. In my experience, it’s perfectly normal to not get carbon-copy results with every measurement. Here are a few examples culled from my files. [Edit - traces are separated for the sake of clarity.]

Four REW Measurements Taken Over a 30-Minute Period

These were taken in the same session, over a 1/2-hour period. Notice that response between 20-40 Hz varies quite a bit. The depression at ~55 Hz varies in depth and the center shifts a little. With the sawtooth section between 55-90 Hz, the differential between the high and low points is a few dB greater with the lower two traces, especially the dark blue one. The 90 Hz peak is at least a couple dB lower in the gray trace compared to the others, as is the 100 Hz peak

The next graph shows measurements where each was taken two month intervals, in Sept. and Nov. last year, and Jan. this year.

Three REW Measurements Taken Over a 4-Month Period

The next graph shows two measurements taken on the same day, an hour apart. The two traces are “as is” readings, not separated by the Trace Offset adjustment like the graphs above. Notice that the light blue trace is at least 1-dB higher in overall SPL level.

Two REW Measurements Taken 1-Hour Apart

Naturally, REW is going to offer different filter recommendations for differing traces, and different waterfall readings as well (another reason I don’t take much stock in them for filter setting purposes).

I’m using the RS digital meter. If some here actually are getting perfect carbon-copy results, I expect they’re using the Behringer mic? The latter has a much smaller diaphram, which I expect would make it less susceptible to variations in air pressure and other atmospheric changes, which can alter measured results.

Regards,
Wayne

[jpv]

Thanks for the reply, I see the traces are at diff volumes then each other and your last graph they are the same. I did not realize the program will seperate the diff traces. It looked like they were just getting louder. How do you turn that feature off or on? also how many traces do you do to make a good judgement of the room. Do you take an avg over many days?

[OvalNut]

Something as simple as a change in barometric pressure and/or temperature can affect sound pressure readings. Also, any furnace/AC fans blowing or not blowing at the time.

...oh, and the curvature of the earth.


Tim

[Wayne A. Pflughaupt]

jpv wrote:

Thanks for the reply, I see the traces are at diff volumes then each other and your last graph they are the same. I did not realize the program will seperate the diff traces. It looked like they were just getting louder. How do you turn that feature off or on?

Sorry for the confusion, jpv. On the left side of the REW screen you’ll see the “Trace Adjustments” button. Under it is a function called “Trace Offset.” All I did was separate the traces by 8 dB in the top graph and 12 dB in the second, for the sake of clarity. Otherwise, it would be difficult to precisely see the differences from one trace to the next because they would all be on top of each other.

also how many traces do you do to make a good judgement of the room. Do you take an avg over many days?

Actually, I’ve been using the same EQ settings since September (all those graphs are unequalized baselines), and nothing has changed audibly, despite the differences the measurements show.

What you might do is save your baseline readings and take some additional baselines from time to time to see if anything drastic appears that wasn’t coming through before, and perhaps re-tweak the equalizer if you see it regularly re-occuring. But truth be told, you probably won’t hear a difference one way or the other. IOW, don’t loose any sleep over it.

OvalNut wrote:

Something as simple as a change in barometric pressure and/or temperature can affect sound pressure readings. Also, any furnace/AC fans blowing or not blowing at the time.

...oh, and the curvature of the earth.

You bet.

Here is another graph. I was taking some measurements one morning earlier this month at 2-minute intervals to see what kind of differences I would get between them. It was cold outside and I had the furnace running and a thermometer was measuring 73° inside the house. Just for grins, I wondered what I would happen if the SPL meter was warmer. So I set it on a window sill where there was streaming sunlight. The thermometer showed 86° in the sun. After 30 minutes I took a few more REW sweeps, and the graph below shows the results.

The blue trace is a 73° reading, the red after the meter set in the 86° sunlight for half an hour. As you can see, the “hot” trace (no pun intended) registers 1.5 dB higher overall. (The receiver’s volume control was not changed.) I had five readings at 73° and four at 86°; all the 86° traces were 1.5 dB higher. For illustration purposes, the two I picked here were the most closely identical in appearance.

Regards,
Wayne

[DrWho]

Wow, errors of that magnitude should be telling you something is terribly wrong with the measurement! Keep in mind that this is a dB scale...if you're off by 3dB, then you're off by 200% - just imagine a volt meter that reads 6V when it's really 12V.

Errors under 1dB would be acceptable...maybe a bit more if you can verify environmental noise is tainting the measurement, but if noise is the problem, then the measurement should most certainly NOT be used for any form of decision making about the system. In other words, you should be ignoring the regions where the measurement isn't valid.

[g3kko]

If older hardware seems to be struggling a bit, has anyone tried raising the priority of the java.exe task in Windows to "realtime"?