[Jonas_h]

I have just bought a M-Audio FastTrack Pro USB sound card, and have just made a calibartion. Use RCA output to unbalanced line input and follow the instructions on the REW calibration guide.

When done, i get this measure:

Hmm I have inserted an image, but it wont show?
See it here: clickit-systems.dk/measure.jpg

Sample rate: 48000Hz
-3dB points: 2.9Hz, 4.499kHz
Input device: Line 1/2 (M-Audio FastTrack Pro
Input: No input selected
Channel: Right
Input volume: no control
Input RMS target: -6.0dB
Actual RMS at 1kHz: -7.1dB
Output device: Line 1/2 (M-Audio FastTrack Pro)
Output: No output selected
Sweep level: -6.0dB
20Hz .. 20kHz flatness: +2.6, -36.0dB


At first glance, I think it looks pretty bad when made with a sound card which costs $250! I though I would use it for full range measurements too, but already at 400hz the graph starts going crazy!

Is this a normal response? I can still use the money back guarantee from the store, so I hope I can get some fast response from you guys

 

[Jonas_h]

It seems that I have solved my problem The recording-level in windows mixer was set to 100% which made it go crazy.. Lowering to 70% has resulted in a completely flat calibration!

 

[brucek]

Is this a normal response?

No, it's not.

Did you get the correct input / output levels in the VU meter when you ran the soundcard routine.

It looks like you have some feedback. Do you have a monitor mode turned on.

brucek

 

[Jonas_h]

No, it's not.

Did you get the correct input / output levels in the VU meter when you ran the soundcard routine.

It looks like you have some feedback. Do you have a monitor mode turned on.

brucek

I think I had the monitor turned on too on the first measure. But the problem was the level of the recording device in windows. The meter jumped +-2db up and down all the time.

My new measurement looks like this: www . clickit-systems.dk/measure_good.jpg (Put the link together yourself. I'm still not allowed to use links)

I would think this calibration is good enough!

 

[Jonas_h]

And know the calibrations are crazy again :S So it was not the level problem. I am really confused now! Haven't touched anything at all since the last good measurement!

EDIT: It is like the VU meters are jumping up and down (like +- 1.5db).. Any idea why this is happening??

 

[brucek]

The soundcard cal looks fine, so use that one.

You are allowed to post pics after 5 posts, so you should be OK.

brucek

 

[Jonas_h]

The soundcard cal looks fine, so use that one.

You are allowed to post pics after 5 posts, so you should be OK.

brucek

Shouldn't I locate the problem instead of just using the one succesfully created file? When trying to verify the cal file (guide in the help section) the graph goes crazy again and I cannot make a new calibration. The same problem as I had in the starts is again an issue.

When I just have started the calibration process and I have to match the levels of the input and the output, the VU-meter for the right channel is jumping up and down - not just standing still. What can the reason for this be?

I use RCA output to 1/4" input..

 

[brucek]

Shouldn't I locate the problem instead of just using the one succesfully created file?

Yes, but the soundcard calibration is fine, so you use that file and leave the loopback in place, and take a measure (or as many as you like) of that loopback cable. The result of the measure (if the soundcard file is correct) will be a flat line from 0-20000 Hz.

Use the loopback test measures to solve the problem.

You Run the Check Levels routine,

Then run the Calibrate SPL routine,

Then Measure.

The levels on the VU will jump, since the test signal is pink noise.

brucek

 

[Jonas_h]

Yes, but the soundcard calibration is fine, so you use that file and leave the loopback in place, and take a measure (or as many as you like) of that loopback cable. The result of the measure (if the soundcard file is correct) will be a flat line from 0-20000 Hz.

Use the loopback test measures to solve the problem.

You Run the Check Levels routine,

Then run the Calibrate SPL routine,

Then Measure.

The levels on the VU will jump, since the test signal is pink noise.

brucek

I did take the measure with the loopback cable, and it was not accurate at all... So thats why I wan't to find out why it doesn't make good calibrations anymore.

Could latency have anything to do with it? I can change this setting in REW and in the control panel of the sound card..

 

[brucek]

Could latency have anything to do with it?

Yeah, you could be getting some dropouts in the signal. You can try setting the latency to the highest setting. Also try increasing the replay and record buffer sizes in REW to see if that helps. You could also try a different sample rate than you used to do the soundcard cal the first time (44.1 or 48).

brucek

 

[Jonas_h]

Nothing helps The weird thing is, that if I run 10 calibrations in a row without touching any configuration, the graphs looks very different! Some of them are almost flat, and then the next one can be completely crazy!

I'll have to take my computer to the store and try some different cards tomorrow.

Could it be the fact that i'm running Win7?

 

[brucek]

Could it be the fact that i'm running Win7?

Not many have tried Windows 7, so it's a possibility. Perhaps the Windows 7 driver is suspect....

have you listened to a swept signal from REW? Is it smooth or broken up?

brucek

 

[Jonas_h]

Maybe this could be the problem: http://forums.m-audio.com/showthread.php?t=8981

Sounds possible because the graph is sometimes flat - so if I measure the exact time no other program tries to use the sound card.

 

[brucek]

Maybe this could be the problem

Yeah, certainly if your card is being interrupted while REW is performing its sweep, it can cause a dropout and the graph will show like yours does. That's why I was asking about listening to a swept signal through your speakers by using the REW generator swept signal. If there's any hitching or dropouts, then something else is taking up the CPU's time. Just select the generator and pull down Meas. sweep and set a start of 20-10000Hz (or whatever you prefer) and listen.

brucek

 

[Jonas_h]

Yeah, certainly if your card is being interrupted while REW is performing its sweep, it can cause a dropout and the graph will show like yours does. That's why I was asking about listening to a swept signal through your speakers by using the REW generator swept signal. If there's any hitching or dropouts, then something else is taking up the CPU's time. Just select the generator and pull down Meas. sweep and set a start of 20-10000Hz (or whatever you prefer) and listen.

brucek

I just played a sweep through the soundcard to the speakers, and there were popping and weird sounds all the time. So I have found my problem!

I am going to the store and exchanging it now. Hopefully they will let me try out some different cards in the store on my compyter so I can see which ones are suitable.