Foray into REW... questions galore....

[glaufman]

Hi all... so back in December the wife bought me some toys for xmas... the ECM8000, EuroRack, and SoundBlaster 24bit Ext, exactly 2 wks after I lost my job and had to give back the company laptop I was going to run REW on, so they've been collecting dust until yesterday when I caught a deal on a computer (she wouldn't let me move her desktop downstairs to play around...) so I just HAD to start running scans immediately despite not having the SPL to calibrate them... for this and other reasons I may/not get into later, these scans aren't completely "legit" so lets not spend TOO much time on them, I'm really just looking for a little help in learning how to read these, to see if I'm on the right track, and to ask about some ideas/issues I had while running REW...

First and foremost, I have at best modest electronics, a SONY AVR. my front speakers are Klipsch RF-3 towers (approx 27Hz - 20kHz)... no sub, so lets not comment on the roll off under 30Hz in these scans... first the freq response: (1/3 smoothing)
bode.jpg
the impulse
impulse.jpg
and the LF waterfall
waterfall.jpg
So, I see from both the bode and the waterfall that I have some bass issues to deal with... interesting note, though, is that running the scan showed some rattles I didn't know i had, I'll track down, but until I do, how much could these be affecting the scans? Possibly responsible for some of the peaks/ringings?
On the waterfall, am I correct in thinking that 45-50 db is a good floor, or should it be lower, say 35 db?
At what point do we consider signals low enough to ignore?
The bode shows peaks around 39, 97, and 253 Hz...waterfall shows ringing around slightly different frequencies... is one more significant than the other?
I know I'm losing low F resolution in favor of the full range scans here, but I'm doing that for a reason, what do we think about the HighF pix?
So again, I havenot calibrated the SPL, as the friend with whom I share equipment owns the SPL and isn't around right now to borrow from, but the shapes of the curves are still valid, aren't they?
How's this impulse response look?
I'm running Vista Home Premium, with SB Live! 24bit External, and sometimes when I went ot check levels in the measurement window prior to a scan, the output VU went all the way to 0 db, but no sound came out... I had to play around with the settings, never sure quite what did the trick, seemed to be something differet each time, but eventually I would get it working, and run a quick scan, and then when I wanted to scan again I'd have ot go through the same rigormorall sometimes, but not others, any known-stupid-things I might be doing wrong?
Lastly, at least for now, I'm interested in hearing all kinds of thoughts on line out vs digital out from the SB LIVE, as I plan to do some playing around... I've heard ALL digital outputs from sound cards stink, but I need to see for myself, anyone ever do a cal file on the SB using this output? Otherwise I'm going to have to get clever in isolating it's effects vs the effects of whatever spdif input I use etc...

[brucek]

So again, I havenot calibrated the SPL, as the friend with whom I share equipment owns the SPL and isn't around right now to borrow from, but the shapes of the curves are still valid, aren't they?
How's this impulse response look?

When you say you haven't calibrated the SPL, I assume you mean that since you're using an ECM8000, you don't have a means to establish a known SPL level of 75dBSPL at the listening position, and as such can't match the REW SPL meter to that know level? Not a big deal as long as you feel that your level set at the listening position for taking measures is reasonable and then you set the REW to 75dB anyway to make REW believe it's calibrated.

I also assume you did a proper soundcard calibration and stored that file in REW for use and that you downloaded and stored the ECM8000 cal file from the site into REW....

Anyway, the impulse looks fine and so does the measurement.

You have a few low frequency peaks as you've already determined. You could address the one at ~40Hz with EQ or placement and the other two could helped perhaps with placement. You might try the phase control on the sub to see if it's causing the problem at ~100Hz.

The waterfall looks quite good in fact. You can see the likely room mode ringing out at ~40Hz.

Forget about using the digital out while running REW. There is no way to loop back to create a soundcard cal file. Use analog only.

Yeah, we have reports of quite a few member with Vista problems. I guess it takes a bit of messing around..

brucek

[glaufman]

Wow, that was fast, thanks Bruce! You've pretty much echoed my opinions on the scans, I'm not sur eif I'm elated or disappointed at how not-too-bad they look, considering:
a) I like playing around
b) this was an input I never use
c) I never set up the eq's on this input
d) it's analog which I never use
e) etc...

So... yes, you're right on the SPL cal, which was a little frustrating (noone's fault but my own) in that once I set to what my ears told me was 75db, the check levels said it wasn't high enough, so I had to tweak values, and then went back and reset to 75db, probably not necessary, and probably worthless, if you're not going to cal to a real standard, why do it at all, but I did it, and I slept last night...

On the sub phase, I must've forgotten to mention, I have no sub, these scans were just the two towers... hmmm. I wonder what they look like separately...

Yes, I did a proper cal for the card and downloaded the ECM cal, I took the curves out of the pix for clarity but could repost if you're interested...

So what would a "bad" impulse look like, would it simply decay slower? Or would it have muktiple peaks?

If you say the waterfall looks quite good, maybe I'm misinterpreting it, I thought it looked like it needed a lot of work below 100 Hz... and then again between 180-220Hz... I know you're better off not Eq'ing every little peak and dip, but what do we look for? when you say ~40Hz, I assume we're talking the same one that I meaure at 33Hz... we're not concerned with 68Hz, 77Hz, or 100Hz? Why? If htis is quite good, what does a mediocre one look like? How about a poort one?

Reason I want to use the SPDIF out form the sound card is I use the SPDIF inputs on the AVR... Noone's figured out how to do a cal file yet? Noone's even faked one?

As for Vista, I don't mind struggling if it's a known issue, as long as I know it's not something stupid I'm doing...

[brucek]

So what would a "bad" impulse look like, would it simply decay slower? Or would it have muktiple peaks?

You have a good noise floor at -75dbFS and not a lot of distortion with a smooth decay that isn't too long..

I have no sub, these scans were just the two towers... hmmm. I wonder what they look like separately...

Just be sure to have the receiver in stereo mode and don't turn on any soundfields or effects. I usually use both mains as you do...

I thought it looked like it needed a lot of work below 100 Hz... and then again between 180-220Hz... I know you're better off not Eq'ing every little peak and dip, but what do we look for? when you say ~40Hz, I assume we're talking the same one that I meaure at 33Hz... we're not concerned with 68Hz, 77Hz, or 100Hz? Why? If htis is quite good, what does a mediocre one look like? How about a poort one?

You have a 300ms scale and mostly the signal has dropped away by that time - that's good. Actually I like the horizontal scale to be set to LOG rather than LIN as it then matches the response and you can run the slider slices from 1 to 30 and see how the signal drops away on each slice. You can see likely the room mode ~35-40Hs area that is still ringing after 300ms - it's still around 60dB at that time, so it wouldn't hurt to eq or use placement to try and reduce it a bit. As you say, it wouldn't hurt to try and lower the peak at ~100Hz also...

Reason I want to use the SPDIF out form the sound card is I use the SPDIF inputs on the AVR... Noone's figured out how to do a cal file yet? Noone's even faked one?

Well, for the amount of time using REW, it's just easier and more accurate to simply hook up the system in analog.

brucek

[glaufman]

Sure, it's easier to hook up the system in analog, but I assume it's improper to "assume" that the analog input has a similar characteristic to the spdif, see I'm hoping to use this to at least look at my system response, not just the room response...

[glaufman]

BTW, I need a gut check on something... when I did the cal for the sound card, I included the Eurorack in the loop... sound card output went to input on the mixer, with the mixer's output going back to the sound card's input... I left the phantom power off on the mixer as I was not using the mike.... anything inherently wrong with this?
One other thing, rather than use the same cable with an adapter, I used similar but different cables, i.e., when doing the cal I took a mini to phono cable from the card to the mixer, but when taking the scan I used a mini to RCA cable, same both production made, both Y type (splitting the tip/ring) ,and both the same manufacturer... any issues, other than the RCA cable was abit longer than the phono cable?

[glaufman]

Here's the waterfall with the log scale...
waterfall log scale.jpg
And here's the same one moved upwards to a 40db floor to show some more decay... most of this isn't significant because it's below 48 db or so? (excepting the obvious ones that aren't....) Is 48db a fair cutoff there?

waterfall log scale 40db floor.jpg

[brucek]

but I assume it's improper to "assume" that the analog input has a similar characteristic to the spdif, see I'm hoping to use this to at least look at my system response, not just the room response...

The goal of course, is to remove as much influence from the test equipment as possible. That's standard practice. Unless you have a digital out and a digital in on your soundcard that a calibration file can be created, then it will never be as accurate as the analog out and in. I'm not saying you shouldn't use the digital. It just won't be as accurate as the analog in this situation.

when I did the cal for the sound card, I included the Eurorack in the loop

Yep, that will remove inaccuracies - most people do this. It doesn't completely remove the first mic stage from the measurement, but it's a big help because it allows compensation of the line output amp and any response problems with the three band equalizer in the 802 mixer. I certainly use a cal file that has the 802 included.

any issues, other than the RCA cable was abit longer than the phono cable?

No issues here. The cables aren't long enough to consider a problem....

Is 48db a fair cutoff there?

For sure. Go ahead and play a 30Hz 48dB signal at the listening position. Can you hear it? No.....

brucek

[glaufman]

Ok, I felt the need to prove your point, WOW, the scans through the spdif area mess!... I've posted the files below so other people can see... I suppose a cal file COULD be made from these, but I doubt it would be any good...
What I did was, for argument sake, was feed the analog line out from the sound card to the receiver's analog input, and take the receiver's preamp output back to the line in on the sound card, pretty straightforward, here's what it looked like:
bode, soundcard line to rcvr preamp to soundcard line.jpg
then I took the coaxial spdif output from the sound card and fed it to the same (but spdif) input on the receiver, again taking the receiver's preamp output back to the line in on the sound card... i got similar results using the optical out from the sound card, so I wont' post that picture...
bode, soundcard spdif to rcvr preamp to soundcard line.jpg
QUESTION: is this what you'd expect to see, or is this showing a feedback loop?

[brucek]

I felt the need to prove your point,

I'm not sure I understand what you're attempting to prove here? Forgetting about the fact that the digital out appears to be creating an oscillation on loopback, you can see that the low frequency response of the analog and digital (when compared are different - no surprise). This results in inaccuracies. Not huge ones, but it's there.

As I said before, you don't need to prove anything here. The digital will not be as accurate as the analog, since the analog allows for a proper calibration file. If you can live with small inaccuracies, go ahead and use the digital output. No big deal.

burcek

[glaufman]

I guess what I'm trying to determine is, feedback loop aside if I can get rid of it, the only reason we say the digital doesn't allow a calibration file is there's no way to feed it back in without using another device, which then wouldn't be present when we take the scans?

[brucek]

the only reason we say the digital doesn't allow a calibration file is there's no way to feed it back in without using another device, which then wouldn't be present when we take the scans?

Correct...................

[glaufman]

OK, that kinda makes sense, except, (and pardon me because I havne't fully thought this through...)

If the goal is to see room acoustics, then feeding the spdif into the avr, and feeding back the preamp out to the sc as a cal loop takes part of the AVr out of the equation... basically it takes the errors in the AVRs DAC out of the equation, so after cal adjustment, the measurements you see are TRUER representations of the ACOUSTIC response of the room, no?

[brucek]

If the goal is to see room acoustics, then feeding the spdif into the avr, and feeding back the preamp out to the sc as a cal loop takes part of the AVr out of the equation...

No, since the AVR (and its associated response anomolies) is part of the 'unit under test' you have to exclude it from the 'test equipment'. The goal of a calibration file is to make the test equipment as neutral as possible - not the equipment you're testing and using. Then when you test the 'devices under test', they are compensated with the equalization filters.

brucek

[glaufman]

Sure but by the same token, would you then advise to always use the analog inputs in practice if that's what we "have" to use during testing?

[brucek]

would you then advise to always use the analog inputs in practice if that's what we "have" to use during testing?

It's the soundcards digital out versus analog out that is in question. We don't use the soundcard when we play the system normally.

Computer soundcards are often quite poor and so we want to find a way to reduce the variables that enter into the test equipment. In this case using analog out and in offers the best chance to that end. You can't really hope after that to compensate for every type of input used on your system.

brucek