Home Theater Forum and Systems banner

1 - 17 of 17 Posts

·
Registered
Joined
·
509 Posts
Discussion Starter #1
hi all

again getting wacky results, this on top of the funny stuff last time with multiple sweeps.

Just dialing in the new subs, and to my pleasure found that I had them nice and flat.

I was measuring at the 256 k rate, and decided to check with the 1M rate.

As you will see, when ending with 20 000, the 128 and 256 rates are pretty well identical, the 500 drops quite a bit and the 1M got an error message of to low a reading!

20 000 end at various rates.jpg

As I was only looking at the relative levels of the sub and mains, I also did sweeps with an end point of 800 hz, and with this one the 128, 256 and 512 were all pretty identical, only the 1M being off.

800 end at various rates.jpg

I went woo hoo with the first measurments, and then boo hoo with the others ha ha.

Any ideas??
 

·
Premium Member
Joined
·
6,711 Posts
Did the sweeps sound continuous, i.e. no clicks/gaps in them? If you look at the scope trace does the captured sweep look continuous or does it have gaps or discontinuities in it? On some PCs captured audio can have clicks/pops/noises from other activity (wireless LAN cards are a prime cause, or other applications placing brief heavy demands on the processor), the longer the sweep the greater the chance of ending up with some corruption in the captured audio if that problem exists. The same would apply to multiple sweeps, one or more of the sweeps might suffer corruption/interference from whatever is causing the problem. In those cases single short sweeps would give more reliable results, but best is to try and track down the cause and remove it (e.g. disabling a wireless card, disconnecting from the internet while measuring, not running other applications etc).
 

·
Registered
Joined
·
509 Posts
Discussion Starter #3
Hi John

no obvious discontinuities in the sound of the sweep, and no clicks or pops.

Umm, the sweeps on the scope all look the same to me, I ran thru all the settings (the first three were very consistent, 1M again was 'off', only below 200 hz BTW) and I clicked on scope and whatever it was that came up first does not seem to change even if I click on a different measurement, and also unlike say the waterfall where it needs to be generated, this one stays the same??? Am I overlooking something?

Anyway. maybe they don't look the same to you so I'll include both, and I changed the axis limits to show the whole sweep. hope that is what you wanted.

I only took a measurement up to 2000 hz in these cases.


the scope sweep at 128k
scope sweep of 128k.jpg

the scope sweep at 1M

And lastly, to show what the scopes are for, the all measured for all options, I cut the graph off low as after 200 hz they all merge again.

multiple rates for john m.jpg

EDIT
for some reason they got out of order, but the first is 128 and the one on the bottom is 1M
 

Attachments

·
Registered
Joined
·
7,514 Posts
Umm, the sweeps on the scope all look the same to me, I ran thru all the settings (the first three were very consistent
The scope signal updates after every sweep. You can't take a bunch of measurements and then look at each one's scope. You would have to look at the scope after every sweep since only the last measurement is shown.

It is supposed to show the sweep signal normalised to 100% of the signal (100%dBFS) and then overlayed on that is your measured signal. Your signal should be completely inside the sweep envelope to show no clipping.

My question is......where is your sweep signal?

It should look like this: See the sweep and the measurement overlayed with it.....



brucek
 

·
Registered
Joined
·
509 Posts
Discussion Starter #5
Hi Bruce

well, I guess that means I'm not a complete idiot, they do look the same ha ha.

I unchecked the sweep box, cause I thought John only wanted to see the measurement. Sorry.

And if it only displays the last measurement taken, then it was neither the 1M or the 128, it was the 500 !! just to ensure everyone is well and truly confused.

Just re-did the 1M sweep, here it is.

1m after bruce.jpg

Did you want anything else??
 

·
Premium Member
Joined
·
6,711 Posts
The noise level looks very high in those scope plots, were the measurements in a very noisy environment? To more easily see whether all is working correctly with the measurement process use a loopback connection rather than making measurements via your speakers and microphone, much easier to spot something amiss that way.
 

·
Registered
Joined
·
509 Posts
Discussion Starter #7
The noise level looks very high in those scope plots, were the measurements in a very noisy environment? To more easily see whether all is working correctly with the measurement process use a loopback connection rather than making measurements via your speakers and microphone, much easier to spot something amiss that way.
You got it in one John, done at night (freezing here so fires blazing and many fans in a vain attempt to warm the place up).

I can see the difference myself in the scope when i ran it today when quiet.

2000 hz end, 1M

1m no noise.jpg

As it has cleaned up i didn't do the loopback, can if you wish. (that is the output to the input yeah?)

As a sidecheck, the 1M results are still different from the other three, so no need to repost them.

Anything else you'd like me to do?
 

·
Premium Member
Joined
·
6,711 Posts
It would be interesting to see the Impulse Response plots for the 1M sweep and the 256k sweep.
 

·
Registered
Joined
·
509 Posts
Discussion Starter #9
happy to oblige John, especially as it's night and I don't need to re-run the sweeps in a noisy environment ha ha.

Even I seem to see major differences of a qualitative nature between the 1m and the others, could be imagining things though of course.

128

impulse response 128.jpg

256

impulse response 256.jpg

512

impulse response 512.jpg

1M

impulse response 1m.jpg
 

·
Registered
Joined
·
509 Posts
Discussion Starter #11
You guys have to treat me as a dumb machine, it can only do what it is told as it is too stupid to know any better!!

128

db fs 128.jpg


256

db fs 256.jpg

512

db fs 512.jpg

1M

db fs 1m.jpg

I hope you guys are only 'chasing' this for your own information or that it is important somewhere, I'm personally quite happy to use the shorter measurement lengths..unless this is showing somehow that on my computer they are untrustworthy - then I AM interested ha ha.

I only posted because I thought it might be showing up a bug or something, if I can do fine on the shorter measurements cool, but I'm more than happy to help you chase it down if it is indeed important.

Does no-one else get these anomalies??
 

·
Registered
Joined
·
7,514 Posts
I hope you guys are only 'chasing' this for your own information or that it is important somewhere
I think it's important. It speaks to the consistancy of REW, that is only in question when there is some sort of problem. I would like to know what your problem is?

Your waveforms all show an unusual amount of noise.

The impulse responses appear different for each measure. I wonder if you have an unusually small amount of RAM in your computer. The longer the sweep time, the more RAM you'd require.

When you increase sweep time, the results should be identical except that the noise level should decrease about 3dB each time.

Below are graphs of my cheap sub connected to the computer I'm using right now. It will provide you with a bit of a reference of an average system. Actually my really old 350Mhz K6 that I use for REW normally with its old soundblaster card is much quieter than this system.

I have overlayed the responses of the 4 different sweep times.

See how the 4 responses are identical in the response plot. In fact they're so close it's hard to see that I am overlaying 4 measures.

Also see how the 4 overlayed impulse responses are near identical except that for each higher sweep time, the noise reduces. The sweep times from low to high were blue, purple, green, yellow. See how the noise for yellow is the lowest and blue is the highest. The noise does indeed drop about 3dB per measure.

I also include the scope of the 128 measure to show the noise level. Remember how high yours showed.

Can you determine first why there is so much noise in your system? You almost have no signal above the noise.

all response.jpg
blue=128, purple=256, green=512, yellow=1M
all impulse.jpg

scope.jpg

brucek
 

·
Premium Member
Joined
·
6,711 Posts
Terry,

The plots show a lot of noise but also look to have quite high levels of distortion. Might be interesting to try playing sine waves through the system at the same level as you are using for the sweep and see what the 2nd and 3rd harmonic distortion levels are like.
 

·
Registered
Joined
·
509 Posts
Discussion Starter #14
Your waveforms all show an unusual amount of noise.
Yeah, I can see that from the examples you posted. Mine in comparison hardly stick their little heads up.

Hmm, at first when john mentioned it I thought of noisy environment so did it again when much quieter, now I'm very strongly beginning to suspect some sort of internal computer noise, trouble is I'm an idiot when it comes to computers. The signal certainly gets through the speakers easily enough.

Just to maybe help narrow it down, I use an ECM 8000 with the shack generic file, running into a xenyx 802. The soundcard is creative soundblaster, what model i don't know.

The impulse responses appear different for each measure. I wonder if you have an unusually small amount of RAM in your computer. The longer the sweep time, the more RAM you'd require.
The computer is about three years old now, and was built with the help of a friend who knew his stuff (I assume ha ha) and was pretty 'modern' at the time. Tried finding how much RAM I have, thought I might get it by right clicking on the desktop but couldn't find the info, the stuff in the control panel didn't jump out at me as having the data either.



Can you determine first why there is so much noise in your system? You almost have no signal above the noise.

brucek
Just ran a few sweeps again in the day when no external noise, the obvious noise floor is still evident, and the first three sweeps were still consistent with the 1M the odd one out.

John (how do I quote from two different posts???), did the distortion test, was looking good till I got down low then aarrggh. Which to be frank is REALLY annoying as I've just spent quite a bit of work building two new subs (which to my extreme displeasure 'clack' all too easily, at levels I'd normally listen too when my PHL's don't miss a beat).

They are four peerless xls 10's per side in a sealed box, I just assumed that they would be up to the task, now I'm not so sure.

200 hz 1.69 <0.01

100 hz 1.43 <0.01

80 hz 1.5 <0.01

60 HZ 1,79 <0.01

50 HZ 2.0 <0.01

40 HZ 1.77 <0.01

30 HZ 2.8 <0.01

20 HZ 105!!!!!! <0.01

17 HZ 52!!!!!!! <0.01

i'M CERTAINLY NO EXPERT (sorry) but that can't be right can it??? and if the 2nd %'s are correct would it not also show in the 3rd??

I now know why the shack has included this fella!...:wits-end::wits-end::wits-end:
 

·
Registered
Joined
·
509 Posts
Discussion Starter #15
Ok, ran a few more. First off, I kept lowering the x-over point to the sub until the 'clacking' disappeared, I can at least add more to the mains to fill in any holes.

I had the subs crossed at 69 hz, when I got down to 40 hz I could have full volume and no clacking...which confuses me because shouldn't it be the LOW frequencies that cause it??

Re-did the distortion measurements with sub only (mains off)

40 hz 1.61 < .01

30 hz 2.05 <.01

27 hz 1.45 <.01

26 hz 12 <.01

25 hz 16 ,>01

20 hz 81 < .01

and 17 hz 43 <.01.

Maybe we suddenly hit the noise floor?? although it's an old house and rattles do occur.

Re-did some sweeps at the different rates and you wouldn't believe it they came out the same!

So here are the four sweeps, and I decided since the 1M was finally the same (I've made no changes other than measure sub only) I'd do a multiple (four) sweep at the 1M setting, and you will see it no doubt

sub only, plus multiple 1m sweep.jpg

And this is the impulse of the four 1M sweeps, and even I can see what is wrong with it, and yes I did get the message that the impulse peaks were not where they should be.

impulse multiple 1m sweep.jpg

hope this helps rather than further confuses you, one of us totally confused is enough for all of us!!
 

·
Premium Member
Joined
·
6,711 Posts
Very strange, some kind of soundcard buffering problem maybe? I'd say just stick to 256k sweeps :) On the sub distortion questions best head over to the DIY subwoofers forum where there are folk far more knowledgeable in that area.
 

·
Registered
Joined
·
509 Posts
Discussion Starter #17
can't help you with computer questions john, but i must say I do like the sound i'm getting so it can't all be bad.

All pretty odd eh?, but I'm fine on sticking with short measurements ha ha. Thanks for staying with me all the way, and of course bruce too.
 
1 - 17 of 17 Posts
Top