REW repeatability?

**robertcharles123** · **10-01-09, 02:26 AM**

Quote:

elee532 wrote:

I gave this a try on another computer... 0 to 200Hz sweep using 128, 256k, 512k, and 1M sweep lengths (graph 1, seperated traces is turned on). Graph 2 shows 0 to 20,000Hz sweeps at 128k and 1M.

Computer 1 is my HTPC... 2.5ghz dual core, Gigabyte GA-E7AUM-DS2H motherboard using the onboard Realtek Digitial Input and Output.

Computer 2 is a Lenovo ThinkPad X61T, Core2Duo 1.6Ghz, using USB Soundblaster X-Fi.

As you recall, my initial question surfaced when my results from a 0 to 200Hz sweep were different from the results from a 0 to 20,000Hz sweep between 0 and 20Hz (graph 1 below for reference).

Anyway, comparing these with the earlier readings, it looks the computer was the source of the difference and not the sub or pre-pro. Is my understanding correct? (I suppose if I was really motivated, I'd connect the X-Fi to Computer 1... just don't have the time or energy right now.). Any thoughts on why this might be the case?

Thanks!

This is a perfect example of what I have been saying for quite some time. I just do not trust all of the measurements I see from some of the subs being tested. I just cannot beleive that at 75 dB we can get so much room gain from some of the systems being tested. I would think that most systems would somewhat resemble the results of the modeling software- within reason. Some room gain should be expected, but for the graphs to roll up below tuning is hard for me to beleive. That is why I keep asking about repeatability with different computers and equipment. I am super happy you got some good data. Repeatability is my biggest problem with this setup. I am getting close with my REW setup, but I want to be able to judge my results against another independent tester. That was the purpose of my GTG, but little interest has been given to the idea. Maybe at a different time of the year.

Good job finding out about the measuring problem.

Thanks,

Robert

**Wayne A. Pflughaupt** · **10-01-09, 08:59 AM**

Quote:

I just cannot beleive that at 75 dB we can get so much room gain from some of the systems being tested.

SPL level has no correlation with room gain, but I'm intrigued by your issues with repeatability. Could you elaborate?

Regards,
Wayne

**robertcharles123** · **10-01-09, 09:36 AM**

Well, I feel that if you measure the driver and the spl of the signal- at the driver is 75dB, you are going to get a different reading concerning room gain than if you measure listening position of 75dB. It has to make a difference. My room is 4800 cf. When I put the mic at 1/2" from the cone and point the mic directly at the cone, I get completely different graphs. I am, at that point measuring the driver itself. As I back the mic up so it won't hit the cone, my graphs change shape and then I start getting room interactions, but nothing like a slope going up below the tuning. For instance, At the GTG, a few months back, we used 2 different microphones at 2 different locations and got pretty much the same looking measurements. The room was causing some differences, but a roll off at 20- 25 Hz measure at 3 feet for one particular sub did not result in a graph at the opposite wall measure showing a rise beyond the tuning. The graphs just showed less rolloff per octave. In the graph from the previous post, I would almost certainly think that the measures were in error due to the fact that they look so different and are from the same position.

I need to go, but I need to get a handle on this, so please bare with me.

Thanks,

Robert

**robertcharles123** · **10-01-09, 02:27 PM**

This was my first measure at listening position. I know it is not really scaled correctly, but disregarding that, if I accepted the response measure as "good", I would be seeing extension down to 2Hz. I did not believe this, so I kept measuring and checking the adjustments and calibrations.

Now my graphs are different again, but very closely resemble what was modeled when designing the system.

Here are one driver and top and bottom passives measured 1/2 inch from the cones. My newest graphs at listening position are resembling this general shape-give or take, but no upslope below the cabinet tuning. In order for me to believe that that can occur, I would have to put my sub in the room where this occurs and use it against the sub that is tested there now and run multipleREW setups on different computers to see if it is a computing error or measuring error or actual reality. I have yet to experience it myself and I wat to do just that.

Any better? I want to understand How this works and do it correctly. I want accurate and repeatable results that can be done by anyone and the result remains the same.

Thanks,
'
Robert

This is a measure of my subs at listening position-basically in the center of the room.

I could and can't repeat this measure now, although I was getting this measure or something like it for quite some time.

**Wayne A. Pflughaupt** · **10-01-09, 05:26 PM**

Quote:

robertcharles123 wrote:

Well, I feel that if you measure the driver and the spl of the signal- at the driver is 75dB, you are going to get a different reading concerning room gain than if you measure listening position of 75dB. It has to make a difference.

Well sure, close-mice measurements remove most of the influence of room gain.

Quote:

For instance, At the GTG, a few months back, we used 2 different microphones at 2 different locations and got pretty much the same looking measurements. The room was causing some differences, but a roll off at 20- 25 Hz measure at 3 feet for one particular sub did not result in a graph at the opposite wall measure showing a rise beyond the tuning. The graphs just showed less rolloff per octave. In the graph from the previous post, I would almost certainly think that the measures were in error due to the fact that they look so different and are from the same position.

Certainly, and elee532 told us he had found that his computer was the source of the problem.

Regards,
Wayne

**Wayne A. Pflughaupt** · **10-01-09, 05:35 PM**

Quote:

robertcharles123 wrote:

This was my first measure at listening position. I know it is not really scaled correctly, but disregarding that, if I accepted the response measure as "good", I would be seeing extension down to 2Hz. I did not believe this, so I kept measuring and checking the adjustments and calibrations.

Due to the extremely low level (30-40 dB??) this graph would be essentially useless – mostly noise from the system and room background noise. Plus, I don’t think anyone would say the Radio Shack meter (if that was what you were using) is going to be accurate much below 20 Hz.

Quote:

Now my graphs are different again, but very closely resemble what was modeled when designing the system.

Here are one driver and top and bottom passives measured 1/2 inch from the cones. My newest graphs at listening position are resembling this general shape-give or take, but no upslope below the cabinet tuning. In order for me to believe that that can occur, I would have to put my sub in the room where this occurs and use it against the sub that is tested there now and run multipleREW setups on different computers to see if it is a computing error or measuring error or actual reality. I have yet to experience it myself and I wat to do just that.

I’m afraid you lost me as to exactly what the various measurements in that graph were…

Quote:

This is a measure of my subs at listening position-basically in the center of the room.

I could and can't repeat this measure now, although I was getting this measure or something like it for quite some time.

What do your current graphs look like?

Regards,
Wayne

**robertcharles123** · **10-01-09, 08:05 PM**

Quote:

Wayne A. Pflughaupt wrote:

Due to the extremely low level (30-40 dB??) this graph would be essentially useless – mostly noise from the system and room background noise. Plus, I don’t think anyone would say the Radio Shack meter (if that was what you were using) is going to be accurate much below 20 Hz.

I’m afraid you lost me as to exactly what the various measurements in that graph were…

What do your current graphs look like?

Regards,
Wayne

The graph with the three measures are the driver and the upper and lower passive 1/2" from the cone.

I will get connected tonight and run some stuff and see if I can gleen some knowledge from a few guys here to get some descent results.

**robertcharles123** · **10-01-09, 08:27 PM**

I am using a cm 140 not a radioshack meter

**DrWho** · **10-01-09, 09:00 PM**

Quote:

robertcharles123 wrote:

This is a perfect example of what I have been saying for quite some time. I just do not trust all of the measurements I see from some of the subs being tested

Dynamic range (the volume difference between the loudest signal and the noise floor) of the measuring system can play a huge role when measuring low frequencies. If you look at all the calibration files, there's a lot of internal digital gain to compensate for the natural roll off of electronics and of the measurement microphones being used. While this gain is trying to offset what the system is doing, it is also going to increase the low frequency content of the noise floor of the system. So if your measurement's output is near the noise floor of the measurement rig, then you're going to see energy at the low frequencies that didn't come from the speaker itself.

This is of course assuming that all of the calibration was done correctly. I personally prefer to run without calibration files (although I'll post my results with the calibration files in place) because I can learn to recognize the behavior of the microphone and distinguish it from the behavior of the system I'm trying to measure.

Also, the purpose of the frequency response graph is not to tell you that your system sounds good. There are all sorts of ways to have a system that measures flat, but that doesn't mean all those ways are going to sound good. It's also interesting to observe just how much low frequency extension is overstated in its effects on perceived sound quality.

And lastly, the room gain of every room is very different and is a function of the room volume, how rigid the walls are and how well sealed the room is.

**robertcharles123** · **10-02-09, 01:02 AM**

Quote:

Wayne A. Pflughaupt wrote:

Due to the extremely low level (30-40 dB??) this graph would be essentially useless – mostly noise from the system and room background noise. Plus, I don’t think anyone would say the Radio Shack meter (if that was what you were using) is going to be accurate much below 20 Hz.

I’m afraid you lost me as to exactly what the various measurements in that graph were…

What do your current graphs look like?

Regards,
Wayne

Red line is listening position with no mic cal file.

Blue line is listening position with mic cal file

I am running my cabs through a crown ITech 8000. It has a built in HPF at 8 Hz. It is not defeatable. I am getting a couple of FACE amps soon, and I will measure with them and see if any changes occur.

Any thoughts?

Here are my LFE cabs.

**Ricci** · **10-02-09, 10:50 AM**

Quote:

DrWho wrote:

Dynamic range (the volume difference between the loudest signal and the noise floor) of the measuring system can play a huge role when measuring low frequencies. If you look at all the calibration files, there's a lot of internal digital gain to compensate for the natural roll off of electronics and of the measurement microphones being used. While this gain is trying to offset what the system is doing, it is also going to increase the low frequency content of the noise floor of the system. So if your measurement's output is near the noise floor of the measurement rig, then you're going to see energy at the low frequencies that didn't come from the speaker itself..

Great post.

I've been trying to educate people on this, but I don't think I've been able to word it so succinctly.

20db of calibration for your mic/meter plus another couple of db from your soundcard added to 45 or 50db of low frequency background noise = Voila! 75-80db of apparent <15hz response and..."Wow my ported 150w 12" White van sub is really digging deep!"

**DrWho** · **10-02-09, 11:59 AM**

Quote:

robertcharles123 wrote:

I am running my cabs through a crown ITech 8000. It has a built in HPF at 8 Hz. It is not defeatable. I am getting a couple of FACE amps soon, and I will measure with them and see if any changes occur.

Any thoughts?

You are 10dB down by 34Hz...I wouldn't worry about the 8Hz highpass right now.

**Wayne A. Pflughaupt** · **10-02-09, 12:19 PM**

Quote:

Ricci wrote:

Great post.

I've been trying to educate people on this, but I don't think I've been able to word it so succinctly.

Yes indeed, that was an excellent post. Filed away for future reference.

Quote:

robertcharles123 wrote:

Any thoughts?

Only that I’m confused about the repeatability question. Are you telling us that these readings were taken in the same location, with the same sub? I would have a hard time believing that...

Regards,
Wayne

**Ricci** · **10-02-09, 02:54 PM**

Robert,

How about a little more background on these measurements? Just the LMS subs, or with mains, or other JBL subs? What crossover freq? Any EQ enabled in those Itech's or in your processor? Audyssey? What microphone do you have? I'd highly suggest getting one of the calibrated ones available here from Cross Spectrum labs if you haven't already for about $90. What sound card?

The 35hz and over response is not too bad with just the 90hz dip. The action in the 16-32hz octave is bad though and this is where those big subs should be at their best. You drop some 20db from 38hz to 28hz and then staying at 10db or more lower overall level than the 80-35hz range. This is a strange variation in response to me. Almost like you have some cancellation, or EQ going on.

**robertcharles123** · **10-02-09, 07:53 PM**

Those measures are taken at listening position with the soundcard plugged directly into the amplifier. Nothin else in the mix at that point- just the Itech8000 to the cabs.

What do you think could be the problem?

**robertcharles123** · **10-02-09, 07:54 PM**

No crossover- just the 0-200 sweep into the amp.

**robertcharles123** · **10-02-09, 07:57 PM**

I honestly do not think my computer is powerfl enough to handle the setup. It seems like it glitches at times when it is sweeping. That is the reason why I want to use another person's computer who has good repeatable measurements in order to compare graphs. I will say, regardless what these graphs say at this point, I have never heard or felt better bass in a home environment or theater for that matter, so I am satisfied either way. I just want to make sure I am getting the most out of the setup and learn to use these programs correctly.

Any ideas?

**robertcharles123** · **10-02-09, 08:03 PM**

Quote:

Wayne A. Pflughaupt wrote:

Yes indeed, that was an excellent post. Filed away for future reference.

Only that I’m confused about the repeatability question. Are you telling us that these readings were taken in the same location, with the same sub? I would have a hard time believing that...

Regards,
Wayne

That is what I have been trying to tell you. I got some really good measures that were relartively flat at one time, only to lose the use of the program to a virus and now you see what I get. I am saying that all of those graphs are from the same spot where my head would be. the mic is on a stand and held in the same spot throughout.

Any thoughts as what to do to get the measurements to be reasonable and believable?

**robertcharles123** · **10-02-09, 08:14 PM**

This is a graph of listening position the night I got REW working again. This was about 2 weeks ago.

**Wayne A. Pflughaupt** · **10-02-09, 08:17 PM**

The only time I've seen response differences that drastic has been when people moved the sub to a different location. If everything is sounding good to you, I'd sure be questioning what your getting from REW. You might download some sine waves from the BFD Guide for 20 Hz and 40 Hz. If you register a 10-12 dB difference between them that would confirm whether or not REW is in the ball park.

Regards,
Wayne

**robertcharles123** · **10-02-09, 09:24 PM**

Cool. There really is no moving my cabinets. They require lots of planning in order to get from one point to another. I was thinking of moving them a little, but other things are going to be in the way, so there they stay. I will try what you said. I might need a little coaching. Thanks.

Thanks,

Robert

**DrWho** · **10-02-09, 09:58 PM**

Try turning off your wi-fi when taking measurements...or anything else that does automatic stuff in the background.

**robertcharles123** · **10-02-09, 10:13 PM**

I turned off the internet and the connection last night for the last measures. Funny you say that about the noise. I had a lot of noticeable noise when the rca cable was plugged in, I took it out and used another, I am going to have to use a good plug and see what the story turns out to be. I will try that later on.

**robertcharles123** · **10-02-09, 10:44 PM**

So I should play a 20 Hz sine wave and get a SPL.

Then, measure a 40 Hz sine wave and get a SPL.

No changes should be made in the setup from measure to measure-whatsoever?

I will do this.

What about the tone generator in the REW?

**robertcharles123** · **10-03-09, 12:47 AM**

Quote:

DrWho wrote:

Try turning off your wi-fi when taking measurements...or anything else that does automatic stuff in the background.

You know, I bought the crowns as b stock. They might have a filter turned on somewhere, I just do not know how to tell. I am going to run another amp to the cabs and see what happens. Sooooon!

Advertisement