REW repeatability.

**Chrisbee** · **06-15-07, 04:25 AM**

Hi

I wondered about variability between tests when using REW.

I was seeing considerable differences between my curves particularly at the bottom end of the frequency range. I wondered what I could do to improve repeatability.

So just out of interest I kept repeating sets of 9 REW sweeps without changing anything.

Here is my IB calibrated to 75dB on the Galaxy 140 with 75dB target level in REW: Headroom average 25dB. REW warnings of low readings on each test.

Only 5-20Hz shown. Galaxy meter at listening position at ear height.

Below 10Hz the curves seem almost random with slight variations above.

Here I recalibrated to 80dB on the meter but without changing REW's 75dB target setting: Headroom average now 12.5dB.

Here I recalibrated the meter to 85dB leaving target level at 75dB. Headroom average now 9.2dB.

It seems if you want improved repeatability it pays to give REW enough signal to work with.

**JimP** · **06-15-07, 06:11 AM**

I wonder if this is background noise in the electronics?

What happens if you go lower with the SPL? does it scatter more?

**Chrisbee** · **06-15-07, 06:25 AM**

Quote:

JimP wrote:

I wonder if this is background noise in the electronics?

What happens if you go lower with the SPL? does it scatter more?

I suppose I ought to try it to find out.

At these very low frequencies perhaps atmospheric pressure plays a part in the signal to noise ratio?

It is also quite windy here today and the windows are open "on the catch" for ventilation. The IB enclosure is not airtight either.

This may all help to explain the increased scattering below 10Hz with lower SPLs.

I thought it fascinating how the bundle of curves tightens with increased signal.

**JohnM** · **06-15-07, 07:33 AM**

You can also try using longer sweeps, which should similarly improve consistency and help in rejecting low frequency noise. Headroom would ideally be around 12dB for room measurements.

**Chrisbee** · **06-15-07, 07:56 AM**

Thankyou for these ideas, John.

I'll have play around and see what happens.

It's probably all quite academic on real programme material.

Though it's remarkable to be able to see such close results at these extremely low frequencies. Probably much of what the meter is reading is harmonic distortion.

My attempts to push 10Hz too high <95dB(C) produced a clearly audible higher harmonic. I need more displacement!

**JohnM** · **06-15-07, 11:47 AM**

One of the characteristics of the log sweep measurement technique is that harmonics are easily rejected, REW excludes them from the displayed frequency response. You can get a feel for how much the harmonic distortion content is increasing by looking at the portion of the impulse response that lies before the main peak (i.e. at times < 0), the harmonics appear as scaled down versions of the main impulse response. You can also use the percentage 2nd and 3rd harmonic displays on the SPL meter when playing sine wave test tones with the signal generator.

**Chrisbee** · **06-15-07, 01:25 PM**

Quote:

JohnM wrote:

One of the characteristics of the log sweep measurement technique is that harmonics are easily rejected, REW excludes them from the displayed frequency response. You can get a feel for how much the harmonic distortion content is increasing by looking at the portion of the impulse response that lies before the main peak (i.e. at times < 0), the harmonics appear as scaled down versions of the main impulse response. You can also use the percentage 2nd and 3rd harmonic displays on the SPL meter when playing sine wave test tones with the signal generator.

Fascinating. Thankyou, John.

Here's my impulse response. There seems to be very little happening before the "0".

EDIT: Sorry, forgot about SPL meter under Tools in REW.

**Chrisbee** · **06-16-07, 12:13 PM**

Here are some interesting VLF curves: Sweep length now 1M. (Adjusted in details in Test box)
2-20Hz graph limits. 75dB REW target level. SPL meter reads 80dB(C)
Half the sweeps with and half without the C-weighting box ticked in meter settings.
The more horizontal curves below 6Hz are C-weighting. The Galaxy SPL meter cal file stops at 5Hz.
Amazing consistency above ~6.5Hz!
4 x 15" IB opposed driver manifold. Galaxy 140 meter, Fast, C-weighting, Mid SPL range.

I'll try the REW signal generator next for distortion measurements using the REW SPL function.

**Chrisbee** · **06-16-07, 12:47 PM**

REW Signal Generator using Frequency follows cursor function.

REW SPL function reading dB and 2nd harmonic distortion.

REW SPL function calibrated to 75dB.

Hz.....dB.....2nd%

50.....75......0.15
40.....75......0.2
30.....75......0.2
25.....75......1.2
20.....75......1.2
15.....74......0.5
10.....74......1.0
9.......69......1.5
8.......62......3.0
7.......60......5.5
6.......57......7.0
5.......40......90

Figures for 2nd harmonic distortion are rough averages.

3rd harmonic readings remained unchanged throughout.

These distortion figures look too low to my uneducated eye.

Perhaps I should up the levels?

**Chrisbee** · **06-16-07, 12:58 PM**

Front door slamming just as I hit "Measure".

C-weighting ticked.

**Ilkka** · **06-16-07, 01:47 PM**

Quote:

Chrisbee wrote:

Perhaps I should up the levels?

Yes, try for example constant 90 dB at every frequency. Also post 3rd HD figures.

**Chrisbee** · **06-16-07, 03:05 PM**

Quote:

Ilkka wrote:

Yes, try for example constant 90 dB at every frequency. Also post 3rd HD figures.

Hi Ilkka

I'll try again tomorrow.

**terry j** · **06-16-07, 10:25 PM**

wow, still finding out about the capabilities of REW!

So, we can see the distortion figures for the measurements as well?? Won't be doing any measurements real soon, but am I to take it we can simply read those distortion figures from the SPL meter box???

will be watching with interest.

**Chrisbee** · **06-17-07, 02:05 AM**

Quote:

terry j wrote:

wow, still finding out about the capabilities of REW!

So, we can see the distortion figures for the measurements as well?? Won't be doing any measurements real soon, but am I to take it we can simply read those distortion figures from the SPL meter box???

will be watching with interest.

I'll try not to disappoint.

**Chrisbee** · **06-17-07, 03:30 AM**

Here's the latest run using 90dB target level.

4 x 15" IB. REW generator sinewaves: F follows cursor

REW SPL meter readings: 2nd harmonic readings. WYSIWYG

Hz..SPL(REW).2nd%

50.....92.7......0.27
40.....93.5..... 0.4
30.....92.9..... 0.25
25.....93.4..... 0.4
20.....93.6..... 1.3
15.....91.6..... 0.4
10.....91.9..... 0.9
9..... 86.........1.2
8.......80.2......2.3
7.......77.........~3
6..... 74.7......~7
5.......52.......~200 *

*REW repeatedly locked up at 5Hz with loss of test tone.

I had to restart my computer 3 times.

**JohnM** · **06-17-07, 11:05 AM**

Quote:

Chrisbee wrote:

*REW repeatedly locked up at 5Hz with loss of test tone.

I had to restart my computer 3 times.

In "freq follows cursor" mode the generator should be limited to 10Hz minimum, as it is when setting the frequency directly. I'll fix that for the next release.

**Chrisbee** · **06-17-07, 12:35 PM**

Quote:

JohnM wrote:

In "freq follows cursor" mode the generator should be limited to 10Hz minimum, as it is when setting the frequency directly. I'll fix that for the next release.

Aha! Thankyou, John. I was pushing the boundaries a bit too much then.

**Chrisbee** · **06-18-07, 04:27 AM**

The deafening silence concerning the 90dB IB distortion figures prompted me to repeat the exercise at 95dB just to spice things up a bit.

95dB target level. 4 x 15" IB.Galaxy 140 SPL.
REW sinewave generator. REW SPL calibrated at 95dB.
2nd harmonic read directly from REW SPL function.
Distortion readings are roughly averaged as they fluctuate considerably during the test. 3rd harmonic did not register at all.

Hz.....SPLdB.....2nd%

50.....97.6.....0.33
40.....98.0.....0.41
30.....97.5.....0.46
25.....97.9.....0.48
20.....98.5.....1.0
15.....96.3.....0.5
10.....96.8.....2.0
_9.....91.4.....2.5
_8.....84.6.....2.5
_7.....81.4.....2.3
_6.....79.0.....4

I wonder whether I'm brave enough to try testing at 100dB?

**Chrisbee** · **06-22-07, 05:52 AM**

JohnM

Would it be possible to add a moving bar (or even a numerical indicator) so that REW could show the actual frequency (moment by moment) of the test sweep?

It was great fun listening to the very slow extended sweep (1M) climb from the subterranean depths but I had no idea which frequency was playing at any particular moment.

The slow sweep was a great way to discover previously unknown rattles in the doors, windows, walls, floors and objects in the room. The standard sweep is travelling a bit too fast to excite such rattles to a noticeable level. On programme material these rattles might well be excited. These effects might still be so short in duration that it would be difficult to use them as a diagnostic tool for finding unwanted noises in the HT or AV room. Once the annoying frequency was better pinpointed in the slow sweep the Frequency follows Cursor option could be used to close in on the rattle.

**JohnM** · **06-22-07, 02:33 PM**

Way ahead of you

Just leave the generator window open during the sweep, the frequency display is updated as the sweep progresses.

**Chrisbee** · **06-22-07, 02:43 PM**

Quote:

JohnM wrote:

Way ahead of you

Just leave the generator window open during the sweep, the frequency display is updated as the sweep progresses.

So it does! Excellent!

Thankyou.

**terry j** · **07-19-07, 12:27 AM**

just a quick 'tickle' on this one, I too would be interested in knowing what are good distortion figures and what are bad distortion figures for the bass.

I would assume that the distortion would tend to go up as we go down. And looks like a good/reasonable level to do the test at is around 85-90 db.

I noticed something interesting last night when I had a quick fiddle with it, will mention it when I investigate it a little more.

I do realise that Ilka has probably covered this in great and educational detail elsewhere, so please if someone could give a quick link to it I'd appreciate it.

thanks

Advertisement