Room measurement, what to learn?

[Rodney gold]

Its really impossible to say anything on how a measurement will sound on any micro scale.. yes you can point out gross abnormalities..but thats all, when it comes to the nitty gritty and small things that make a system, speaker and room come together and synergise..the best tools are your ears

 

[BE718]

Yes, they show where there are resonances. But it is the frequency spacing between *series* of such resonances that help distinguish room modes/effect from speaker effects. And my impression is that a linear frequency scale makes this easier to spot.

A speaker issue is not going to have a long "tail" and decay time, but yes the spacing will help confirm.

Also REW room simulator (or any of the countless mode calculators out there) will predict and provide a very good indication where the modes will be.

 

[nobeone]

But it is the frequency spacing between *series* of such resonances that help distinguish room modes/effect from speaker effects. And my impression is that a linear frequency scale makes this easier to spot.

I used to work for a company that made vibration monitoring equipment, a lot of the output of which are spectra showing the frequencies some machine is vibrating at. Spotting harmonics is important in "understanding" the spectra. I would agree it is easier to do by eye with a linear frequency axis. Vibration monitoring tool vendors and I would guess spectrum analysers provide various tools to help, does this REW software have a cursor to allow you to read off the magnitude at one chosen frequency? Harmonic cursors (i.e. not a single cursor line a one frequency, but a number of cursor lines at multiples of the frequency of the main cursor can make identifying harmonics much easier. Taking a Fourier Transform of a spectrum (cepstrum) can help too.

 

[Bemused]

Why is the decay time so long up in the extreme treble? I've not seen that before. Is it perhaps something non-system related being picked up e.g. a fluorescent light?

I'll stick a waterfall up on mine at some point once I've figured out how best to set the parameters - my 48Hz bump is pretty slow so needs some fitting in!

Yes I suspect it is a laptop problem, I used the same microphone as you with REW on Win7

 

[Bemused]

I find the differences between the two individual channels and the plot for both driven very interesting. I tried an average graph of L & R and it was very different to 'both'. I was expecting the additional 3db overall gain in the 'both' trace but not the increase in certain nulls.

Perhaps looking at the phase plots will show why the channels don't simply sum.
Its been some time since I switched on REW but seem to remember there was phase information.

 

[Tenson]

Apologies for not reading all the thread, but if Tony is interested in my initial thoughts on his data:

Firstly consider the speakers as operating together below about 1KHz and individually above this. So don't worry about the unevenness above 1KHz for the trace of both speakers together.

What does immediately stick out to me is the right speaker having a significant and broad boost above 7KHz. I can't really tell if it is the room or speaker but warrants some investigation.

Looking at the 'speakers together' trace below 150Hz the response is pretty boosted. You must have a good solid 'whack' on kick-drums

Everything else looks good - no serious nulls or peaks in the modal range and the midrange is very even for each left and right speaker.

 

[Tony L]

1) Look at plots with a *linear* frequency scale, not the usual log scale. In partticular, do this for the range from LF up to, say, a few kHz.

That makes no sense to me as music is logarithmic, e.g. 80Hz is an octave above 40Hz, 160Hz two octaves above. The frequencies translate to musical notes far easier this way and that is far easier for me to understand. Making it linear would compress all the musical information into the left of the chart and put the right half or so as cymbal sheen most folk over 50 can't hear anyway!

What does immediately stick out to me is the right speaker having a significant and broad boost above 7KHz. I can't really tell if it is the room or speaker but warrants some investigation.

I'm not sure at this point if that is the room or whether I need to slightly re-centre the compression drivers (something that is always off to some degree with vintage Tannoys IME). I notice REW has a real time analyser of some kind so I may have another go at some point now I've got a really good mic stand and can position the mic very accurately exactly on driver centre and a short distance away. The compression driver is attached with four bolts and needs to be carefully centred in the magnet gap. You can hear microns of difference here, tiny changes in bolt-tightness too! The other thing to note is as mentioned earlier I do have a slight hearing imbalance that I'll have to live with for the rest of my life so having just a little more top end on the right actually helps - to me the central image sounds rock solid and stable as-is and the right side does not sound brighter to me, i.e. I may actually screw it right up by fixing it! If I play a mono jazz album all instruments from the bass up to the cymbals sound nicely centred and stable, so it is not a huge issue. I'm certainly not hearing mono cymbals or higher trumpet notes pan to the right as one might expect from looking at the graph.

 

[Tony L]

Here' s a real close-up left and right measurement:

The mic is about 10cm from the grill cloth and aimed right down the horn at the compression driver. I set it so the feet of the mic stand were resting against the bottom of the cab so I could be sure of getting exactly the same position on each side. Even so I don't trust this as gospel and wouldn't base an actual compression driver realignment on it as I know the vintage Mullard tubes in my preamp aren't a perfectly matched pair despite being NOS and having the same date codes (right one is very slightly louder, which helps me), plus I'd not trust my original unmolested Tannoy crossovers to be anything like perfectly matched after 45 years or so! If I was going to embark on a compression driver realignment I'd use the same crossover and amp channel for both measurements.

Putting the accuracy of the above to one side there is a lot there I really do not understand, e.g. what the hell are those deep nulls about? The measurements were both taken one channel driven, so there is nothing arriving out of phase from the other speaker, and I'd have thought I was too close to the driver to get any real room effects.

PS I don't think the volume scale is accurate in this measurement, I had to drop the mic level in System Preferences to avoid clipping. I usually have the System Preferences mic slider up full, which as I understand it is how to use it with REW, but I'm really not sure to be honest. The sound in room was not loud at all, felt like about 70-75db.

 

[Tenson]

I guess the nulls are reflections from the horn throat and other diffraction bits. If so, the frequency of the nulls will move as the mic goes slightly off axis.

The broad boost seems to have diminished this close.

 

[Tony L]

Horn diffraction etc makes sense above 1Khz, but what is the stuff around 600Hz about. That is really odd. I don't really understand Tannoy time-alignment, though I guess like most speakers that may need some distance to get its ducks in order, so some nulls might be phase cancellation or something. I know the compression driver and bass unit are wired out of phase to compensate for the former sitting a fair bit behind the latter. The bass driver is the one out of absolute phase, i.e. moves back on a positive voltage. I guess the mic is far too close to get anything from the aperiodic/port loading.

 

[Tenson]

Given concentric alignment it shouldn't need any distance to get ducks in order. I expect the lower null is cabinet diffraction while the higher ones are horn throat. It's more obvious close-up.

I.e. don't worry about it. An RTA would reveal a lot I suspect as you can see what changes when you move the mic around.

This section about Power Cepstra might be somewhat interesting with the plots further down - https://books.google.co.uk/books?id... response tannoy&pg=PT491#v=onepage&q&f=false

 

[Tony L]

There is a known horn resonance somewhere that is dealt with a notch filter in the crossover, so I guess one of the two deep dips (2.6k or 4.2k) will be that.

 

[cooky1257]

Could the one around 600hz be a cavity resonance/ cancellation due to the fact the driver is front mounted?
I do recall commenting on LR balance being a bit off when I heard them but couldn't say I noticed one being brighter than t'other.

 

[davidsrsb]

600Hz is about 50cm wavelength, so a reflection path 25cm longer would give this notch. As you are on axis, there could be a strong reflection from the cabinet cut out. Moving the microphone away a few cm would change the notch frequency

 

[Bemused]

Here' s a real close-up left and right measurement:

Putting the accuracy of the above to one side there is a lot there I really do not understand, e.g. what the hell are those deep nulls about? The measurements were both taken one channel driven, so there is nothing arriving out of phase from the other speaker, and I'd have thought I was too close to the driver to get any real room effects.

Its caused by a reflection from the single speaker arriving back at the microphone latter (out of phase) than the primary signal.

PS I don't think the volume scale is accurate in this measurement, I had to drop the mic level in System Preferences to avoid clipping. I usually have the System Preferences mic slider up full, which as I understand it is how to use it with REW, but I'm really not sure to be honest. The sound in room was not loud at all, felt like about 70-75db.

You did load the microphone callibration file?

I recall that I had problems with gain of both microphone and REW output signal. At one point REW was so loud it fried my tweeters. I never really resolved the issues other than setting output with a voltmeter and cross checking mic gain with a sound meter.

I suspect there is far more going on here than we amateurs can comprehend.


Stefan (orangeheart) always advised doing a gated measurement first.

 

[Tony L]

Could the one around 600hz be a cavity resonance/ cancellation due to the fact the driver is front mounted?
I do recall commenting on LR balance being a bit off when I heard them but couldn't say I noticed one being brighter than t'other.

FWIW I'd term it rear-mounting, i.e. the driver is mounted from the rear to the rear of the baffle. May well be diffraction I guess, given the frequency it is to do with the bass driver not the horn anyway. Could possibly be a room cancellation I guess, even at that distance. I pretty sure the imbalance you heard was simply down to my turning the preamp down so we could talk and doing so inaccurately as the two individual (Croft style) volume knobs on the Verdier pre are fairly sensitive, a degree or two in movement is enough to shift the centre image.

You did load the microphone callibration file?

Yes, and for normal listening seat measurement it works, though again I don't think the db scale is accurate as it looks louder than it sounds. I need to look into this aspect of REW as I've no idea how one accurately sets the level so x db actually reads as such. Anyone got any tips (Apple Macbook Pro)? The mic level check button before measuring just seems to deal in -x from 0db, i.e. headroom, it doesn't give a db level that can be checked independently.

 

[cooky1257]

FWIW I'd term it rear-mounting, i.e. the driver is mounted from the rear to the rear of the baffle.

This one has tied me in knots for years, baffle pov or driver pov?

I need to look into this aspect of REW as I've no idea how one accurately sets the level so x db actually reads as such.

I nearly deafened myself running my first FR plot in REW.

 

[mudlark]

Mr Pedant here. Do call damped rooms semi-anechoic. There's no way you can make your floor very absorbent.

If you want a semi-anechoic situation use your garden, but only play loud stuff when the neighbours are out. In fact it would be very interesting for most people to play some stuff inside and do the same thing outside. The spl will be much less. what you miss outside is your room; no boundaries means RT60 is buggerall and you have a semi-anechoic lab and annoyed neighbours.

 

[Purité Audio]

A straightforward guide to REW,
http://audiosciencereview.com/forum...om-measurement-tutorial-for-dummies-part-1.4/

There is also a part 2.
Keith

 

[Tony L]

A straightforward guide to REW,
http://audiosciencereview.com/forum...om-measurement-tutorial-for-dummies-part-1.4/

There is also a part 2.

He says in big bold letters "Attempting to get accurate SPL levels is the biggest mistake users of REW make!". I'm sorry but I don't agree. If the vertical db scale is out so will be the severity of the dips and peaks, i.e. something that gives the impression on the chart as being plus or minus 5db at an average level of 85db, but was in reality measured at 65 or 70db may only be a 2db swing. Inaccurate measurements are arguably a lot less useful than no measurements at all. I'd certainly hate to hear a system using electronic room correction arrived at with inaccurate measurement data. Science is worth doing right! I want to learn how to calibrate this thing properly so it is a real measurement tool and I can trust the result.