Interpreting R+D results

[jono]

Hi there -

We're finishing construction of a new studio, and we've been working to get some measurements on the control room. It's been trapped heavily, early reflections covered, cloud, etc. etc. Using R+D to measure, but struggling with what the data means from the various charts, and where we should really be focusing our efforts. Below are 4 charts, all taken from a single monitor with mic position from what we believe to be the optimal listening position (measured at around 38% room depth, and adjusted a bit based on multiple measurement and ears): and ETC chart, a bode with a gate of 15ms, a smoothed 1/10 octave at 150ms, and a chart with the psychoaustic filter applied.

If you look at the psych graph - you'll see that the chart shows roughly a 15db decline starting at 175hz and ending at 70hz. Looks pretty good other than that. If you look at the 1/10th octave over 150ms - you see less of a variance at the low end, but a bump of 6-7db between 150hz and 275hz. Not sure what to believe. the 1/10 octave has a very long gate time associated with it - the only way to see any data down low - so I'm not sure how valid it is.

I will say, listening to music thru both monitors from this position sounds really good....but the low-end decline in the psych chart is a bit concerning. we also ran a sine wave test from the listening position, and we definitely do see a decline in the low end from 175 on down as we move the sine wave lower.

Any reactions to the charts, and suggestions as to where we should be focusing our efforts? thanks, jon

[SAC]

The ETC is wonky.
Until you improve the integrity of the measurements themselves, there is not much sense analyzing them.

Be sure to use hardware loopback correction for the soundcard/preamp

[jono]

Originally Posted by SAC

The ETC is wonky.
Until you improve the integrity of the measurements themselves, there is not much sense analyzing them.

Be sure to use hardware loopback correction for the soundcard/preamp

I should have mentioned that the ETC is filtered to see only 50-1050hz...but I'm not sure what you mean by wonky? looks very similar to the examples that come with the documentation from the software....

[SAC]

You don't filter it (and for whatever reason you think trying to look at the specular behavior of modal frequencies will benefit you) and it appears that you are looking at a 'modulation envelope' rather than individual reflections.

And if you think the ETC response reasonable, would you mind telling us to what is particularly useful about almost millisecond wide reflection arrivals... Much greater resolution is required for small acoustic space use.

You might want to start with basic responses and learn how to interpret them before you start massaging responses and expecting others to intuit whatever clever(sic) responses are possible but not necessarily useful.

Also. looking at the response AFTER all treatment has been done is only going to tell us what remains and is contributed to by the treatment - not what problems existed in the room itself that required remediation. In fact, except for a 0-~300 linear scale waterfall, the frequency responses are worth little at all.

And all looking at the frequency response does is tell us that SOME superposition of some reflections superpose to create anomalies. The provide us with no information as to what specifically might be causing it. And you might want to change the display to linear from log as detail helps and reduce the smoothing. And what gate times are you using, as we really don't need something like 20ms where the resolution will be limited to 50 Hz increments.

The mic position should be the actual listening position. With the ETCs generated, one for each individually driven loudspeaker.

And to the degree that the current ETC is valid you have some prominent early reflections.

[DanDan]

Hi Jon, I don't know ETF/R&D so I won't delve in there.
Some general observations instead.
Windowing can affect the FR graph so I wonder if that LF shelf/roll off is a true picture.
Assuming it is, try the other speaker. The ear/head/brain sums.
The FR graph, as it stands, suggests a small speaker. If it is active with LF control, turn it up! Otherwise, bigger speakers, subs.
Have you tried the speakers closer to the front wall, for a LF boost? Listener position and speaker width also affect LF. Take a look at Thomas Barefoots Wall Bounce Calculator 2D for some virtual fun.

On the plus side.
The vertical axis seems very fine, the variations are small and the slope is very much to my taste. i.e. a nice downward slope towards HF. The old Bruel and Kjaer curve again.
IMHO such a slope represents a 'tonality'. In this case a pretty nice one. This, and the brain, can over-ride the light bass, particularly if the bass is consistent with position changes.
This agrees with your Ear, which presumably you use everyday for 'sound' judgements.

Waterfalls are generally the most informative of the graphs in a broad sense, strange not to see them here.

Or more succintly, pretty good, get some lows somehow.

DD

[jono]

Originally Posted by DanDan

Hi Jon, I don't know ETF/R&D so I won't delve in there.
Some general observations instead.
Windowing can affect the FR graph so I wonder if that LF shelf/roll off is a true picture.
Assuming it is, try the other speaker. The ear/head/brain sums.
The FR graph, as it stands, suggests a small speaker. If it is active with LF control, turn it up! Otherwise, bigger speakers, subs.
Have you tried the speakers closer to the front wall, for a LF boost? Listener position and speaker width also affect LF. Take a look at Thomas Barefoots Wall Bounce Calculator 2D for some virtual fun.

On the plus side.
The vertical axis seems very fine, the variations are small and the slope is very much to my taste. i.e. a nice downward slope towards HF. The old Bruel and Kjaer curve again.
IMHO such a slope represents a 'tonality'. In this case a pretty nice one. This, and the brain, can over-ride the light bass, particularly if the bass is consistent with position changes.
This agrees with your Ear, which presumably you use everyday for 'sound' judgements.

Waterfalls are generally the most informative of the graphs in a broad sense, strange not to see them here.

Or more succintly, pretty good, get some lows somehow.

DD

Thanks DanDan. Came to the same conclusion today regarding the lo end...we have been trying to see how far we would get without the subs...given the data, we decided to fire them up today. We also moved the monitors forward and back and found what seems like a good spot. New charts below - the first reflection is stronger due to the fact that we also removed a table that was in the spot where the console is going to go. it is now gone, so it's completely open in front of the listening position facing the monitors. I ran another test after putting a large pillow on the floor between the listening position and each speaker. The pillows killed that first reflection off of the floor almost completely. The gate time on the lo-end chart is 150ms, and the gate time on the mids-chart is 5ms. Waterfall gate time is 150ms. Making progress. Thanks for the input, and any other input greatly appreciated. jon

[Hannes_F]

Hi there, your charts look a little too smoothed imo. I would try to find the equivalent to 1/24 or 1/48 octave smoothing.

[SAC]

You are right Hannes.
I think what is happening is that thee is too much 'neat' processing trying to be done with the band limited testing which is then convolved into general results rather than to take broadband measurements and then properly window them in post.

Band limited and/or results are not appropriate for extrapolation into general trends. They are appropriate for drilling down - not 'out or up'.

[jono]

I had no smoothing on before. but, is this more appropriate? 1/10 octave smoothing. the amount of data shown is dependent on gate times, which are listed on the respective charts.

Just to be clear - I had a local Austin gentleman help me with the design of the room. Given the logistics of the project - I travel a LOT and I'm in and out, and he has lots of other things going on (he services a large portion of the vintage analog gear here in town), there was no way to iteratively measure/treat/measure/treat. We couldn't have kept the contractor on schedule and busy with this approach. So my partner had to make decisions about treatment, which we went ahead with. We designed it to enable us to back the treatment down, or add more very easily.

He is old-school (as opposed to me - I'm just old), and has been measuring with a neutrik analog chart plotter. He actually still has thermal paper for it. We listened to the room and charted it on the plotter, and it looks really good right now. There is no ability to vary gate times, f_ck with filters, nothing. However, neither of us have used this software before - So I'm looking for any help in interpreting the results of measuring with software. Right now, it appears to chart perfectly with the plotter. The references for us are the manual, and Gervias' latest revision which has a section on R+D software. No one is manipulating data here (if that is the implication- a thousand apologies if it is not). If room measurements, pics or anything else would be helpful, let me know. I realize that this is a forum, and all help is free - so I'm not expecting in-depth consulting here. Just some guideposts to know if there's anything we're missing.

[DanDan]

jon, the waterfall looks excellent to me.
Since you have not gone very far on the learning curve, may I suggest you download and try REW.
Or if you have a Mac, FuzzMeasure is very easy to use.

DD