SPL Meter Calibration and FM Curves

[WMcD]

I see that there is discussion of the use of SPL meters. Also, at least one site has suggested calibration tables for the RS meters.

Attached is some text and charts which reveals something of interest. The A, B, C, and D "weighting" scales (at least usually A and C) are used on SPL meters, of course. The weightings are actually filters which correspond to various phon levels on the Fletcher Munson equal loudness contours. You'll read that this is actually inverting the F-M contours.

It appears to me that we can start by assuming a given SPL meter is accurate while set to a given weighting. It may be easiest to use the C because it corresponds to the 80 phon level, which is not really flat but flatter than some others. If you read the FM chart, you can "calibrate" the meter readings on C to come out like a unweighted meter. That is all the purported calibration tables are doing, IMHO.

The implication is that because the C scale is not flat, the RS meter while set to C is inaccurate. Well, true if you're expecting flat; but it is not supposed to be flat. No meter set to C weighting is flat. [Of course a given measurement instrument may be just not accurate -- that is a different issue.]

It is interesting to see that the 100 phon level contour is almost exactly flat below 1000 Hz.

Best,

Gil

SPL ABCD scales and F-M Curve.pdf

[Guest IVstringer]

First, it should be noted that A/B/C/D weighting scales serve a very important purpose in regards to OSHA standards, noise control, and the auditory sciences. If you wanted to measure your sound "exposure" your ears receive to determine whether your hearing is at risk, you DEFINITELY need the A weighting, because A weighting corresponds to your ear at "normal to loud" levels. C weighting is applicable for "jet engine" type levels, and whether your speakers can play that loud, OSHA standards would still use the A weighting at all levels.

Regarding loudspeaker output measurements... the weightings are irrelevant. Maybe this is what you refer to as "inaccuracy." In fact, they are not "inaccurate." Rather, they serve a purpose that is not applicable to this measurement application. Or, you might say the measurement you made on a C weighting may not correspond to a non-weighted measurement.

Ideally, if your SPL meter has a "bypass" or "flat" setting, you would like to use that. At that point, your SPL meter is just a pure RMS pressure meter. Many cheaper SPL meters tend to only come with A or C weightings. In that case, you should use the C weghting for measuring a loudspeaker system's output, as the C weighting has a flat response for a larger frequency region. In fact, the C weighting is flatter than you might guess from looking at the 100 phon curve. The actual C-weighting curve has -6 dB points at 20.6 Hz and 12.2 kHz, so it's flat for a pretty large chunk of the auditory frequency range. If you measured pink noise output through your system, I would say that you can "believe" the number within probably 1-2 dB or so. If you are trying to get a "frequency response" using various pure tones, you will not really have to correct for any measurement between 60 Hz and 4 kHz (the -1 dB points).

Any time you are trying to make a RMS measurement with pink noise in a reverberant room (read: any normal room), it is usually best to make several measurements in different locations around the room, as room modes can affect the RMS reading. For measuring the noise output of appliances (for example), a microphone is placed on a rotating boom which continuously moves around the room in a circle, and the final measurement is a result of the average level after one circle around the path.

I added a graph I made of the C-weighting response. Don't freak out! Check the y-axis scale, it's in 1 dB increments. Hope this adds some more info regarding this topic.

Cweighting.pdf

[Mallette]

Good stuff!

Thanks,

Dave

[Islander]

Those curves may not make much difference when setting levels among speakers, but the curve at low frequencies could make readings in the bass range hard to interpret. Since SPL meters are often used to help integrate subwoofers into music systems, that could be a problem. Should correction to "actual" flat readings be used in that case, since it seems like using a C-weighted meter would introduce a fairly coloured bass response?

[mas]

For measuring relative levels where absolute levels are not critical, how much does it matter?

[Guest IVstringer]

Exactly.

[Islander]

Perhaps I wasn't clear in my question. When matching levels of main, center and surround speakers, the meter's response curve makes little difference, if any, so I'm in agreement with you there.

However, if the meter does not have a flat response in the bass area, when setting the sub's level and high-frequency roll-off, it would be easy to dial in a noticeable dip or peak in the area where the main speakers and the sub overlap.

You'd think you could trust your ears, but I've noticed some peaks and dips when the meter reading is constant, which makes me think that my ears' frequency sensitivity is very far from lab-standard. A particular test tone, say 85Hz, might sound quite a bit louder to me, even though the meter reading remains at reference level. Since it's likely that not everyone has the same quirks in their hearing, I'd prefer to have a bass response that doesn't favour any frequency, but remains flat all the way down until the subwoofer starts to roll off.

[mas]

Just a thought as you specifically mention the "overlap"...

Dependng upon the frequency(ies) that the SPL meter samples and its manner of processing, you can potentially encounter situations where the meter sees similar energy levels but where you ears detect a null at particular frequencies due to the superposition of the source signals. In this case, an RTA might be a 'more better' indicator. It would at least tell you what the sprectral balance was at a particular point in space, but it would not tell you the how or why (the 'detail') that was contributing to such a behavior.