The correct way to record SPLs

[Buckle-meister]

Dear all,

I have recently purchased the Radioshack (analogue) SPL meter. However, after reading both the manual, and Audioholic papers, I have a number of questions that perhaps someone may care to clarify for me.

1. In the paper Basic Audio & Video Set-Up Guide, the graph shows that for a c-weighted curve, the response is not flat outwith 100Hz to 10kHz. However, the last page of my manual states "...if weighting is set to c, the meter's frequency response is flat from 32Hz to 10kHz (+-3dB).". With especial regard to frequencies less than 100Hz, which statement is correct?

2. The manual states that one should not position oneself directly behind the meter itself, yet I think I remember reading in a thread/paper that for HT purposes, you should sit in your listening position because you do want the influence of your body to be taken into account. Is this true?

3. Should the meter be held at ams length, or close to ones body. Again, I think I seen it written in a thread/paper that the meter should be at ams length, yet if that is so, why? I thought that the idea was to get SPL meter readings at the listening position.

4. The above noted paper also states that the meter should be held with the microphone pointing diagonally upwards (but towards the speaker being SPL'd). Why not directly at it?

5. Finally, presumably the meter should point at each speaker it is SPLing?

Thanks folks.

 

[WmAx]

Dear all,

1. In the paper Basic Audio & Video Set-Up Guide, the graph shows that for a c-weighted curve, the response is not flat outwith 100Hz to 10kHz. However, the last page of my manual states "...if weighting is set to c, the meter's frequency response is flat from 32Hz to 10kHz (+-3dB).". With especial regard to frequencies less than 100Hz, which statement is correct?

You can find a graph of standardized response weighting curves at this link:

http://www.sfu.ca/sonic-studio/handbook/Sound_Level_Meter.html

2. The manual states that one should not position oneself directly behind the meter itself, yet I think I remember reading in a thread/paper that for HT purposes, you should sit in your listening position because you do want the influence of your body to be taken into account. Is this true?

The reference to positioning your body is only relevant for midrange and treble frequencies. Your body is not large enough to impede the bass wavelengths to a degree to cause measurement error(s).

3. Should the meter be held at ams length, or close to ones body. Again, I think I seen it written in a thread/paper that the meter should be at ams length, yet if that is so, why? I thought that the idea was to get SPL meter readings at the listening position.

Ideally, you should get a microphone stand or other positioning tool/jig and adjust the height/position to be in the same place as your head normally resides during listening.

4. The above noted paper also states that the meter should be held with the microphone pointing diagonally upwards (but towards the speaker being SPL'd). Why not directly at it?

This upwards pointing position is probably compensation for the microphone response at high frequencies. Irrelevant for low frequency measurement.

5. Finally, presumably the meter should point at each speaker it is SPLing?

Just set it up in position as I previously instructed. It does not matter(for low frequencies) whether the mic is pointing up, down, left, right or backwards, so long as it is in the approximate position that your head would normally reside during listening.

-Chris

 

[Buckle-meister]

Thanks WmAx,

I think that that apart from question 1, everything has been answered satisfactorily.

Regarding question 1 however, I still dont know whether to assume the manual is correct, and make no adjustment to SPL readings, or to go by the charts, and make the adjustment.

If I choose the former and it should be the latter, Ill be recording erroneous data. Obviousely the converse is true also.

Do you (or anyone else out there) actually have your own manual close to hand to see whether it is the same as my own? I have noticed that my Radioshack meter is very slightly different in layout from the picture in the paper I quoted. Could it be that my own meter is a newer version, and so perhaps more accurate than a previous version?

Regards

 

[WmAx]

Thanks WmAx,

I think that that apart from question 1, everything has been answered satisfactorily.

Regarding question 1 however, I still dont know whether to assume the manual is correct, and make no adjustment to SPL readings, or to go by the charts, and make the adjustment.

If I choose the former and it should be the latter, Ill be recording erroneous data. Obviousely the converse is true also.

Do you (or anyone else out there) actually have your own manual close to hand to see whether it is the same as my own? I have noticed that my Radioshack meter is very slightly different in layout from the picture in the paper I quoted. Could it be that my own meter is a newer version, and so perhaps more accurate than a previous version?

Regards

The RS SPL meter uses an approximation of the C weighting, when it is in C weighted mode. A flat response would be unweighted response. Additionally, there is a standard deviation compensation chart for the analog RS SPL meter(to account for other response errors that are typical of this model). Unfortunately, I don't know where the chart is at the moment, or I would provide a link or copy via e-mail to you. I don't use a SPL meter for frequency response measurements(it's been years--I use a PC measurement system now) now, I currently use SPL meter only for SPL measurement.

-Chris

 

[Tom Andry]

You can also get the Rives Audio Test CD - it has tones adjusted for the RS meter

 

[WmAx]

You can also get the Rives Audio Test CD - it has tones adjusted for the RS meter

Wow! Cheers to Rives. Buckle-Meister: I have to second the recommendation to get the Rives CD.

-Chris

 

[Buckle-meister]

Quick question:

Is it the c-weighting that gives a flat response between 100Hz and 10kHz?, i.e., without the weighting, the response would not be flat between the aforementioned frequencies.

 

[WmAx]

Quick question:

Is it the c-weighting that gives a flat response between 100Hz and 10kHz?, i.e., without the weighting, the response would not be flat between the aforementioned frequencies.

Without weighting, the response would be flat. Weighting has a non flat response. The graph I linked to earlier demonstrates the frequency response of various weightings.

-Chris

 

[Buckle-meister]

Ok, so if I understand things rightly...

If the SPL meter is set to c-weighted, and that curve takes a 'standard' form (flat between 100Hz and 10kHz), then the meter's manual is incorrect to state that its response is flat down to 32Hz and I should (assuming I dont purchase the Rives disc) use the correction charts after all?

With regard to the Rives disc; I did see it mentioned in a paper, but only after having ordered the Avia disc. Is it still recommended to get the Rives disc, or just to use the charts to correct my meter?

Regards

 

[WmAx]

If the SPL meter is set to c-weighted, and that curve takes a 'standard' form (flat between 100Hz and 10kHz), then the meter's manual is incorrect to state that its response is flat down to 32Hz and I should (assuming I dont purchase the Rives disc) use the correction charts after all?

Their is only one 'C' weighting. The weightings would be meaningless if one company could call something flat to 32 Hz, "C", and another could call something flat to 100 Hz,"C".

the Rives disc: I did see it mentioned in a paper, but only after having ordered the Avia disc. Is it still recommended to get the Rives disc, or just to use the charts to correct my meter?

Depends. Is it worth the cash(to you) to save time by directly recording the correct SPL values to paper? The Rives disc will make this possible. Otherwise, you have to record the values and then add the correction value to each frequency.

-Chris

 

[MDS]

I have the digital SPL meter, which should be calibrated identically to the analog - they differ mainly in features.

My manuals says the same thing as yours: 'C-curve (flat) frequency characteristics...causes the meter to respond mainly to frequencies in the 32 to 10,000 Hz range.' The specifications page says the accuracy is +/- 2 dB at 114 dB SPL.

The various correction charts you can find on the net (like svsubwoofers.com) are mainly used to correct for inaccuracies of the meter at frequencies below its stated accuracy range. If we take the manual at face value, the meter should be accurate to +/- 2 dB between 32 Hz and 10 kHz. For all other frequencies you would need to use the correction charts (and assume they are correct) to determine the real SPL at those frequencies.

Mainly the charts are useful for calibrating your sub because the meter is known to be inaccurate at low frequencies. On average the meter is off by 2-3dB at low frequencies so you could still get very close to a perfect match between sub and the other speakers by simply calibrating the sub 2-3 dB below the other speakers; ie calibrate the sub to 72 dB and the other speakers to 75 dB because the meter on average reads 2-3 Db too LOW for the bass frequencies.

 

[Buckle-meister]

Mainly the charts are useful for calibrating your sub because the meter is known to be inaccurate at low frequencies.

I understand that there is a lot of marketing hype, propoganda, snake oil etc 'out there', but normally the specifications of electrical equipment are accurate aren't they?

Doesn't anyone think it strange (and therefore unlikely) that Radioshack have knowingly published inaccurate performance characteristics of their product?

 

[Ethan Winer]

Chris (and all),

> Without weighting, the response would be flat <

Yes, assuming the microphone element itself is flat. Electronics is easy, transducers hard.

Stop me if you've heard this one before. Below are two ETF graphs - 200 Hz and below, 200 Hz and above - showing the response of my particular RS meter (digital) compared to my high-end calibrated AKG omni measuring microphone. Below about 800 Hz the RS meter is dead on, and no "correction" is needed. But above that the two mikes deviate significantly, and even with the so-called correction curves the responses differ by 3 dB and more. (Which is not so bad in the grand scheme of things.)

--Ethan

 

[WmAx]

Chris (and all),

> Without weighting, the response would be flat <

Yes, assuming the microphone element itself is flat. Electronics is easy, transducers hard.

While RS meter does not demonstrate a very linear response, especially into higher frequencies, it is sufficient for basic LF response(as you stated) measurement. I assume that the user was only intending to do lower frequency measurements.

But, it is not true that it is difficult for microphone transducers to be relatively flat. For example, a $3.00 Panasonic electret capsule(that is popular with DIYers) has a response that is typically flat +/- 1.5dB(some product examples may go out of this typical tolerance, but not commonly), from 20-20,000 Hz. The same goes for the Behringer ECM8000 measurement microphone. You can aquire basic microphone calibrations for about $40 from Kim G.(reachable on the Madisond forum), but this will only shave off +/- 0.5dB, to bring accuracy within +/- 1dB. Hardly worth the effort, unless you need better accuracy into high frequencies(>15kHz), where the uncalibrated mic may start to deviate by several dB by 25kHz, for example. Also, further accuracy(can be achieved using laser calibration services to calibrate frequencies over 10kHz, approx. $100) is not of practical use, except for labratory reference purposes. Even with a lab-grade calibration, frequency response errors of up to 1dB can still occur(not inherant of the microphone), due to change of air pressure, humidity and temperature that varies from day to day. If absolute accuracy is required, you must have (1) laser-calibrated microphone (2) record pressure, temperature and humidity (3) compensate for environment conditions on the final response curve(s) using the environmental/climate measurements.

+/- 1.5dB variations are of not significant, audibly. While they can be audible(if of sufficient bandwidth) in direct A/B comparison, the difference is subtle.

-Chris

 

[Ethan Winer]

Chris,

> it is not true that it is difficult for microphone transducers to be relatively flat. <

Agreed, but they're never as flat as electronics. That's all I meant. Also, the actual measurement of higher frequencies in a room is difficult because of all the comb filtering present from nearby surfaces. Even the back of a leather chair or couch creates major ripples at higher frequencies.

> +/- 1.5dB variations are of not significant, audibly. <

I agree with that too. I see people obsess all the time over tiny variations, all the while blissfully unaware that their room has a half-dozen peaks and deep nulls spanning 30 dB (sometimes more) below 300 Hz.

--Ethan

 

[WmAx]

Agreed, but they're never as flat as electronics. That's all I meant. Also, the actual measurement of higher frequencies in a room is difficult because of all the comb filtering present from nearby surfaces. Even the back of a leather chair or couch creates major ripples at higher frequencies.

At this point, we may have different things in mind(I mention HF, but remember that many of my purposes involve gated measurements, as used to simulate anechoic measurements, for use in loudspeaker tuning/design --- as such, they(gated measurements) are not affected by HF reflections from walls, etc.).

-Chris

 

[Ethan Winer]

Chris,

> gated measurements are not affected by HF reflections from walls, etc <

Indeed.

--Ethan