Proving SPL meter roll-off

ChrisBee · 04-26-2004, 01:41 AM

Hi

I've been using my RadioShack/Realistic analogue SPL meter for about 8 years in several different listening rooms and with several different sub/speaker systems.

But it was only recently that I discovered the "well-known" bass roll-off and "necessary" correction figures.

Before that I was always rather pleased with my flat frequency responses of my subwoofer & satellite systems.

But by adding in the correction figures my nice flat response curves are trashed! :b

So why would I want to spoil my graphs by using the correction figures?

The question is: Is there a cheap & fool-proof way for an amateur enthusiast to check if his SPL meter has the typical bass roll-off response?

Or not?

Or is it simply that ALL of them bass shy?

Thanks
ChrisBee

Edward J M · 04-26-2004, 04:41 AM

The meter is C-weighted. Without going into great detail, it will read progressively lower than the actual SPL as you go deeper in frequency.

The meter also has some inherent inaccuracies that cause it to read even lower than the C-weighted curve would suggest. Provided below is a table with the frequency, the theoretical correction factors for the C-Weighted curve, and the commonly used RS Correction Factors. As you can see, there is a difference between the two and it grows larger as the frequency drops.

Whether each RS meter will display the exact same offsets is up for discussion, but my hunch would be no. If you really want to know what the meter is reading at any particular frequency, get it professionally calibrated.

Of course, professional calibration will cost you as much as the meter itself....so most people just use the RS correction factors as "good enough". Unless you are into serious audio measurements, I tend to agree.

Frequency / C-Weighted CF's / Common RS CF's
10.0 / 14.3 / 20
12.5 / 11.3 / 16.5
16.0 / 8.4 / 11.5
20.0 / 6.2 / 7.5
25.0 / 4.4 / 5.0
31.5 / 3.0 / 3
40.0 / 2.0 / 2.5
50.0 / 1.3 / 1.5
63.0 / 0.8 / 1.5
80.0 / 0.5 / 1.5
100.0 / 0.3 / 2

Mark Seaton · 04-26-2004, 12:06 PM

While the meters are reasonably accurate for broad band pink noise measurement, the response varies quite a bit unit to unit so far as frequency response. This is not a hunch, but what many in the pro community have measured. With suitable microphones coming down to less than $100 (Superlux is one of my prefered units) I hope we continue to see better measurement setups available for reasonable money. Considering many spend more on a set of video cables, I don't consider the investment out of line for a hobbyist who wants to actively work to improve their system.

Edward J M · 04-26-2004, 01:45 PM

Agreed, Mark. I've tested a few RS meters myself and they are all different, especially in the deep bass regions.

The factory calibrated B&K meter I own for basic SPL monitoring is far better, and follows the c-weighted curve almost exactly down to 10 Hz.

With burned test tones on a CD, the RS correction factors, an Excel spread sheet, and the RS meter, I think the enthusiast can get a "rough" idea of the in-room FR, and where large peaks and nulls are located. I wouldn't rely on that type of method for precise measurements, because (as you said), each RS meter will react a little (or maybe a lot) differently.

My test rig didn't really cost me that much. Mic, phantom preamp, cables, adapters, mic stand, a decent multi-meter, a factory sound calibrator, software, professional calibration of the mic/preamp, and an audio-grade sound card set me back around $800. Well worth it for the serious tweaker wanting to get the most out of his system, or for equipment reviews and comparisons where the data will be more closely scrutinized.

Speaking of mics, I was amazed the ECM8000 measurement mic tested so darn flat right out of the box when I had it professionally calibrated (along with the phantom pre-amp as one unit). The correction factors for the combined mic and premap were never larger than around 2 dB (and often much less than that) from 10-20,000 Hz.

I also noticed you recently recommended WinMLS (or something to that effect). I saw it can measure things like group delay and impulse response (nice). For the hobbyist who already has RTA software, which version of the MLS software would you recommend for some basic extras like GD and impulse? TIA.

Regards,

Ed

04-26-2004, 02:59 PM

As you may or may not know, there is a new version of the RS analog out (there was a thread here somewhere that brought that to light).

I did take a look at there web site, and it is similar, but does have an updated appearance.

I wonder how long before we get a new set of correction factors for the new one?

BGL

**HTF Ads**

Edward J M · 04-26-2004, 03:33 PM

Quote:

I wonder how long before we get a new set of correction factors for the new one?

I'd bet the guts are the same. It is definitely still c-weighted, so at a minimum apply the c-weighted correction factors I provided above. Anything else you apply beyond that is a bit of a crap shoot, I'm afraid.

ChrisBee · 04-26-2004, 04:27 PM

Hi,

Thanks for the interesting responses.

I still can't get my head round the basic problem:

If the RS meter is SO inaccurate, then why does the needle return to EXACTLY the same spot for so many sinewave test tones?
Yesterday the needle dropped to zero between 1Hz stepping test tones and returned within a hairsbreadth all the way from 43-70Hz @ 84dB. I then re-adjusted the PCi crossover to 60Hz. To remove a trough at the previous 40Hz setting. Resulting in a flat response according to the RS meter from 15Hz to 70Hz(+-2dB)

But then of course I had to add in the meter correction values and it all went North as I went South!

I'd post the graph if I could find a way that didn't involve a website URL. I haven't found an image "browse" button yet to upload an image direct from my computer. :b

ChrisBee

Edward J M · 04-26-2004, 05:08 PM

There is a difference between accurate and precise.

You weigh 225 pounds. Your scale reads 217 pounds. Every day you step on the scale (at your weight of 225 pounds), the scale reads 217 pounds.

You lose 13 pounds and now weigh 212 pounds. Your scale now reads 204 pounds. Every day you step on the scale (at your new weight of 212 pounds), it reads 204 pounds.

The scale is precise, but it is not accurate. See the difference?

The RS meter may very well be fairly precise in its ability to repeat data or convey changes in SPL. But it is not accurate in an absolute sense, both due to the c-weighted filter, and compounded by its own inherent additional inaccuracy.

Kevin C Brown · 04-26-2004, 07:27 PM

And, to add on to what Ed said about accuracy and precision (I was just about to post my own similar version) is that because of this:

Quote:

While the meters are reasonably accurate for broad band pink noise measurement...

Remember that most of us simply use the Radio Shack meter for balancing levels between channels. That is precision, and the meter is actually very good for that. But if you want to know what the absolute freq response of your sub is, not gonna do that very well.

But I have my own theory too. For example, I use the Radio Shack meter and ETF software with my PC to do rudimentary measurements for my ht. I can certainly believe that there are differences *between* meters at different frequencies. But I wonder about the scale of the inaccuracies of any one individual meter *within itself*. What I mean by that, is that let's say I measure 20 Hz to 200 Hz. Meters will read differently. But is it more of a trend? Or more of problems with discrete measurement frequencies?

I.e.:

Freq Hz....True dB....Meter X.....Meter Y
..20............75.........73...........73
..40............77.........76...........79
..80............78.........76...........77
..160...........77.........75...........80

I believe real meters act more like Meter X than Y. (Look at the "trend" for X vs the "haphazardness" of the errors for Y.) So even though a lot of people put forth the inaccuracy of the Radio Shack meter as an issue, I believe that you can still use one meter by itself for a lot of interesting things in a home theater.

MikeTz · 04-26-2004, 09:06 PM

Chris:

If you can truly calibrate the frequency response of your room to within 6dB from 20Hz to 1KHz you are blessed. Enjoy that acoustically flat room! Most people have rooms that vary more than 20dB below 200Hz. Sounds like your room dimensions and speaker placement are spot on.

Get a drink, fire up the music, and smell the roses.

MT