Measuring Frequency Response using an SPL Meter

[uglybassplayer]

I was thinking of doing some informal frequency response tests of my Schroeder 1210 bass cabinet using an SPL meter. I was able to download some third octave-wide warble tone mp3 files (about 20 seconds for each tone) from the HSU Research Forum for the following frequencies:

250hz, 200hz, 160hz, 125hz, 100hz, 80hz, 63hz, 50hz, 40hz, 31.5hz, 25hz, 20hz and 16hz.

My plan was to play the 250hz tone through the cabinet and set the volume level so that my SPL meter gives an average reading of 90db (C weighting, slow response) when I'm standing about 6 or 7 feet away from the cabinet (similar to where I'd normally stand while playing) as a starting point and then continue recording the SPL levels for the rest of the frequencies. I could then plot those levels on a graph to give me some idea of the cabinet's response.

Does anyone know if this would yield a fair representation of the real world frequency response of the cabinet, or are there better ways of doing it? Please keep in mind that I don't have an RTA, just my cheap and trusty R/S digital SPL meter.

Any suggestions and ideas would be appreciated!

- Ugly.

 

[Wes Whitmore]

I would think that you need to test it at 1watt/1meter. It woud also be more accurate if you played an audio track with test tones and recorded your measurements. Also, it should really be done outside without boundry reinforcement of walls. You could also do it inside to compare what kind of gain you are receiving when you sit it close to a wall.
Wes

 

[uglybassplayer]

Thanks for the suggestions, Wes.

I understand the 1 watt at 1 meter statement in order to be consistent with standards, but using my amp head or a power amp, how would I be able to determine that I am pushing exactly 1 watt into the cabinet? Could I use a multimeter to check that? If so, how?

Is there any merit to recording the response at stage levels and at multiple distances (like 5 feet, 20 feet, 40 feet, etc) to give me an idea of how the SPLs compare for each frequencies from different perspectives (say the stage, versus the audience)?

Regarding the audio track... I've burned the test tone mp3s onto a CD, so in effect that's what I'd be doing.

EDIT: While we're on the subject, I found some posts on other forums, and this [Invalid or Expired Link Removed] that states that that you have to pad the readings of certain frequencies because the R/S meters have a low freq rolloff... the extreme low freq is rolled off because if you have flat response to 0hz, anything would be setting the meter off, its own movement, etc. Can anyone confirm this?

Thanks,

Frank.

 

[Wes Whitmore]

You convert wattage to voltage. 1 watt =2.81 volts ( I think). I''' get you some info and copy it over to this site, if someone doesn't beat me to it.
Wes

 

[Jerrold Tiers]

While we're on the subject, I found some posts on other forums, and this [Invalid or Expired Link Removed] that states that that you have to pad the readings of certain frequencies because the R/S meters have a low freq rolloff... the extreme low freq is rolled off because if you have flat response to 0hz, anything would be setting the meter off, its own movement, etc. Can anyone confirm this?

Frank.

That depends on the "weighting" of the SPl meter. That is the amount by which it is intentionally not reading actual flat spl over the frequency range.

If it is "A" weighted, you betcha....that is adjusted to the ear's response, and is an approximate "loudness" rating. it is rolled off low and high.

You want a "flat" weighting, actually "NO" weighting, for frequency response measurement.

Many meters, including the Radio Shack meter, have a "C" weighting" available, which is "flatter". It would be better, but probably not perfect.

I don't recall the exact response for the "C" weighting, but I believe it still has a rolloff at some low frequency. if that is in your range of interest, you have a problem to solve.

Measuring low frequencies is a pain, as the wavelength is so large and you need a large space to do measuring in. And, you get different results for "half space" (speaker laying on back in large hard surfaced open area) than for "1/4 space" (next to a large rigid reflective vertical wall in a large open area), etc.

The directivity of a speaker varies with frequency. At lows, where the wavelength is large with respect to the cabinet, it tends to be omnidirectional. Then 1/4 space will have double the power density of 1/2 space, because all the power is "squeezed into" a lesser volume of space and hence there is a higher SPL.

If there are obstructions, you can get standing waves, and poor results. A warble tone averages them out to a degree, but it is difficult to be sure of what you have unless you do the "open field" measurement.


Oh, yeah, the area also needs to be quiet, any extra noise will be added into the reading. If loud enough it will show up as an error. If noise is 10 dB lower than your test signal, the error is no more than 1 dB. The meter's self noise is less than that for any reasonable test level.

 

[uglybassplayer]

I figured I've give it a run and see how the numbers play out. Please keep in mind that I realize that these results may have little meaning because of my test environment as well as my method of testing, but I'm still hoping that someone can make some sense of it .

Here was my test environment: Finished basement with acoustic drop ceiling a little over 7 feet high, fairly large room, about 20x25. CD player (EQ set flat)--> Samson mixpad mixer (all EQ set flat) --> ART SLA-1 poweramp --> Schroeder 1210 (sitting directly on carpeted floor 3 feet away from the nearest wall). All measurements are using the C weighting with slow response.

Test #1: R/S meter mounted on tripod 5'10" high, 5' in distance directly in front of the Schroeder pointing at it. I set the volume so that the meter was averaging 100dbc with the 250hz test tone (using this as my reference tone).

Test #2: R/S meter mounted on tripod 5'10" high, 20' in distance in line of sight of the Schroeder, but also 5' to the left of the cabinet. Everything else left the same.

Test #3: R/S meter mounted on tripod 5'10" high, 1' in front the cabinet and 1' to the left of the cabinet with the mic facing towards the "audience" (approximating where I stand while playing a lot of the time). Everything else left the same.


Freq. TEST #1 TEST #2 TEST #3
250hz 100dbc 70dbc 100dbc
200hz 97dbc 69dbc 101dbc
160hz 102dbc 69dbc 96dbc
125hz 97dbc 69dbc 93dbc
100hz 100dbc 70dbc 102dbc
80hz 92dbc 82dbc 94dbc
63hz 90dbc 78dbc 95dbc
50hz 90dbc 79dbc 95dbc
40hz 90dbc 89dbc 93dbc
31.5hz 84dbc 80dbc 85dbc
25hz 80dbc 77dbc 82dbc
20hz 68dbc 70dbc 76dbc
16hz 59dbc 60dbc 63dbc

So, what does this mean? Beats the hell out of me ... like I said maybe someone else here can make some sense of it BTW, the readings DO NOT include any of the adjustments suggested by the Audio Innovation website.

- Ugly.

 

[Petebass]

It means the cab has a -3dB point of about 95Hz, which is not that deep. I'm guessing this is a pretty efficient speaker?

It's good to see I'm not the only one who tries to do his own frequency graphs. I often measure -3dB points that are almost a full octave above what the manufacturers claim. For example my Nemisis 4x10 claims a -3dB point of 38 Hz. My home-made JBL 15 "sounds" deeper but it has a calculated -3dB point of 52Hz. How could that be? I did similar tests to find out.

The JBL 15 did indeed roll of at 50Hz. The Nemisis, despite the 38Hz claim, began to roll off at 60Hz.

The other interesting thing about this experiment is that you stopped testing at 250Hz. If you can set it all up again, test right right up to 20K. You'll be very surprised what you'll find. I found that all my 10" loaded cabs have a hump at 1K-2K. Some also have a dip at 250, which explaines why I have been boosting that frequency for many years. Most commercial cabs, yours included, have a bit of a kick at 100-150Hz.

Forgive me for dragging this up again but in the Schroder mega-thread, I asked for -3dB points and kept getting told to go away. You have in effect given me the answer to a question I've been asking all along. Thankyou.

 

[Joe P]

I figured I've give it a run and see how the numbers play out...
...So, what does this mean? Beats the hell out of me ... like I said maybe someone else here can make some sense of it BTW, the readings DO NOT include any of the adjustments suggested by the Audio Innovation website.

- Ugly.

So who's going to plop the numbers into Excel, and Make a graph? If I don't see one show up in a few hours, I should have some time at work by then to whip one up.
Here's one I did once for an old Cerwin Vega PA speaker...

 

[billfitzmaurice]

To determine if you have one watt you measure the signal voltage. The square root of the speaker impedance is the voltage for 1 watt. 8 ohms is 2.83v, 4 ohms is 2.0v.

Go to the AudioAsylum site and do a search for RS meter correction factors. Measure in C only, A is worthless for frequency response. Only measure outdoors, with no buildings or walls within 50 feet.

That plot of the CV is typical of musical instrument/PA gear. The obviously poor result compared to hi-fi gear is one reason why manufacturers don't make these available.

In the Schroeder tests above the results are corrupted by room interactions. Indoor testing must be done with the mic element practically touching the driver to take the room out of the equation, and then mathematically extrapolated to get a 1w/1m approximation.

 

[Petebass]

So who's going to plop the numbers into Excel, and Make a graph? If I don't see one show up in a few hours, I should have some time at work by then to whip one up.
Here's one I did once for an old Cerwin Vega PA speaker...

Something like this?

 

[Joe P]

Like this?

 

[Passinwind]

Ampeg Insider: I don't recall the exact response for the "C" weighting, but I believe it still has a rolloff at some low frequency. if that is in your range of interest, you have a problem to solve.

Invalid Link Removed

When comparing manufacturers claims, make sure they're spec'ing the same weighting (or none), or it's apples and oranges.

 

[billfitzmaurice]

Loudspeakers should never be measured with any weighting, but since the RS meter doesn't allow otherwise you have to add correction. Amp and other electronic processing gear signal to noise ratios are sometimes expressed 'A' weighted, but shouldn't be, as that hides poor figures in the low and high frequencies. 'A' weighting is properly used pretty much only in industrial noise applications as it measures frequencies where high SPL levels can lead to ear damage.