Subwoofer EQ test cds (1 Viewer)

[Kevin C Brown]

Paul- Funny: I've also done my own "adjusting" of the "std" cal adjustments. Makes the adjustments curve smoother. The RS meter is an analog device. There's no reason to expect large changes between freqs that are relatively close together. Wayne kind of alluded to this. (I will need to copy your figures and plot them for kicks. )

Correction at 75 dB vs 85dB. No, I don't think that the adjustment should be the same. It should be more like a percentage. I.e., at 40 Hz, I add 13.67% to the actual reading on the meter. But we all take "x" to be a fixed number of dB. Plus, 85 dB is substantially louder louder than 75 dB. dB are on a log scale. If 6 dB is twice as loud, than a 10 dB difference is more than 3 times louder. So I can't see that the correction should be the same.

Shoot, maybe I just realized something. Because dB is a log value, but in truth, dB are actually dimensionless but really a ratio of two sounds to each other, therefore x at 85 dB is actually quite a different number than x at 75 dB. Neat. It *is* a fixed number. Never mind! Wow, that's powerful...


Shoot, I still want to hear more from Bruce about Fletcher-Munsen and C weighting and how best to make use of the RS meter for all this...

 

[Brian L]

You know, a few posts back, I actually thought I had a handle on things!

OK, someone needs to throw me a bone here...

A weighting, vs. C weighting vs. U (which I had never heard of before this thread).

In very simple terms, I was under the impression that A weighting rolled off the low end of the meter's response significantly, while C weighting does not. The point is to have greater accuracy depending on the frequency content you are measuring.

Thats why we all are using C weighting to adjust sub channel balance relative to the main channels. If we all has system with very limited low end response, A weighting would be fine.

OK...correction factors. My impression of these is that putting the weighting aside, the RS Meter is known to be inaccurate in the low frequencies. The published factors are intended to give you an accurate SPL reading across all frequencies, so that when the meter says X dB at X frequency, you add/subtract what you need to get to the truth.

And I always thought that those factors are based on the meters response in the C weighting position. So, apply the factors to readings taken with C weighting, and you now know the REAL SPL's and can use that data to plot your in-room FR, tweak your EQ, impress your non-audiophile friends, pick up better looking women.......

What am I missing in all of that? Why does BruceD say that using corrected values does not provide true readings? Why does Wayne argue that the factors don't matter? BruceD and Wayne...those comments are not intended as a challenge to what you have posted, but a cry for knowledge on my part!

To me, if I am tweaking with my BFD, I need to know the true magnitude of a peak/null. If I am using uncorrected values, my cut/boost will be wrong. A reading taken at 25 Hz will require a correction of +5 per the published factors. Thats not a small amount.

If the RS meter is in fact off by anywhere near the published correction values, how could I ever get good results without the factors. For me, its all about accuracy, or getting as close as I can with the tools at hand.

And while I am OK with the notion that you want to set your overall sub level by ear, you can not adjust a BFD,or any EQ really, by ear. You must measure, and at the end of the day, I want the most accurate measurements I can get short of the calibrated mic/PC method.

Sure, the factors vary depending on who did the testing/interpolating, and my RS meter may differ from your RS meter, and maybe the type of test tones may cause the factors to vary a bit if I am using a different type of test tone, but at the end of the day, it does appear that it is agreed that the low end response of the RS meter does roll of, and that to arrive at the truth, you need to take your readings and add 'X' to what you see.

And when you are done, your plot should represent the true in-room resonse at the listenning position (or as close to true as the above mentioned inaccuracies allow).

So, having said all that, I will now shut up so someone can try to drive some of this through my skull!

BGL

 

[PaulT]

You forgot D Weighting

http://www.norsonic.com/web_pages/correlation.html

http://www2.sfu.ca/sonic-studio/hand...vel_Meter.html

http://www.prorec.com/prorec/article...256904007DFFC0

 

[Brian L]

One can not forget what one never knew to start with

Some very interesting stuff in those links Paul....I think I need to crawl into a hole and do some reading. Perhaps the answers to some of my questions/confusion can be found there.

BGL

 

[BruceD]

Actually, the mono RCA output of the RS SPL meter is a preamp output, not a mic output. I've heard conflicting statements about whether the "C" or "A" curve weighting is applied to this output. When I first used the RS SPL meter with the ETF program (5 years ago), the meter's 75dB SPL reading matched the ETF program's 75dB level with "C" weighting enabled (but IIRC not with the "A" weighting enabled). I quickly went to a calibrated mic with preamp and never looked back.

The issue about what weighting to use depends on the source signal being used:

1) pink noise : A type of random noise which has a constant amount of energy in each octave band (and for subs is typically a limited band of low frequency octaves called "shaped pink-noise").

2) white noise: A type of random noise which has equal energy at all individual frequencies.

3) Sine Waves: Usually equal energy at all the individual frequencies.

4) MLS-based tones: I think these are similiar to pink-noise and shaped pink noise.

I would say for sure that "C" weighting would be needed for measuring items 2 and 3 above, because as frequency goes lower you actually need a higher actual SPL to sound equal (i.e. the meter's "C" weighting means you need to increase actual SPL to get a flat curve).

 

[Kevin C Brown]

OK, that helps. If the RS meter puts out a signal with C-weighting already applied to that output, that *would* explain why the correction factors on ETF's web site do not include any low freq numbers. Already been applied.

 

[PaulT]

As an interesting 'aside', if you are Electronically inclined there is a mod for the 33-2050 to make it read flat across the low freq region so you don't have to apply the correction values:

Link Removed

The section about an external mic mod may also be of interest to you.

BTW - I believe we have hijacked Ryan's thread, but it has sort of followed through on the original question

 

[Brian L]

Another great link Paul. I had seen this somewhere before, but had not read through to understand what was done. Looks pretty easy to do, but I still am clinging to the thought that precompensated test tones is easier still.

As for the OP's inquiry, we at the HTF pride ourselves in giving you MORE than you paid for

BGL

 

[BruceD]

Paul,

Eric specifically says that this mod eliminates the "C" weighting factor, which essentially makes it a "U" weighted (unweighted) response, see below :

 

[Brian L]

OK, perhaps a light bulb has just turned on in my head....

Are the correction factors there to simply get rid of the weighting, or are they there to correct inaacuracies of the meter, exclusive of the weighting?

Methinks they are there to eliminate the weighting, yes?

Which is what the meter mod does, or even using compensated tones attempts to do.

Or am I still hopelessly lost???

BGL

 

[BruceD]

That is the crux of the problem!!!!
It is typically not clear what most of the correction charts available on the web truly adjust for.

We do know the HW corrections mentioned by Eric only adjust the "C" weighting to unweighted and do not correct for the mic response.

Again, I think the important point to take away from this discussion is that if you want to use Sine-wave tones to measure response, you need to use a "C" weighted measurement methodology that starts with a reference SPL measurement (75dB or 85dB) at 1kHz that is then used for all frequencies below 1kHz.

I don't know how accurately the RS SPL meter adheres to a strict "C" weighted curve, but they do give a response graph in the manual.

If anybody has a hankering for a DIY project, Eric's same website has a really GREAT microphone-wand and battery operated mic-preamp. It is what I built and use with my ETF5 program. I had the wand calibrated (test-graph printout and diskette with correction values) by the person referenced in the article.

 

[Ilkka R]

Check this. Scroll down and you can find a graph posted by trilogy showing differencies between C-weighting curve and RS-meter correction curve. As you can see RS-meter needs some additional boosting. So those values what we all know as RS-meter correction values includes C-weighting plus additional boost for RS-meter.

Jack output gives exactly the same values as you can see on meters display. So it's C-weighted and you need to apply those same correction values, if you want flat response. Most programs (ETF5, TrueRTA) have some kind of mic calibration file, so this is quite easy to do.

 

[Lee-c]

Yes, I'm almost positive that the Radio Shack SPL meter correction values for
low frequencies that one sees posted here on the forum are assuming that you're
calibrating your subwoofer with the Radio Shack SPL meter set to "C" weighting when
using the Avia calibration disk or the Video Essentials calibration disk. Which
both use, I think, pink noise for Dolby Digital Reference Volume Level calibration of
the main speakers, surround speakers and subwoofer.

The RS SPL meter compensation chart for subwoofers listed below was
posted by GeorgeJM here on the HTF.

10hz.....+20db
12.5hz...+16.5db
16hz.....+11.5db
20hz.....+7.5db
25hz.....+5db
31.5hz...+3db
40hz.....+2.5db
50hz.....+1.5db
63hz.....+1.5db
80hz.....+1.5db
100hz....+2db
125hz....+.5db

 

[Brian L]

The latest disptach from the "I can't leave well enough alone" dept.....

Mom and daughter were out this afternoon, so I thought I might revisit my sub calibration, but this time, I used my Infinity RABOS meter and RABOS Test Disc. This meter is said to be optimized for bass measurements, although I don't think I have ever seen any published correction factors (so, you might say, ignorance is bliss!). I did e-mail Infinity, who replied that the meter was "accurate" up to 1K, whatever "accurate" means.

FWIW, with these warbled tones, I found that I still had a pretty sizeable (>10dB), broad peak centered around 34Hz, and a mild dip (about 4 dB) around 60 Hz, amomg other minor aberrations.

So, a couple house later, I was able to use four filters on the BFD to get a RABOS measured response of +/- 1.5 dB from 25 to 100Hz (not sure why is was flat out to 100, as I had a 12dB/80 Hz crossover dialed in, but thats a question for another day).

Now, my previous measurements were with pure sines and using the RS meter with published factors from Snapbug. After that session, I was more or less flat from 25 to 80.

Obviously, RABOS did not agree. But like a man with two watches who does not know the time, I know have two conflicting plots, and I have no way to determine which is "right".

Anyway, I did keep the old curve in the BFD (gotta love that product!!!!!), so I will now probably just do some subjective listening of a variety of sources, and wee if I like one more than the other.

BGL

 

[Kevin C Brown]

Ilkka - Great job finding the difference between C-weighting and the additional adjustments for the RS meter.

This has ended up to be a great thread.

 

[PaulT]

Ilkka, yes, thanks for that link. For those who don't want to read through it, one of the statements from trilogy is quoted:

here's the link to Eric's main audio homepage refered to by BruceD above:
Link Removed

Have you guys had any further success burning 'corrected' tones. It would be interesting to plot your freq response using same gain test tones and applying the corrections in the spreadsheet (as I believe most of us have been doing), then comparing this to the plot done using the same tones gain adjusted when they were created and forsaking the corrections in the spreadsheet.

Since it appears that the 'published corrections' correct both the C weighting 'and' and discrepancies in the meter, then any errors (besides mic placement) should be limited to the interpolation of the corrections depending on the tone fequency used (and we have seen that that can be upwards of 0.5dB depending on the upper and lower frequency limits used in the interpolations) and the range of cut or boost (0.5dB increments for example) available in the tone generation software.

Hmmm, starting to get anal about all of this

 

[Kevin C Brown]

Shoot, I just had a thought. And I think this is what Bruce has been pointing out.

There is the Fletcher-Munsen curve. Basically, human hearing isn't so sensitive at lower freqs. C weighting is an attempt to account for this. (Right?)

So we all blindly use the "published corrections" that essentially get us to a truly flat freq response. But maybe that isn't what we want. Maybe we only want to apply the particular part of the corrections that compensates for the RS meter's inaccuracies, and leave the C-weighting in there. Then, we get to a flat freq response according to how we hear.

And, I think the "house curve" that Wayne talks about is that very attempt to boost lower freqs to account for how we actually hear.

For example, I have my own test disc disc now. 40 pts from 10 Hz to 160 Hz, 1/10th octave spacing or better. So after I apply the corrections, it tells me that the 20 Hz tone I just heard is just as loud as the 40 Hz tone. But my ears and brain tell me that isn't the case. The 40 Hz was louder. Because that's actually how I perceive it...

And then, we all know that we should adjust our subs 3 dB lower than what Avia tells us. Because the meter isn't as sensitive at lower freqs. But a lot of people run their subs hotter. I've seen up to 8 dB hot. Accidentally, that is a brute force attempt to compensate for how we hear.

Hmmm...

 

[Kevin C Brown]

Hee, hee, I just had one more thought for the evening. We really do want both corrections in there, to get to a truly flat response.

Why? Because if I'm a soundtrack or music engineer, I'm going to master the sound, a) on a system that has been calibrated to a truly flat freq response, and, b) I'm going to master it so that it sounds good to me. That means that compensation for how we hear, C-weighting and Fletcher-Munsen, is already "included" in the mix.

OK, I can go to sleep happy now.

 

[PaulT]

That has been my understanding as well, and found that when I EQ'ed my Sub to be 'flat' it actually sounded like that - flat. When I EQ'ed with a house curve (which I think of as the inverse of C-Weighting - or removing the C-Weighting of the meter) it sounded a lot better to me.

Took a break typing and did a bit of reading and find that the "C-Weighting Network on a Sound Level Meter was derived as the Inverse of the 70dB Fletcher-Munson Curve", so the above definately rings true.....Google "Fletcher Munson" or "Equal Loudness Curve" for a ton of info.

 

[Eric C D]

I agree with that thought (but that's only this amateur's opinion - I'd LOVE to hear from a professional in the recording industry). HOWEVER, there is one more fact about the Fletcher Munson curves: they vary by volume. At louder volumes, the curve is flatter than it is at low volumes. (this is the reason for "Loudness" controls on some playback systems)

What this means is that if you are going to play the music at the same volume it was mastered at, then you should want a "truly flat freq response." But if you're going to play it quieter, you'll want to tip up the low freqs (and perhaps the highs!). The converse is true as well if you're going to really crank it up.

For me right now, this just means I add a few dB to my sub (at my receiver, since it's repeatable) when I'm playin music quietly. I'm expecting my next eq to accept multiple curves, so I'll have a quiet music curve, loud music curve, and probably one or two HT curves.

enjoy!