I was reading an article about loudness measurement http://productionadvice.co.uk/lufs-dbfs-rms/

It says some meters use the wrong method to measure RMS for music. Importing the test pink noise file from the web site - which is supposed to be RMS -11.5, it showed as -14.5 in Sonar. 
 
It also showed as -14.5 in Sound Forge.
In Fab Filter Pro L limiter it showed -14.5 using normal -32db meter, but -11.5 using K-14 ...
 
Ummm ............. totally confused.  Can anyone shed any light?
 
 

The K-meter does not read RMS. It reads something else. RMS is a measure of the power of the signal. The K system measures how loud it sounds and involves psycho-acoustics. These are two different measures of signal strength and are not expected to give you the same result since they use different criteria.

 
I'm not an expert, but the 3db difference in Sonar has been mentioned often.
Here's one example.  And the four posts in another thread starting with this one.
 
 

The K meter does in fact measure rms voltage so fret_man is incorrect.  The Sonar rms reading is in fact 3 db lower and therefore is also incorrect.  But in a way the Sonar rms reading is also correct.
 
By that I mean that true rms readings are 3 db down from the peak or the very top of a sinewave which is actually right. (by electrical engineering standards) But the standard for rms readings in signal levels in DAW's is that the actual value of the rms reading for a continuous sine wave for example is the level right at the top of the very sinewave which is 3 db higher.  (DAW industry has agreed that this is so, nearly every other DAW measures the very top of the sinewave but Sonar has not changed this in years)
 
So to sum up Sonar's rms readings will always be 3 db lower than what they should be.
 
In order to get a more accurate rms reading you need to invest in a separate VU meter plugin such as the Klanghelm meter (or others) when set for calibration to a speciafic level eg -14 then the needle will show 0 dB VU when -14 rms signal is present.
 
I don't like reading rms levels way down a scale anyway it is not very helpful.  A VU meter translates the rms ref lel (eg -14) into an indication that is right up high on the scale (0 dB VU) and therefore is much easier to read.  It is good that you are interested in rms values of signals.  It is very important and actually tells you much more about the actual level of the signal.
 
But VU meters don't respond to fast peak transients so it is also important to use your peak metering in conjunction with rms readings especially on drums sounds etc..But for most signals the rms values are important and very useful.  Any peaks or attack transients will take care of themselves and be handled by the headroom you have chosen to work with. eg -14 ref levels mean peaks can go as much as 14 db higher than the rms level and it won't be clipped.
 

also, look at your pan laws.
 

batsbrew is also correct. Mono test signals can cause grief as well when panned centre. I find it is best to always work with interleaved stereo test signals. It eliminates this issue but yes it is also another source of error. (if you are using a mono test signal try panning it to extremes and observe what you see, often it will change compared to pan centre)
 
But what I think you are seeing is the result of Sonar's rms readings are just 3 db lower than what the industry standard is.

Agreed, I am incorrect. Do not listen to me. Being a EE I thought RMS was RMS, but apparently not in the the music industry.

The short answer:
 
Sonar does it the mathematically correct way. Period. 
 
However, many in the audio industry have adopted standards that essentially mean "calculate the RMS correctly and then add 3dB".
 
Since most people will never have to worry about the math (and the one's that do understand what's going on), there's no compelling reason for anyone to change anything. And for many "audio people", the mathematically incorrect interpretation seems to make more sense to them, and perhaps better correlates with other traditional non-digital measurements that they work with.
 
It is what it is.

fret_man
Agreed, I am incorrect. Do not listen to me. Being a EE I thought RMS was RMS, but apparently not in the the music industry.

You stepped in where older farts tend to stand back...

RMS is RMS, VU is VU, and so on - the fact that most software developers have chosen to redefine them, well, it is what it is. There is a tremendous amount of confusion, mis-information, and more confusion on the topic. I think it starts with the occasional mis-use of RMS and VU... but that's just a guess.

RMS, in simple terms, is the average, "corrected" to account for the fact that  audio is generally bipolar<G>. It is an electrical measurement of equivalent power - how hot would a light bulb get was the example I remember from college. (it is not an entirely inaccurate analog, but we aren't going there!)

Volume Units (VU) is an arbitrary system that depends as much on the ballistics of the meter movement as anything else, and while it is - sort of - an RMS measurement, "sort of" causes problems!
 
With respect to Sonar - I haven't run a signal through Sonar to check the metering accuracy in eons. It might be right, it might not be. I use external, mechanical VU meters when I really want to know  what is going on.

Anyway, have  fun...

I agree with Bill in that I also have a pair of high quality VU's outside the system and when I really want to know what is going on they are what I am looking at for sure.
 
But there are some great VST's out there that do it too.  I have invested in quite a few and done some rather exhaustive tests and found the Klanghelm VU meter to be pretty decent and close ballistics wise to the real thing.  Not perfect but very good.
 
The good thing about a VU VST is that all you need to do is set the ref level and away you go. The meter itself just takes in the spec to show the correct reading for the very tops of the sinewaves as the level, not 3 dB down from that.  And also you are looking at something now that is centering around 0 dB VU which is easier than trying to read rms levels low down on a meter scale.  Sonar pushes the rms mark a little lower.