Now I know my old ears have been through a lot over the years but ...  I have noticed on some mp3's, I have exported from Sonar using Lame (192bps, constant bit rate), some of the quieter background parts get even quieter after the export.  Even to a point where they are almost gone??  Anyone else hear this?  When I export I am never sure about dithering but I export Power2 16bit out of habit.  Is this me or mp3?  Hey good song title.

Most likely it is MP3. It is, after all, a lossy compression algorithm, and it removes certain frequencies that the designers thought would be lost in the mix, and so not noticed when they are gone. One thing to try is to turn the MP3 back into a wave file and play that wave file in the same application that is playing the MP3. If the lost power persists in the wave converted to MP3 and back to wav compared to the original wav you have pretty much solved the question, since once converted to MP3 those frequencies are gone forever. Or you can just compare a wav export to an MP3 export. 

try CDEX

I use the Cakewalk MP3 encoder and find that the resulting files are quiet. It's been that way forever on my system. I tried a number of export settings in the encoder but it doesn't change the outcome. MP3 files have a low volume when exported from my Sonar system.

Or you could just get with the times and crush the life out of the song until there are no quiet parts to worry about. I think the example above is using SONAR's "sausage" setting... LMAO!

mmmmmmmmmmmmmm sawsedges

Hi,
 
Can't say I have noticed that but now you got me thinking ... I usually just mp3 the CD's off Windows Player on my computer, and so far, all the rips have been fine, as far as playing them in my car is concerned, including Tangerine Dream, but I have not done any Klaus Schulze in my car, for example.
 
I imagine that I would have noticed something like that a long time ago, as that is not good ... and I find it totally strange that Sonar/Cakewalk, would use a cheezy mp3 tool, and do something that poor ... you're talking about a major program/DAW, using a sub-quality tool! That doesn't add up, and I wonder if something else is the issue ... now, I'm even concerned that using Sonar is not a good idea, considering how much mp3 stuff I do, from my CD's. I'll have to try it with a few pieces of my own stuff ... but I find this very scary! And bad news!

I will go back and compare the wav to the mp3.  When using Lame I also save the wav file.  It also may be the difference between listening environments.
 
Moshkito, I do not think this is a Cakewalk issue.  More likely a mixing problem caused by me. Will look into this further.

I've got some old ears, too. And a high frequency hearing loss in the left ear (thank you, Led Zeppelin ), so take the following with a grain of salt...
 
I was doing some tests ripping the same CDs and converting the WAVs to MP3s at different sample rates (LAME codec) and playing them back in different environments. I wanted to see if I could hear differences.
 
For me, anything below 192K (all constant; I never use VBR) lost way too much source material. Note that I don't listen to simplistic pop vocal type material.
 
I noticed some more subtle differences between 192K and 256K. 320K seemed to add nothing except larger file sizes, until...I played it in the truck I owned at the time that had a great sound system and a really good (fast/accurate) subwoofer. THEN I noticed the difference between 256K and 320K...the low subharmonics were significantly attenuated at 256K compared to 320K.
 
There's also the "deconvert" aspect. I use LAME to convert to whatever deconverter is in the player I'm using (my Android phone or built in to the CD player in the vehicle (playing a data CD with MP3 files in folders)). I've gotta wonder what that does with the dynamic range vis-a-vis the original. I suspect the decoder brings up the noise floor. It's almost like we need a Dolby codec like the original one for cassette tapes...record with a bias and push that bias down (along with the noise floor) during playback.
 
Or am I really from outer space?

OK, let's start with a very broad and simplistic overview of how an mp3 encoder works:
 
1. It slices the audio up into tiny pieces by both frequency and time.
2. It determines psychoacoustic masking thresholds and for each little time/frequency slice it replaces stuff below the threshold with noise.
3. It checks to see if it went over the specified bit rate and if so it loops back to step #2 but acts progressively more aggressively until it meets the specified bit rate.
 
Now the key point here is that the compression is constantly changing with both time and frequency. The result of this is that at higher bit rates you don't get broad overall compression effects - what you get is specific artifacts at specific points in the audio that vary with time, with frequency, with the encoder used, with the specific settings for the encoder, with what happened right before, with what is happening at other frequencies at the same moment, etc, etc, etc...
 
So if you are hearing, or someone else is describing, broad, consistent, generalized effects that don't vary over time and/or with material, I am skeptical that what you/they are hearing has anything to do with lossy compression.
 
The recommended procedure here is to be very careful about not introducing other changes to the audio and then doing ABX (or other double blind) testing.