I always (well unless for technical reasons) master over a full mix.  I don't see the point in the two phase mix approach, although I will get slaughtered, it allows me to do slight adjustments to the mix if needed.  Sometimes this is important, sometimes not...it just depends.
 
I had to do some adjustments last night on a mix/master, because I'm getting use to a new passive mastering EQ (Waves Abbey Road), and some new cans (AKG 712's).  But whatever works :).
 
Ben  

Ben I think the main reason is because you have been hitting your ears over a long time with so much sound you may not be in the best position to master well.
 
So in the mastering chain you might have an EQ first and this is one to watch out for.  You may not make the right decision there and then.  All I know when I leave it a week that first EQ ends up being something different entirely!
 
With compression some say the mix changes when the compression is being applied so in that case by all means leave it on and adjust your mix accordingly but take it off before you print a pre master.  And it is a good idea to have a totally pre mastered print as well.  Gives you different mastering options.
 
Funny though when I do a great mix I find the mastering compressor actually does not change anything for me.  The mastering compressor should not be working that hard (in my opinion) for the mix to change.  A great mix still sounds great to me with our without the mastering compressor.  (I think Danny Danzi agrees with me on this one too) It just seems to add some glue as they say and just add a nice professional sheen over the whole thing.  That is about all I can describe about it. (lately I have been into Tokyo Dawn Labs Kotelnikov and it is just sublime)
 
The mastering limiter (PSP Xenon that is) also does not alter my mixes either, just makes them louder so I am happy to do that stuff later on.  My limiter is only adding 3 to 4 dB of (rms) gain so it is not working very hard either.  None of the mastering processes should be working hard.  It should be a combination of all three that give you the sound you are after.  As soon as you start putting too much work onto just one of those processes it all falls apart for me.

Jeff Evans
Ben I think the main reason is because you have been hitting your ears over a long time with so much sound you may not be in the best position to master well.
 
So in the mastering chain you might have an EQ first and this is one to watch out for.  You may not make the right decision there and then.  All I know when I leave it a week that first EQ ends up being something different entirely!
 
With compression some say the mix changes when the compression is being applied so in that case by all means leave it on and adjust your mix accordingly but take it off before you print a pre master.  And it is a good idea to have a totally pre mastered print as well.  Gives you different mastering options.
 
Funny though when I do a great mix I find the mastering compressor actually does not change anything for me.  The mastering compressor should not be working that hard (in my opinion) for the mix to change.  A great mix still sounds great to me with our without the mastering compressor.  (I think Danny Danzi agrees with me on this one too) It just seems to add some glue as they say and just add a nice professional sheen over the whole thing.  That is about all I can describe about it. (lately I have been into Tokyo Dawn Labs Kotelnikov and it is just sublime)
 
The mastering limiter (PSP Xenon that is) also does not alter my mixes either, just makes them louder so I am happy to do that stuff later on.  My limiter is only adding 3 to 4 dB of (rms) gain so it is not working very hard either.  None of the mastering processes should be working hard.  It should be a combination of all three that give you the sound you are after.  As soon as you start putting too much work onto just one of those processes it all falls apart for me.

Hi Jeff, Actually if you listen to this https://soundcloud.com/aaudiomystiks/higher  I think you will find, I'm pretty close.  This is the track I was talking about in my previous post...which I finished last night.  It's coming through on my laptop speaker amazingly.  Having 500 dollar cans helps too I've found lol.  I missed so much detail in my Beyerdynamic DT990s.
 
It takes a while to train your ears, and you're right...sometimes I mix and master when I'm too fatigued, in which case I use visual clues...meters and needles to judge things.  Also, yep getting the mix right should be the first step...but understanding the mastering chain and the effect is also important, and hence I fiddeled a bit with the mix in mastering...no change in compression mind you, just a couple of minor volume changes, esp on a couple of the verbs, and I changed the EQ slighty on the kicks, but I could trust myself in this regard because I knew I'd push the curves a bit too far and was creating low-mid issues.
 
My mastering chain, and you're correct I do use an EQ first...but only to remove sub-frequencies and too much top end, and maybe low-mids to clean the track.  I then re-gain stage the track, because unlike you and Danny...I don't care about being in the red...within reason when mixing.  This is becuase I don't go back out of the box when mixing and mastering, which is a big no if you do...but does not matter if not, and indeed it helps create a fatter warmer track if you're using emulator tech.
 
I then feed the master into Sonar's tape sim to even the transients just slightly, and to make sure the VU on sim sits between -18 to -12.  Which is then fed into the console emulator to do a little magic to said transients.
 
Now, this is why I was fiddling around with the mix in mastering last night, because I was incorporating a new mastering EQ into my chain (I don't fiddle often), but I was really struggling with the Waves RS56...it works great off the bat for sound art stuff mind you, but for the life of me I could not get the mix to sit with my normal mastering chain.  Normally after the console emulator, I would use the Pro Channel EQ, and I was trying to do the same with RS56.  Being a complete passive type unlike the Pulteq...makes it harder to set IMO.
 
Eventually I gave up, and continued with normal chain so...Pro Channel EQ set to pure but leaving room for the RS56 later, then I use the Waves H-Comp as my mastering compressor...unlike you though I don't push the signal into and make the compressor do the hard yards...for me its small increments and gain staging.
 
Then I used the RS56, and this is where it worked in the chain...mind you it took some tweaking.
 
After this, I use Waves Master Tape...and no this is not an advert for Waves ;)...Sonar is still my go to piece of software tech...but I gain stage between RS56 and the Master Tape...adding in an extra DB of volume after the gain stage, and use the wow and flutter to dial in some punch, and tape hiss helps the top end come through...I know but I can hear it...and the AKG K12s aren't bad cans.
 
Finally I use the concrete limiter with soft-clip on, and set the release manually...and you can hear the difference if you don't set it manually.
 
If you listen to that track Jeff, and you do so in cans...you will notice the verbs and delays wrap around the back of your head...I use a 360 degree panner to create fake binaural type mixes...we are the headphone generation.
 
Cheers Ben 
 

BenMMusTech
 
Now, this is why I was fiddling around with the mix in mastering last night, because I was incorporating a new mastering EQ into my chain (I don't fiddle often), but I was really struggling with the Waves RS56...it works great off the bat for sound art stuff mind you, but for the life of me I could not get the mix to sit with my normal mastering chain.  Normally after the console emulator, I would use the Pro Channel EQ, and I was trying to do the same with RS56.  Being a complete passive type unlike the Pulteq...makes it harder to set IMO.
 
Eventually I gave up, and continued with normal chain so...Pro Channel EQ set to pure but leaving room for the RS56 later, then I use the Waves H-Comp as my mastering compressor...unlike you though I don't push the signal into and make the compressor do the hard yards...for me its small increments and gain staging.
 
Then I used the RS56, and this is where it worked in the chain...mind you it took some tweaking.
 
After this, I use Waves Master Tape...and no this is not an advert for Waves ;)...Sonar is still my go to piece of software tech...but I gain stage between RS56 and the Master Tape...adding in an extra DB of volume after the gain stage, and use the wow and flutter to dial in some punch, and tape hiss helps the top end come through...I know but I can hear it...and the AKG K12s aren't bad cans.
 

IMHO you are "sound shaping" your master bus, or whatever other name one could find for throwing a ton of FX at the master bus.
 
There's nothing wrong with that from a creative perspective - if it sounds right, it is right (and what sounds right is very flexible in the creative domain).
 
Yet, this has nothing to do with classical mastering where you finish the piece of art first, then have it polished so that it sounds as good as possible across playback systems. For some reason they still build speaker into cars ...

Rob[atSound-Rehab]

BenMMusTech
 
Now, this is why I was fiddling around with the mix in mastering last night, because I was incorporating a new mastering EQ into my chain (I don't fiddle often), but I was really struggling with the Waves RS56...it works great off the bat for sound art stuff mind you, but for the life of me I could not get the mix to sit with my normal mastering chain.  Normally after the console emulator, I would use the Pro Channel EQ, and I was trying to do the same with RS56.  Being a complete passive type unlike the Pulteq...makes it harder to set IMO.
 
Eventually I gave up, and continued with normal chain so...Pro Channel EQ set to pure but leaving room for the RS56 later, then I use the Waves H-Comp as my mastering compressor...unlike you though I don't push the signal into and make the compressor do the hard yards...for me its small increments and gain staging.
 
Then I used the RS56, and this is where it worked in the chain...mind you it took some tweaking.
 
After this, I use Waves Master Tape...and no this is not an advert for Waves ;)...Sonar is still my go to piece of software tech...but I gain stage between RS56 and the Master Tape...adding in an extra DB of volume after the gain stage, and use the wow and flutter to dial in some punch, and tape hiss helps the top end come through...I know but I can hear it...and the AKG K12s aren't bad cans.
 

IMHO you are "sound shaping" your master bus, or whatever other name one could find for throwing a ton of FX at the master bus.
 
There's nothing wrong with that from a creative perspective - if it sounds right, it is right (and what sounds right is very flexible in the creative domain).
 
Yet, this has nothing to do with classical mastering where you finish the piece of art first, then have it polished so that it sounds as good as possible across playback systems. For some reason they still build speaker into cars ...

Yep, you're right in regards to mix must sound good across a number of systems, and from experience for my stuff...if it sounds right on crap speakers...it will transfer to other system.  I might have left too much low-mids in the track I posted a link to in last post...again I'm getting use to a new EQ.
 
But the OP wasn't really talking about classical mastering...he was asking from his perspective, and so classical mastering isn't really relevant is it?
 
And if you think about it...what is classical mastering?  I mean there's the 60's and the aforementioned use of the RS56 for LP, and i'm sure mastering for cassette tape requires a different type of mastering, as cd does, and now the digital medium does too.  As an example and even though Jimmy Page was doing the latest remasters for cash too, it was also to remaster for the digital medium.  Even he gets the digital or to put it more accurately the digitised analogue medium has to be treated in a new way to all forms of analogue mediums, LP and Tape and even early digital i.e. CD.  And for once, I'm actually pretty qualified to explain the difference in mediums...medium specificity is a part of my research degree. 
 
Whilst all classical engineering techniques, esp when getting the signal into the box and back out if this how you mix and master apply...and depending on the type of digital mixing paradigm you use i.e. the hybrid type, where the signal goes in and out of the box and therefore you need to always watch the levels and actually negate both the benefits of analogue and digital, or total digital where you get the benefits of analogue emulation technology by staying in the box once its in, this means emulating the signal level of analogue too, and using an excess of what looks like superfluous and excess amount of plug-ins to emulate things like the fattness of transients.
 
Sure, and again I'm qualified to say this, if I was mastering someone else's material I would judge it on it's merit, and be judicious in my choice of mastering equipment...that's what a good mastering engineer does.  Again, the OP from what I could tell wanted to know from his perspective and classical mastering technique would probably  not apply :)
 
Cheers Ben 
 
 

Bristol_Jonesey
But I also have a Pre Master bus which I use for exporting individual songs which are then re-imported into my album project with the same mastering chain.

 
Interesting. I'm trying to picture how you're routing the tracks and buses. Are the pre-master bus and master bus in series?

henkejs

Bristol_Jonesey
But I also have a Pre Master bus which I use for exporting individual songs which are then re-imported into my album project with the same mastering chain.

 
Interesting. I'm trying to picture how you're routing the tracks and buses. Are the pre-master bus and master bus in series?

should work either way - parallel or in series. either way you can export both without negative impact ...
 
---|---> master
   |---> premaster
 
---> pre-master ---> master
 
 

Tracks > Busses > Pre Master > Master > Mains

Boy, I wish I knew more about mastering . . .
 
. . . which is why I don't really do it. I have tried both approaches being discussed. I think if you master while you mix, with Ozone or a stack of plugins on the master bus, then what you are doing is essentially mixing with Ozone. 
 
If you value "setting it aside" and coming back with a two-step process - then you must have the time for multiple re-renders so you can make inevitable mix corrections.
 
In either case, you can always come back the next day with fresh ears, so the two-step process is not your only opportunity to do that.
 
I was an Ozone abuser on the master bus for a while, but I just got the feeling that it gave me 10 options, two of which were good choices, but you must go through all 10 to get to the two.
 
Nowadays I mix to the best of my ability, put FabFilter Pro-L limiter on the master bus, maybe add a db of gain with that, and call it a day. It keeps my song cleaner - because I've found trying to master without real knowledge is the ultimate rabbit hole, with a screwed-up song at its bottom.
 
-Tom