Yes, that's what I'm talking about. Everything else is quite obvious. (as is the point of using M/S; as described above in a discussion you obviously haven't read but replied to...)

You must have a differenet massive passive to me then- it's definately not transparent, even using the unbalanced outs. In fact the THD is goes up by 1% as soon as the signal goes through the tubes.

Also, what M/S encoder/decoder are you using? Even the simplest M/S matrix in the analogue realm is adding more op amps or transformers to the signal path. So for this reason it's not quite as transparent as stereo.

But you have already made up your mind regardless of those facts. I'm not sure who you are trying to convince.

You have to get your massive passive fixed.
this is an old graph a made some years or so ago: (I don't remeber why I did it at 10K)
http://www.grandqvick.se/bilderhifi/[email protected]

did you listen to the files I made for you? I think they pretty much says it all.

M/S done ITB.

I just tried an experiment in Cubase, and the results perplexed me.

I split the channels of a stereo mix and sent both panned to the center to one group, then I took the L-R and the R-L (panned left and right respectively) and sent them to another group. I started with both channels set at -6 dB as Tom V suggested, but it sounded nothing at all like the original mix. So then I sent a phase reversed copy of the original mix to another group to see what I had to do to get it to null my attempt to encode M and S. The closest I could get was with my M fader set at -7.85 and my sides set at -10.77 dB. I tried making my M buss both stereo and mono, and it seemed to make no difference either way. The stereo pan law default is -3dB, so that's what I always use.

Can anyone explain why I can't seem to get these to null?

What the??? There is no R-L. It's L-R for Side and L+R for Mid. So your panned group is Mid. The Side needs to go to both channels too for recreating L and R.

There are good articles about m/s around the net, for example here
AEA Big Ribbon Mics ™ Bringing You the Future of Ribbon Microphone Technology

Matti

Sigh. Are you telling me you put L-R on both sides? That would be silly, unless, of course, you reversed the phase of the side for the right side. What do you suppose you get when you reverse the phase of L-R?

Meanwhile, can anyone answer my question?

Are you trying to encode to Mid/Side or encode & decode passing through a Mid/Side state? I can't figure out from your post. If you want to listen to side material, that's strictly L-R. Sent to both channels if panning hard L/R.

At this point
(L+R)(3/4) is being sent to the Left and Right (1/4 less due to panning law)

So here you're sending
(L-R) to Left
(R-L) to Right

Why both channels? It's wrong here because of the panning law. Right now you have going:
Left channel: (3/4)L + L +(3/4)R -R = (7/4)L -(1/4)(R)
Right channel: (3/4)R + R +(3/4)L -L = (7/4)R -(1/4)(L)

Raising or lowering the output channels at this point isn't ever going to null out with the original signal because you have phase inverted "other" channel info in both sides already (since you haven't canceled it out at equal levels.) The easy fix is raise your panned center L+R by +3db and reduce the combined output at the end by -6db (to match original signal levels)

I am trying to derive and manipulate the mid and side information from a stereo mix. You seem to be contradicting yourself with your last two sentences, but otherwise you're making sense, and I appreciate the attempt to explain!

I've already tried raising the center relative to the sides, and it doesn't work - I tried it at varying levels; perhaps my description of that was what you found confusing about my initial post.

At any rate, if I do the same process as I had described above (L-R to the left, R-L to the right and L+R to the middle) , but turn off the pan law, I should end up with 2L on the left and 2R on the right, correct?

Well, I tried it, but I still couldn't get the result to completely null the phase reversed original mix.

Andy,

Not sure if you have tried this. Try sending the L+R to two tracks panned left and right with the setup described in your last post above and see if this nulls. Also make certain that the two sources that you are using for the null test have no latency issues.

The matrix above however is not what I would suggest since there are two S encoded channels, one inverted and one not.

Fair enough. The massive passive I was working with had a clear difference when switching between L/R and M/S. It was more obvious the harder your signal hit the unit. It is of course entirely possible that it was faulty.

Just a disclaimer here though. One persons "transparent" is another persons "damn, something changed!" so stating that I am wrong without you knowing me and me not knowing you is a bit strange, no?

Cheers!
bManic

.. but this is a bit rude, no? I don't know about your monitoring or how you listen to things but I can hear a clear difference between a minimum phase equalizer and a linear phase equalizer when I'm doing mid/side equalizing.

I'll try to describe what I hear:

Example: I use a wide bell, on the side channel, at around 200-300hz to give some nice body to a nearly hard panned palm muted guitar. If in this case I use a minimum phase implementation I hear the sound becoming slightly "detached" from the middle. Like there are now suddenly two different elements, the opposite to "glue". Using a linear phase implementation I instead hear a slight widening, like simply raising the volume of the side channel, without the "detachment". All the original "glue" is still there, the guitars still feel like a part of the center.

I've always assumed that this is due to the phase warp that happens when using a minimum-phase design (digital or analogue, it doesn't really matter.. doing it analogue will warp the phase even more because it is impossible to create two identical analogue paths. There will always be small variances in each channel in the frequency domain and phase). Sure I might be completely wrong about the technical details of why I hear this "detachment" but you calling my statement complete bull is a bit lame.

Why do I call this "detachment" changing the mix? Because it is. It creates two separate feeling elements from one whole and clearly changes the perceived mix balance and coherence in a way that stereo linked equalization does not.



EDIT: btw, I listened to your audio files (at home on headphones) and I can not hear any difference so at unity gain, your unit is very transparent indeed. Do you have the massive passive with the built in mid/side switch or are you running it externally encoded/decoded? We had an external mid/side encoder/decoder which was at the start of the chain and at the end (basically forcing us to run all units in mid/side mode if that was chosen. Yeah, I know, stupid setup but that's how it was). Assuming that manley built their mid/side encoding and decoding right before and after the passive stage, before amplification, it might be pretty damn transparent (though those added components would bring up the noise once amplified).

I'll try to remember to take some examples with me next time I get a chance to work at the facility again so that you get to hear what I was hearing.

Cheers!
bManic

Eh, no. It changes the stereo image. Who mentioned "ruining"? Why suddenly jump to extremes? I only stated MY own reasons for doing mid/side EQing ITB (or with a digital outboard but that's the same thing really).

The subtle stereo shift/spread that happens as soon as you plug in any of your analogue equipment is part of that analogue "mojo" that so many like but it is also a shift that is not always wanted on some projects and hence another approach is used instead.. just like there actually CAN be too much sweet harmonic distortion from sweet sexy tubes.

Cheers!
bManic

I guess the point is that this shift/spread is not always part of the "analog domain."

If fact, it can be completely inaudible.

Think of all the records made completely on analog gear, with no stepped switches anywhere -- eyeballing the console knobs for matching was the closest they got -- and the stereo image can be totally stable and convincing.

Tubes can be nearly free of distortion when implemented properly, but it is sometimes assumed that they are used as an effect.


DC

Easiest way to do m-s that i've found is download voxengo msed for free, duplicate the channels, turn mid down to negative infinity on one and side on the other and you have mid and side with no hassle.

Yes, sorry. I get easily pissed off when someone is questioning my monitor chain, it's my pride and joy. I've been stating how important monitoring is in so many threads now I thought your comment was offensive to me. Sorry.

yeah - but you think a massive passive is transparent!! Only ribbing you. thumbsup <-- people should use this combo more often.

thumbsup ...way too often, in my experience usually by electric guitarists.