Regarding the LA2A Leveler

[stephentayler]

Somewhere on this thread

http://www.motunation.com/forum/viewtop ... =0#p285986

Cheers

Stephen

[Shooshie]

I haven't tried Magic Dave's instructions yet. I'm looking forward to doing that.

Yeah, me too. I searched this post and Dave's other posts, but I can't find these "...instructions...".
Sorry to be a pain, but has anyone got a reference?

I don't really remember. If/when I ever create a "tutorial" for this, I'll have to search the whole forum for any reference, because I think Dave's suggestions were actually spread out over two or three posts. Each was complete in its own context, but together they gave a more complete picture.

Shoosh

[Eleventh Hour Sound]

I haven't tried Magic Dave's instructions yet. I'm looking forward to doing that.

Yeah, me too. I searched this post and Dave's other posts, but I can't find these "...instructions...".
Sorry to be a pain, but has anyone got a reference?

I don't really remember. If/when I ever create a "tutorial" for this, I'll have to search the whole forum for any reference, because I think Dave's suggestions were actually spread out over two or three posts. Each was complete in its own context, but together they gave a more complete picture.

Shoosh

That would rocketh!

Let us know when you take a journey into the heart of that optical circuit!

[BKK-OZ]

I've done that - I went thru about 25 or so of his last posts, but I couldn't find it.
I might be a dummy tho, so maybe someone else can get the job done.

[wurliuchi]

Here's what Dave said (pasted in):

The MW Leveler behaves exactly like a real LA-2A.

"Warm up time" is important to understand. "Cool off time" is also important to understand.

The MW Leveler models a photo emitter and sensor circuit. The way that circuit works is that input volume over time will saturate the photo sensor. The amount of saturation is based on a number of variables, including how much gain reduction you are applying, the transient nature if the input signal, average signal strength, and time. That's the way the LA-2a worked and that's how the MW Leveler works.

The LA-2A is always adjusting to input, so it is always in "warm up" mode when signal is going through. When signal stops going to input on the LA-2A, the photo sensor starts to cool down. This cool down can last as long as a minute or more. Again, how long depends on the characteristics of the signal before signal stropped coming in.

With DP 6.02, we added a feature to make this a little more managable. The added feature was completed just as DP6.02 was ready to ship, so we didn't get a chance to include documentation on the feature.

If you Control click or right click on the MW Leveler meter, you get a sub menu. The sub menu includes the option to save the T4 state. When you select the save option, you are saving the state as it exists at that moment. So, if you play back one bar of signal through the Leveler and save the state at that point, that would not be the same state as if the Leveler had been getting signal for the length of the song. Remember, the Leveler is constantly adjusting to input signal over time.

One thing you don't do with a real LA-2A is non real-time bounce to disk.

Once you have saved a T4 state, that state will be reset in the Leveler under specific conditions. If you bounce to disk, that causes the Leveler to reset at the start of the bounce. That way, if you bounce from a cold start, the Leveler will already be "warmed up" for the bounce.
If you close the Leveler window, that will reset the T4 state.
If you automate a bypass on the Leveler, the T4 state is reset when the plug-in is un bypassed. So for example, let's say you have a vocal track that comes in half way through the song. You can play back however much of the vocal track you want to get the Leveler warmed up. Save the T4 state at that point. Now automate a bypass for the Leveler that unbypasses just before the vocal comes in. The Leveler will be warmed up when the vocal track comes in.

It's important to understand that the MW Leveler is not an imitation of an LA-2A. It is an actual model of the hardware. It does work the same way as the hardware. If it didn't, it would not sound and behave like the original hardware.
In order to eliminate the warm up and cool down times, that would drastically change the way the Leveler reacted and
sounded. Remember, the way the Leveler is reacting in the first bar of music is not necessarily the way it is reacting after 3 minutes of saturation. Even if we instituted some kind of lookahead feature (which would add significant CPU overhead), where would you want to look ahead to? 30 seconds into the track? Two minutes?

The MW Leveler is a musical instrument. It has it's quirks, and they are identical to the original hardware. If you understand how the device works, it is a very powerful musical instrument. The LA-2A is not the famous device that it is for no reason.

The Save State function will give you consistency for bounces and track playback without having to "warm up" the cell every time.

Dave

After having lots of problems with levels jumping, I had inserted some standby automation for the plug in thinking that would solve the problem. (Can't remember where I read it, but I understood that keeping it in StandBy mode would keep it "warmed up"

No, stand-by mode does not keep the Leveler warmed up when the track is not playing.

If you bypass the Leveler (with no saved T4 state), the cell starts to cool off, even if there is signal present. Therefore, if you are comparing dry to processed signal, every time you bypass and then unbypass, the T4 cell is cooling off then warming back up again.

If you put the Leveler into standby mode, signal is still sent to the T4 cell, while the track is playing, although you are hearing the direct input signal at that point. Stand-by allows you toggle the Leveler on and off without it having to warm up every time you turn it back on.

If you have saved a T4 state, bypassing then unbypassing the Leveler resets the T4 to the saved state.

Regardless of whether you have a saved T4 state or not, going in and out of stand-by does not reset the T4 cell.

Dave

Dave,

Please tell us about the pre-checked 'automatic memory restoration' option.

This is what loads the saved T4 state when you unbypass the Leveler, or Bounce to Disk. If the option is unchecked, the only way a saved state gets reloaded is if you manually select that in the options window.

You also don't want to use a saved state from one model of the T4 with a different model of the T4. In other words, if you save the state with Fast Modern as the chosen model, you won't like the results if you change the model to Slow Vintage and then restore the saved state.

It's useful to know that if you save an MW Leveler preset that includes a saves T4 state, the saved state is part of the preset.

Remember, there is no generic "warmed up" mode. When signal is hitting the T4, the T4 is constantly adjusting it's response.

Let's say that you have a vocal track that starts with the singer screaming loud, but at the end of the song, they drop down to a quiet performance for the last few bars (or maybe the opposite - the singer starts soft and gets loud at the end). In such an example, it would make a big difference where you chose to save the state of the T4.

Dave

Thanks. So, if no state has been manually saved, can 'automatic memory restoration' do anything at all?

If no state is saved, there is no memory to restore.

Sounds like one would need to dump any saved states included with saved presets, when using them on a different session. Is that correct?

I agree with that. The saved state is specific to the track content and the specific place where you stopped and saved.

Dave

We were working on getting levels for BGVs that only appear in the choruses of the song. When we were setting levels we were playing the chorus over and over, (which I guess saturated the MWL to a certain degree) but when I did a bounce to disk, it seemed I was getting different levels. (Maybe because the saturation was different since there was no signal for a minute or so up until that point)

-Vincent

When you bounce, the state is reset at the start of the bounce. Regardless of how much saturation you do before the bounce, you will get a reset when you do the bounce. That's the biggest reason to be able to save the T4 state.

Dave

It would be nice if the saved T4 state would be saved along with the sequence and could be saved as a normal DP Preset for the MWL..

Where would the automatically saved state come from? One minute into the song? End of the song? If you tried to take an average of the input signal from beginning to end, that would mean the state wouldn't represent any specific part of the track at all.

The best approach is to understand how the Leveler is working, and then make it work for you.

Dave

I discussed with an engineer in a studio full of outboard equipment.

We were both laughing about it.

Here's the problem with the premise of the plug-in.:

If you want to emulate a real studio processor, the plug-in should start warming up from the moment you turn your computer on.

Nope. Go back and reread my first post.

The "warming up" that you describe has nothing to do with powering up an LA-2A. There is no "warm up" involved at all. We are not talking old MiniMoogs here.

Over time, the T4 cell reacts to it's input. If the T4 cell sits there for three days powered up, but with no signal going through, it is not "warming up" in any way at all.

Once signal hits the T4, the T4 starts to react to that signal. As long as the T4 is receiving signal, it is constantly adjusting it's response. When the signal ceases, that T4 cell starts to "cool down".

At any given point in a track, the T4 will be responding differently, based on all previous input up to that point. The best way to understand this is to think of glancing up and looking at the sun for a quick moment. You look away and see spots until your eye recovers. If you spend longer looking at the sun, your eye takes longer to recover when you look away. That's exactly how the T4 works.

If you changed the characteristics so that there was no memory in the T4 cell, it wouldn't sound like an LA-2A.
Anything that doesn't work the same way as the original T4 can not claim to be a true model of the T4.

One suggestion I saw involved an analysis of the entire audio track that could be used as a reference. That is still making the assumption that you want the Leveler to sound as if it were starting from a cold start on the track. What is more likely is that you will want the leveler started when the cell is already in a reactive state. Therefore, you have to decide what that reactive state will be. Do you want the state the cell was in at the end of the track playback? Do you want the state the cell was in after a bar of playback from the beginning of the track?
This method would also require that the entire track be recalculated every time you changed a parameter in the Leveler or any plug-in in line before the Leveler.

Here is a metaphor:
A guitar has tuning pegs. If the tuning pegs are not set correctly, the guitar is out of tune. Therefore, does that mean that tuning pegs are a bad idea and should be replaced by a system that automatically tunes the strings? Well, that sounds like a nice idea until you examine the question of alternate tunings, the ability to grab a tuner and crank it while playing, etc. In other words, the technology as it exists offers specific capabilities. If you change the technology, you change what that technology can do.

Just because Gibson makes a robot guitar, that does not mean all my other old guitars are unusable now.

Dynamics processors all have their flavors. The MW Leveler is a model of the LA-2A. If you don't want that flavor, there are other choices. If you do want that flavor, you get it with the MW Leveler.

Dave

If we accept, for argument's sake, that the model does work like the real thing, then compare people's experiences with the real hardware, we can only assume there's some other intermittent bug that's not flushed out yet. You really don't hear people complaining about hardware LA-2A behavior.


Someday I'll get the pair of hardware LA-2A's I've built fully wired into their case, and do some comparisons. My T4's are a recreation built by someone who's reverse engineered their own version, using correct materials. They may well have different response times than any of the four provided models.

Two notes to add here.

The MW Leveler is not just one model. We found that from year to year the T4 cells differed significantly. My understanding is they were hand made components. The different models in the MW Leveler are very different in their characters. Switching between the models while using the same parameter settings will give very different results. That's not something you can do with the hardware model.

One of the advantages of the LA-2A is that you can use it for drastic gain reduction with very good results. Try 20db of gain reduction with a fixed-attack compressor and see what happens. You won't like it. You can do 20db of reduction with the MW Leveler and get great results. But if the cell is not already in reactive mode when the signal comes in, the Leveler has to get back to a saturated state before it can do that amount of reduction without a volume bump at the start. I suspect that the people who notice a volume bump on the Leveler are using it for extreme gain reduction.

Note two:
It's not just the T4 cell that makes the sound of the LA-2A. The T4 is part of the gain reduction circuit, but it is not the only contributing factor to overall sonic response. For example, take a look at the UAD LA-2A plug-in. See the graphic of the set screw on the front panel? Nice graphic emulation of a parameter that is found on the real thing, but the graphic is just that - a graphic. It doesn't do anything. The MW Leveler, on the other hand, has the actual Response knob, which does do what that parameter does on the hardware unit.

We also used original vintage copper in the wiring of the MW Leveler, but I can't tell you too much more about that...

Dave

Dave, would you concede that people should be told that Bounce To Disk does not work when using the MW Leveler?

Shouldn't there be a little warning in the Bounce To Disk window that says "If you are using any instances of the MW Leveler, this bounce will not sound like what you have been listening to."?

I mean, isn't this a fair thing to alert the consumer to, since Bounce To Disk is touted as a viable way to print mixes?

I bounce all the time using the Leveler. I just finished a 50 song remaster for Sony that used the Leveler (with no saved states). So no, I don't think people should be told that bounce doesn't work with the Leveler.

Yes, the more documentation the better. As I mentioned earlier, the new save state features in the Leveler didn't make it into the 6.02 update notes because of the timing of the release.

Dave

That's an interesting idea, and falls in the category of "I would never think of that because I would never do that". I use an LA2A for 2 things; barely bump it to get maybe 3db of comression/limiting, or smash it all the way to mix in parrallel. I would never use it for heavy steady compression on a vocal for instance, the volume will be all over the place. The middle range is no man's land for me on that unit.

That's making me thing that maybe some people just aren't used to how an LA2A works. If you're putting it on a vocal put it in limit mode and hit it lightly.

bb

Very useful comment. Thank you!

Dave

Whew! Thank you, Dave.

[Shooshie]

Wow. Thank you BOTH! I knew it was more than one post, but I guess I lost count after two or three.

PS: and into the Tips Sheet thread it went! Thanks, Wurliuchi!

[BKK-OZ]

All hail the great Wurliuchi!
Thanx VERY MUCHO!!!!

[wurliuchi]

Happy to help. I can copy and paste with the best of 'em.

[James Steele]

Yes thanks!

[kassonica]

Thanks and thanks again

[busle]

the kidz theze dayz

I was kidding.

[Eleventh Hour Sound]

Thanks!

[kassonica]

ANother tip for the Leveller is the response knob makes a very subtle desser.

It's default is all the way to the left which as I understand it limiting in some way the high frequency content.

You turn it all the way to the right as you hear the high frequency's content a little more.

Of course it's dependent on the amount of gain reduction but on highly sibilant vocals I've found a very subtle deessing function with the response know about 11 o'clock and maybe one or 2 DB of gain reduction either in the vintage fast mode of the slow modern mode.

[EMRR]

All the way to the left (7 o'clock) should be full range, anything more clockwise is an increasing high pass filter on the side chain, which results in 1) bass notes causing less compression, and 2) less overall compression occurring, unless a corresponding increase in gain reduction is made to counter the level difference in the side chain.

The LA-2A manual and schematic are both online, for any who wish to understand the inner workings of the beast.

[rthorderson]

All the way to the left (7 o'clock) should be full range, anything more clockwise is an increasing high pass filter on the side chain, which results in 1) bass notes causing less compression, and 2) less overall compression occurring, unless a corresponding increase in gain reduction is made to counter the level difference in the side chain.

The LA-2A manual and schematic are both online, for any who wish to understand the inner workings of the beast.

Are we sure about that? According to the LA-2A manual (since we have no other manual really), it says...

"This potentiometer is factory set for a “flat” side-chain response (clockwise). Increasing the resistance of this potentiometer by turning it counter clockwise will result in compression which is increasingly more sensitive to the higher frequencies."

If I am reading that correctly...that is backwards to how you described, right? I wonder which way is right...