Important: 4k bugfix update

Yes foge1 please make a shortlist, so anyone could reproduce it following those steps.

Thanks, Boele

Connect the outputs of your mpc to something you can meter the amplitude of the signal leaving the mpc i.e. a mixing desk.

i) okay assign a snare drum to a pad
ii) set that pad to 16 levels of velocity using the 16 level function
iii) go to the part mixer screen
iv) now set the part volume to max
v) play the snare at maximium velocity pad 16.
vi) meter the output on your analague mixer i.e not mpc.
vii) turn the part mixer volume down (in the mpc) whilst still playing pad16 I see no change at all in output for the first third of the fader movement.
viii) move the part mixer fader down until you can meter a change in output (still playing pad 16) now move it up just a little until you are back in the area where there is no drop in output level.
9) now leaving the part level roughly a third down at the point where the output from pad16 is still almost max. Now play Pad12 (which will be same snare lower velocity). Meter the output its alot lower! now turn the part mixer fader up till max and I meter at least a 7db change in output.

So the part mixer is not a mixer it seems dependent on velocity!
Across the same range of moving the part mixer fader the highest velocity sees almost no change across the first third, however at a lower velocity level I see what I would expect to see i.e. the amplitude going down evenly across the first third.

I stumbled across this because I was mixing my drum performances and whilst mixing could here changes in velocity relationships.
i.e. whilst turning a drum sequence down I was noticing the loudest velocity hits staying really loud whilst the lower velocity dropped in level. The disturbing part being the loud hits remaining the same level even though I had turned them down a bit.

Like I said let me know if anyone else gets this.
G

Thanks. I will take a loot at this tonight.

foge1, I have been moaning about this issue since 2002.

The whole gain architecture of the MPC4000 is always hitting "ceilings" which means that you can tweak come parameters for ages and nothing happens.

It is this age old thing of there seemingly only being one gain stage, and all the following parameters just act as offsets to it.
The combined dynamic range of all those parameters must be about 400dB or sthg (lol I haven't counted) but the actual dynamic range of the gain stage is sthg like about 60dB.

The parameters include:
Sample amplitude (via mod matrix, log/lin and cc07)
zone level
keygroup level
filter attenuation
program level
part level
(have i forgotten some?)

These all adjust the SAME gain stage, and if it's already up full, then you won't hear any change in (in this example, part level) until you turn it way down.

This is absolutely and totally freaking insane.

If a part mixer cannot act as a part mixer then that as far as I am concerned is a HUGE bug, as who is to say when u hit that gain ceiling.

What do u do just listen out oh yeah thats that bullXXXX gain ceiling again.

That means when programming you have to create your programs with the part mixer set to half way to reliably hear that the patches you are creating are true to what you are doing otherwise any level adjustments you make you may not be hearing!

Thats just about the final straw!
Why on earth would someone design something sooooo badly.
Its just not funny any more, the 4k as anything other than a sequencer has so many flaws u have to work around its just crazy.

g

aye.
Well in 2002 when I first started programming drum machine stuff on the MPC4000,
I was going nuts cos I'd put different drum hit velocities into my patterns,
and there'd be no volume change.
And finally I figured out that I had to turn combinations of part, program, etc etc levels way down, and then the Note On velocities would actually begin to change the drum levels in a linear way from v=1 to v=127.

Trying to program a whole live set was a terrible pain in the @ss cos it was really hard work getting all the levels to be right from song to song -
you couldn't just "turn stuff up" cos it, well, wouldn't.


I've moaned about this from day 1.
Occasionally there'd be vague mutterings from Akai support about it was so you could have 96-voice polyphony or sthg without distortion.
It used to be even worse when the max level you could get back from the FX bus was 12dB quieter than from the main part.

Its the same thing I said earlier
The gain architecture is a mess

This is therefore a bug that needs addressing the gain architecture simply needs more head room and as there is a master pad this is surely easy to achieve without causing distortion through the output stages.

As far as I have read this is the single BIGGEST BUG as you can't even create a program, or mix a program without running into this unnecessary foolishness.

If akai address anything this has to be at the top of the list.

How did they not notice theirs a problem?????????????
G

I wouldn't call this a bug. But I understand the problem (theoretically that is, I still have to test this myself).

And I think this is not as easy to solve as it seems.

By changing levels you will also change the levels of all *existing* songs on mpc 4ks. For instance, it does matter how much of a signal is going into the FXs. Changes there can have drastic results.

Will come back to it. We need to discuss this more extensively. I don't think there is an audio diagram in one of the manuals that describes the flow of the audio on the 4k is there?

foge, you could also turn down the volume settings of each program and multi you create. thats what Im doing for years and it works decent except you have to turn down the global volume each time you edit a sample or audition a sound from the HD because otherwise you'll become deaf. And the MPC4000 is not noisy so you'll still have a clear sound.

I have a folder in the root of the mpc drive called # Templates and its at the top. It contains various multi-, program- and sample-templates. Every time I create a new program, Ill load a template. The volume is mostly set to minus 18, and about minus 10 for each layer. This gives enough room to add gain.

^ That is the obvious work around, I also wouldn't call it a bug (although I'm sure they could have implemented it better). I wouldn't be setting anything to + decibels. You are better off cutting volumes rather than boosting them (esp on a 24 bit machine). If they did allow for the extra headroom, everything would be distorting, or you would have to have your global volume at -36 to compensate.

It seems that many of the gain settings go by DB, however CC 7 acts more like a %. I've found that using CCs to modify amplitude in the mod matrix barely does anything, that was the most annoying part to me.

Complexity is a small price to pay for flexibility. If we didn't want to deal with it, we would have been with those that sold our 4000s years ago and are posting in the 5000 forum waiting for it to come out.

So your template the program level in the part level in the multi window (doing this from memory as I am away from my mpc.) is set to -18db and for each sample your default level you have set is -10db.

And with these setting using the mixer at any setting you never experience the gain ceiling problem? i.e. this is enough.
I wonder how many dB of headroom these setting give u? but then I suppose these are offset values and probably knowing akai do not even correspond to real measurements.

Its crazy when I just want to get on with the music that i have to worry about this.
I wonder if the level you set the master out (global) affects the available headroom too? and I don't mean the headroom before u clip the D/A stage.
I will investigate.

Reminds me of Lauryn Hill singing 'it could all be so simple, but you'd rather make it hard'

I wonder if this poor gain staging affects other mpcs?


g

OK, I've been looking into this too and it is obvious that the gain architecture of the 4k is at least "strange"

Like McSmooth says, running out of headroom seems to be the initial basis for this design. And the various level parameters around the machine makes it complicated.

So, let's think from the coders point of view here:

What could be done to improve this? I am not a "gain specialist" so I have no expertise in this really. Lowering the global volume seems like the easiest way out. Or an adjustable global volume? Does that make sense?
Making connections between the level parameters could be another, though that sounds complicated to me.

Shoot.

I have not a clue about code dudes, however if in logic you can overload individual channels by crazy amounts before any distortion occurs because each channel has a huge amount of headroom.
This can then be compensated for by simply checking that the mix buss isn't clipping and turning the mix bus down.

the same should be possible with the mpc.
The headroom needs to be increased so that only when every gain stage is at its maximum value do you actually hit the maximum output for that part. This increased output could then be compensated for at the mix buss output stage in the global output pad.
So the facility is already there. Give us more headroom and then compensate at the output stage to prevent clipping.
g

@ scd What is the word from your contact @ AKAI? Do they plan on fixing any of the bugs?