MPC 2500 Midi Beat Clock slave to Pro Tools Critical test

Got a very important question and hope some smart folks here can help me. This i scientific so don't hate if you dont get the purpose of this exercise.


I am considering a 2500 BUT need to find out how it behaves when slaving to Pro Tools via Midi Beat Clock.

If someone works in that way can you please check what happens when you loop 2 bars in PT and run a 2 bar pattern in MPC:

Can you tell me what happens when you hit play in PT, MPC should start and loop around. Does the timing start drifting after a few loops or does it stay tight ?

when running a click on both and monitoring the click from MPC via Audio in there should be a delay (midi + audio latency) this can be measure and compensated for in PT via Midi Track offset in Events menu. Adjust depends on HW buffer size and midi interface latency.

I am really interested in how stable this setup would be , I know the MPC5000 has really bad midi beat clock and is useless in that type setup.

MPC4000 and MPC3000 are meant to shine here.

Thanks for the time.

tx . will PM him lets see if he gets bak.

I have a 2500 synced to Pro Tools. The timing sounds ok to my ears but I never got scientific with it. I totally get why you would want to know the latency, and it's probably something i should check out myself. So I'll get my lab coat on tomorrow and post back in this thread.

Thats sounds really good, Thanks.

Ok so let's get scientific.

First off my spec. I'm running Pro Tools 10.1 on a Mac mini with OSX 10.7.3. My audio/MIDI interface is a PreSonus StudioLive 1602 (a 'king beast if anyone else is in the market for an overpriced interface), which goes in through the FireWire bus. I used 24/44.1k and a Buffer Size of 128, and no Delay Comp. My MPC is a 2500SE (maxed out memory etc) running the free JJOS, but I think everything we're doing here is the same between AkaiOS and all the JJs.

Usually I work with the MPC as the MTC Master. I *think* this way preserves the MPC Swing, but don't quote me on that. But it was easy enough to swap around and make Pro Tools the Master.

So I set up PT to bus a Click track to a mono audio track, and to record the MPC's click on another mono audio track. Then I hit Record, recorded a couple of bars, and zoomed in. See the screenshots below.

Note that there is always a slight latency between Pro Tools instrument tracks (Click is an instrument) and external audio tracks. This is Pro Tools' fault, and Avid and users are aware of the problem, as seen here and here. However...










At every beat, the Pro Tools Click is right on the grid, sample accurate, as you would expect. The MPC drags by a varying amount every time. Some of this will be latency, either between the MIDI TC out or the audio in from the interface, which is acceptable. However, it varies by such a wild amount every time that they aren't keeping good time with each other at all.

I think it's wrong to assume that this means the MPC's clock is junk. Just because Pro Tools can keep time with itself (the click matches the grid) doesn't mean it can keep time with reality, or anything "outside the box". Whichever system is to blame doesn't matter, though, as I don't think PT is going to be a good master of the MPC's clock.

I'm going to post this now because I'm sick or storing drafts, but in my next post I'll make Pro Tools the slave to the MPC's clock and see how that changes things. So stay tuned.

Hi Earhacker,

Thanks for doing this, can you just do a chop and a spot to see how many samples we are talking about, on those hits that are off. We can clearly see that there is a jitter, some are on beat and some are off.

Latency is not so much an issue as it can be compnesated for using the offset, its the jitter that throws things off as its random.

I assume as some of these you are showing are spot on you have applied a midi beat clock offset ?

What happens when you let the MPC click and PT click loop lets say 2 bars, does it stay locked in or start drifting on the start of the loop, there was this issue where PT sends a stop and a start message and that throws things off when looping a section.

Curious to see how many samples you are off on this test where MPC was salve to PT.

THX !!!!

e9000k wrote:can you just do a chop and a spot to see how many samples we are talking about, on those hits that are off. We can clearly see that there is a jitter, some are on beat and some are off.

Actually, not much at all. The one at 1|4|000 looked the worst to me, but it's only out by about 72 samples (Tab to transient didn't see the MPC click, so I did it by hand). At 44.1k that's less than 2ms.

Latency is not so much an issue as it can be compnesated for using the offset, its the jitter that throws things off as its random.

As far as I know, Midi Track Offset only affects incoming MIDI notes, and has no effect on outgoing MTC. That's the one in the Events menu. Am I looking in the wrong place?

I assume as some of these you are showing are spot on you have applied a midi beat clock offset ?

All the shots are from the same two bar loop, and there's no Offset turned on apart from any defaults I don't know about. I'm pretty confident there's nothing on.

The click at 2|0|000 being bang on the beat, I think that's just a fluke. I just ran it again and it was off again.

What happens when you let the MPC click and PT click loop lets say 2 bars, does it stay locked in or start drifting on the start of the loop, there was this issue where PT sends a stop and a start message and that throws things off when looping a section.

I'm not sure how I could measure this accurately. I can't hear any drift, but we're talking about differences of ~2ms, I wouldn't hear it! Gimme a step by step and I'll run it for you.

THX !!!!

No problem

Earhacker wrote:

e9000k wrote:can you just do a chop and a spot to see how many samples we are talking about, on those hits that are off. We can clearly see that there is a jitter, some are on beat and some are off.

Actually, not much at all. The one at 1|4|000 looked the worst to me, but it's only out by about 72 samples (Tab to transient didn't see the MPC click, so I did it by hand). At 44.1k that's less than 2ms.

That is correct, with midi beat clock the first beat will be off as the MPC takes a moment to understand the tempo that the DAW is sending. This is common to beat clock. Ignore the first everything thereafter should be fine in a perfect world.

Latency is not so much an issue as it can be compnesated for using the offset, its the jitter that throws things off as its random.

As far as I know, Midi Track Offset only affects incoming MIDI notes, and has no effect on outgoing MTC. That's the one in the Events menu. Am I looking in the wrong place?

Yes you are right, only the offset in the Midi Beat Clock settings box affects midi clock. The other stuff is just for normal/manual midi delay compensation

I assume as some of these you are showing are spot on you have applied a midi beat clock offset ?

All the shots are from the same two bar loop, and there's no Offset turned on apart from any defaults I don't know about. I'm pretty confident there's nothing on.


Ok, you can use the offset in the "Midi Beat Clock" menu where you select the interface to which Midi Beat clock i send by PT - its a minus amount to send the clock signal early so the machine then works "in time" with PT


The click at 2|0|000 being bang on the beat, I think that's just a fluke. I just ran it again and it was off again.

Agreed, just a lucky one.

What happens when you let the MPC click and PT click loop lets say 2 bars, does it stay locked in or start drifting on the start of the loop, there was this issue where PT sends a stop and a start message and that throws things off when looping a section.

I'm not sure how I could measure this accurately. I can't hear any drift, but we're talking about differences of ~2ms, I wouldn't hear it! Gimme a step by step and I'll run it for you.

Yes the only real way to measure it is to record the mix out of PT to a 3rd device and then import that wav into PT and tab to transient for example. When you set the correct offset up in PT you will hear that the two clicks overlap and sound virtually more like 1, with no or hardly any flams. Last time I did this I adjusted the offset so majority of the notes were on time. then it would loop around and fall out of time every few repeats, 2 repeats on time , 3rd repeat totally off and so on... I was told by some synth dudes that its because PT is sending start and stop messages at the repeat sections and, it did not do this in version 6.4 downwards.

THX !!!!

No problem

Please see if you can adjust the midi beat clock offset so it compensates for the majority of drifts, there will still be the odd ones due to the nature of MPC listening to the not so accurate clock send through the MAC.

Keen to find out if you can also notice those fall offs when looping. This would be a big reason not to buy one..

THX again, this is really helpful, bottom line so far is that its at worst 72 samples out, this is not all so bad for the MPC2500 and JJOS - may be different with AKAIOS

E900k,

I just saw your PM so apologies for the delay in getting back to you.

I have never used a 2500 and subsequently have not been able to test its clock. But I can safely say that no MPC ever made has an abysmal clock such as the one inside a TR909 and many other hardware sequencers boxes out there. Some MPCs internal clocks are tighter than others though. The tightest being the 4k, 3k, and the 60 and im pretty sure the 5000 is also up there under some circumstances. The others are all tight though, its just those 3 mentioned are a few samples tighter than the others. The 4k, and the 3k are at near sample accuracy tightness which is quite stunning really. However, if any of those machines are sent jittery midi clock from a Daw they lose their tightness because their clock is now controlled by an external source.


Your absolutely correct about the random nature of sending midi clock from your Daw to your MPC or any other gear for that matter. You , me or anyone else will never get stable, locked on sync from any Daw sending midi clock, or midi time code. But I dont deny some people get by quite well and are happy enough with a little bit of slop here and there. Im not, im obsessive about this

There are some things you can do to help the situation such as not using a USB midi interface which can compound the problem. If at all possible use a PCI card midi interface. Although with many folks using laptops, and Imacs that really isnt possible at all.. Another option is to send midi clock or time code from your MPC to your Daw. Providing the MPC can send stable midi clock or midi time code you can get on pretty well with that system. However, this has some drawbacks I dont like the smell of. Anyone doing this will notice the BPM counter on their daw fluctuating. In some cases it can be pretty wild. Its not just a little jitter from the MPCs, its also (again) the midi which is now sent to our Daws, and again it will be worse via a USB interface. But at least the MPC is playing in time. Hmmm, well yes and no. Remember that fluctuating tempo display. That means the audio recorded is kind of being recorded with a constant run of subtle tempo changes. It may sound Ok to your ears, but after subsequent layers of tracks the effect in my experience is the same as sending the MPC midi clock. So, if you then go to use a VST which is sample accurate it will always be in time and would fit perfectly on the Daws grid/time line but your audio recorded from the MPC wont be. Hence why the majority of people seem to end up editing audio. My biggest beef with this is dont mess with your Daws sample accurate audio clock. Sending the Daw midi clock is effectively doing just that, and it can cause some audio issues which are to much to go in to here

So, how do we get 100% stable midi clock from our Daw. Well some genius inventor who is also obsessed with timing and hated the fact that in this day and age we cant get stable midi clock from our Daws came up with a solution. This solution works 100% and it means your MPC receives sample accurate midi clock from your Daw. Thats no lie! However, its not free. It comes at a price. Its called Innerclock Sync Gen 11. IMHO, its the most marvellous box . I never ever have to edit audio to fit the time line now. I sync my MPC4k with my Daw and its just bang on 100% all the time. I dont work for the company, have no agenda or anything like that but if sync is important to you (as it should be) you owe it to your self and your MPC to check this box out. www.innerclocksystems.com

I would also recommend having a good read through parts of the site. The owner has very good insight in to why we need accurate sync and the dramatic effect all those little nuances of timing drift can gave on our creativity and of course the music.

For me it was about integrating my Daw with my hardware. I gave up years ago trying to deal with sync issues with my Daw and hardware. So I sold all the hardware and just worked in the box. When I discovered the Sync Gen I started buying hardware again. It was the sole catalyst for me doing that. I think that speaks volumes for the sync gen.

Sorry, it was long one...hey I cant help it Im passionate about timing n shit

Another thing I must mention.

Lets say you make Hip Hop (A lot of folks around here obviously do)

Now lets say your making a track where much of the content has been sampled and your triggering loops from your pads and its 90bpm. Now lets say you have sent the MPC jittery midi clock. Now, whilst some of those notes(ie loops) wont trigger exactly where they should and some are 40 samples late or early. The chances of you noticing that are quite slim. So, you could be OK with your Daws midi clock if you make beats mostly based on loops. However, lets say you programme all your drums, bass line, and just use the odd musical loop here and there. Will you hear any of those drums a bit out? I think that would depend on the song adn the persons ears writing it. Now lets say you make every part your self and each part is your own midi notes. I think you then need a Sync Gen.

Very small timing errors are more noticeable in faster syncopated rhythms. In my case thats what I generally make and I dont use much loops in my music so I really need tight sync.

labcoats, Thanks for the detailed breakdown. Looks like the 4000 and SYNCGEN is the way to go.

The innerclocksystems seem is breaking it down scientifically, the 4k is clear winner when synced externally, 3k and 60k also solid but 5000 is a big fail....

[url]http://www.innerclocksystems.com/New%20ICS%20Litmus.html
[/url]

Does this box also send song position pointers, I could imagine that u drive your MPC off the DAW and move around in long tracks how does your MPC know where to start?

I dont know about that because its not a function I use. I would hasten a guess that sending your Daw midi machine code even with the Sync Gen connected you would be fine because their just start/stop and follow messages. I use the daw transport as the master and the MPC follows the song no problem. When looping Daw/MPC its flawless,

The 4k is an excellent machine, I couldn't be happier. I would have no problem using a 60 or 3000 either. For me though their sequencers would have been lacking for the way I work. If I wanted their sound I would get an S900/50 and S3000. The 5000 if connected to the sync gen tightens up in song mode, but I agree, its clock isnt the best. I've been trying to get a hold of a 2500 to perform the Litmus test but have had no luck so far. I've tested an MPC2000xl which produced 46 samples of maximum jitter. Not bad at all.

Thats some interesting stuff.

If you could merge the midi of the SYNC Gen with the Midi of standard Midi interface, send song position through standard midi and beat clock through sync gen. Probably get best of both worlds. Unless MPC4k can accept diff things thru diff ports or even merge itself?

Could you explain your workflow with the 4000 and DAW, would you work on loops in the DAW and then pull up the matching sequence on MPC. Then in song mode its obvious that one would more work across longer parts of the song.

Break it down for me

THanks

Roughly $700US for roughly 2ms of jitter, plus shipping from Australia? No thanks. I don't know of any musicians who can play their instruments to sample accuracy. Why should my sequencer?