MPC 2000XL tempo runs slower than DAW

Recently I came to record multitracks from my MPC 2000XL (8 individual outs + main L&R outs) into Logic. I did a single pass of recording all the MPC outputs. I noticed that the multitracks start out in time with Logic's metronome, and then slowly drift away from it. By the 3rd/4th minute the recorded audio was audibly out of time with Logic's metronome. Specifically, the MPC's metronome fell behind Logic's metronome. This was quite unexpected to me, considering that the MPC's BPM was set to the exact same value as Logic's BPM.

The TL;DR of this post: I am keen to know from fellow 2000XL owners, whether this flaw is present on your 2000XL units too or just mine in particular?

Some setup info
I'm syncing my MPC to Logic using Linear Timecode (LTC), which is listed as "SMPTE" in the MPC's MIDI/SYNC menu.
This means the MPC is chasing Logic. Logic = master, MPC = slave.
The LTC audio file was generated from https://elteesee.pehrhovey.net/, choosing 25FPS, 44.1kHz, 16-bit.
I dragged the LTC audio file into Logic and routed it out of my audio interface (RME Fireface UFX II) to the MPC's SMPTE input.
The MPC is configured to receive SMPTE in at 25FPS.

I understand that no two clocks will ever be perfectly in sync and eventually start to drift from each other. But I am a bit disappointed at how quickly the MPC started to lag behind Logic's tempo. As a control, I ran a metronome app on my phone alongside Logic and they did not drift apart at all for as long as I ran them (~10 minutes).

To rule out the possibility of this being a quirk with the method of sync I am using, I set up an experiment to test all the possible sync methods (including no sync): I created a 2 bar sequence (drum pattern) at 90BPM in the MPC. Sequence uses kick, snare and shakers. Only 1 shots, no loops. As a bonus, I also applied a swing of 63%, sounds nice and swung. This way I could also test whether the swing is impacted by different sync methods. Went into song mode and made a song arrangement by repeating that 2bar sequence for ~90 times. Added a 1 bar empty sequence at the start of the song to give the MPC time to latch onto whatever sync data it receives. It equated to a song duration of around 8 minutes. For anyone who wishes to check the results for themselves, I have uploaded the Logic session to my dropbox for a limited time (will need to remove it soon due to storage limitations): https://www.dropbox.com/scl/fi/acrf3wdblgdu9i26fkign/MPC-Sync-Affecting-Swing.zip?rlkey=lxsyljiu6mf0cbpujj4682guz&st=93j5rg4h&dl=0. Non-Logic users could also check this out by dragging the audio files into your DAW of choice. I think the audio files are named appropriately.

Results
NO SYNC: Groove/swing sounds totally preserved (as it should be since there's no sync happening, it's the raw sequencer). Since there's no sync, once printed, I lined up the first drum hit with Logic bar 2 so that it lines up with the synced passes below (remember bar 1 is reserved for pre-roll). MPC's drift from Logic is the worst here. Already at 1:30, the MPC sequencer is lagging behind Logic's 90 BPM metronome. By 7:00 it's a train wreck.

SYNC USING MTC @25FPS: Groove/swing sounds totally preserved here too to my ears. MPC's drift is bad here too but not as bad as NO SYNC. By 7:00 it's still a train wreck though.

SYNC USING LTC @25FPS: Groove/swing sounds totally preserved here too to my ears. If I play this on top of SYNC USING MTC @25FPS:
(1) They never drift apart. Meaning they both drift from Logic's tempo at the same rate.
(2) A lot of hits don't quite hit perfectly together, though they're very close. This is probably just revealing the MPC sequencer's jitter, rather than the groove/swing being changed.

SYNC USING MIDI CLOCK & SPP: No drift, perfectly in time with Logic's metronome. Groove/swing has clearly been bastardized. It's quite obvious that the kick sounds straighter (less swung) despite Timing Correct being the exact same. I think the swing may have been messed with due to the jittery MIDI Clock from Logic. Apparently it is common for DAWs to output jittery MIDI Clock due to CPU load, background tasks etc. One solution I see online to counteract this is using hardware which outputs a jitter-free MIDI Clock from your computer. But I do not have such hardware so cannot verify this myself.

At this point I have presented my findings and invite everyone to check out the session files for themselves and give their 2 cents. I hope that if anyone is using MIDI Clock to sync their MPC, they are aware of the perils!

Hi Quentin,

Thanks for sharing this post and your findings. I will see what I come up with, but it may be a minute. I just received my MPC2000XL and am on a bit of a learning curve.

Some initial thoughts that might help here. Posting what computer (mac/pc), operating system, sound card, and DAW could help target potential known issues with any hardware or software. Have you troubleshooted any potential known issues first?

Have you thought about trying an external clock such as the Sim'n Tonic Nome II? The reason I am bringing this up is that there are particular pieces of gear that don't play nice when slaved to other devices. An external clock might help tighten things up.

My last thought is that it may take recording into your DAW of choice and nudging the tracks to get things to line up.

I will dig in more at a later date when I have some additional time to think this through and see what I come up with.

Here is what I came up with:

Firstly, what I am running:
mpc2000xl running 1.20
Mac mini m4 running Bitwig Studio and Cubase.
SSL Big Six as mixer/DAW audio interface

First test: Record a 4/8 bar house loop
mpc2000xl as master; Bitwig Studio as slave; sync method - midi clock

Results: When I tried recording a 120 BPM 4/8 bar loop from the mpc2000xl into Bitwig Studio I encountered what I believe is MIDI clock jitter. The recorded audio was off and I can't quantify how many milliseconds it was off, but enough for me to believe it was a fail.

Second Test: Record a 4/8 bar house loop
Bitwig Studio is master; mpc2000xl is the slave, sync method - midi clock using CME Pro U6MIDI Pro Smart MIDI Interface.

Results: I recorded an 8 bar loop into Bitwig Studio and cut it down to a 4 bar loop. Recording using Bitwig as the master and mpc2000xl as the slave produced the desired result.

Concluding thoughts:
I found that the MPC2000XL wasn't necessarily the best master when recording all 8 outs into Bitwig. When I slaved Bitwig to the MPC's MIDI clock, the BPM display would constantly fluctuate and recorded loops didn't line up as cleanly over 4 or 8 bars. When I switched things around and made Bitwig the master and the MPC the slave, everything locked up much better and recording became a lot easier. The MPC still kept its groove and swing, but the DAW timeline stayed rock solid.

For a hardware-only setup I'd have no problem using the MPC as the master, but for multitrack recording into a DAW, I can definitely see the advantage of letting the DAW run the clock. I can also see where something like an E-RM Multiclock would make sense if you're trying to keep multiple hardware sequencers, drum machines, and a DAW tightly synced while still preserving a hardware-centric workflow. In my case, simply making Bitwig the master solved the problem, but I can understand why people running larger hybrid setups invest in dedicated clocking solutions.

In my experience, the 2000 and 2000XL clocks were always a bit like this.

I learned it the hard way years ago. I'd record a stereo mix into the DAW so artists could track vocals, then later come back and dump the individual outs for mixing. The new pass would never line up perfectly with the original stereo recording. It was close, but close enough to create audible phasing if you layered them together.

That was enough to convince me that the timing wasn't reproducing exactly the same from pass to pass.

I vaguely remember experimenting with SMPTE as well, including recording the SMPTE signal into the DAW and using it for sync with the XL. It's been a long time, but I eventually stopped chasing perfect synchronization and switched to a simpler workflow: record the core sequence, line it up in the DAW, then build the arrangement there.

I also remember some very thorough timing tests on Gearslutz years ago, possibly from E-RM. If I'm remembering correctly, the MPC 4000 actually came out on top, with the MPC 60 and MPC 3000 very close behind it. The 2000-series did not fare quite as well once people started measuring things objectively instead of relying on reputation.

Interestingly, that lines up with my own experience. I don't remember running into the same issues on the MPC60, MPC3000, or MPC4000. Those machines seemed much more consistent when doing multiple recording passes.

Based on your testing, nothing here jumps out as a faulty unit to me. It sounds pretty consistent with what I remember from the 2000XL.

Great work on the testing, by the way. It's refreshing to see actual measurements instead of folklore.

My 2000 doesn’t have the individual outs so I have to clock mine synced to the daw. I use a nome 2. It’s solid. The groove is retained.

I did one take and soloed each sound every 4 bars and then I did a few more takes with the samples where I messed with the filter.

Came out sounding nice.

But if I had the 8 outs I’d probably skip the sync and just record in unsynced to the daw. 1 take, all outputs printing at once. Then I’d grab the wavforms, and line them up on the grid. No need for sync since I’m not going to record any more takes on top.

I’m going to do that same technique with the mpc x. Just use the outs and do one take. There is no real need for slaving the mpc in this scenario, imo. I do the same with the sp1200.

There is a pretty awesome tape mixer daw by gulf coast synthesis that is amazing for this workflow. Hit record, press play on the mpc and now you have all the outputs recorded separately plus it imparts a hyper realistic tape sound as well. Worth checking out.