I’ve been using an Atari 1040STe with Cubase 3 as my main sequencer for many years, but I would like to get something more modern and immediate. As much as love my current setup, it is slow and tedious. For me, MIDI timing is number one and though I would love to get a Sequentix Cirklon which has tighter timing than the Atari, I’ve been on their wait list for over two years and who knows when one will become available so I’ve decided to look into other options. I’ve looked into the Squarp Hapax, but the PPQN is 192 compared to the Atari which can go up to 384 PPQN. It would be nice to find out what the timing is on the Synthstrom Deluge as I am intrigued by this. Of course the Octatrack Mk II is another option, but I have zero interest in the sampler because I don’t work with anything under 96 kHz, but I’ve heard it’s very tight timing wise and easily available. I used to own both the Machinedrum as well as the Monomachine and I loved using the Monomahine to sequence external gear so part of me feels that I would be right at home using the Octatrack as a MIDI sequencer, but once again, I cannot find any information on the PPQN or latency of it.

There is an extensive list of gear on the Innerclock Systems website which had been tested to determine MIDI latency and the Atari ST outputs 0.28ms on both Cubase as well as Notator. Sadly they have yet to test the Octatrack, Deluge, or Squarp Sequencers. I’ve looked into possibly getting an older piece such as the Yamaha QY700 which can output 480 PPQN, but of course there is the risk of age related issues and continuing to use floppy disks like I am currently with the Atari. I’ve also considered the Akai MPC X or MPC 4000 as well since I don’t own a solid sampler yet so I could kill two birds with one stone on that one. I know the timing of the MPC 4000 is rock solid at 0.06ms, but I could find no information about the timing of the X.

Essentially, I have it honed down to the Elektron Octatrack Mk II, Synthstrom Deluge, and the Akai MPC 4000 or X. Once again, MIDI timing for me is number one. On one end, it’s wonderful to have so many options, but on the other end, it's become somewhat overwhelming.

I would love to hear other opinions and advice about this.

If you go with one or the manufacturer solutions, like the eMagic Unitor for logic (Apple) or the Midex 8 for Cubase, they have specific drivers that add a small look-ahead midi buffer to the ports, which fires off midi data with far tighter timing.

I own a Midex for use with Cubase, which I believe still supports the LTB protocol for the buffered timing.

This would allow you to use a modern DAW, which is Atari Cubase on steroids.

I’d recommend plugging the midi box directly into your computers motherboard, & not via a USB hub.

you could fit a Gotek USB floppy.

an alternative to QY700 is the RM1x - but i think both involve a certain amount of stop/start. apparently the RS7000 can keep running going from play to record. never tried one, currently have RM1x - which is like a more fun version of QY (have QY100 also), with more sections per style, and the ability to mix 'phrases' of different lengths (up to 256bars iirc), plus some groovy MIDI modifier controls that can be useful to explore. never save, just power cycle and you're back where you were. (could run that creatively with an alesis mmt8 synched?)

I'm in a similar situation and have no substantial data to add.
But I 've ruled out the Octatrack and the Synthstrom, just one midi out, too little for my taste.
So that leaves MPCs, I have no experience with the new ones.
But unlike you, I mainly need it for live situations.

You might also want to checkout a Roland MV8800. If you only want a midi sequencer and the sampler is the bonus, then it's a great machine. As a modern production centre, the load times are too slow. But as a sequencer it's pretty tight and swings nicely.

It has a VGA output, so it's like using the Atari in some ways. The tactile feel of using the unit is great and it's a different experience from a conventional DAW.

As you can see from the video underneath, it also has a drum grid function for programming, not only drums but synths if you like that way of working.



I paid 400 euros for my MV a while back, which is pretty good value for what it does.

Octatrack doesn't have a traditional MIDI sequencer, I wouldn't recommend it. The tightest MIDI sequencer I currently own is the Retrokits RK008, timing on it feels very tight to me. I used to own an MPC 4000, RK008 feels almost as tight. AFAIK the RK008 uses MIDI running status, which might explain some of its tightness

The modern atari would be the Cirklon IMHO.

Not sure LTB it is still supported in Cubase 8+, but a WinXP retro-computer running VST5, SX3, or a Win7/10/11 machine with Cubase 6.5 or 7.5 could be an acceptable workaround. VST5 even works on Win10x64, actually.

https://www.soundonsound.com/reviews/steinberg-midex-8

That's consistent with my own findings using a MidiGAL MidiClk test.
Best timing I've ever had on a PC computer, including ISA/PCI based MIDI adapters on W98SE machines.

Depends on how "modern" one wants to go, though.
Cubase can feel pretty dated compared to Ableton or Bitwig.

And the Midex8 can be temperamental sometimes.
Depending on the number of MIDI ports needed, the Midex3 is good (and has a funny relay based auto-switching THRU mode when no MIDI software is used).

Hi!

I've been through this exact process and here is my advice.

What you are thinking is actually leading you up the wrong path. The Atari ST isn't amazing because it's 'tight' or because it has such low microscopic latency levels on the MIDI port. It sounds 'right' because of its feel which includes a certain amount of jitter. I used one for many years with Dr. T's KCS and did have strived ever since for the same timing.

You can get absolutely perfect sample based timing by placing sampled drums on a grid in any DAW. This is as close to perfect as you will get. But? It sounds rubbish. Sloppy. Because 'perfection' isn't 'tight'. It doesn't sound right. So all this talk of ms latencies and so on is a red herring - ignore it. You want musical 'looseness'

This is why so many people like machines like the 808 - its not just the sounds, but the 'feel' which is caused by inherent microscopic timing inaccuracies. Try programming one using samples on a DAW grid and you'll see it feels 'wrong'.

The Elektron stuff like the Octatrack is horrible and nothing like the Atari ST. It may even be technically 'tighter' but it sounds rubbish. I've tried them. The Digitakt in particular - horrible stuff.

I've even tried the zero jitter solution by using the Expert Sleepers SPDIF to MIDI stuff - and it all sounds rubbish (not to mention the hassle and rubbish software).

Also remember most of your outboard MIDI synths have their own inherent inaccuracies when dealing with MIDI IN no matter how tight the MIDI you send to them is.

To save you all your trouble - the best thing you can do in my opinion is to use Ableton Live (or another Groove supporting DAW) and then find some good groove templates. (Samples from Mars do some brilliant ones).

Limit yourself to plugin instruments and drums as far as you can.

Experiment with the grooves applied to your patterns, and you will get very very close to the Atari ST.

I've several custom groove templates, some of them directly taken from the Atari ST.

But to be honest, the ones of drum machines like the Linn Drum and 808 in the Samples From Mars collection, with very very subtle swing (maybe 1% away from 16ths) are the closest I have ever been to that sound you are after.

Everything else is expensive, a hassle, and frankly unnecessary.

Do not listen to any of this.

Why?

Roland MC-80 seems to have good timing (480ppqn). I recently got an SSD working in mine so it's pretty immune to old data storage issues. I've never used Cubase on an Atari so I can't speak to the subjective feel of the unit in comparison though.

Timing wise, any Daw is fine today with a half decent PC made over last 15 years. Groove templates are the place to look. I do love the simplicity and grooves on cubase on the ST though. But if i was looking for a quality musical , groove sequencer, id go for an RM1X.

Two things I was told when I met the Dog of wisdom were

if you're ever unsure if adding something makes it better or not,
try leaving whatever it is set at 1%

if you try to pick up a ball and it doesn't fit in your mouth
it's not your ball

I think that it's important not to conflate PPQN (pulses per quarter note) (=resolution) with 'timing', which often has to do more with jitter/(pseudo-)random timing variation.

There is nothing musical about a drunken robot.

I think there’s some legs to this, which also gives me an idea...

Would it not be possibly to actually capture that jitter feel of the Atari by playing its midi into a DAW, recording it & then extracting that jittered feel as a swing template???

If someone puts this together, please do share with the community, would be very interesting to experiment with!

I’m afraid you are talking nonsense.

Throw $1000s at devices in the belief they have ‘zero jitter’ and that’s the reason why they sound good if you wish to delude yourself.

Have done this with some success… but it’s dependent on the hardware you are triggering also and many other factors. The ideal would be to convert the output from the MIDI ports into some kind of audio signal and extract the groove from there but it’s complex…

Just grab yourself a Mac Mini late 2012 with SSD for peanuts, Emagic Unitor for peanuts, Reaper DAW and enjoy 0.054 ms of MIDI jitter whilst having 8 MIDI outputs at once. Tested, works perfectly.

For comparison Atari is 0.026 ms, MPC 2500 is 0.4 ms MIDI jitter, while average PC with USB-MIDI interface is 1-8 ms of jitter.

This comes from someone who has equipment to measure these things. Yeah that MPC 2500 is a kickass alternative to staring at the computer screen, superb sequencer and UI. So, check in there as well.

The only way you're using AMT with a Unitor is to stick with Win 98/2000, or MacOS 9 and run eMagic's Logic 4-5, not Apple's Logic Pro. I am pretty sure Steinberg no longer supports anything to do with the Midex either..

You can run any of the old systems on Mister FPGA, they run the exact same latency as the old computers - Amiga 500, Atari ST

Try Bars and Pipes or Cubase :

https://www.youtube.com/watch?v=WiYBDQQIgEM

https://www.youtube.com/watch?v=DP3FtNpEfBI

Loads really quick in comparison and miding timing is on point

My idea is that tightness has something to do with it.
I did some measurement.
The Atari was less precise than most of my hardware sequencers and more precise than all the other PC platforms I had at the time.
In the meantime things have changed and there have been improvements and other effective solutions, as you know.

But I tend to agree, meaning I have the "feeling" that perfect timing might not be the perfect solution.
The stereotypical but true example, I recently returned to my 909 after years of hiatus, that thing surprised me again, far from perfect but what a groove.
Gives her best as a master clock, I forgot how enthralling it was.

Within the Atari world I measured Cubase, Notator, KCS, the results were in the same order.
Cubase was the tightest, but Notator is said to have the best feel.
I can't comment on this, I've only used Cubase in actual production, but it wouldn't surprise me if this rumor has any basis.


..I don't know about that.

One would naively think that the two things are somehow related but it is not so or not always.

For example both the MMT-8 and ASQ-10, 96 PPQN, in my tests measured better than Cubase 3 on Atari which has much more resolution.
The ASQ turned out to be the most accurate of all and, I must say, with a very elegant groove to my ears.

I love the 2500 for UI and still have one but am not to happy with the accuracy or output concersion. I was under the impression that it was less accurate than you suggest. Not doubting you - just curious. This is measured chasing Mac Mini clock supplied through the Unitor? My recollection comes from the 2500 chasing to Innerclock sync gen.

David used to have the 2500 here but it looks like no more:
https://www.innerclocksystems.com/litmus

This is an interesting take on things. It kind of gives both sides a 'get out clause' with regards to the discrepancy of reality that everyone (ok, some people) are dealing with here.

It was only on this very website a few days ago I was reading a post or two about how sloppy the Atari timing was and how 'on the money' the latest Cubase versions were. That's odd. Most people always say the opposite of that. The people posting were high post high value members so it made me prick up my ears a bit.

Whatever I think or might summise matters not. There is definitely two very opposing schools of thought here. And each of them is basing their opinion on using both of these platforms.

I remember musos I worked with back in the day pointing out that the Cubase at that time couldn't hold a candle to the tight timing of a Boss DR-660 drum machine. I tried it. Both seemed in time to me. Yet, every now and again the computer would obviously drift out, but when it did it was obviously very out of time. Not timing errors as such, more, overload on the system.

I never did get to the bottom of it. And so it seems neither has anyone else. Lots of very contradictory viewpoints from people who have the 'ears' and also the gear to judge. Pretty fascinating.

It's probably the case that people hear timing and synchronisation differently as in the way they hear melody or harmony differently, perhaps preferring one over the other. One person's 'perfection' is another person's 'groove'.

Anyway, pdpd, I hadn't heard this contrary viewpoint before so I gave you a like on your post. You seem to have made it in good faith and have gone some way to backing up your assertion. It ties up a lot of loose ends with this argument. But who knows, maybe the 'truth' is even stranger again...

Agree. I have a MiST with din5 MIDI connectors and it does a superb job emulating Cubase on the AtariST core.

I usually try not to think about this. Not only the sequencer side but also the synth side causes latency and jitter, so I think that tightening the sequencer side only would solve half the problem anyway...

I'm pretty sure it's not a constant thing, it varies influenced, for example, by the strain to which the system is subjected.
Even if TOS is an extremely spartan operating system and of course this helps.

I don't think perfect timing was the Atari ST magic either.

Only the main midi output had tight timing. If using c-lab unitor to get additional midi ports, the timing on those was much worse. Using the printer port for C-lab export was even worse.

If one stuck with just the main ST midi port and daisy-chained everything, timing would not be great either due to the serial nature of MIDI. Also, some of the older 80s/90s synths and drum machines had latencies of many ms themselves. Vince Clarke has that famous quote about "midi is crap" (or something similar) and went to CV for everything for a period.

For those unaware, midi being a serial protocol means note data follows one after another. Each note's data sent over midi takes about 1ms (actually 960 microseconds) to transmit. If you're playing ten notes at the same time through the same midi port, the last note is going to be triggered 9ms after the first one. This is before the instrument's own output latency is applied.

A benefit of this serial nature meant that there was inherent variation in midi pattern which came for free to the producer. Play a kick on track 1 and closed hat on track 2 on beat 1 and the closed hat plays 1ms after the kick. On beat 2, play a kick on track 1, snare, hat, conga, closed hat on track 5 and the closed hat plays 4ms after the kick.

Both cubase and notator gave higher midi priority to sequences higher up on the screen. By moving sequences up or down, one could change the feel of a song.

Put 16th notes triggering closed hats on an Akai sampler on track 1 and they should play straightforward and be tight. Put 16th note closed hats on track 32 and they're going to play all over the place, as they're affected by all the other midi notes in the pattern (other drum sounds, bass, chords, synth sequences, etc) that are coming out the same port. The sloppy feel and variation of the 16th notes on track 32 could be a positive and it's going to change as the arrangement of the song plays and other instruments come in and drop out.

If you were only using one port, every sequence had a timing effect on all the sequences below it in the Cubase/Notator display.

Producers could get different "feels" just by drag-and-dropping different sequences in different orders. Hardware sequencers may give tighter timing but it's the ability of Notator and Cubase to easily change midi note priority of sequences which is what I think gave the feel people like.

I don't think it'd be that easy to update a DAW to have a feature like this but a plugin drum machine/sampler with a grid-based sequencer should be able to mimic this. Maybe this already exists.

For a given note, look at every instrument hit at the same time on every track above, and add 1ms delay to this note. Allow drag and drop of tracks. Capture actual atari ST jitter and that could be turned on as an option.