brundlefly
The 1820m has very low MIDI transmission delay compared to most USB interfaces, so assuming the synth has a decent response time (usually no more than a few milliseconds for modern hardware), any sync issue should be mostly due to audio latency. The Timing Offset can be used to address this, but can have undesirable side-effects since MIDI grid is offset for recording as well as playback. Ideally you should just keep your ASIO buffer as low as possible, and it shouldn't be a significant problem.
 
With ASIO buffer at 2ms, the RTL will be under 6ms and the MIDI transmission/response time will be adding maybe 5ms to that. Soft synths will be subject to the same outbound audio latency, so the discrepancy should only be on the order of 8ms, which is more on the order of a large phase error than a timing error. Bumping the ASIO buffer to 3-4ms would only add another 1-2ms to the discrepancy.

Reducing the buffer to 2ms does help, but doesn't completely eliminate the problem.
 

brundleflyif the timing is really audibly off, then something else is going on like PDC. If you're using FX plugins that require PDC, you'll need to enable the PDC Override for live inputs so the hardware synths aren't affected. This really the only thing I can think of that would account for differences between projects too.
 
I loved the performance, flexibility and sound of the 1820m, but I had pretty significant problems with the beta x64 drivers.

I was assuming PDC wasn't a factor since I wasn't using convolution reverbs but maybe that's not the case.  I will try the PDC override you mention to see if it makes a difference here.
thanks everyone!
Dan

If you disable record latency compensation and record the hardware synth driven by existing MIDI and playing a patch with a fast attack, how late are the audio transients relative to the MIDI events in milliseconds with the ASIO buffer at 2ms? And what synth is it, exactly?

Dan_E10

bitflipper
It sounds as though your problem might be that you're monitoring the synth through SONAR, where the lag between generating a MIDI note on your controller and the synthesizer responding is going to be variable due to different latencies in different projects.
 
If that's it, then the solution is to monitor your synthesizer in real time. Most audio interfaces feature a so-called "zero-latency" monitoring mode, wherein its analog input channels are routed directly to the interface's output or headphone outputs. Check your interface's documentation.
 

Thanks bitflipper.  I am monitoring through Sonar.  I'm kind of stuck doing this since I'm using delay and effects a lot at the composition stage.  A zero-delay monitoring mode would bypass all fx applied in Sonar correct?  The midi tracks I'm working with right now have been input with the staff view and PRV so any midi controller contributions to the sync shouldn't be a factor I'm hoping.
Dan

True, that's the downside of zero-latency monitoring. However, many audio interfaces have an effects loop for inserting hardware effects, allowing you to hear them without actually recording them. Because they're not being recorded, you can get away with cheap second-hand delays and reverbs that can be found on Craigslist or eBay for $40-50. Then you'll have the best of both worlds.
 
If that's not an option, consider printing your delays live, old-school. I use this fairly inexpensive unit for that purpose. It's stereo and emulates dozens of different classic devices, including delays, reverbs, chorus, flangers and tremolo. 
 

The 1820m has onboard FX that could be used for direct-monitoring with a delay. But I still think the OP should be able to get this ironed out such that input monitoring doesn't add intolerable - or even noticeable - latency. Something's funky in the setup.

brundlefly
If you disable record latency compensation and record the hardware synth driven by existing MIDI and playing a patch with a fast attack, how late are the audio transients relative to the MIDI events in milliseconds with the ASIO buffer at 2ms? And what synth is it, exactly?

I will try this.  The synth is a synthesizers.com modular analog.  It doesn't have a midi interface, only control voltage inputs on most of the modules.  I'm using a Kenton Solo II midi - CV converter to drive it.  I send midi from Emu's first output port to the Kenton and then wire up patch cables from the Kenton for pitch, velocity, and gate signals used by the modular.  I'm assuming (hoping) the latency of the midi-CV converter to be constant.
Dan

Hmmm... could definitely be some extra latency in that chain. I did a quick check of the Kenton's documentation and didn't see any reference to latency, which is a little surprising and suggests they might not wan't to talk about it. Your test results should tell the tale.
 
EDIT: I found synthesizers.com's own MIDI-CV converter promises latency < 1.5ms, which is very reasonable, but it's not clear how fast the synth responds in turn.

I would have thought response to CV would be pretty much instant, with the MIDI-CV conversion taking up much more time unless the synthesizers.com envelope triggers aren't very quick or are slightly delayed to avoid clicks and pops if the attack is set to zero. Even so 80ms is one heck of a MIDI timing error for anything, never mind the kind of quality that synthesizers.com are known for.

I've a real mix of synths, a mix of digital and analogue with MIDI+CV control, hooked up via MIDI either to a MOTU MIDI interface, the UFX MIDI ports or directly connected via USB plus AD2, Session Drummer and DimPro and my out-of-sync problems against are in the order of a few milliseconds. Small enough to be inaudible. Sonar's transmitted MIDI clock does wander a little (or so my MachineDrum claims) but not enough to notice unless I'm looking at the MDs tempo display. I can monitor through Sonar while putting delays and modulations on stuff without any problems at all with an audio buffer of 48 or 64 samples. Though RME drivers and hardware are particularly food for low latency performance my old UA-101 could manage around 10ms round trips without problems.

I suspect MIDI discrepencies in the order of 80ms has something to do with plugin-induced latency somewhere. Or possibly something about the specific modular patch....

brundlefly
If you disable record latency compensation and record the hardware synth driven by existing MIDI and playing a patch with a fast attack, how late are the audio transients relative to the MIDI events in milliseconds with the ASIO buffer at 2ms?

Assuming I correctly disabled record latency compensation and am measuring the delay accurately, it looks to be a total latency of 6.9 - 7 ms for an ASIO buffer of 2 ms.  The frustrating thing is that tonight, the sync between the outboard synth and the soft synth I'm using is much better.  Still, I was able to hear a slight difference between the virtual and outboard synth when playing 8th notes with sounds set to very fast transients.  After playing around with the Time + field at higher latencies I was able to get the virtual and outboard synths to align fairly easily.  At the 2 ms ASIO buffer, I could not hear a difference in the timing of the virtual and outboard synths.
 
Can anyone tell me what the units of adjustment of the Time + field are for a midi track?  Are they ticks?
thanks,
Dan

Dan_E10
 
Can anyone tell me what the units of adjustment of the Time + field are for a midi track?  Are they ticks?
thanks,
Dan

As I remember it's ticks. 

That makes sense, thanks.
 
So far I'm ready to chalk my problems up to PDC as well as attempting to use diferent ASIO buffers.  From my testing so far, I definitely need a larger time offset on the midi tracks to keep things in sync as the ASIO buffer goes up.  I guess that makes sense as the round trip latency on the outboard synth increases at double the rate of the softsynths if I'm understanding it correctly.  Double the ASIO buffer results in double the latency for softsynths, but 4x the latency for monitored signals.  The Time+ field seems to work a lot better than changing the time offset in preferences since it only affects the desired midi channel.  My midi-CV converter doesn't seem to be a large contributor to the overall latency, adding about 3 ms.
 
From this point forward, I'm going to stop worrying about this and just make some Time+ adjustments or turn on PDC overrride if things start to sound out of sync.
thanks everyone,
Dan