Weird...Is there a polyphony limit to Logic?

Question

OK, I'm stumped.
I've just started a new composition. So far I've used only 10 tracks. Each track is using an instance of EXS24. The tracks are, from 1-10, Piano, Bass (arco), Cellos, Violas, Violins, Soft pad, Celeste, Flute, Clarinet, and Harp.
Measures 21-28 are using the harp, clarinet, flute, celeste, pad, violas and cellos. Starting in measure 29, I have the harp drop out completely, and the violins take over the melody, and I've add arco bass.
Here's the strange thing that's happening:
The problem is that everything except the flute cuts out half-way through measure 30! It's weird. It sounds just like it used to when I maxed out my polyphony on a keyboard -- back in the day when I used outboard keyboards . Everything cuts out, even the reverb. Actually, I think the flute cuts out too, but it probably doesn't seem like it because it's in a moving line, so the next note picks up right away. Everything else comes back in on the downbeat of measure 30, when each instrument has to play the next note.
I think the problem is somehow related to the harp -- even though the harp isn't playing in measure 30. If I mute the harp track, measure 30 plays just fine. Or, if I start the track in measure 27 (2 measures before the harp section ends), it'll play fine. But if I play 4 or more measures of the harp section, measure 30 cuts out completely for a second.
I've made sure that I'm never playing more than 3 notes at a time in the harp track, and have even played with the "voices" setting in the EXS24 instance on that track. But no matter how many voices I allocate to the harp (large or small) the effect on measure 30 remains the same. The only thing so far that solves the problem is muting the whole harp track.
BTW, the polyphony requirement isn't very high in measure 30 -- probably only about 12 or 13 notes, although the damper pedal is used on the piano, so that would raise it a bit, but probably no more than 6 or 10 more notes. Certainly I've taxed sequences much harder in times past. But, I thought that MAYBE I've accidentally changed some global parameter that effects what's going on, thereby screwing up measure 30. BTW, this is the only measure in the song that's effected (although the sequence is only 38 measure long at present).
Does this problem sound familiar at all? Or is this just a very bizarre problem that makes no sense to anyone? It certainly isn't making sense to me.
I would greatly appreciate any advice you could give me.
Thank you!
Kevin
Edited to add: BTW, I don't know if the following symptom is helpful or not, but I'm going to add it anyway, just in case it helps explain something.
As I was playing with the "voices" parameter in the EXS24 instance on the harp track, I noticed that even though there is only 3 notes playing at any given time, and no pedal is being used, the "voices used" number keeps going up and up as the sequence plays the harp section. I'm assuming that the "voices used" number doesn't go down because of a long sample length (I'm using the harp that comes with VSL's Opus series).
After the harp section ends in measure 28, the "voices used" starts going down slowly, taking about 3 measures to make it back down to zero.
Does this make sense? I don't know if this is helpful to my problem, or if it's not related at all. But, I thought I'd better include it just in case.

Answer 1

Sounds to me like you're experiencing "SIDS", or "sudden instrument dropout syndrome", an endemic problem with certain version of Logic. This was supposedly fixed in 7.2.3 (if memory serves), but it seems like you're running into SIDS nevertheless.
There are other bugaboos with Logic (related to the performance of audio instrument plugs) that you might want to know about. For example, some sampler plugs, including instances of EXS-24 itself will not play back correctly if you initiate playback from a downbeat. It doesn't always happen, but then again, it happens a lot to me....
For example, if you were to play a sequence back from, say, bar 54, beat 1, you might not hear some of the notes (or you'll hear bits and pieces of garbled sound), even if those notes are quantized to the downbeat, or, even if chase notes is enabled. But then, when this occurs, if you were to start from an 8th note before that point, playback is fine! It's an inexplicable and confounding problem that Logic can't play back correctly from a downbeat but will play back if you start playback off the downbeat.
Next, polyphony. Logic does not have a polyphony limit exactly... The the number of plugs that can be handled (both sampler and fx plugs) dependent on your computer's horsepower. Taking into account non-Logic-related system requirements, other programs that might be running in the background, and then Logic and the cumulative needs of your plugs, these things all draw on processor resources. It's entirely possible to use up all available resources if you have too much going on. That's what the various benchmark tests are all about -- how far can you push the computer until it says, "uncle"...
But it doesn't sound like you have a very heavy processor load, and as I said earlier, it sounds to me like you're running into SIDS.
Now...
By chance have you done any capture recording? (this is important, please let me know).
Finally, regarding the polyphony count of EXS while only a few notes are being played... you are on the right track: as long as a sample is playing (even if it's nearly inaudible during the last moments of a long sample's release), it's using a voice. So let's say that you're playing a simple 3-note pattern outlining C major (C, E, G). If you play those notes over and over again, and assuming that you're playing a harp sound with a long release, here's the deal:
The first 3 notes will indeed use 3 voices (CEG). Now you play another C, E, G. If those original 3 voices are still ringing out, you're now using 6 voices. Play C, E, G again. If those original 6 voices are still ringing out, you're now playing 9 voices. And so on.
HTH. (Please let me know if you're using capture recording.)
Message was edited by: iSchwartz