Bouncing Adds Silence to Front of Clip [SOLVED]

Hi everyone,
 
I hope this is just simple operator error with some setting I can't pinpoint.
 
When I select a clip and,
 
1) Bounce to Clip, or
2) CTRL copy clip to a new track and Bounce to Clip, or
3) Bounce the single clip slected to track, or
4) Drag the clip to Desktop, or
5) Export the clip as audio,
 
SONAR is adding a period of silence to the result prior to the contents of the actual source clip.  The amount of silence added equals the start time of the clip.  So, if my clip starts at 5 seconds, when bounced (or any of the above) the result includes an extra 5 seconds of silence before the original source content starts, which makes the clip 5 seconds longer as well.  If the clip starts at, say, 30 seconds, the result will contain an extra 30 seconds of silence before the contents of the original clip, and will be 30 seconds longer than the original.
 
I've spent hours now trying to figure this out, and have gotten nowhere.  I need to figure this out, because I have many dozens of individual clips to bounce (sampled mandolin plucks) as is, without additional silence which would have to carefully be trimmed off again.
 
Pete
Platinum 2017.02

brundlefly
I don't see how you would get pre-roll included with all of those methods. Selecting the clip should define the limited time range to be exported, unless maybe you have the track's Edit Filter set to something other than Clips...?
 
If all else fails, try exporting with Source Category = Clips. Before this export option was avialable, I always exported samples by dragging and dropping direct to the file system.

Hi Brundlefly, selecting the clip does indeed define the correct time range - nothing before or after the clip is highlighted in the time ruler.  The Edit Filter is set to clips.
 
With Source Category = Clips, the result is the same (bad).  As it is with just dragging the clip to the desktop.
 
It almost seemed that Broadcast File settings may be coming into play (with SONAR padding the time before the start time with silence), but that does not appear to be it.  The bounced clip actually starts at the right time, starting with the silence added by the bounce and ending in the actual waveform bounced.  Very strange ...

If true, that's totally bizarre. MMCSS is about prioritizing CPU and disk access for realtime streaming. It's enabled by default, and should have no bearing on offline bounce/export processes that don't use the audio driver.