I found the information. First, a correction: Billat, et all did only treadmill tests that I can find, in regards to Vo2 max determination. I found 10 studies done by her (and colleagues). Here are the times reached for VO2 max (minimum velocity):
1) 5.35 minutes
2) 6.57
3) 6.00
4) 6.70
5) 6.73
6) 5.50
7) 5.75
8) 6.12
9) 6.22
10)7.02
I think the most relevant ones were studies 4 & 5 because they were done 1 week apart. It was a reproducibility study.
The study the Perronet and Thibault of Canada referred to was done by Padilla S, Boudin M., Bathelemy JC, et al. "Physiologicaal Correlates of middle distance running performance." Euro J Appl Physiol 1992; 65: 561-6
They found that 7 minutes and 8 minutes 40 (run on a track) were equivalent to the slowest velocity eliciting VO2 max.
I found several others that are in the 5 to 8 minute range (6-7 minutes being most common).
One thing to note; how a protocol is designed effects the VO2 max duration time (called tlim - meaning time limit). Short stage protocols (rapid changes from one workload to the next) result in SHORTER tlims. Longer stages result in longer tlim. So, basically, if the test you are using quickly increases speed or has short time stages the total time you can "hold" VO2 max pace is going to be shorter. If you lengthen the time stages during the stage (slowly increasing the wordload, therefore) the time limit is going to be longer.
Example; if stages are only 30 seconds, it apprears that tlim will 4-6 minutes. If stages or 2 minutes, it appears the tlim will be closer 6.5-7.5 minutes. If stages are longer than 2 minutes, it appears the tlim will be closer to 8 minutes and change. *It would be beneficial to the field if someone did a meta-analysis study and determined the statistical relationship between stage lengths and tlims. Hint, for you graduate students looking for a project!