Even if you might be able to more or less accurately track close to the zenith, field rotation is going to kill you (unless you have a field derotator, and even a good one is probably not going to give great results very close to zenith). Without a derotator, there's nothing that you can do about this.
As far as I know, the Celestron alt-az mounts adjust the tracking rates in altitude and azimuth every 30 seconds to approximate sidereal tracking. This is probably adequate for a lot of purposes, but is unlikely to be sufficient near "gimbal lock" position (i.e. when the optical axis is parallel to the azimuth axis) because the azimuth speed increases rapidly as zenith angle approaches zero. I suspect that even small errors in leveling will cause large errors the calculated azimuth speed near zenith because the mount's idea of the zenith angle will be slightly different than the true zenith angle, and you're calculating the azimuth speed by dividing by the sine of the zenith angle, which is very small, so even small errors become significant.
Possibly the subs that were "streaky" were a result of the 30 adjustments occurring during those exposures.
It's kind of what ai suspected, I was just not really sure how Celestron Alt/Az mounts operate. I was already aware that Alt/Az speeds were constantly recalculated to avoid stepwise Alt/Az movements during tracking.
It sounds to me that, as the zenith is approached, the speed changes are larger and required more frequently than 30 second intervals.
EDIT: I guess also that within the "zenith blind spot" the situation becomes totally impossible, for even the most sophisticated Alt/Az mount.
Edited by Rac19, 03 June 2020 - 06:05 AM.