There shouldn't be any difference in the data the acquisition and recording modes. I think when you start recording after starting acquisition it will flush the event buffer and clear all of the plots. Your new events should then begin appearing as they are detected, but should return to a similar distribution seen during acquisition quickly.

Without seeing it myself, my only suggestion is to check the number of events that are being displayed and captured during recording. If this value is low you might not see a stable distribution appear in your graphs, but it should look fine in your final plots. Generally I just set the number of events to display to the same value I'm looking to capture.

Practically, what you are saying is impossible (unless we see with our eyes). If this happens please share image of both the situations.

Have you checked the plots after the recording is done?

Because when you start recording the events, instrument restart acquisition and your plots only shows a defined number of dots (pre-defined in protocol, under the show event options).

John Hardy Lockhart and Jitendra Kumar Shandilya, thank you both for replying!