Incorrect Burndown chart values

One of our Sprint teams reported an issue with the calculations in the Burndown Chart.

Some issues end up with negative remaining values which are wrong or cannot be explained.

I attached "burndown_chart_table_extract.docx" in which I meticulously collected screenshots of everything that happened to issue NSD-5919. As you can see, at sprint end, this issue has a Estimate Remaining value of -6.75h which seems incorrect. It should be 0 or perhaps 65h - 81.25h = -16.25h.

I also attached "jira_history.docx" in which I meticulously collected all time information for issue NSD-5919 under the JIRA issue history tab. I used this information to try to match and understand what GreenHopper does, and I am pretty sure that GreenHopper goes wrong somewhere, but I cannot pinpoint it.

I also tried hard to reproduce the error in order to give you steps to reproduce but I was not successful in doing so.

Could you please analyze my data and let me know your findings? As the original reporter said "Luckily the discrepancy arose so late in the sprint that it didn't affect planning, if it had happened earlier we may have been led to believe we had extra capacity!".