Not sure if thjis will help, but I have experienced similar issues and found that a stub-stack solved the problem. We basically determined that different signals were being given to the carb at the bleeds due to whatever was happeing with the air in that compressed space at speed..the snub stack helped quite a bit...as a note we experienced this when the carbs were very close to the scoop...worst case it is a cheap experiment..as an alternative, open the back of the scoop ( drill holes / rais rear of hood etc.) allowing for a clean flow of air over the carb...maybe due both...