Jim and Dave. We continued the investigation of the problem of incorrect slow channel data in the DAQ from Front Ends following the upgrade to RCG2.8.

The problem is more widespread that just h1hiitmy. We have also seen it in h1susitmy. It appears that only the non-filter-module floating-point slow channels are affected. The problem is related the channel list defined by the INI file. There is a point in the slow floating-point channel list at which channels above do not have the problem, channels below all have problems. The problem is actually a channel shift. For h1susitmy the channels are shifted by 3, for h1hpiitmy the channels are shifted by 2. The point at which the shift starts varies from model to model. It is after a few hundred floating point channels for h1susitmy, and immediately for h1hpiitmy.

Finding the break point is not trivial. Many of the slow channels are zero, or have rapidly changing signals. We are getting the data from the NDS and comparing it with a direct EPICS channel access read of the data.

We will work on scripts over the weekend to get more data on the problem. Hopefully when Rolf is here next week we can implement a fix. The silver lining is that little data has actually been lost, the data is being acquired by the wrong channel name.