Hi Paul,
errors are in count about 20 per month (average). They occur in morning or evening, typically one or two in hour. For example, this is a list for October:
SAM calculated negative direct normal irradiance -356.634 W/m2 at time [y:1990 m:10 d:1 h:5], set to zero.
SAM calculated negative direct normal irradiance -834.828 W/m2 at time [y:1990 m:10 d:1 h:5], set to zero.
SAM calculated negative direct normal irradiance -3418.52 W/m2 at time [y:1990 m:10 d:1 h:5], set to zero.
SAM calculated negative direct normal irradiance -281.697 W/m2 at time [y:1990 m:10 d:2 h:5], set to zero.
SAM calculated negative direct normal irradiance -861.727 W/m2 at time [y:1990 m:10 d:2 h:5], set to zero.
SAM calculated negative direct normal irradiance -244.557 W/m2 at time [y:1990 m:10 d:24 h:4], set to zero.
SAM calculated negative direct normal irradiance -428.493 W/m2 at time [y:1990 m:10 d:24 h:4], set to zero.
SAM calculated negative direct normal irradiance -1732.08 W/m2 at time [y:1990 m:10 d:24 h:4], set to zero.
SAM calculated negative direct normal irradiance -256.498 W/m2 at time [y:1990 m:10 d:25 h:4], set to zero.
SAM calculated negative direct normal irradiance -698.732 W/m2 at time [y:1990 m:10 d:25 h:4], set to zero.
SAM calculated negative direct normal irradiance -3810.47 W/m2 at time [y:1990 m:10 d:25 h:4], set to zero.
SAM calculated negative direct normal irradiance -329103 W/m2 at time [y:1990 m:10 d:25 h:18], set to zero.
SAM calculated negative direct normal irradiance -281.736 W/m2 at time [y:1990 m:10 d:28 h:4], set to zero.
SAM calculated negative direct normal irradiance -15593.1 W/m2 at time [y:1990 m:10 d:28 h:4], set to zero.
SAM calculated negative direct normal irradiance -291.927 W/m2 at time [y:1990 m:10 d:30 h:4], set to zero.
SAM calculated negative direct normal irradiance -886.767 W/m2 at time [y:1990 m:10 d:30 h:4], set to zero.
SAM calculated negative direct normal irradiance -145.701 W/m2 at time [y:1990 m:10 d:31 h:4], set to zero.
SAM calculated negative direct normal irradiance -587.71 W/m2 at time [y:1990 m:10 d:31 h:4], set to zero.
Errors never occur around noon time, only morning or evening. So I think this is due to calculation on the edge of horizon (sun under/above horizon) and mostly due to very fine 1-minute time interval of data. Or do you think there would be another problem?
thank you and best regards
Branislav