It told us which four heads were starving.
Every head on this machine was given identical feeder settings, across all fourteen. Under those identical commands, heads 6 through 9 filled at roughly one third the rate of heads 4, 5, 10, 11 and 12.
That is not a control fault. A weigher cannot command its way out of product not arriving. It is a mechanical product-distribution fault upstream of the weigher, and it had been silently capping this line's efficiency, invisible in any number anyone was looking at.
The controller pushed every feeder to its limit to keep the line running and reported exactly which heads were starved, so a person could go and fix the actual cause. A second machine on the same line flagged a matching feed asymmetry the same session.
When efficiency is bad, the argument on the floor is always the same: is it the weigher, is it the product, or is it the feed? Everyone has an opinion and nobody has an instrument.
Per-head fill under identical commands settles it in one chart. If four adjacent heads are short and the other ten are fine, the weigher is doing its job and something upstream is not.
Every number we publish was achieved on a machine with a feed fault.
The machine under autonomous control had this mechanical feed fault for the entire session. The +18.65 points, the 1.06 g giveaway, the 84.7 % peak: all of it was achieved while starved on four of fourteen heads.
We say so because it matters for how you read every figure on this site: they were measured with that fault in place.
When your machine is running well, it says so.
Another machine on the same line (Machine 4 on the results page) was well-fed and running clean. Counted from the machine's own totals, it ran at 99.86 % efficiency, and the system's report called it “running optimally”. No alarm, no intervention, no invented fault.
Machine 3 sat in standby all session. The controller correctly stood down and never wrote to it.
A diagnostic that finds a problem everywhere it looks would be worthless, so it matters that this one reported a healthy machine as healthy.
Tells you when the problem is upstream of the weigher, and names the heads.
Efficiency and giveaway derived from the machine's own counters, not estimated.
Shows the real giveaway number even when it's unflattering; shows a dash, never a plausible fake, when data is missing.
If the line is under-feeding a machine, the efficiency number says so instead of hiding it.
Every machine, every head, live: what the line is doing and what the controller is doing.
A self-contained report your supervisors can open, print or email, with no network and no software.
Continuously scores its own predictions against reality and speaks up only on a genuine deviation.
Find out what your line is actually doing.
A line assessment is a measurement, not a pitch. You get the per-head picture whether or not you buy anything.