This is a real finding from one of our own design reviews — caught by Probe, our PCB design-review tool. Here is what it read.

If you have ever set an adjustable linear regulator with two resistors, this one is worth two minutes.

We were reviewing one of our own test boards and came to a 1.8 V rail produced by an AMS1117 in its adjustable version, with a 22 kΩ and a 4.7 kΩ resistor setting the output. The designer had done the courteous thing and written the calculation on the sheet, ending in 1.799 V.

The equation used is the correct one. It is the form the datasheet itself gives: the reference voltage multiplied by the divider ratio, plus the regulator’s own adjust-pin current flowing through the lower resistor. Most people drop that second term, and on a low-value divider they are right to. The note did not drop it. It included the term and gave the current as 0.06 mA.

That number is not in the datasheet.

What the datasheet specifies, in its electrical characteristics table, is an adjust-pin current of 50 µA typical and 120 µA maximum, together with a reference voltage of 1.225 V minimum, 1.250 V typical and 1.275 V maximum. Work the corners with the fitted resistors and the rail does not sit at 1.799 V. It lands somewhere between 1.72 V and 2.11 V.

The reason the spread is so wide is the thing worth taking away. This divider is high enough in value that the regulator’s own bias current, flowing through the bottom resistor, contributes about sixteen per cent of the output voltage. The term the note approximated is not a small correction. It is a sixth of the answer, and its datasheet range is more than a factor of two wide.

There is a second effect pushing the same direction. The rail’s on-board load is well below the minimum load current the regulator’s datasheet asks for, and a lightly loaded adjustable regulator sits towards the top of its range rather than the middle. The realistic expectation is not the centre of that window.

And this rail does not stay on the board. It leaves on two connectors, to an instrument pod and to whatever device is under test. A rail labelled 1.8 V, potentially sitting above 2.1 V, arriving at somebody else’s silicon.

Why nothing flags it

The divider ratio is arithmetically correct for 1.8 V. Anyone re-deriving it with the textbook two-resistor formula gets 1.8 V and moves on, reassured, because the schematic note agrees with them. A rule checker has nothing to fire on: both resistors exist, both are in the bill of materials, both are connected exactly as drawn. There is no violation anywhere. The error lives inside one coefficient of a correct equation, and it is invisible unless somebody opens the datasheet and reads three rows of one table.

That is the whole job. The fitted resistor values came from the bill of materials. The reference tolerance, the adjust-pin current and the minimum load current came from the characteristics table. The load came from the rest of the schematic. No single one of those is difficult. Putting the four beside each other is the part that does not happen when a reviewer has a hundred other things to look at.

A calculation written on a schematic is a claim, not a check. It deserves the same citation as any other.