The same number, two different verdicts
Ask two Steam Deck owners about capping a game at 40fps and you can get flatly contradictory answers. One will tell you it transformed their battery life and barely cost them anything. The other will tell you it looked juddery and they went back to 60. Neither is exaggerating, and neither is describing a matter of taste. They are describing two different configurations that share a single number, which is why the disagreement never resolves.
Locking a Steam Deck game to 40fps gets discussed as a compromise you settle for — a midpoint between 30 and 60 that you accept in exchange for runtime. That framing treats the frame cap as the whole setting. It isn't. The cap is half of a pair, and the half that decides whether it feels good is the one most write-ups mention in passing, if at all.
What a frame cap can and cannot control
A cap controls how often the game produces a frame. It has no say in how often the display shows one. At 40fps the game emits a frame every 25 milliseconds. A panel running at 60Hz refreshes every 16.7 milliseconds. Those two intervals do not divide into each other: 60 divided by 40 is 1.5, and a display cannot show half a refresh.
What happens instead is that frames alternate. One frame is held for two refresh intervals, the next for one, then two, then one. The game is producing frames at a metronomic 25ms, and the screen is showing them for 33ms, 17ms, 33ms, 17ms. The output is a steady frame rate presented at an unsteady cadence, and the eye is considerably better at detecting irregular motion than slow motion. That is the judder the second owner described. Their frame counter said 40 and looked stable, which is exactly why the cause is so rarely identified — the number being monitored is not the number going wrong.
Set the panel to 40Hz and the arithmetic changes completely. The refresh interval becomes 25ms, the frame interval is 25ms, and every frame is held for precisely one refresh. Nothing alternates. The reason 40 works is not that it sits halfway between 30 and 60; it is that 40 divides evenly into 40. The number is incidental. The divisor is the point.
Why the advice is model-specific, and why that matters now
The 40fps recommendation entered circulation for a specific machine. SteamOS gained a refresh-rate control that let the original LCD Deck drop its panel from 60Hz to 40Hz, and the pairing of a 40Hz panel with a 40fps cap is what people were actually praising. The cap was the visible half of the advice, so the cap is the half that got repeated.
Valve lists the LCD model at up to 60Hz and the OLED at up to 90Hz. The OLED's refresh range does not extend down to 40Hz — its low end sits at 45. On that hardware the even divisor of a 90Hz panel is 45, not 40. An OLED owner who follows the well-travelled advice, caps at 40 and leaves the panel where it is, has not made a small error at the margin. They have reproduced the exact alternating-persistence problem the advice exists to avoid, and they will conclude that 40fps is juddery, which is a correct observation about an incorrect setup.
Worth knowing before you go looking for the slider: the 45Hz setting has been reported as not reliably applying on OLED units, with the panel returning to 90Hz instead. So confirming the refresh rate actually changed is not paranoia, it is the step that makes the rest of the exercise meaningful.
When the divisor stops being the thing to optimise
An even divisor guarantees even pacing only while the game holds the cap. Drop below it and the alternation returns, because a missed frame has to be covered by holding the previous one for an extra interval. A 40Hz panel paired with a game that wavers between 34 and 40 is not a smooth 40 experience with occasional dips; it is an irregular cadence with a ceiling. In that situation the useful move is downward — 30fps on a 60Hz panel divides evenly too, and a cap the hardware can actually sustain beats a mathematically elegant one it cannot.
Genre changes the calculus as well. Slower games conceal irregular persistence because there is less motion for the irregularity to attach itself to, which is why the same setting draws opposite reactions from someone playing a turn-based RPG and someone playing a shooter. If a title spends most of its time showing you fairly static scenes, the pairing matters less than the battery saving does.
The order that follows from all of this is not the order the advice is usually given in. First establish a frame rate the game can hold in its demanding areas rather than its quiet ones. Then set the panel to a refresh that the frame rate divides into evenly. The cap is chosen second, and the panel setting is not optional garnish — it is the half that does the work. Do it in that order and you no longer need a recommended number, because you can derive the right one for a game nobody has written a guide about.