There are 2 parallel branches, each has 2x 10R resistor. It seems to me that the resulting value will be 10R, but I can't wrap my head around why would someone do this if it is not meant to dissipate power (it's meant to be a filter I believe).
Context: It is at the power input to a PCB. The source is a 6S Li-Po battery.
Dissipation (4x) or pulse power capability (4x) since cost can go up much faster than those capabilities— or redundancy (no single failure will cause a short or open).
The latter strategy is used in certain high-reliability and safety-critical applications from domestic gas heating controllers to space gear.
But I more suspect pulse handling in this case.
As the others have said, it's most likely about dissipation and pulse power.
But there is also cost.
If you already have a reel of 10R chip resistors of one value loaded on the pick and place machine, and you need a 20R resistor, or a 10R resistor with 4x the power handling... then yeah, you could add one BOM item with a larger, higher power resistor. It is likely to be cheaper than 2 or 4 smaller resistors. But if you only need one per board, is it worth it? There will be extra stocking and handling fees for that extra reel, and that will probably offset the cost savings.
And if all the slots on your pick and place machine are already occupied, then it becomes a lot more expensive: the machine has to process all the boards, then someone has to come in, remove one reel, put the new reel in, and set everything up to make the whole stack of boards go through the machine again. All this just to place one resistor... no way. The 10R reel is already on the machine, if you need more power handling, it's cheaper to just use more resistors out of it.
In fact, even in a hobby project, when you solder your board, if it takes 2 minutes to grab the bag of 10R 0.5W resistors, pick one, then solder it, put the bag back in storage... while you already have 10R 0.125W resistors on your board, so the bag of those is already out... not to mention ordering the larger resistors, keeping them in your stock, not getting quantity discount cause you only buy a few, etc... the 4 resistors also make sense in this case.
Four smaller and cheaper resistors that can dissipate less can be used instead of single larger and more expensive resistor that can dissipate more.
Sometimes that configuration is used to protect against failures in the 4 resistor network, compared to a single resistor.
If you have a single resistor, the primary failure modes are 1) an open or 2) a short (less likely).
This network protects against opens by having two paths in parallel. If one path opens up, the second path is still functional.
This network also protects against shorts by having two resistors in series in each path. If one resistor, connections, etc short, the second resistor in the path is still there.
This configuration is used for other passive components such as fuses and capacitors in critical paths.
In hi-rel applications (space), this is usually part of a single point failure (SPF) fault mitigation effort. This is done to minimize the chance that the mission will be lost because of a single fault.
My thinking might be off here, but would this way of connecting resistors not also average out their tolerances?
A single resistor with a 10% tolerance might be off by the full 10% in either direction. The odds of all 4 resistors being off by the same amount in the same direction are fairly low, and worst case scenario you're no worse off than with the single outlier resistor.
