My take on this subject is if it works, don't fix it!
Having said that and since my eTRAC is out of the warranty period, i took a peak inside just to see what makes it tick.
The most concerning issue is that there are a lot and i mean a lot of electrical connecters involved in the design. For one thing there is a mother PCB and a daughter PCB tied together with connectors. Connectors in my experience can/are/could be the weak link in any electronic design depending on the quality of the connectors used and other factors such as thermal cycling, connect/disconnect cycling, corrosive environment, gold plated contacts, etc.
In electronic circuit MTBF or MTTF calculations, each connector contact in a design can bump down the overall reliability number, how much depends on the quality and the connector's own MTBF specification furnished by the connnector's manufacturer. These numbers are typically derived in a test lab and can be misleading depending on the customer's application and the environment the connector is used in.
Electrical connectors can fail permanently but more typically a flaky connector contact causes intermittent issues.
One example i can give is my 1st job out of college was a field engineer for a large main frame computer company. The designers for purposes of field trouble shooting and ease of field repair decided to use chip sockets for each and every IC chip. Further, for ease of field upgrades, they also decided to wire wrap all electrical connections to each socket.
This mainframe computer was a reliability nightmare with many many intermittent failures some of which required to replace the entire and very expensive PCB rather than field repairing it. Sometimes we got lucky and just re-seating the chips fixed the problem.
Towards the end of this computer's manufacturing period, the PCBs were re-designed with all chips soldered in and the wire wraps were replaced by PCB traces.
The result? Those last main frames were very reliable.
As a design engineer before retirement, i tried to avoid connectors and chip sockets as much as possible in my circuit designs. Sometimes its just not possible to avoid them. The age old design dilema......Manufacturability OR Reliability OR Ease of field troubleshooting/repair.
IMO, my bottom line would be to only send in a detector if its either permanently broke or has ongoing intermittent issues.
For intermittent issues its extremely important to try to recreate the problem at will, then relay that failure procedure to the repair facility.
That will help prevent a NPF (no problem found) when the detector is returned.