Does anyone know how accurate the Temp sensors are on the AP’s. I have seen ranges from 120 - 150 degrees on cloudy days with air temp about 70 degrees. I am concerned about the summer months when the temp will be in upper 90’s and the chances this will lead to problems. Any Comments or expirence?
I think the temp is meaningless. I have seen SM’s @ 185 deg last summer and stood watching expecting it to slide off the bracket like butter on a hot skillett. Never did tho, and still works.
BTW - did you notice the temps are gone on 7.07 ?
We are running 7.07 on all of our equip and we get the Temp via snmp. These are 900mhz AP’s, if it makes any difference. Glad to hear someone else sees this and do not see any problems. I hope it is meaningless.
The temp sensor is likely on the silicon wafer in the processor. These things can run very hot. Heat is generated both within the silicon of individual transistors and the metal traces (wires) connecting them.
A bit of trivia:
I learned years ago that within the aluminum connecting layers of a chip the current density in a single trace can be high enough to cause impurities in the metal to actually move and collect in kinks in the trace. Expansion and contraction – from powering on and off – eventually causes the kink to break because the trace has such a high concentration of impurities at that point.
Imagine running everything in your house – including A/C, stove, clothes dryer – through a single light-weight extension cord.
As you have seen, the temperature shown on the Expanded Stats page previous to R7.0 is not very accurate. It is a rough approximation of the board level temperature, derived from the data available to the FPGA. Due to the way it is calculated, it is not very accurate and also can vary widely between two modules that are actually at the same temperature. It was novel that a temperature could be calculated, but was removed from the GUI because it wasn’t that accurate and could be easily misinterpreted.
It is not ambient temperature, which is what the design was tested to, and what it is specced to, and has performed very well to.
Some operators may be monitoring individual modules over time using SNMP (comparing a module against itself, not against other modules), and in that case monitoring module “temperature” may give some indication of changes in module health. However, received power (dBm) and jitter are probably much better indicators to monitor, and leave the temperature as a novelty.
I’ve heard that the temperature of canopy units is not measured, but rather calculated from transmit power, etc.
it is, indeed, meaningless. I’ve seen several hundred units survive in above 100F weather, under direct sunlight.