XR5 low-temperature performance

[cgervasi]

Has anyone tried connecting a good heat sink to the XR5? We have a metal plate that goes from the bottom of a mPCI card to our enclosure, providing good heat sinking. With the XR9, it doesn’t work down to -40C, even with self-heating from having it running as you reduce temperature, but it does work down to -30C.

The XR5, however, starts having problems at +10C [50F]. The TX power drops significantly when I attempt to send large amounts of data at that temperature. Above 15C, it works great. This problem is so severe, it makes me wonder if something’s wrong with my test setup. It’s amazing, though, how it consistently works up past 70C but won’t work below 15C.

[cgervasi]

I repeated my tests without the heatsink. I can start it after soaking for an hour at -40C, and it works fine. If there is a heatsink attached, problems begin around +10C [50F]. The problem seems tightly related to temperature, making me suspect the electrical presence of a large piece of metal is not affecting the circuitry.

It appears if you really sink away the heat well, XR5 does not work well.

[UBNT-Mike.Ford]

Hey Cgervasi,

I have been running my XR5's here at temps down to -40C and beyond without issue.

Can you please describe your test setup? Are you exposing the bare board to the temperature? and if so are you getting condensation or ice on the board?

Thanks,
Mike

[cgervasi]

Can you please describe your test setup? Are you exposing the bare board to the temperature? and if so are you getting condensation or ice on the board?

I am not having a noticeable problem with condensation. The setup is a Gateworks platform inside a metal enclosure. The enclosure has a metal heat sink that is coupled to the XR5's heatsink by a thermal pad. So the heat sink goes under the XR5, between XR5 and the host board. This heatsink is thermally connected to the enclosure as well, dissipating radio heat into the enclosure.

In the exact same enclosure without the heatsink, i.e. nothing but air between the XR5 and the Gateworks board, the XR5 is fine. With the heatsink, it appears that the heatsink is too good at removing heat. I'm not sure if that's the mechanism, but with the heat sink, it won't work below +10C.

[UBNT-Mike.Ford]

Hey Cgervasi,

I do not have a way to replicate this here, however I would not recommend adding or subtracting the XR5 heatsink.

Thanks,

Mike

[cgervasi]

It turns out the sensitivity to low temperatures I observed was only a problem on one unit. I tried two other XR5 cards, and they work fine down to -45C with or without the heatsink.

With the heatsink, the chilled XR5's RSSI increases about 10 dB, even though the radio it's linked to is outside the chamber.

With the heatsink XR5 works well at -45C and up to 85C. You can go to 90C, but the link gets unreliable if the heated unit is transmitting at its full data rate for more than 5 minutes. Without the heatsink, in my application which has XR5 plugged into a Gateworks IXP board, I can go up to 65C with continuous TX and 70C with light data traffic.

XR5 does not have a low-temperature thermal problem as I had thought. The problem was just with one unit, which I may have damaged myself.

[UBNT-Mike.Ford]

It turns out the sensitivity to low temperatures I observed was only a problem on one unit. I tried two other XR5 cards, and they work fine down to -45C with or without the heatsink.

With the heatsink, the chilled XR5's RSSI increases about 10 dB, even though the radio it's linked to is outside the chamber.

With the heatsink XR5 works well at -45C and up to 85C. You can go to 90C, but the link gets unreliable if the heated unit is transmitting at its full data rate for more than 5 minutes. Without the heatsink, in my application which has XR5 plugged into a Gateworks IXP board, I can go up to 65C with continuous TX and 70C with light data traffic.

XR5 does not have a low-temperature thermal problem as I had thought. The problem was just with one unit, which I may have damaged myself.

Hello,

Thank you for the update.

Mike