LimeSDR-mini: What needs to be cooled and to what temp?

**frog** · October 6, 2019, 7:12pm

I’ve changed the title and content of the original post, which was about my LimeSDR mini not receiving below 31mhz when the temperature of the LimeSDR mini as measured on Lime Suite was above 45C.

After more experimentation, it appears that the temperature readout of LimeSuite doesn’t really make sense to me. At a measured temperature of 45C in LimeSuite, a FLIR infrared camera measuring the temperature of the FT601 and Altera FPGA chips puts the surface temperature of the IC’s at approximately 158 F (70C). I have a heatsink over the can on the LMS7002M, so I can’t really accurately measure the LMS7002M surface temp.

Therefore, I can only imagine that when my LimeSDR mini was having low-freq receive issues, the board temp was well over 70C.

So maybe I need to ask a better question - which ICs need to be cooled on the limesdr mini, and what is the recommended surface temp?

**ultrajv** · October 6, 2019, 1:54pm

Something is wrong there. I dont need cooling to receive under 31Mhz. I can go to 7Mhz just fine with none attall.

**frog** · October 6, 2019, 1:57pm

not sure you understand my post. you can go to 7mhz when the fpga is over 45C temp?

**ultrajv** · October 6, 2019, 2:00pm

Ive never used cooling of any sort and am able to receive 7Mhz. Maybe your enclosure dosnt have adequate ventilation?

**frog** · October 6, 2019, 2:06pm

My board is in a sealed enclosure because i have an ip54 requirement, which is not unusual.

**ultrajv** · October 6, 2019, 2:13pm

Ok, if you look at the enclosures that are available for LimeSDR mini. Most have plenty of ventilation holes.

**andrewback** · October 11, 2019, 7:26am

Indeed and I’m guessing the SDR and its host is just heating up the air in the box. Depending on ambient temperature and workload/throughput, this could become an issue.

**kbthomas** · October 12, 2019, 2:56pm

I found that I need to cool both my LimeSDR-mini and LimeSDR-USB in order to keep the gain stable and at a reasonable value as I work 1420MHz for radio astronomy work. Heatsinks won’t do it. I use an mineral oil bath at 10 degrees (0 for the mini) for that. It’s important not to get too low in temp else the clocks stop working properly.

**kbthomas** · October 12, 2019, 2:57pm

Sorry- that’s 10 and 0 Celsius

**frog** · October 14, 2019, 1:50pm

I soldered a 2-pin header to the J10 holes on the PCB, and attached a 5V fan. I did not add any heat sinks to the ICs (yet)

The fan now blows on the board and the on-board temp sensor reads 36-38 deg C in a 24C ambient room.

Receive performance is good and does not falter now after being on for 48 hours

I think it should be mentioned somewhere that adding a fan to the limesdr mini to regulate the board IC temps is highly recommended!

**wswat** · October 17, 2019, 8:07am

Hi, we have some relevant feedback here…

We have systematically tested ten brand new LimeSDR-USB because we were observing self-test failures.
Each PCB was tested unboxed at an ambient temperature of 22 Degrees C, no heatsinking, no moving air.
Results:

9 of the 10 units failed the self test on the high-band Rx sensitivity (the on-chip temperature sensor indicated 58 - 60 Degrees on all units at the time of failure).
One unit passed the self test when the on-chip temperature exceeded 60 Degrees.
Remedial Action:
We have glued a variety of heatsinks to the various failed units and will run further tests.
Comment:
In my opinion, this exercise is indicative of an inherent thermal design flaw in the chip/PCB. Temperature is clearly an issue for the device. Given the existence of other SDRs on the market using the same LMS chip (e.g. Fairwaves) which are rated over an industrial temerature range, it seems the PCB needs a makeover from a thermal design perspective.
We LOVE the LimeSDR features and price point, but this issue makes it untenable as the basis for a commercial product for us.
Some feedback from Lime would be appreciated here

**9a4db** · October 18, 2019, 9:44am

We have the same conclusion on the subject.
Regarding the cooling you need to have at last two heatsinks on the top of LMS & FPGA or
one with two levels and whole bottom of the USB board on the solid oversized heatsink
or metal case. Good luck.

**frog** · October 18, 2019, 12:20pm

I guess the real question is whether or not anyone has successfully run the limesdr mini or limesdr long-term on passive cooling only.

This is currently what my setup looks like. With a small 5V fan, the on-chip temp without heat sinks is 36-38C, which seems to work well for the mini.