Someone by email asked:
I see issues with the BB PLL starting properly at temp (70°C). Do you guy test this?
The part works over the specified operating range, ambient of -40C to +85C. The device was tested over temperature.
Is 70C ambient temperature or internal junction temperature? What hardware is being used?
In higher power modes the part can easily exceed maximum 110C junction temperature at 70C ambient. If this is the case, thermal management needs to used (heatsink, airflow and PCB thermal design techniques).
If you are testing a whole assembly over temperature, AD9361, FPGA etc. it is possible that other devices are causing issues at 70C.
As Phineas would say "Why, Yes - Yes we do"
We have done some temp testing - on a ZC706 + FMCOMMS3. We have an oven, that we insert the FMCOMMS3 + ZC706.
Why 706? The Zynq on the ZC706 is the XC7Z045-2FFG900E - which is spec'ed to the "E" temp grade (Extended Tj = 0°C to +100°C). The DIMM on the board is Micron's MT8JTF12864HZ-1G6G1 - which is spec'ed for Commercial (0°C ≤ TA ≤ +70°C); and the chip down - MT41J256M8HX-15E is Commercial (0°C ≤ TC ≤ +95°C). The Zedboard is much lower specs, and I dont think it has been temp tested at all.
The AD9361 temp is spec'ed to Operating Temperature Range −40°C to +85°C (Although it doesn't say if this is junction, or ambient, and you need to back junction off from this -- Tomas - any insight?)
Our tests results are fine: The running temp inside the Zynq is 74.5 & the AD9361 temp is running at 70.0 - bottom left side of the screen.
The part starts up (both from power cycle, or reboot) as many times as we tried it.
-40C to +85C Operating range is ambient temperature. Maximum junction temperature is 110C (in most cases maximum junction temperature is limited by package glass-transition temperature, but in this case the constraint is electromigration).
The datasheet provides Thermal Resistance information in Table 12. With the thermal constraints provided, ambient and junction temperature specifications, thermal resistances from the datasheet and power dissipation for a desired part configuration the user can arrive at a thermal management solution required not to exceed maximum junction temperature for a given operating mode.
Ok - maybe mhennerich can describe the data in/out (looks like 2 x 2, 15.72 MSPS @ 2.4GHz - CW tones?)
Ok - some clarification on the problem statement.
To clarify – it wasn’t the high temp that made us lose lock, but a change in temperature from high to low.
We should be able to set the oven up to cycle temperatures a few times (70°C -> wait, power on, calibrate -> 10°C -> wait, check for proper operation, power cycle, calibrate -> 70°C -> wait, check for proper operation, repeat a few times).
I'm not familiar with the oven to say what sort of condensation control it has, or how loud it is, so it's kind of unknown how fast we can cycle temperatures (or how long we can do it without annoying everyone in the office). It's not going to be as fast as hitting it with a can of freeze spray if that was your test.
Michael -- do you have time to re-test tomorrow?
Just an update - Some folks missed this, and we will do the test early next week.
A quick update - today I run two tests, using different temperature cycle profiles.
Profile start 70°C -> 10°C -> 70°C:
AD9361 calibrated at 70°C ambient, running LTE20 profile 2Rx2Tx: CW 10MHz TX Attenuation -10dB.
Loopback Frequency spectrum observed with MAX HOLD, and BBPLL Lock in 0x5E:8. No unlock over entire range observed.
70°C -> 10°C -> 70°C -> 0°C -> 70°C:
During temperature maxima re-calibrate AD9361 (@ 70,0,10): No unlocks or other issues observed.
Other conditions like 1).
May I ask about more details - what is the exact temperature profile I should use?
Please provide temperature calibration points and exact device configurations for each calibration point.
Retrieving data ...