This thread has been locked.
If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.
Hi,
On about 50% of the parts, when performing FSCAL in temperature region of +70C to +80C, the devices become intermittent (communication).
When the ambient temperature goes above 80C or below 70C, devices go back to normal operation. If we do not perform FSCAL, this problem does not occur: devices work perfectly fine to >+90C.
We tried using different PPM crystals, as well as a precision external oscillator (outside of temperature chamber), with same results. So seems like using the internal oscillator or not using the internal oscillator does not make a difference for this issue. To reiterate, there is no reference (40MHz) frequency drift when this issue occurs. What can this be? Does this indicate a defective CC1200 device? Thank you!
Hi,
Can you check if the PLL is out of lock in this temperature range? In SWRU346 (CC120x Low-Power High Performance Sub-1 GHz RF Transceivers User's Guide): https://www.ti.com/lit/swru346 :
If you connect the board(s) using SmartRF Studio and control the device using it, do you see the same behaviour?
Regards,
Zack
The PLL (FS Out of Lock) is not out of lock (still locked) in this range, when this occurs.
The device is already mounted on our production board, but all register values are from the SmartRF Studio.
If you program the chip to send continuous, random modulated data, how does the spectrum look like after you have done a FSCAL outside and inside the problematic temp range?
Yes: Blue = normal (no FSCAL), Black = after FSCAL
So the CF shifts down quite a bit after FSCAL... (Done with reference X-Osc, so no crystal drift.)
Hi Alex,
If you use the standard 500 ksps 4-GFSK PHY from SmartRF studio, do you still see this behavior?
Regards,
HG
I also have to mention:
If I don't do FSCAL, the spectrum does not shift and everything works well to 90C and above...
- I would assume that you get the same frequency shift even if you send a CW? If that is the case, could you measure the frequency shift you see? And also the delta between the expected frequency?
- Not sure if it's clear from the above: Does the same happen regardless of which temperature you did the last FSCAL on? And is the frequency shift the same?
This only happens if FSCAL is performed between 70C and 80C ambient temperature. Outside of this temperature region, FSCAL brings it back to correct frequency. I'm also not exactly sure if this is an actual frequency shift or distortion of some kind.
Tis is the same problem with 4GFSK/500ksps. Looks very weird...
Again: Blue is normal (no FSCAL)
Black: Problem (FSCAL performed at 70C)
When FSCAL is performed under 70C or over 80C, everything returns to the Blue trace and operating as normal.
(This is with external reference oscillator. No crystal.)
I believe you didn't answer my question.
- Does this happen if you do a calibration at, lets say, 65 degrees and then again at 75 degrees? What I'm asking about if the delta in temperature between the last successful calibration and the calibration between 70 and 80 C.
- Are you able to do the same measurement using a carrier wave? Using a CW will show better the frequency shift and also the degradation. From the plots it looks like the spectrum is no longer symmetrical.
- It's very few external components that can affect this:
-> The 40 MHz xtal beeing outside the spec stated in the datasheet. If the frequency is on the programmed frequency at room temp I doubt that the external xtal cause this.
-> The cap between LPF0 and LPF1 being the wrong value and or the wrong part number.
If I FSCAL at 65C, operation is Normal. If I then FSCAL at 75, the operation is Intermittent (plots above). Then if either the temperature goes below 70C or above 80C, and I FSCAL, it will return to normal operation (blue traces). So this only happens if FSCAL is performed at 70C to 80C. FSCAL functions as prescribed outside of this temperature range (tested from -55C to +90C).
My CC1200 is in the link configuration already. How can I use CW?
What do LPF0 and LPF1 do exactly?
Those settings only impact phase noise (you will see the difference only when testing with a CW or narrow band setting), I doubt that they will impact anything.
LPF0/ LPF1: Sets a pole (I think) in the lookfilter in the synth. If the wrong part is used (at least value) it can impact the spectrum.
The CC1200 has a lower allowed minimum RX bandwidth compared to the CC1201, otherwise they are the same.
Regards,
Zack
Are you able to try with 1.5 nF? Double-checking several reference designs, the CC120x devices have a recommended value of 1.5 nF.
No... I tried with 1.8nF and 1nF... Same result. But when I went down to 100pF, it didn't work at all.
Is there an actual publicly available selection guide for this?
For this, I'm using external Rohde & Schwarz bench top reference source, outside of the temperature control. (I tried 10 to 100 ppm crystals, with exactly the same results). Crystal is not connected for these plots.
Update:
Here's the same spectrum for CC1201:
4GFSK, 625ksps, FSCAL at 70C (also tried with FSCAL from 30C to 80C), ECS-400-10-37B2-CKM crystal.
Works as prescribed, almost 100% TX spectrum correlation vs. temperature.
CC1200 and CC1201 are equal, the only exception is fewer RX BW options for CC1201.
How many boards have you tested with CC1200 and are all CC1200 you have tested from the same batch (same marking)?
Any chance of bad soldering of the ground pad?
I'd say, I got >50 CC1200 like this. We X-ray every board, but there is always a chance, of course... And yes, I only tried 1 CC1201 (building 6 more units, and will update on results). I understand that CC1201 is a bit more expensive. Is there any chance CC1201 are from the center of the wafer, or something... Maybe better screening on these?..
Update: I built 7 boards with CC1201s and they all have good performance. No problems described above are exhibited.
Hi,
The radio on these two devices are identical.
Could you do the test with the 50 kbps PHY, and also with a CW?
Could you also upload your schematic here?
Regards,
HG
As I posted above, changing the symbol rate does not solve the issue.
(I cannot test with CW, since it only shows this in a 2 endpoint link.)