CC1125 output power (433MHz band)

How can you conclude that the problem is related to CC1125 output power ?

Have you measured the conducted output power from the CC1125 on your boards ?

The gain parameter alone in the RFFM6403 varies 2 dB.

Regards,

    Richard

The UHF signal is "sniffed" along the UHF path (from CC1125-PA pin to RFFM and beyond) with a passive tip, with 1K series resistor to avoid major perturbation of the circuit. The absolute power reading is not very precise, but relative levels can be compared.
So far, this has been the method to compare the prototypes and the production boards.

Right, RFFM gain is very dependent, expecially on temperature (and not only), but again, the level sensed with the passive tip at CC1125-PA pin is already very different between the boards.

Can you please clarify which is the expected relation between register 0x2B (PA_CFG2) and output power in 433MHz band on 50 ohm resistive load? What is the expected tolerance?

Kind regards,
Renzo

Thanks Richard,
the provided tolerance clarifies that we should have a HW problem somewhere. The production batch of CC1125 is different btw our protos and production units, but the effect should not be so dramatic as we observed.
Final question: given your provided tolerance, can I refer to figure 4-10 of CC1125 datasheet also for 433MHz band?
Kind regards,

With CC1125 programmed with 0x2B (PA_CFG2), the typical output power is around 6 dBm.

If you are targeting 30 dBm output power from RFFM6403; then the output power from CC1125 should be increased to about 10 dBm due to the gain. To avoid any gain variations in RFFM6403, then it could be better to increase to 12 dBm.

Figure 4-10 can also be used for 433 MHz as well but the output power for 433 MHz will be slightly higher than for 868 MHz.

Dear Richard,

I made some more tests around CC1125 and I have some nice discoveries to share.

The attached XLS summarizes the measures done directly at CC1125 PA pin, leaving only the polarizing inductance (56nH in series with 18R//47pF to +3.3V) and a 39pF ac-coupling ceramic. The output is directly soldered to a semi-rigid coax and observed with a spectrum analyzer. The goal is to bypass all the tolerances induced by the RFMD amplifier chain and observe the CC1125 PA output.

Three board have been transformed in this way:

Proto #3 (marking "CC1125 TI 49I AX55 G4") <= quite "strong" output

Prod .102 (marking "CC1125 TI 55I CGYK G4") <= very "weak"

Prod .103 (marking "CC1125 TI 55I CGYK G4") <= slightly "weak"

What we can observe is that the CC1125 output:

a) was quite consistent (+-1dB) between the 5 prototypes (marking "CC1125 TI 49I AX55 G4") - even after cumulation with the tolerances of the RFMD amplifier chain

b) is very part-dependent for CC1125 in the second manufacturing lot (marking "CC1125 TI 55I CGYK G4"): 3 dB difference ?!? between prod .102 / . 103 and up to 5dB compared to previous manufacturing lot.

I am quite out of ideas. Any hint is welcome.

Kind regards

RF tolerance - serie production.xlsx

The way you measure makes it a bit difficult to comment. The point where you are measuring is not a 50 ohm point and by soldering a semi rigid in here gives the PA not too ideal impedance.

Sirs,

Resuming: I have 5 prototypes, output power quite coherent.

Now I have additional 45 boards, which exhibits quite variable output power levels. The only way to correctly and reliably pin down the problem is to perform "conducted measures", as suggested by Richard in a previous post and as I did, finally.

The impedance: from CC1125 datasheet, Fig 4-16: inside a 10dB return-loss circle, output power is +13/+14dBm; moreover, pg. 17, optimum load @ 433 MHz = 55 + j25 ohm. I present to the three boards the same 50 ohm load. If I can measure otherwise (or better), please advise.

The datasheet (pg.17) does not specify MIN nor MAX, only TYP values for transmitter...
The only tolerances I ever saw have been given by Richard in a previous post: "Typical tolerances for a fixed voltage and temperature from batch to batch would be approx +/- 1 dB." Good news, but unfortunately I observe far more than that. Should I blame the PCB or the soldering process? From visual inspection, all seems fine.

Citing my very first post:
"Datasheet states (pg. 17) a minimum output power of -11dB (typ., although at 869.5MHz), and a maximum output of +15dBm (typ at 433MHz), which is roughly consistent with "output step size" = 0.4dB (typ) => [63(max)-3(min)] * 0.4dB ~ [+15 - 11]dB
Note, however, that this is NOT consistent with the User Guide formula (pg. 44 and pg. 88) "output power = [(PA_POWER_RAMP+1)/2] - 18" which gives the extreme limits [-16dBm, +14dBm] and a step = 0.5dB."

WHAT IS THE CORRECT VERSION? Datasheet (and my personal extrapolations - first paragraph) or User Manual?


I suspect we will finally calibrate individually the boards in order to offer our clients a consistent output power. But it is very important to know if the output power of the CC1125 can be relied on... or not.

- How does the interface between CC1125 and FEM look like? Knowing this will it making it easier to recommend how to measure.
- To be able to state min/ max data in the datasheet we would have to implement screening in production test which is fairly difficult to do due to the way the test hardware is designed. To find the expected output power for a given PA_CFG2 setting, use SmartRF Studio. These values are based on lab measurements. The equation in the UserGuide is not accurate and has limited value. Note that the variation of the output power is dependent on output power. The variation is larger for low output levels.

Thank you for the provided information.

I'd be glad to send you schematics, photo of the testbed, etc. in a private manner. My email is renzo.posega@livetools.tv
Please drop me a line and I'll send you more data.

Kind regards,

Renzo