CC2540 & CC2541 TX power level

Jerome,

The BLE stack sets the different output power levels as you correctly states as -23, -6, 0 and +4.

In the case of the CC2540 you will have all these output power levels: -23, -6, 0 and +4.

In the case if the CC2541 you will only have three power levels: -23, -6 and 0

The CC2541 cannot physcially achieve the 4dBm output power typically and the stack should not allow you to set the output power to 4dBm and if it does, it will default to its max power of 0dBm.

Thanks, but just to be sure, the CC2540 will accept the -23dbm even thought it's data sheet says it cannot go lower then -20?

Yes. That's fine. The stack should be able to set the output power lower.

(I'm attaching my question as a reply to a closely-related base thread. This is my first post here so if I violated some etiquette, please be kind. ;-) )

We're using the CC2541 and our EMC testing has revealed that even with the power level set to -23 dBm, we're about 5 or 6 dB too noisy to fit under the RTCA DO-160G limits for a "Category M" device (near electromagnetically-significant apertures but not Line-of-Sight with airplane antennae). We'd rather not turn off the radio entirely and wonder about true capabilities of the TXPOWER register. The data sheet lists the following legal values:

Table 1. Output Power(1) (2)
TXPOWER Setting | Typical Output Power (dBm)
0xE1       0
0xD1    –2
0xC1    –4
0xB1    –6
0xA1    –8
0x91  –10
0x81  –12
0x71  –14
0x61  –16
0x51  –18
0x41  –20
0x31  –23

This SUGGESTS that there are at least a few more values (0x21, 0x11, and possibly 0x01) yielding lower power, but one knows not to violate the data sheet without good reason. Do these hypothetical lower-power values exist and does the chip work stably at these power levels?

Also, why does the SDK only materialize in its enum a small subset of the available transmit power settings?

#define LL_EXT_TX_POWER_MINUS_23_DBM 0
#define LL_EXT_TX_POWER_MINUS_6_DBM   1
#define LL_EXT_TX_POWER_0_DBM                  2
#define LL_EXT_TX_POWER_4_DBM                  3

#define HCI_EXT_TX_POWER_MINUS_23_DBM LL_EXT_TX_POWER_MINUS_23_DBM
#define HCI_EXT_TX_POWER_MINUS_6_DBM    LL_EXT_TX_POWER_MINUS_6_DBM
#define HCI_EXT_TX_POWER_0_DBM                   LL_EXT_TX_POWER_0_DBM
#define HCI_EXT_TX_POWER_4_DBM                   LL_EXT_TX_POWER_4_DBM

Okay, it looks like the undocumented three lowest values of the TXPOWER register DO work. Here’s the data I just took with a spectrum analyzer:

                             Relative Power
TXPOWER                 at my
Setting                Analyzer's Input

0xF1                       -27.78 dBm
0xE1                       -27.82 dBm
0xD1                      -29.80 dBm
0xC1                      -31.80 dBm
0xB1                      -33.65 dBm
0xA1                      -35.68 dBm
0x91                       -37.60 dBm
0x81                       -39.50 dBm
0x71                       -41.67 dBm
0x61                       -43.59 dBm
0x51                       -45.28 dBm
0x41                       -47.36 dBm
0x31                       -49.51 dBm
0x21                       -51.26 dBm
0x11                       -53.04 dBm
0x01                       -55.25 dBm

I did this testing using a CW Tone (via CI_EXT_ModemTestTxCmd( HCI_EXT_TX_UNMODULATED_CARRIER, CW_FIXED_CHANNEL ); ) but an earlier confirmatory test seemed to show the same general results with the ordinary Bluetooth stack up-and-running. Unfortunately, my work area is too radio-noisy to take fine-grained measurements across a wide spectrum hence the CW tone. (The spectrum analyzer input was a Bluetooth antenna very closely coupled to the antenna of my D.U.T.)

Hello,

Yes they should work and were designed in to do just what you observe on the bench.  The issue with using them today is that they are not production tested by TI and therefore we don't have them in the datasheet nor can we guarantee them. 

Regards,

BK

BK:

Thanks for your feedback! I've since done more testing (including some with the spectrum analyzer directly coupled to the CC2541's output filter in place of our D.U.T.'s normal Bluetooth antenna) and everything continues to look good: with the radio advertising (that is, in normal advertising operation rather than a CW tone), the attenuation was as expected and no funny spurious frequencies were observed at any significant amplitude.

One of our engineers asks whether the attenuation is likely to be stable over temperature. If you have any data about this, I'd love to hear it. Otherwise, we'll probably conduct this round of experiments again with our D.U.T. in a thermal chamber.

The documentation makes sense when you consider that the Bluetooth special interest group made standards that cover transmit power. Part A chapter 3 of the core specifications states that the minimum power setting for transmit power is -20 dBm. If you transmit at a lower setting, you aren't in compliance and can't claim you have a BLE device.

Atlant, would you mind letting me know what register you are using to set the transmit power? I can't seem to find any reference of it in the user's guide. 

Is there the equivalent table available for CC2540 ?

Fred Visticot said:

Is there the equivalent table available for CC2540 ?

Check the chip's data sheet! But given how similar the CC2540 and CC2541 are, I'd certainly expect that there is an equivalent (and probably equal) table.

Peter Borenstein said:

The documentation makes sense when you consider that the Bluetooth special interest group made standards that cover transmit power. Part A chapter 3 of the core specifications states that the minimum power setting for transmit power is -20 dBm. If you transmit at a lower setting, you aren't in compliance and can't claim you have a BLE device.

Atlant, would you mind letting me know what register you are using to set the transmit power? I can't seem to find any reference of it in the user's guide.