problem on freq drift of LO in bluetooth system

[sunjiao3]

Hi, friends, I met a strange problem!!!
In bluetooth sytstem, the tolerance of freq drift of fc for a packet with 1/3/5 time slot is 25KHz/40KHz/40KHz. And the test result verifies that the longer packet with 3/5 time slots would have higer freq drfiter than the one with only 1 time slot (625us). But why? I think, the longer the time spent, the more stable the LO would be and thus the LO frequency should have been more stable with less drift.
Would any one of you please enlight me in this question?
Thank you in advance!!!!
Kind regards!!!

 

[RFCMOS]

Hi,

The more time slots you have, the higher is the average power you transmit and then the higher is the temperature drift on your LO.
Hope this helps.

 

[sunjiao3]

Hi, RFCMOS, thank you for your answer, but I haven't understand it well. 1 time slot equalls only 625us, so, and would the possible temperature drfit be so evident to change the frequency drift? And how could I verify that it is caused by the temperature? Would you please give some examples or details?
Thank you in advance !!!!!

 

[ge]

sunjiao3,

I believe what RFCMOS is implying is that over time, the temperature could be varying your LO frequency. I uncertain about this theory. To verify, rerun your test in a temperature controlled environment (0C, 30C, 50C).

Other possibilities is LO phase noise or something modulating the LO. Longer time frame gives more time for noise and modulation to add up.

Assuming you do not have a phase noise test set, take a look at your LO on a spectrum analyzer. How clean is your power supply? Is your LO being fed from a switching regulator?

 

[sunjiao3]

Hi, ge,thank you for your answer, the temperature coefficient is a problem. In the schematic simulation, I've found that a temperature drift of only 0.015 degree may lead to a frequency drift of 15KHz. But, the transceiver is working in full-duplex way in TDD mode. That is, whether it is in receiver mode or transmit mode, the LO didn't stop to work. So, I wondered what RFCMOS said, why the more time slots the packet is composed of, the more power it consumes?

As for what you said, ge, longer time frame gives more time for noise and modulation to add up? As I said above, the lo keeps working in receive/transmit mode. Thus, the time duration of packet counts?

 

[Abhishekabs]

Is your power supply stable?
Sometimes this happens if you are not providing stable bias.

 

[RFCMOS]

Hello,

On your board, the power amplifier delivers more power when you have 3 TS compared to 1 TS. Therefore, the temperature should be higher in this case which could induce a drift of your reference oscillator.
To demonstrate this, you can use an external reference generator.
I hope that I am clear and that it could help

 

[ge]

sunjiao3,

You say the LO is operating full time... Is the transmitter transmitting full time or only during timeslots?
RFCMOS is suggesting that the RF section is only transmitting during the time slots, hence the relationship between timeslots and the amplifiers producing heat to heat the LO source.

I misunderstood and was picturing the cumulative addition of noise.