Because of the Thanksgiving holiday in the U.S., TI E2E™ design support forum responses may be delayed from November 25 through December 2. Thank you for your patience.

  • State
  • Locked Locked
  • Replies 6 replies
  • Subscribers 47 subscribers
  • Views 481 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

TRF37A75: Phase Noise of TRF37 Family of RF Gain Blocks

FarronDacus
Intellectual 750 points
Part Number: TRF37A75
Other Parts Discussed in Thread: LMX2594, , TRF37B75, TRF37B73, TRF37C73, TRF37C75, TRF37D73, TINA-TI

Hello:

I'm working a low noise synthesizer development using the LMX2594 fully integrated synthesizer, and the LM2490 with low noise external VCO's, locked against a very low noise 100MHz reference. 

With the resulting low phase noise, a possible issue is the "phase noise" of the buffer amplifiers.  Texas Instruments makes a very nice family (the TRF37 line) of moderately low noise buffer amps at frequencies up to 4GHz and 6 GHz with the convenient feature of gains available over the range of 12dB, 15dB, 18dB, and 20dB.  The 15dB and 18dB gain parts also come in two power levels, so are convenient for sometimes saving DC power.  These parts and major specs are shown in this table:

Part

Gain

Noise Fig

Freq

Supply, P1dB

Cost

TRF37A75

12dB

4dB

40-6000MHz

5V/80mA/18dBm

$0.52 at 1k

TRF37B75

15dB

3.8dB

40-4000MHz

5V/85mA/17.5dBm

$0.52 at 1k

TRF37B73

15dB

4dB

1-6000MHz

3.3V/52mA/15dBm

$0.43 at 1k

TRF37C73

18dB

3.5dB

1-6000MHz

3.3V/55mA/16.5dBm

$0.43 at 1k

TRF37C75

18dB

3.5dB

40-4000MHz

5V/85mA/19.5dBm

$0.52 at 1k

TRF37D73

20dB

3.2dB

1-6000MHz

3.3V/63mA/16.6dBm

$0.43 at 1k

Nice performance at low prices.  The only issue with using these as low noise synthesizer buffers is that the "phase noise" is not specified.  This is the noise close to the carrier that results from the low frequency flicker noise of the devices mixing up around the RF carrier to create a phase noise like total noise.  The parts are SiGe bipolar, which tends to have a low flicker noise, but it is not specified.  It could easily be a problem for buffering ultra-low noise crystal references, and possibly ultra-low noise synthesizers as well.

Whatever the noise is, it has to be lived with since there is not any way I am aware of to reduce it through external means.  With a discrete transistor buffer, this noise can be reduced through emitter degeneration.  To my knowledge, no such options exist for this part. 

Has Texas Instruments ever measured this noise?  If not, could it be measured? 

A discussion of this noise and how it is measured is at:

https://www.markimicrowave.com/blog/how-do-amplifiers-affect-signal-phase-noise/

A detailed article by phase noise expert Enrico Rubiola is attached.

Thanks,

Farron

RubiolaAmpPhaseNoise2012.pdf

  • 0
    TI__Genius 15555 points

    Hello Farron,

    I couldn't find the measured data you are looking for, however you should be able to conduct a noise analysis of the part on TINA-TI for an estimation. 

    Best,

    Hasan Babiker

  • 0 in reply to Hasan Babiker31
    Intellectual 750 points

    Hello Hasan:

    Thanks for the prompt answer. 

    The advice to try TINA for noise presentation is good, IF the TINA model accurately represents the part (often they are macromodels that do not full capture a part) and if the transistor 1/f noise parameters for the process have been carefully measured and entered into the SPICE models.  Can you say if those two conditions are true?

    If they are not true and we are misled by the TINA results, it could lead to a very negative program outcome.

    ...Farron

  • 0 in reply to FarronDacus
    TI__Genius 15555 points

    Hello Farron,

    The model states that the noise figure is accurately represented which will mean the part will need it will need to provide a proper spot noise response as well. I can't, however, confirm that the model closely matches the 1/f noise found in the part. 

    One thing to note if you would like to use the TINA model to get a response is the effect of resistor noise on your total output noise. One way to get rid of this effect is to change the temperature of the resistors to -273 C. 

    Best,

    Hasan Babiker

  • 0 in reply to Hasan Babiker31
    Intellectual 750 points

    Hasan, I think using SPICE to try to get 1/f noise translated to the carrier frequency (spot noise figure) is going to be highly problematic. 

    First,  in addition to needing the process 1/f noise parameters taken and entered in the SPICE models, since TI does not have actual measured noise data, there has been no opportunity to converge the macro-model noise behavior with the actual noise of the part. 

    Then, it has to be considered that a linear noise analysis (the normal form) would show only the 1/f noise at baseband.  The up-conversion around the carrier is a non-linear process and is power level dependent, so  the analysis would have to be harmonic balance or transient to fully capture the non-linear up-conversion of noise.  The FFT to spectrally display the noise would have to be to a very high dynamic range to capture that, on the order of -170dB at a minimum. 

    Is there a transient noise capability in TINA SPICE?  If so, can TINA SPICE transient analysis and FFT operate to that numerical precision?

    I am guessing here that the most viable way forward would be for TI to measure that data.  It would be a valuable selling point for the parts to have that information in the data sheets, showing that the TI TRF37 buffer family is a good match to the outstanding TI low noise synthesizer line.  Is there any interest there in TI measuring that information?  Texas Instruments does have the necessary phase noise test ability, so should be able to take it. 

    Regards,

    Farron

  • 0 in reply to FarronDacus
    TI__Genius 15555 points

    Hello Farron,

    Those are valid points, and I agree using the SPICE model may not be precisely accurate when translating to phase noise. There are EVMs available for these parts that can be used to collect this data.  

    Best,

    Hasan Babiker

  • 0 in reply to Hasan Babiker31
    Intellectual 750 points

    OK, I'll have to tell my client that to use these we either have to get a phase noise test set and take the data, or just take a chance.  Our other option is go to the trouble of designing discrete amplifiers using transistors for which the noise data is available. 

    Thanks,

    Farron