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.

LMLH6702 Distortion measurement

Guru 20090 points
Other Parts Discussed in Thread: LMH730316, LMH6702, LMH730227, LMH730216, LMH6629

Hello,

Is LMH6702(SOT23 PKG) datasheet specification tested with LMH730316?

At the customer evaluating of LM6702(SOT-23) with LMH730227, the distortion data is worse about 10dBc from the datasheet typical value.
Could you please let me know the reason of this difference?
Is this difference influence of EVM ?
Please see the follwing as the detail.

<Measurement Result>
Inverting circuit, G=-1 , Scheamtics is same as Figure 25.
5MHz,10dBmOut(RL=10ohms), HD2/HD3 = -72dBc/-86dBc
In the datasheet P.5 HD2/HD3 = -100dBc(typ) / -96dBc(typ) 


Best Regards,
Ryuji Asaka

  • Hi Ryuji,

    The LMH6702 distortion data were generated on the predecessors of the current LMH730227 (SOIC) and the LMH730216 (SOT-23) which had the "CLC" prefix. I believe the LMH730316 to be a good EVM for testing a low distortion device like the LMH6702 (based on good distortion we measured with the LMH6629 using the same board at a much later release date).

    Here is another E2E post about distortion measurement precautions, and which has a link to a document useful in making distortion measurements:

    http://e2e.ti.com/support/amplifiers/high_speed_amplifiers/f/10/p/275045/962620

    Here is the document I am referring to:
    http://m.eet.com/media/1134294/15200-476907.pdf

    Almost all of the LMH6702 data is with a lighter load (RL=100ohm) whilst the data you've shared is with RL=10ohm even though you've kept the power at 10dBm (0.89Vpp or 89mApp output current). I'm curious to know if the customer's setup can test with datasheet load (RL=100ohm) so that we can establish a baseline to know if their setup is matching ours or not? The page 5 distortion numbers you've quoted correspond to much lower output current (2Vpp across 100ohm or 20mApp) or 13dB lower output current than the customer's setup. From Figure 12 shown below you can see that increasing the x-axis by ~13dB affects HD2 by more than 10dB (granted Figure 12 x-axis increase involves higher output swing as well as higher output current- I cannot be certain which change accounts for the biggest impact?):

    Please review my comments and let me know if you have any questions? I'll look for your response back to make sure we get better distortion data matching.

    Regards,

    Hooman

  • Hello Hooman san,

    Thank you for the reply and sorry...I made a typo.
    Actual RL is 100ohm. Not 10ohm.

    If we measured LMH6702 SOT-23 PKG specification with LMH730227, is the specfication worse than datasheet specification which was tested by LMH730216?

    Best Regards,
    Ryuji Asaka

  • Hi Ryuji,

    I expect better distortion results from the LMH730316 than from LMH730227. Some of the effects of the layout on distortion, with a highly-linear device like the LMH6702, was something that we "learned" as we went along and we got better at it later at which time we developed the LMH730316. So, some of the plots in the LMH6702 may have been optimized individually for the test frequencies being considered, because at the time the CLC730227 layout we had was less than optimal.

    The most critical thing is the return path from the decoupling capacitor "ground" lead AC current back to the power supply and away from the sensitive input nodes. The capacitor AC  current contains harmonics which corrupt your data with a less-than-optimal layout.

    Also,as you might have seen in the link I sent before, copied below, the way you configure your load (RL=100ohm) also matters since you like to minimize the return current return path length flowing through the cabling tying your DUT output to the Spectrum Analyzer, as shown in Figure 2 from this same link showing a 1kohm load with minimal current flow through the load-side transmission line:

    http://m.eet.com/media/1134294/15200-476907.pdf

    Good, clean input excitation (using sharp / deep filtering) also could affect your readings. Spectrum Analyzer distortion is also something you may have to ensure against for a more reliable data by using appropriate attenuators / notch-filter without affecting dynamic range.

    Here is an article from Xavier Ramus which might also help out:

    http://citeseerx.ist.psu.edu/viewdoc/download;jsessionid=5080F69C6E208D7597AD7F61C6C99448?doi=10.1.1.138.299&rep=rep1&type=pdf

    Regards,

    Hooman

  • Hello Hooman san,

    Thank you for the reply.
    I understood that there are the layout effect for the distortion.

    The customer using LMH730227(for SOIC pkg EVM) with LMH6702 SOT-23 PKG for their evaluation.
    And there is the difference about 30dBc between their measurement and the datasheet.
    I understood that the layout and measurement method are important for distortion.
    However, this 30dBc is big.

    I arranged the LMH7320216 and LMH730316 to the customer and they will evaluate LMH6702 with these boards.
    I will let you know when I got the result from the customer.

    Best Regards,
    Ryuji Asaka

  • Hello Hooman san,

    Sorry again.

    Could I have the test circuit and measurement structure for distortion parameter of LMH6702?
    Is this same procedure with your information of EDN (1kohm load with minimal current flow through the load-side transmission line) ?
    I would like to know the actual measurement method of the Distortion and noise response in the datasheet.

    Best Regards,
    Ryuji Asaka

  • Hi Ryuji-san,

    The test circuit for measuring the LMH6702 distortion is as follows:

    RF= RG= 237ohm (for Av=+2), RL=100ohm total (generally 50ohm series feeding a 50ohm input of either an SMA attenuator which then feeds a 50ohm notch filter). I don't think we used the technique in the EDN article to reduce the load current return path length at the time.

    Test setup In General (I don't have access to the specifics of the setup, but I can offer general guidelines):

    1. Input low pass filter of test frequency (50 ohm setup) used to get a clean input sinewave and eliminate source distortion terms.
    2. Output passed through notch filter (of test frequency) before being applied to the spectrum analyzer. Use attenuator before the notch filter (to de-sensitize the DUT loading due to erratic notch filter input impedance at notch frequency). May have AC coupled the entire load path.
    3. Spectrum Analyzer span reduced, input attenuation increased as much as possible to reduce Analyzer distortion. Start at maximum attenuation and reduce until reading gets "unexpected".
    4. Use supply decoupling cap between V+ and V-. If you can affect distortion this way, chances are your layout is not optimal and you are seeing the effect of supply decoupling cap return path current affecting results.

    I don't remember much other details.

    Regards,

    Hooman

  • BTW, the low pass filter on the input and the notch filter at the output that I have noted are part of our test setup and they are self-contained / custom-built filters that we have for measuring distortion. They have an input and an output port each (SMA or BNC type).

    Regards,
    Hooman