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.

tlv3491

Other Parts Discussed in Thread: TLV3491, TLV3691

1321.TLV3491.pptxHI,

    We met some problem of TLV3491, as attached, pls advice how to optimize it.

regards,

haroad

  • Hello Haroad,

    My assessment of the TLV3491 comparator circuit leads me to believe that the oscillation-like waveform at the output is the result of noise modulating one, or both, of the input levels. The output is merely switching state in response to the noise-related amplitude changes at the inputs. When the adjustable resistor is adjusted such that the two input dc levels are identical the comparator will then respond to noise as if it were an intended input level change on an input. This is a common response for analog comparators and not the fault of the comparator. It's doing what it supposed to do - comparing two input levels and making a decision where to set the output state.

    A solution to the response is to reduce the noise sensitivity of the comparator circuit. That is done by adding hysteresis to the circuit. It is an application of positive feedback that sets up two different input threshold voltage levels; one for an input voltage transitioning from a lower to a higher level, and another for one transitioning from a higher to a lower level. The difference between those to threshold levels is the hysteresis voltage. For very noisy circuits the needed hysteresis voltage may be hundreds of millivolts. For a less noisy circuit a hysteresis voltage of a few millivolts, or few tens-of-millivolts may be sufficient.

    There are some standard circuits for adding hystereis to a comparator, but because of the very high 10 Megohm resistors in this circuit the hysteresis feedback resistor was calculating in the Gigohms. Such resistors a hard to obtain and expensive. I modified the circuit to divide down the effect of the hysteresis and that allowed for lower, more practical resistor values to be used.

    Below you can see how I modified the customer's circuit to add +/-5 mVpk hysteresis to the input threshold. More, or less, could be added as needed to reduce the noise sensitivity.

    Let me know how this works.

    Regards, Thomas

    PA - Linear Applications Engineering

  • Thomas,

        Thanks for this valuable information. We'll try it.

       another is, we also tested TLV3691, it's ok with the original circuit. Why they are different? Can you confirm that if TLV3691 has internal  hystereis function?

    regards,

    haroad 

  • Hi Haroad,

    The TLV3691 does have internal hysteresis. Please see section 7.3 of the datasheet. A distribution of the hysteresis is provided in Figures 25 and 26 of the datasheet.

  • Hi,
    As the TLV3691 has internal hysteresis, how can i get the internal hysteresis SCH in IC? I want to add a external hysteresis and figure out the result of combing internal hysteresis with external hysteresis.
  • Hi Hollis,

    We cannot post any part of a design of an IC to the public because it is confidential material. The external hysteresis should dominate any hysteresis interal to the IC.

  • Hi Timothy,

    Thanks for the valuable information.
    The detail is in the attachment ''TLV3691 test for e2e.pptx'', could you help to verify that?
  • Hi Hollis,

    I am reviewing your circuit using the TLV3691 with external hysteresis. One thing I have noted so far is that the TLV3691 model's internal hysteresis is closer to 30 mV, not the 17 mV typical listed in the datasheet. That will make some difference.

    I'll let you know what I find.

    Regards, Thomas

    PA - Linear Applications Engineering

  • Hi Hollis,

    I made calculations to determine the hysteresis voltage switch points initially assuming an ideal comparator and the resistor values used in your circuit; noteably with the 22 MEG feedback resistor. The ideal comparator's output would switch between 2.65 V and 0 V. Using superposition I determined that when the output is high (+2.65 V) the voltage at the non-inverting input would be 96.556 mV. When the output is low (0 V) th evoltage at the non-inverting input would be 66.398 mV. Thus, the external hysteresis provided by the 22 MEG resistor in conjunction with the other resistors is the difference between those two levels, or 30.2 mV.

    I mentioned when I tested the TLV3691 simulation model's internal hysteresis it was found to be about 30.5 mV. This can be seen in the circuit shown directly below.

    Using TINA to analyze the circuit containing the 10 MEG and 260 k resistor, but no 22 MEG feedback resistor, it was found that the internal hysteresis alone produced the following results:

    Internal Hyseteresis

    Vin (non-inv input)         Max 116.86 mV

                                           Mid   66.97 mV

                                           Min    17.07 mV

                                           Max - Min = 99.79 mV            Which is just about the 100 mV being applied at the input by the external generator.

    High-to-low transistion voltage     54.361 mV

    Low-to-high transistion voltage    84.948 mV

    VHysteresis = 84.948 mV - 54.361 mV = 30.6 mV            Which is the same result as for the simple test circuit shown above. 

    Added External Hysteresis

    When the 22 MEG feedback resistor is added the effects of both the internal and external hysteresis are seen.

    Added External Hysteresis

    Vin (non-inv input)         Max 145.87 mV

                                           Mid   81.32 mV

                                           Min    16.72 mV

                                           Max - Min = 129.15 mV             Which is the 100 mV from the input generator, plus about 30 mV of the jump seen in the waveform added by the external hysteresis.

    High-to-low transistion voltage     54.439 mV

    Low-to-high transistion voltage    114.068 mV

    VHysteresis = 114.068 mV - 54.439 mV = 59.6 mV            Which indicates that the internal and external hysteresis of about 30 mV each have added.

    If the internal hysteresis had a negative sign, and the external has the positive sign it appears they would have cancelled. But that isn't the case here.

    Do you have a different amount of external hysteresis in mind? If so, then calaculations can be done to set it to a different level. Keep in mind that the 17 mV typical hysteresis is more likely what an actual TLV3691 will exhibit.

    Regards, Thomas

    PA - Linear Applications Engineering