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.

TL431LI-Q1: Spice Simulation Model : Worst-Case + Propagation delay issue

Part Number: TL431LI-Q1
Other Parts Discussed in Thread: TL431LI, ATL431LI, TL431, TLV3012-Q1

Dear Ti Community,

For one application, I plan to use the TL431LI part in open-loop mode (for voltage comparison) and, by the way, need to perform many simulations.

So, I have several questions :

  1. What are the differences between the "Transient" and "Average" Spice models for this reference ?
  2. Then, I want to perform simulations with all worst-case conditions including manufacturing tolerance on some parameters (like the Vref dispersion accross the full temperature range with min. and max. values). Is there any Spice model available to simulate worst-case behaviour of the chip ?
  3. Last item is related to a very simple model I have made and for which I have question because it is not well-document within the datasheet.
    • Did someone has any explanations about the propagation delay of 50µseconds ? It seems that this delay cannot be shorter !
    • In the datasheet (Figure 21 / Page 15), there is no 50µseconds of propagation delay I have noticed in my simulation.
    • So, every comment to my concern will ever be welcome !

Kind Regards,

  • 1. The following is for the TL431LI models. The TL431LI models are behavioral models. The transient pspice models are used to test transient conditions or DC operating points but they do not have accurate AC modeling if you want to do loop stability tests. The average pspice model is used to test for loop stability.

    2. We do not have a model to perform worst case simulations.

    3. The models do not have accurate propagation delay. The propagation delay should be faster and in the micro second time frame.

  • Hello Marcoo,

    First, thank you for your answering.

    Regarding your suggestion of using the ATL431LI instead of the TL431, I would rather stay with the TL431LI as it is the only one which has a grade-0 option (+150°C max ambient operating temperature).

    Then, if I switch back to my initial request, in fact, I would like to perform a very accurate study with the use of the TL431LI (the propagation delay matter to me a lot).

    I am a bit surprised that Texas Instruments does not have a model for worst-case simulation because in automotive field, this would be usual to have such kind of study.

    Would you have any suggestion to get or develop a worst-case model for the TL431LI (full temperature range, Vref drift and dispersion due to manufacturing tolerance, propagation delay,...) ?

    Kind Regards,

  • Hi,

    From my experience, the TL431LI and TL431 have similar propagation time. While this is not an idea solution, the TL431 model has more accurate propagation delay around 1-5us with no load cap. Regardless both device models do not have parameters embedded for worst case simulations such as monte carlo simulations.

    At the same time the TL431 propagation delay is not characterized so information on it is limited.

    Are you using the TL431 for a comparator application?

  • Hello,

    Indeed, I would like to use the TL431 in comparator (open-loop) mode.

    The fact that this device has a built-in reference voltage was quite interresting to get good accuracy on the comparison threshold voltage but the propagation delay (in first simulations with 50µseconds min.) is very limiting.

    The aim would be to remove steady-state on a PWM signal (low frequency <400Hz) still with good accuracy on duty cycle (about 0,25%) using integrator and comparator.

    Kind Regards,

  • Hi,

    Very interesting application. Would this require very sharp switching? The TL431 is good comparator but it does conduct near its threshold that cause it to be semi active and not as sharp.

    We do have a large selection of op amps with integrated references as one chip solutions that might interest you with sharper switching.

  • Hello,

    Thank you for the link to op-amp parts with integrated reference voltage.

    In fact, my technical needs are listed below :

    • Automotive qualified part (AEC-Q100 mandatory)
    • Operating from a +12Volts battery supply (range +8,0Volts to +35Volts)
    • No additional regulated power supply available (no +5,0Volts nor +3,3Volts)
    • Temperature working range : up to +140°C as ambient temperature !

    Among the part list you have provided, the TLV3012-Q1 is the only part that can address automotive application.

    In addition, it needs a regulated power supply and cannot cover an operating temperature up to +140°C.

    Would you have any other parts that could suit best to my needs ?

    Kind Regards,

  • Hi,

    I see the difficulties with those design requirements.

    How about a multiple device solution.

    TL431LI -> LP2903E-Q1

    The LM2903E-Q1 is 30V which is not 35V but we can add a zener or possibly another TL431LI to protect it from 35V. I assume 35V is not a typical condition so the Zener will usually be off and not consuming current. Then we have a AEC-Q100 grade 0 comparator + VREF solution.

    Is this for overvoltage/undervoltage monitoring of the battery?

  • Hello,

    Indeed, the technical requirements are very demanding in addition that we are in a very space constraint design too !

    I would avoid as much as possible a multiple device solution if we cannot meet all the requirements (TSSOP-8 is too big).

    Then, you are right, the +35Volts won't be a typical supply voltage. This requirement is related to ISO 16750-2 pulses, like the pulse 5b (centralized load dump).

    My application is not related to battery monitoring system (for under- over-voltage) but rather related to some kind of "dead-line" monitoring on a PWM signal.

    The PWM signal shall operate with a fixed frequency of 200Hz (typically) and the duty cycle provides a setpoint for powering a load.

    If the PWM signal remains stuck at high level longer than one period, then we have to power-off the load as quick as possible.

    Currently, I am working with an integrator stage and a comparator but others solutions (with less drawbacks) might exist.

    If you have any suggestion, I would be glad to hear from you again !

    Kind Regards,

  • Hi,

    That is good information. Thank you.

    I can test the TL431 response time in the lab across temperature if this will help you understand the typical performance. I do expect it to be 1us to 10us. Would this be useful to you?

  • Hello,

    Of course, I would highly be interested by your suggestion of performing some tests about TL431 response time.

    For the testing, could you, please, study the response time according to the following settings :

    • Part reference : TL431LIAEDBZRQ1 (ideally)
    • Power Supply Range : +8,0Volts / +9,0Volts / +13,5Volts / +16,0Volts / +17,0Volts / +35,0Volts
    • Temperature Range : -40°C / +25°C / +120°C / +140°C / +150°C
    • Pull-Up Resistor on the Cathode : 1kOhms / 4,7kOhms / 10kOhms / 47kOhms
    • Cycle Count (for each given configuration) : > 1k Ton/Toff (in order to draw bar chart)

    In addition, during testing, I would be interested to record : propagation delay, turn-on and turn-off times.

    This is a quite huge wishlist but I really need this data !

    Kind Regards,

  • Hello Marcoo,

    I come back to you to know if you would have some news to share with me regarding my last request ?

    Did you have opportunity to perform the expected test ?

    Kind Regards

  • Hi Arnaud,

    I managed to squeeze some lab time in.

    I cant perform all the tests requested as I do not have that capability on my bench I took a subset of the tests focusing on some of the extremes and typical. I also removed tests that could damage the device.

    I hope this is sufficient. the typical transition of the TL431LI with high currents ~800ns. At low currents this can be 8us. This time covers the main switching. There is still a settling time for the voltage to reach a stable state. So i provided a zoomed in screenshot and a zoomed out one so you can see the behavior. The effect of temperature is very minimal.open loop switching.zip.