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.

ATL431: Reference Current

Part Number: ATL431
Other Parts Discussed in Thread: TL431,

I am looking to use the ATL431 as a comparator, as shown in SLVSCV5D Figure 30. In this application, the "Vin" of Figure 30 is a higher voltage signal, so "Rin" is actually a resistive voltage divider to get the voltage down to the 2.5V reference when the signal exceeds a threshold.

Paragraph 9.2.1.2.3.1 states "Because Iref can be as high as 0.15 μA, TI recommends to use a resistance small enough that will mitigate the error that Iref creates from Vin. Also, the input resistance must be set high enough as to not surpass the absolute maximum of 10 mA."

1. The first part of the sentence "Iref can be as high as 0.15 μA", this is taken from 6.5 "Electrical Characteristics" and Iref uses Test Circuit Figure 23 to establish the 150nA. However this aligns with the shunt regulator schematic of Figure 32, not the comparator schematic of Figure 30. Therefore, is this still an appropriate number to use for the comparator application? If not, what should I be using instead?

2. The second part of the sentence "to not surpass the absolute maximum of 10 mA" does this suggests the Iref is dynamic? How can Iref become 10mA? Or, what is the 10mA referring to?

The purpose of these questions is determine how to calculate the Vref tolerance, given the comparator application.

Thank you.

  • Hi SRW,

    I have answered your questions below. I believe there is always some confusion regarding '431 devices in comparator mode vs closed loop.

    1. You can consider the REF pin regulation to be current based. The ATL431 requires a 150nA (max) of current to regulate properly in closed loop. This current also serves as a threshold for when the comparator will trigger and switch between high and low states. Keep in mind that the device will not consume 150nA until the REF pin reaches 2.5V so this is why we typically refer to 2.5V as the threshold.
    Make sure that at your desired voltage threshold The REF pin has is enough current (150nA) to trigger the state change.
    Another way to look at it is (Vtrigger - 2.5V) / Rin > 150nA due to the IR drop on Rin.

    2. When the voltage on the REF pin is higher than 2.5V the current consumption of the device will be greater than 150nA and will increase based on the voltage. If there is no resistor Rin (Figure 30 from Datasheet SLVSCV5D), then the IRef can increase past 10mA and damage the device. We recommend to always use a Rin or voltage divider to ensure that there is a series resistor with the REF pin to limit the current. Rin needs to be sized according to the worst case high voltage to ensure that (Vhigh - Vref)/Rin is smaller than 10mA. While 10mA is the ABS max, I recommend to size it to ensure that its not greater than 100uA to 1mA.

    We also have an app note on this on the TL431 but it applies for ATL431
    www.ti.com/.../slva987.pdf

    If you are considering ATL431, also consider ATL431LI. The ATL431LI is an alternative to ATL431 with a higher unity gain bandwidth with and larger stability region.
    www.ti.com/.../atl431li.pdf

    -Marcoo
  • Thanks so much for the very quick response Marcoo.

    1. " When the voltage on the REF pin is higher than 2.5V the current consumption of the device will be greater than 150nA and will increase based on the voltage."
    - Do I need to include this in my error calculation, since it will divert current from the lower resistor of the voltage divider, lowering the voltage on the FB pin? If so, is there information on how to account for it? Or can I simply use the 150uA through Rin?

    2. From SLVSCV5D, 9.2.1.2.2: "For applications where ATL43x is being used as a comparator, it is best to set the trip point to greater than the positive expected error (that is, +1.0% for the A version). For fast response, setting the trip point to > 10% of the internal Vref should suffice."
    - Does this mean I should just expect a fast response once the signal has reached its threshold +10%? For example, lets say I use an input voltage divider of 2, such that if my signal hits 5V, the FB pin will be at the 2.5V threshold. However, while this technically is my trip point it would result in a slow response, so I should expect fast reaction once the signal hits 5V + 10% = 5.5V?

    Thanks again!
  • Hi SRW,

    1. I would include the IR drop as an error. So the trigger worst case voltage would be approximately 2.5V + Rin*I(Rin). The IR drop will cause the REF pin to see a lower voltage. As you mentioned if a voltage divider is used, then the resistor in parallel will have a lower voltage. To calculate this, I would just make an equation using the trigger point to see what the max values of the resistors you can use based on the error then go off of that. The IR drop is typically very small compared to the 2.5V and thus it is not give a significant error.

    2. This section applies to the overdrive voltage. Figure 31 shows that the transition period and if the trigger point is not sufficiently above the VREF then the response can be slow and inaccurate. The +10% comes from making the trigger be sufficiently large to make the response fast so it is not stuck at unwanted voltages. Keep in mind that this graph is based on Rin of 100K Ohms and thus the transition is slightly exaggerated.

    In your example a voltage divider of 2 with a 5V rail can have a slow response if the resistors are large since there is no room for overdrive. If you use 5.5V with the same divider as above then there is overdrive voltage and the transition will be faster. If you want 5 to be the trigger then I would use a voltage divider ratio so when the Vin is 5V then the voltage seen at the REF pin is +5-10%. The issue is that +10% is very inaccurate, so it is a trade off between the response time/ accuracy/ resistor sizing. Smaller resistors have a smaller IR drop that cause the overdrive voltage requirement to be smaller.

    -Marcoo
  • Marcoo, thanks so much! Perfectly answers my question on all accounts.