TXS0102-Q1: TXS0102-Q1: one-shot issue(Consultation on trigger conditions of one shot)

Hi Jacky，

      Thank you very much for your prompt reply and explanation. We understand the one-shot threshold is a region and We can't change to LSF0102-Q1, because  their footprint is not PIN to PIN  compatible. we need your help to explain the below question for more details. We are confusing about the definition of rising edge. Please help to explain the below questions. 

Q3: As the below pass waveform shows, we can see the rise edge from 548.82mv to 812.14mv, rise time is about 10ns. Why doesn't this rising edge trigger one-shot?

pass waveform:

Q4: How to determine if an rising edge is valid? What is the detection capability of TXS0102-Q1 for rising edges?

Hi Jack,

    As the below pass waveform shows, we can see the rise edge from 37.5mv to 762mv. Why doesn't this rising edge trigger one-shot?

Hi Jack & TI team，

    As the below waveform show  pass VS fail, we can see the rise edge from 37.5mv to 762mv. Why doesn't this rising edge trigger one-shot?

Hi Hans,

I am still taking a look into this question- and will provide response early next week after 9/1 (US National holiday). Thank you for your patience here.

Regards,

Jack

Hi Jack,

     Is that any response today?

Is that any Schmitt trigger circuit or hysteresis in the one-shot circuit? or is that any deglitch feature in the one-shot circuit? How about their feature? 

HI Hans,

Thank you for your patience here, the one-shot does not have any Schmitt-trigger/ hysteresis built in, the one shots are activated once the input signal reaches a certain threshold region.

Are the I2C signals being driven over long distances i.e traces/ connector lengths? If so the undershoot/ overshoot can potentially trip the one-shots, and would further recommend using dampening resistors at the outputs of the TXS to help reduce ringing if so. 

As for the waveform provided, can you help to clarify which signal is input and output for the below:

Please note that the host output low voltage of 540mV is a pretty high VIL voltage into our device. As the TXS is a switch based translator, the VOL will be heightened due voltage drop across RDSon, and therefore 540mV would be the absolute minimum VOL voltage. If the host sits at the 3.3V side, then this means that the potential 0.54V being output into 1V8 side is near the max I2C VIL threshold of 30%. Is it possible to lower the VIL into our device? 

Regards,

Jack

Hi  Jack，

        Thanks for your details response. The host(sever) is using PCA9617A and its output  level is 3.3V driving VOL at about 540mV to TXS 3.3V side, so the TXS's input is 3.3V side. And we can't lower the VOL (540mv) from sever, because the sever from customer has been produced.

we just need to explain the one-shot logic for more details. such as the above fail waveform: Why is it that sometimes one-shot is triggered and sometimes it isn't? such the rise edge from 548.82mv to 812.14mv with rising time 10ns doesn't trigger and the rise edge from 37.56mv to 762.50mv with rising time 1.27us triggers one-shot.

If there is not any Schmitt-trigger/Hysteresis built in, is that any deglitch feature bulit in? If the rising edge is very short such as from 540mv to 812mv with rising time about 10ns, the TXS can filter the abnormal rising edge?

Hi Jack,

        We understand that reducing the level of the low voltage is crucial, but the current design is unable to achieve this. All we need is to be able to answer the customers' questions as below. So we need your assistance to provide us with a better and more reasonable explanation. Thanks!

we just need to explain the one-shot logic for more details. such as the above fail waveform: Why is it that sometimes one-shot is triggered and sometimes it isn't? such as the rise edge from 548.82mv to 812.14mv with rising time 10ns doesn't trigger and the rise edge from 37.56mv to 762.50mv with rising time 1.27us triggers one-shot.

Hi Jack,

         So, you mean that 548.82mv did not fall within the VOL range, so the transition from 548.82mv to 812.14mv would not trigger the one-shot function, right?

         But, we also observed some waveforms rising from 775mV to 1.8V. The rising edge of the waveforms indicated an acceleration, suggesting that the one-shot has been activated. I mean that the triggering condition for the one-shot function does not seem to be activated only through the use of VOL. It seems that there are other triggering conditions. How can this phenomenon be explained?

 This below waveform is from the Failure Injection Verification process. The rising edge is from 775mv to 1.8V, the rising indicates an acceleration.

Hi Jack，

         Thanks for your detailed and patient explanation. I have understood that "one-shot threshold" refers to the concept of a region.

I also conducted tests on one-shot threshold using four txs0102-q1 samples. From the waveform, it can be seen that the threshold values were all close to 700 to 800 mV, no lower threshold was observed.

Could you please help confirm whether the one-shot threshold is a bit higher than 30% and closer to 40%? For example, the target is greater than 700mv or greater than 690mv, etc.

We have solved this SI issue and can continue to use txs0102-q1. Now it is necessary to clarify the issue of this threshold region. We need to confirm that it is the answer that is closer to 40% or higher.

Below is the waveform that we have measured.

sample#1, such as: 800mV

sample#2, such as: 800mV

sample#3, such as 715.5mv

sample#4, such as 725mv

sample#2, 0~693mV, The one-shot has not been activated.

sample#3, 0~681mV,The one-shot has not been activated.