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.
Hi Sir,
We design TCA4311 enable when power on directly as below diagram we have in our datasheet.
My customer is asking if they unplug the HOST board, what the behavior in JBT side TCA4311? will the SDAIN/SCLIN be always low?
And how the TCA4311 operate when HOST board re-plug?
Could you please help to explain more details on TCA4311 behavior?
Thanks
Anne,
------------------------------------------------------------------------------------------------------------------------------------------------
I HIGHLY suggest moving from TCA4311A to TCA9511A as it is our newest I2C hot insertion buffer. It provides higher Tj ranges, wider Vcc ranges, improved rise time accelerator triggering algorithms, and a lower dynamic offset voltage than TCA4311A.
------------------------------------------------------------------------------------------------------------------------------------------------
"My customer is asking if they unplug the HOST board, what the behavior in JBT side TCA4311?"
I assume the JBT is a daughtercard without onboard power. In this cause, the daughtercard would just unpower itself and everything downstream from the TCA4311A would also float to GND.
"will the SDAIN/SCLIN be always low?"
Without power, the IN-SIDE of TCA4311A would most likely be GND.
"And how the TCA4311 operate when HOST board re-plug?"
If set up properly (staggered connection from GND, VCC, thern SDAIN/SCLIN), the IN-SIDE will power up to 1-V before it make connection to the backplane (HOST in your example) and then once IN and OUT sides of the TCA4311A go above 70% of Vcc the device will connect out side to in side.
"Could you please help to explain more details on TCA4311 behavior?"
http://www.ti.com/lit/pdf/scpa058
An app note discussing hot insertion I2C buffers was written above^^
behavior will be similar to TCA9511A
-Bobby
Hi Bobby,
Thanks for your feedback and we have more questions from our customer need your help to answer it.
Thanks
The power supply design are separate for both HOST board and JBT board which means when HOST board unplug, the JBT board still powered. And the HOST board and JBT board connect by a cable which only include I2C SDA, SCL, GND signal.
Q1. When HOST board unplug, we will see the TCA4311 on JBT board SCL_OUT goes low and other devices on this path can’t be recognized.
And after we re-plug HOST board back, it will no issue. May I know the reason of why the SCL_ out will goes low?
Below I the details TCA4311 schematic on JBT board.
Q2. There is only pull up resistor in the TCA4311 SDA_OUT/SCL_OUT local side, is it be fine for the design?
Q3. Is the red box is SR latch? And what is the stop bit and bus idle function?
Anne Wu said:Hi Bobby,
Thanks for your feedback and we have more questions from our customer need your help to answer it.
Thanks
The power supply design are separate for both HOST board and JBT board which means when HOST board unplug, the JBT board still powered. And the HOST board and JBT board connect by a cable which only include I2C SDA, SCL, GND signal.
Q1. When HOST board unplug, we will see the TCA4311 on JBT board SCL_OUT goes low and other devices on this path can’t be recognized.
And after we re-plug HOST board back, it will no issue. May I know the reason of why the SCL_ out will goes low?
[Bobby] It looks like this is because the pull up resistor is on the Host and there is no pull up resistor on the 'in' side of the JBT board. So the IN side of the device may just float to GND.
Below I the details TCA4311 schematic on JBT board.
Q2. There is only pull up resistor in the TCA4311 SDA_OUT/SCL_OUT local side, is it be fine for the design?
[Bobby] It sounds like you intend to use the device while not connected to the main board which means pull up resistors need to be on the in-side if there is another master on the daughter card who needs to communicate on the I2C bus. Alternatively, the master on the daughter card could also just disable the device by pulling the enable pin low when it is not connected to the main board.
Q3. Is the red box is SR latch? And what is the stop bit and bus idle function?
[Bobby] stop bit and bus idle are part of the I2C spec and the in-side of the 4311A looks for these conditions inorder to connect the two sides together after a hot insertion event occurs. And yes, that should be some kind of digital latch.
Comments in bold above.
-Bobby
Hi Bobby,
Thanks for your quick feedback and one more question from customer.
Thanks
Q4. About the enable pin, should we need to use the enable/disable function to connect to the cable present for Hot-plug detect?
Or we can keep enable pin always high no matter cable is plug or unplug.
Please help to advice it.
Hi Bobby,
We have following questions on Q1 & Q3 after your reply and add one more question on Q4,
Please see below and help to advice it.
Thank you.
Q1. When HOST board unplug, we will see the TCA4311 on JBT board SCL_OUT goes low and other devices on this path can’t be recognized.
And after we re-plug HOST board back, it will no issue. May I know the reason of why the SCL_ out will goes low?
[Bobby] It looks like this is because the pull up resistor is on the Host and there is no pull up resistor on the 'in' side of the JBT board. So the IN side of the device may just float to GND.
[Anne] Customer is asking our application diagram in our datasheet didn't mention about the pull up resistor in JBT side, but only HOST side.
If customer needs to add pull up resistor in both HOST and JBT side, how could they design?
Their current design need to support hot-plug when TCA4311 power always on and enable pin will always high.
Please help to advice it.
Q3. Is the red box is SR latch? And what is the stop bit and bus idle function?
[Bobby] stop bit and bus idle are part of the I2C spec and the in-side of the 4311A looks for these conditions inorder to connect the two sides together after a hot insertion event occurs. And yes, that should be some kind of digital latch.
[Anne] If we connect In/Out inverse, what will happen?
Q4. About the enable pin, should we need to use the enable/disable function to connect to the cable present for Hot-plug detect?
Or we can keep enable pin always high no matter cable is plug or unplug.
"About the enable pin, should we need to use the enable/disable function to connect to the cable present for Hot-plug detect?"
If you only have a pull up resistor on the out side and not the in side and want to use the device when unplugged, then you need to disable the device when unplugged. The backplane should enable the device when hot plugged. (Hotplug-->enable)
"Or we can keep enable pin always high no matter cable is plug or unplug."
Leaving the enable high is likely the reason why you were seeing this problem "When HOST board unplug, we will see the TCA4311 on JBT board SCL_OUT goes low and other devices on this path can’t be recognized."
" If customer needs to add pull up resistor in both HOST and JBT side, how could they design?"
If you do this, then you lose the 1V precharge feature. The point of the pull up resistors on the backplane is to allow the device to power up before it makes connection to the backplane and you do not want to pull the voltage up to Vcc before it makes connection to the backplane.
"Their current design need to support hot-plug when TCA4311 power always on and enable pin will always high."
This device was not designed to support a powered on hot insertion and then unplug while being enabled at all times. If you populate the in side with the pull up resistor on a powered on daughter card you lose the 1V precharge during hot insertion. If you have the enable high while the in side is not HIGH due to no pull up resistors, then you risk the inside floating to low and driving the out side low. The best solution to this is place a pull down resistor on the enable pin which gets enabled after a hot insertion event occurs.
"[Anne] If we connect In/Out inverse, what will happen?"
Both in and out side are then connected together. Based on your design, when you remove the daughter card, then in side can float low since there is not a pull up resistor on that side and can drive the out side low.
-Bobby
Hi Bobby,
How will be the impact about the 1V precharge?
And could you please help to explain more details about why the input is floating will cause the output goes low?
Thanks
user4979899 said:Hi Bobby,
How will be the impact about the 1V precharge?
The 1V precharge on the in-side of the device turns on when the device itself powers up for the first time and only works when there is no pull up resistors on the in-side. It is meant to be used to help the signal on the backplane see 'less' of a glitch when you insert additional capacitance to the live backplane.
And could you please help to explain more details about why the input is floating will cause the output goes low?
If you have an input and it is not referenced/biased then the signal could change due to parasitics in the PCB due to crosstalk or even from the environment due to the electromagnetic force being induced into the copper to generate temporary current and building a charge onto the pin. Typically an input will float to GND because it discharges itself through air/another medium. This means the input can temporarily see a HIGH or LOW. If the input goes low, then the output also goes low. This is assuming you are leaving the in-side without a pull up resistor and the 1V precharge has already been deactivated after inserting and removing the daughter card.
-Bobby
Thanks