PCA9306: Can run 1MHz between the master and device when disabling PCA9306?

The blue boxes are not devices, but bus segments with possibly more than one device. The left box contains the master and at least another 1 MHz device. The right box contains one or more 400 kHz devices.

=> Okay, understood but still have the further question. Will the left box communicate to the right box in 1MHz operation when disabling PCA9306?

When PCA9306 is disable, can the device side still work in 3.3V based on the understanding from the video?

I am attaching my circuit snapshot in below for your reference.

Let me try to revise my previous comment to simplify the question.

In my design, both of master and device sides can operate in 1MHz.

When I disable PCA9306 (EN pin tied to GND), will both side operate in 1MHz?

In this case, is the circuit operated as the expected? meaning 1.8V in the master side and 3.3V in the device side.

THanks,

Yihua

In that application example, the purpose of the PCA9306 is to disconnected the two bus segments, so that the 400 kHz devices does not see 1 MHz signals.

This is not what you want to do in your application. To get level shifting, you do not need to disable the PCA9306. Why do you want to disable it?

Okay, now I understand the intention of that application example.

In my application, we want both of master and device sides can operate in 1MHz thru PCA9306.

sounds to me is PCA9306 is not expected for this purpose, right?

Thanks,

Yihua

Clemens and Danny,

Thanks for sharing more information.

Since you confirm PCA9306 is the capable device for 1MHz operation and looks like it's possible after recalculating the knee frequency (4MHz based on 120ns rising time (fast mode plus)), but we received the different confirmation from the local TI FAE.

Anyway we are receiving an instability issue from our customer when they run the long run testing with 1MHz and the issue is intermittent (1 out of 2xx cycles or 1 out of 3xx cycles). The max bus capacitance in our design is around 350 pF and current PU resistors in the left (Device) and right sides (Master) of PCA9306 are 1K ohm. Even we tried the stronger PU resistor (smaller resistance), the issue is still observed but low fail rate.

Hence, I would like to ask you what the recommended pull up resistor is for our current configuration?

Thanks,

Yihua

350 pF is quite a lot, and requires strong pull-up resistors. The signal quality would be visible on an oscilloscope trace.

It might be a better idea to use a level shifter with buffers, e.g., the TCA9617B. But it is possible that you need to reduce the frequency anyway; see Maximum Clock Frequency of I2C Bus Using Repeaters.

Yes, 350 pF is quite a lot but still meet the spec requirement. Since PCA9306 is capable for 1MHz operation, what's the recommendation for the pull up resistance based on this configuration?

BTW, Replacing PCA9306 by TCA9617B is not possible for now because the board is already TO. We will need to find the pin/package compatible part for the replacement. The local FAE recommends another part is LSF0102. What's your suggestion for this recommendation?

Begin with 470 Ω resistors on both sides (this results in a cut-off frequency of about 2 MHz, but digital signals need overtones). Look at an oscilloscope trace, and adjust as necessary.

The PCA9306 and LSF0102 are, for practical purposes, identical.

If you have to keep the board, your only way is to adjust the pull-up resistors or the frequency.

Yihua,

It is likely that your FAE was also giving you correct information. Please note that the PCA9306 supports fast mode (400 kHz) as noted in the datasheet. I am assuming that this is the information you were provided by the FAE.

In addition to fast mode, also noted in the datasheet section 9.2.2.3, "The maximum frequency of the PCA9306 device depends on the application. The device can operate at speeds of >100 MHz given the correct conditions. The maximum frequency is dependent upon the loading of the application."

You will notice as shown in Figure 3 that the device reaches -3 dB around 250 MHz.

The high capacitance of the bus that you mention provides a source of concern, but it is not disqualifying for the device.

Best,

Danny

Clemens,

Thanks for the recommendation.

May I know why both sides' pull up resistor need to be consider to change? So far we only tried to change one side (device side after LS)'s pull up resistor. This is very interesting recommendation and would like to know the reason. Meanwhile, may I think PCA9306 and LSF0102 are the similar parts for my design concept?

=====================================================================

Daniel,

The original design engineer doesn't support my project. So I didn't now whether he has been informed PCA9306 support the fast mode (400KHz). Somehow the datasheet brings the confusion to my team now. It claims it support the standard/fast mode in Feature section however sounds to me it's also capable for higher I2C operation after receiving your comment. Why isn't the higher speed support been listed in Feature section too?

Yes, I understand the bus capacitance in our design is a lot as Clemens also mentions but it still meet I2C specification, right? Or does somewhere in the datasheet claims PCA9306 can't support the bus capacitance over 370pF?

Anyway, thanks for both of your support here. Your comments are very value and It will be very appreciated to have your support continuously.

PCA9306 and LSF0102 are identical.

They are passive switches; when a line's voltage is low and the switch is closed, the two sides are connected, and the pull-up resistors are in parallel.

Thank you, Clemens.

We will try to change the pull up resistance on both of sides as the current suggestion.