PCA9306: PCA9306 series application

Hello,

The usage you described should be OK.   Please make sure the schematic is correct in tying EN and VREF2 together then pulling them to 5 V via a 200-kOhm resistance.

Assuming the schematic is correct, we should take a look at the I2C waveforms on each segment.  Since the PCA9306 does not re-drive the signals, the loading (pull-up resistances and parasitic capacitances) can act in parallel.  If you find that the "low" level voltage is too high on the waveforms, you could try increasing some of the pull-up resistance values.  If you find that the rise times are too slow (e.g., due to higher parasitic capacitive loading), then you could try reducing some of the pull-up resistance values.

Regards,
Max

Hello MaxPCA9306 application_20200727.pptx

We changed the designed structure,please help to review and there is  still some issue.

The PCA9306 and LSF0102 had be pulled 200K ohm at EN pin.

We tried to adjust pull high resistor,but we can't find the optimized  resistor value.

Could you check the waveform and give some suggestion?

Regards,

Elvis

Hey Elvis,

Max is currently out, I am one of the engineers who provides support for the I2C portfolio (which includes the PCA9306).

Comments based on your powerpoint:

• PCA9306 (and LSF0102) are pass through element translators, this means they do not redrive the signal
  • Slide 2 shows the 5V side does not pull all the way up to 5V but rather leakage current from side1 pushes the device up
  • This is because the pass FET is trying to regulate itself back into the cutoff region
• The sharp rising edges you see (like what you marked as 1.840V in slide6) is the ACK hand off between master and slave
  
  • This shouldn't cause problems since the clock line should be low during this time
• Adding/changing the pull up resistors will change the VoL of the devices on the bus since left side and right side are not redriven but are pass through.
• There is some crosstalk in some of your signals but from what I can see, it shouldn't be an issue

Can you tell us what device the light sensor is? (A datasheet would help). I'm curious what the ViL is for the device. If this is confidential, you can send it to duynguyen@ti.com instead of posting it online.

Since you are saying the light sensor is "losing data," and the slides state the I2C measurement is made from the INA220 lines.Can you also tell me what data you are expecting to see and what you actually see? From slide 6 it shows address/byte 0x45 followed by 0xB9 and then 0x8A. Is this not expected?

Can we look at the light sensor's lines (measure I2C from the light sensor's SDA/SCL) and also increase your point of measurement from 10k to something like 100k.

Thanks,

-Bobby

Hi teams,

Please refer to the ALS spec and schematic.

Our experiment is showing as the attached.Please help to check how to adjust I2C waveform

ALS schematic.docxISL29023IROZ-T7+SPEC.pdfPCA9306 application_20200728.pptx

Thanks

Elvis

Hey Elvis,

I believe the occurrence where you 'sometimes lose ALS data' is due to the ViL spec of the ALS sensor: (it does not meet the I2C spec of 30% of Vcc = ViL)

The VoL seen by the device may be close to this limit and the ALS may not recognize its slave address in some of the transactions.

When you use a 6.8k pull up resistor, you shift up the VoL and sometimes get marginal success/failures. When your pull up resistor is 10k+, your VoL is likely low enough that you meet the ViL threshold for the ALS in these cases. (Recall that in my last post I mentioned PCA9306 is not a redriver, this means you have 2.2k parallel to 10k/100k parallel to 6.8k)

Slide 6 and 7 do not have the I2C transaction. My initial guess here is the VoL from the INA devices may be too large for the master to accept as a valid low.

One work around you could try is to replace the PCA9306 on the power board with a TCA9517 (B side face the INA devices). This would separate the 5V pull up resistor from the 3.3V ones and help provide a lower VoL. When the INA ACKs the VoL on the 3.3V side would be lower and may help to have the master recognize the low (assuming my guess about slide 6/7 is correct).

-Bobby

-Bobby

Hi Bobby,

In PCA9306 case,there is not pull-up resistance at INA220 I2C bus.

Do you think what the design is Ok?

from attached,do you think the I2C waveform can be met I2C spec?

PCA9306 application_20200729.pptx

And what reason is it caused the I2c waveform is not smooth?

Because we worried the I2C of INA side is only approach 3.3V.

Although It's enough level(VIH min=0.7*Vs).

Thanks

Elvis