Post Go back to editing

LTC4317 Communications to Honeywell pressure transducer IC (LTC4317IDHC#PBF)

To whom it may concern,

I have previously utilized the LTC4317IDHC#PBF without issue but I am having difficulties in communicating with several Honeywell pressure transducer ICs (LTC4317IDHC#PBF).  For this project, I have 8 of the pressure transducers connected with I2C communications going through 4 LTC4317 devices.  The schematic is shown below:


When I communicate directly to the sensors by bypassing the LTC4317 (I lifted up the legs up of Honeywell sensor and attach SDA and SCL from MCU eval board directly to device.  These I2C com lines are also connected to the input connector J1 as well) I get the following waveform with the sensor acknowledging it's device address:

1V/div, 5ms/div

When communications go through the I2C address translator I get the following:

I think the issue is that the pressure sensor is reading that first data pulse (that scope isn't including in the address) and thinking the address is 0x78.

I tried communicating with another I2C device which was an RTC with I2C address 0x68 through the same circuit and didn't have issues.  (attach RTC I2C lines in parallel with a honeywell sensor):

Do you have any suggestions?  We would like to use this part instead of transitioning to an I2C multiplexer type device.

  • Hi Trevor, 

    The captures above don't have voltage levels, but I do see the RC constant changing between an ACK response and the NACK.  

    The LTC4317 is not a bus buffer; It is a pass switch only (with the exception when its actively doing a translation). The point to note is that the electrical characteristics of the bus will change when each LTC4317 is enabled or inserted into the I2C bus. That’s because with each LTC4317 there is a 10kΩ pull-up on each OUT segment of the bus (per the schematics above). The additional 10kΩ resistors will need to be driven by the I2C devices and may inadvertently be making a pretty stiff pull-up.

    My first suggestion would be to check the low level on the bus at the Honeywell device, and see if logic low may be too high for the device to detect the signal. 



  • Jason,

    Thanks so much for your prompt reply.  The Honeywell sensor is the HSCMAND060PA3A3.  It looks like the high and low logic levels are 80% and 20% of the supply voltage respectively(3.3V).  

    Digital Operating Specifications pg.10

    All the scope graphs are 1V/div, 5ms/div.  Do you think it would be worthwhile to increase the value of the pullup resistors?  Do you have any idea what that short first data pulse is in the graph labeled COMS to Honeywell through I2C translator send 0x5F?