SDA waveform distorted when using LTC4311 evaluation board

Hi David,

   What are you using as the "LTC4311 Evaluation board"?  I'm asking because while there isn't a specific dedicated LTC4311 evaluation board there are other boards, including 3rd party boards, which can be used to evaluate the LTC4311.  It may be helpful to know what your system consists of since I believe what you're seeing is more a property of the rest of the system which the LTC4311 is connected to and not the specific behavior of the LTC4311.  The particular eval board you're using may be important if it includes things other than just the LTC4311 part.

   A good representation of the LTC4311 is a switched pull-up current source which is only active during the rising edge of the signal line(s) it is connected to.   Normally the current source is "off" and so the LTC4311 is effectively an open circuit.  When the LTC4311 detects a rising signal edge, it turns on its pull-up current source until the signal level approaches within 400mV of the supply voltage at which time the current source turns off.  The current source remains off until the signal line falls below the input threshold and an another rising edge is detected.

   Thus, the LTC4311 has no impact on the falling edge trajectory.  This trajectory is entirely determined by the interaction of all of the other components connected to the bus.

Eric

ps - Looking at your waveforms, I find it interesting that there is a step in the SDA low voltage.  I've seen this type of waveform characteristic when there are different devices pulling low, especially if there is series resistance present in the bus between the two pull-down devices and the measurement point is located on the bus closer to the pull-up source.  When the remote pull-down device is active, the measured low voltage is slightly elevated due to the voltage drop in the bus series resistance adding the the VOL of the device driving the bus low.  When another device located closer to the measurement point is active, the voltage divider structure changes and the measured voltage is closer to that device's VOL since there isn't the voltage drop from the series bus resistance.   Not knowing what your bus structure is, I don't know if this scenario is possible in your situation but I suspect that something like this may be happening...

Hi Eric, many thanks for your swift reply! I should have mentioned that we are using the Adafruit LTC4311 active terminator/extender board to try out the LTC4311 (see https://www.adafruit.com/product/4756). Assuming it works well, we will add the LTC4311 device to one of the boards on our bus. The Adafruit board has only the LTC4311 part on it, no other devices:

Our product is a VR game treadmill (see short intro video here: https://youtu.be/Zi8rjwtZ_cs?si=TpmBgJxZTWD8CWbK). The I2C bus starts from a PCBA inside the base platform and runs up the arm that supports the user. Each of the three hinges on the arm has either one or two I2C devices:

The product is in production and shipping. The bus works adequately well already. We want to improve rise time and also improve noise immunity in the presence of conducted and RF interference. The LTC4311 seems like a good option for achieving these goals.

Thank you for explaining how the LTC4311 works. The presence of the LTC4311 should indeed have no effect on fall trajectory. We will try to figure out what is happening. We did modify the Adafruit board to add a 3.3V regulator (our bus cable carries 12V). Maybe our rework caused the trouble somehow.

You are correct in noticing the step in the I2C waveform, which is normal in our system for the reason you stated (series resistors on the I2C signals form a voltage divider with the pull-up resistance). Here's another waveform where it's more pronounced:

Hi Eric, thanks again for your help. We found that adding a TCA4307 bus buffer alongside the LTC4311 device fixed the problem with SDA fall trajectory. Do you have any concerns about using TCA4307 together with LTC4311? Here is the TCA4307 datasheet:

PDF

This engineer on the Adafruit forum reported good results from combining LTC4311 with TCA4307:

This would not be my first choice or recommendation for a bus buffer.

I would instead, suggest the LTC4300A-1 which incorporates both the rise time accelerator and a bus buffer into a single device.

Eric

Thanks Eric! We will try the LTC4300A-1 immediately and report the result. Do you offer an LTC4300A-1 evaluation board? If not, we'll make one ourselves.

Hi David,

   Unfortunately we don't offer an evaluation board for this part.  I'd suggest using one of the many MSOP-8 breakout boards that are available for prototyping.  These are much lower cost than any evaluation board we could offer.

    Depending on how/where its being added into the system, the pull-up resistors may already be present elsewhere on the bus segments and so all you'd need to add to the breakout board is the LTC4300A-1, a decoupling cap and the wires to connect it in.  

Eric

No problem, Eric. We will make our own evaluation PCB with connectors that allow it to connect cleanly to our cabled bus. I will post the results to this thread.

Hi Eric, below is the schematic of our LTC4300A-1 test board. Can you please take a quick look and confirm it's correct? Thanks again.

Hi David, 

  The connections to the LTC4300A-1 all look correct/reasonable within the scope of the schematic shown.

Eric