Most rugged I2C bus range extender (LTC4331 not in stock today)

Hi Frank,

Q1:   The LTC4331 or LT3960 would be my recommended parts for your application, but unfortunately they're not currently available right now.  The LTC4331 will extend the I2C bus easily over your application distance with just a single twisted pair while the LT3960 would need two twisted pairs.  The LTC4331 adds useful features such as address remapping for the remote bus, stuck bus protection, SMBus alert pin support and link health detection.  The LT3960 doesn't include these extra features, but it does allow multi-drop topologies.

   Looking at what's available today, given the shorter distance of 2m and depending on your bus speed, my next suggestion would be to try using one or two LTC4307 Low offset Bus Buffers (controller - LTC4307 - cable - LTC4307 - peripheral ).  The LTC4307 separates the I2C bus into two segments and includes rise time accelerators to help with minimizing the effect of the bus capacitance on the rise time.  There are several other bus buffers which ADI offers which also do these functions and could work, but the LTC4307 has a low voltage offset across the buffer.  Depending on the parts in your I2C bus, the VIL levels can be problematic with bus buffers/isolators and the LTC4307's low offset voltage will reduce the potential of this causing you problems.   

Q2: Not necessarily, and in fact, improperly applied isolation can actually create or worsen EMI issues.

Since you're powering the remote board from the same supply as the I2C controller, unless you also include isolated power for the remote board, adding isolation to the I2C bus alone will not provide any benefits since the power connections will effectively remove the isolation.

In terms of if the remote board should be isolated (power and I2C bus) or not, that is going to depend on if the application use case will be benefit from isolation.  Reasons for including isolation might include safety or robustness/protection from transients or addressing different voltage domains.   If, for example, the remote board and controller board had their own power supplies, then isolation for the I2C bus would be likely beneficial.  However, since you've indicated that the remote board is powered from the controller board, there shouldn't be different voltage domains which would require isolation.  I don't know enough about your application's requirements to comment if there are safety or robustness reasons, but I'd assume not since you didn't mention these items as concerns.

Eric

Hello Eric,

Thanks for your reply. 

I don't grasp the way the advantage of the low offset. The datasheet shows on page 12 that the low offset allows cascading of multiple devices.

I saw on an analog video for adum1250 at

https://www.analog.com/en/education/education-library/videos/2013357472001.html

that there is threshold detection strategy to see if the SDA low is propagating from MCU to peripheral or peripheral to MCU. Does the low offset function of the LTC4331 play a similar role?

Frank

Hi Frank,

   Different I2C devices may have different VOL and VIL voltages and for some combinations of parts, the margin between VOL and VIL may be very small.   In these cases, adding a component such as a buffer between the parts can affect the low voltage in a way so that they won't communicate properly. 

   In the case of the bus buffer, it adds an offset voltage between its input and output.  If you cascade 2 buffers so that there is one on either side of the long cable, then its 2x the offset voltage which is added.   If your application has a wide enough effective range between worst case VOL and VIL, then you might be able to tolerate bus buffers with greater offset voltages.  Since I don't know the details of the parts in your application, I recommended a bus buffer with the lowest offset voltage so as to minimize the chance that this would cause you problems. 

Eric

ps -  Since you mentioned the ADuM1250 video and asked about the LTC4331 and wondered about the low offset's role, here's a quick overview of the 3 types of parts with respect to how they prevent latch up / handle VOL.

   The bus buffers and isolators are truly bi-directional in data flow (a controller can be on either or both sides of the buffer/isolator) and they need to know which side(s) are driving the low so that they can prevent a latch up condition.  Incorporating an offset voltage is one way for doing that. The isolator only sends a logic 1 or 0 across the isolation barrier so it needs to use the different VOL voltage on side 1 to prevent latch-up.  The bus buffer knows the actual voltages on both sides and this lets it prevent latch up.

The LTC4331 is unidirectional in data flow and requires any controller(s) to be on the local side with only peripherals on the remote side.  The local side LTC4331 behaves as an I2C peripheral which receives the data and then clock stretches the controller until it can report back the response from the remote side.  Meanwhile, the remote side acts as an I2C controller which repeats the data from the controller and then reports back the response.  This allows the LTC4331 to know which direction data is flowing and so it can drive a low VOL on both sides without needing to have an offset voltage.