LSF0102: Maximum Cable Distance at 100kHz I2C operation

Hi Clemens,

Thanks a lot for your answer. 

I have LSF series and P82B96 available for my design. As I mentioned, VOL on Sx,Sy of P82B96 limits me because I need the conversion from 1.5V to any of 1.5/1.8./2.5/3.3/5V. 

Do you think that using LSF0102 along with P82B96 would work as in the following figure? 

Reagrds,

Selim 

The LSF is a passive switch, so it will not change the V_OL of the P82B96. With only the LSF010x and the P82B96, you cannot solve this problem.

I know that the VOL of P82B96 will not change. I assumed that LSF0102 would have VIL around Vccx0.3 and the diagram would work when Vcc is 5V( yielding 1.5V VIL). But you are totally right this chip doesn’t provide this level. I checked TXS0102 as an alternative but it seems to have VIL max as 0.15V which is far below. 
is there any other level translator that I can combine with P82B96 to overcome VOL issue? 
Thanks,

Selim 

Thanks a lot again. As you suggested, TCA9617B seems as best solution in our case.

Just as an alternative solution in case we operate at a distance less than 10m, do you think that using only LSF0102 on both master and slave side with extra intermediate buffer boards using LSF0102 through short cables as in the following figure would work in the expected way? 

We ensure that VrefB is always higher than VrefA for all LSF0102s. Firstly 1.5V low speed(around 10kHz) I2C signals from board1 are translated to 5V these signals are send through Cat5 cable to an intermediate board (or several depending on the achievable cable distance) and buffered with LSF0102 by translation from 5V to 5V. These signals are send through Cat5 cables to the board2 and translated to desirable voltage with another LSF0102.  

Regards,

Selim

The LSF is not a buffer; it's a passive switch. Inserting an LSF makes the V_OL slightly worse because of the on-resistance of the switch.

Hey Selim,

If an additional buffer is needed, the TCA617B can be cascaded to further separate the capacitance associated with the 10m cable. Note that this can work as long as A side is connected to B side (and never B-side to B-side of the second device). See section 9.2.3 for more information on this implementation.

Regards,

Jack 

Hi Jack,

Thanks for the additional info.

I think this limitation of not being able to use B-side to B-side comes from VOL incompatibility between VOL and VIL of B side.

When this occurs our application loses its ability to convert voltage level below 2.2V on B-side because B side can translate from 2.2V to 5.5V.

Is there any other product like TCA9617B that provides lower level translations on B side?

Thanks

Most I²C buffer have such a voltage offset.

There are I²C buffers with a different mechanism, the TCA9800/1/2/3, but they do not work at 5 V (and have other restrictions).

Hi Clemens,

I have following solution that enables me to carry my signals at 5V to have more noise immunity and provides me desired level translation at both ends. Do you think it is feasible? When TCA9517/9617s are cascaded, they ever see B sides of each other but they see A side on one end. VOL and VIL levels looks compatible when A sides are seeing each other. 

Regards

Besides, I have following configuration which sends the signals at 3.3V but reduces number of ICs used. Our I2C clock frequency is really low and we can operate even at 1kHz. Can you share your thoughts of which of the shared two diagrams would be optimal?

Regards,

Selim 

Thanks again

1) The reason I used TCA980x on first diagram is to have all desired voltages(1.65V/1.8V/2.5V/3.3V) available both Board1 and Board2 since  the single use of TCA9517 on both sides does not provide 1.65V and 1.8V as an output on B side. This can be overcome with use of TCA9517-Q1 as you suggested. Then the configuration would be following:

2) However, with suggested configuration, I am not able to carry signals at 5V over ethernet cable. TCA980x provides maximum 3.6V on SCL SDA while TCA9517 can go up to 5.5V. To be able to carry signals at 5V, following configuration was prepared taking into account VOL,VIL issues coming from the sides of each device. Since TCA908x does not require any VCCA-VCCB relation, I hope this configuration is foreseen to work. 

Both of the two diagrams above use different ICs (1st diagram)on different ends or use differnt configurations of same ICS (diagram2). In diagram 2, ethernet cable sees A side of TCA9517 from one end and sees B side of diagram from other end. This makes it difficult to standardize it for general purpose.

3) As a final diagram, since we want to have exactly same configuration on both Board1 and Board2 to standardize this configuration in case we use it as a module to plug&play on any board. Then I come up with following configuration that enables me to carry signals at 5V through TCA9517-Q1, and to use levels of 1.65V/1.8V/2.5V and 3.3V through B side of TCA980x on both boards. In this configuration, A sides of TCA9517-Q1s sees each other. Does this create any problem?I checked datasheet and VOL VIL levels look compatible. Do you think that this configuration is feasible? 

If we manage to carry signals at 3.3V, we can exclude TCA9517-Q1 and use TCA980x on both board1 and board 2, with additional buffer board intermediate. Does this makes sense? Are we allowed to connect A sides of TCA980x by cables in case of short  distances?

Many Thanks,

But the configuration in diagram 3 would work as expected right? Also the other configurations on diagram 1 and 2 since there is no B sides facing each other VVCA-VCCB problem.

Based on your doubts, do you think that would it be more reliable to remove TCA9517-Q1 and use only TCA980x as following:

Regards,

Selim 

Thanks a lot for your time and information provided.

To be able to obtain lover voltage than 1.65V on B side, would it possible to use TCA9416 instead of TCA980x? I guess this would decrease the chance of operating over long cables because TCA9416 is only a translator without buffer option? Am I right?

Correct; the TCA9416 is a passive switch with integrated pull-up resistors.