ISO1540: SDA1 and SCL1 Low level voltage is too High

Hi Power Guy,

Thanks for using our forums and sorry to hear about the issues you are facing.

I looked at the TMS320 datasheet and unfortunately, it does have TTL thresholds for its inputs, which means the VIL spec is 0.8V max - which explains why the 800mV logic low output (Vol )from the 1540 may not be getting recognized properly as a logic low. Incidentally, this was done on the 1540 to avoid a race condition between the two bidirectional channels - and hence a diode is added internally in the series path as shown in the block diagrams.

W.r.t the level shifter, it may still be able to help here. Have to take a closer look. In general, because the input and output supply domains of a level shifter are relatively independent from one another, it can help address signal swing at both ends. The problem statement thus reduces to having a shifter that will accept the incoming swing [say from 0.8V to Vcc] but output as much of a near rail-to-rail signal [0 to supply] on the output such that interfacing it to a micro and its associated thresholds no longer becomes a problem.

Are you running the MCU and the 1540 @ 3.3V supply? Any other system considerations here? A (simplified) block diagram may help to rule out any other issues with whatever solutions/proposals we may come up with.

Apologies for the rather short/quick reply - we will follow-up when the team is back in the office tomorrow.

Thanks,
abhi

Hi Abhi,

The the MCU and the 1540 both use 3.3V supplies. 

We need to provide isolation for the I2C signal. Our unit will become part of a system that will be sharing the I2C lines too so we can not load the bus down in any way. 

Currently what we have is :

MCU - I2C buffer - ISO1540 (primary of ISO goes to I2C Buffer) 

Communication has been verified before going to the ISO1540, thus I'm 99% sure is the low logic on side 1 of the ISO1540. 
At this point I will have to rework a level shifter in. Can you please recommend at part you think will do the job? 

thank you

Hi,

Thank you for sharing additional information.

The recommended operating conditions given in Section 5.4 of TMS320F28069 do specify the max VIL to be 0.8V for 3.3V supply. This does explain why you are facing communication issue. But the Table 6-56 of the same datasheet specs that VIL max for I2C lines is 0.3*VDDIO which should be ~1V for 3.3V and ~0.9V for 3V supply. So I am not sure if the issue about ISO1540 VOL1 being 0.8V.

To quickly check this, an I2C buffer / level translator between MCU & I2C isolator should be used. I see that you already use a buffer between MCU and I2C isolator, could you please share the part number?

If you can use TCA9800 and connect Side B to MCU (without any pull-ups) and Side A to ISO1540 Side 1 (with necessary pull-ups) then the VOL1 of 0.8V shouldn't be a problem. As specified in datasheet of TCA9800 and also shown in Figure 18 of datasheet, Side B shouldn't have any pull-ups and I recommend connecting Side B of TCA9800 to MCU without any external pull-ups. TCA9800 includes an internal current source that acts as pull-up resistor.

Please do test this device and let us know if it resolves the issue. Thanks.


Regards,
Koteshwar Rao

So, I had some success. 

Connected side 1 of ISO1540 to Micro and communication goes thru. 

I had an LTC4301 buffer in between the ISO1540 and the micro, all using 3.3V. For some reason communication stops if I use that buffer. 

I do not have a TCA9800 readily available but I do have a TCA9406, it should work just like the TCA9800 right ?

I still need to be able to disable communication going to the ISO1540, that was the main use of the buffer. 

So, now that I'm able to talk btw the MCU and ISO1540. I'd like to shift the logic low level of the ISO1540 side one slightly lower.

I'm very close to the VIL limit on the MCU and I fear over a wide temp range I will start seeing communication errors.

Is there a way to shift the Logic Low down on the ISO1540 by around 200mV ?

thank you