I am using the ISO1541-Q1 I2C isolator in an application where side 2 is connected to battery power. In this usage, the isolator in question has the highest quiescent current draw of all my components drawing from battery power, in an application where keeping current draw low is critical.
The I2C messages in general will be used very intermittently and prompted by a main board (side 1) connected to system power. Would there be issues cutting Side 2 power to the ISO1541-Q1, via a control signal on the main board and an opto-SSR when the Isolator will not be in use? Would this cause any issues or lead to a long ramp-up/lead up time when turned back on to send a message?
The original isolator component I had picked was the AduM2251 and I was told that cutting power to side 2 when not in use would not be an issue. I prefer this part for its package and a couple other specs so I just wanted to confirm this would be okay in my application.
Hi kbradley5393,
Welcome to the community!
I talked to an engineer here on this.
Since I2C technology utilizes external pull ups I would be curious if SDL and SDA going high at the same time as VCC if it would be interpretted as a pulse and end up glitching, essentially caused by a brown out situation and undefined behavior.
As far as burning up the chip, I wouldn’t think it would be a problem because there isn’t much of a risk of your violating logic level SDA/SCL lines going high when VCC is low since they’re pull up resistor driven by the power bus. Just be sure to tie those pull ups to the same side of the power rail. You could maybe ignore/lockout any glitches until the control line has come on and given enough time to settle.
I agree with you, AduM2251 does make me feel better about the situation when they indicate the part is hot swappable and is made to filter out glitches.
ISO1541 has a TUVLO, time to recover from Under Voltage Lock Out, defined from 30 to 110us. There would be no guarantees during this period as you have undefined behavior so there would be that to pay attention to.
Appreciate you looking into this for me! I imagine there could be some unexpected behavior on those lines coming live from the pullups at the same time but I could probably build in a wait period during the startup, that’s not really an issue.
Completely overlooked tying the pullups to the same switched power rail so that’s definitely something I will change.
