I'm trying to develop a board that includes I2C and SPI devices that are unpowered for the majority of the time to save power. I found that for these particular devices (Honeywell HSC pressure sensors) you definitely can't have the I2C ones unpowered because you end up sort-of powering them through the I2C lines and getting a weird voltage of around 1V.

However, I have unpowered SPI microsd cards on the same bus as some other things (a RTC etc.) and it seems to work fine.

Both devices say that the peak voltage on any line is something like Vdd+0.3V, which if Vdd is 0V is 0.3V (e.g. for microsd). So my questions are:

  1. Do they really mean Vdd+0.3V even if Vdd is 0V?
  2. Is there any nice way to do this? I've found an I2C buffer chip (PCA9515) but it draws 3 mA all the time which isn't ideal - I'd have to turn that chip off too, which I'm not sure it will like. For SPI I found this 74LCX125 chip, which seems to sip power as long as the inputs are not floating, so very weak pull-downs might help there. It all seems a bit complicated though.

Any suggestions?

  • 2
    \$\begingroup\$ There are hundreds of buffer ic. There should be some with nA standby current. \$\endgroup\$
    – Passerby
    Commented Apr 10, 2014 at 22:55

2 Answers 2


1. Do they really mean Vdd+0.3V even if Vdd is 0V?

Yes. Second guessing datasheets is a really bad idea.

The problem is that most chips will have protection diodes from inputs to the positive supply. These are to keep the device from getting hurt from small static discharges during handling. When the device is powered down, these diodes will either clamp the line externally to one diode drop above ground, or partially power up the device so that it's power voltage is one diode drop below the signal voltage.

Unless you find a IIC device specifically designed to be powered down while not in use, there is no good way around this. IIC is bi-directional, so simply buffering the lines into the device won't work.

The better solution is to find devices that draw very little quiescient current. A IIC device that is not talking on the IIC shouldn't need to draw much current. Some devices may have a deliberate powerdown mode for that reason, which probably keeps the IIC front end powered up but with minimal current draw.


I found a nice solution, apparently there are chips called "signal switches" that I can place on the bus. They have bidirectional "zero"-delay switches. They do draw significant power (1 mA apparently) but have 5V tolerant high impedance inputs in their off state so I can just turn them on or off as desired together with the sensors.


http://www.ti.com/lit/ds/symlink/sn74cb3q16244.pdf http://www.idt.com/document/qs3vh16244-datasheet

I haven't tried this solution yet.

  • 1
    \$\begingroup\$ I used this solution and it worked well. \$\endgroup\$
    – Timmmm
    Commented Jun 4, 2015 at 7:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.