Can I apply voltage to the ADAU1701 SCL/SDA pins when device is powered off?

I believe the I2C specification may implicitly answer my question...the spec says this for Fast Mode devices which I believe applies to the ADAU1701: “If the power supply to a Fast-mode device is switched off, the SDA and SCL I/O pins must be floating so that they do not obstruct the bus lines.”

I request one of the ADI experts to confirm this for the ADAU1701.

Thanks!

-Ram

Hello,

After a quick test on the evaluation board I can see that the SDA / SCL pins appear to go high impedance when power to the device is disconnected.

I would like to add that the absolute maximum ratings section of the datasheet describes that the voltage on any pin should not exceed IOVDD + 0.3 V at any time. So, it is not recommended that you operate the device on the I2C bus when IOVDD is disconnected.

I imagine that if the SDA / SCL lines are high when IOVDD is disconnected, the ESD protection diodes inside the ADAU1701's SDA and SCL pins might be activated, which could effectively shorten the lifetime of the device.

A simple solution to this problem, if you anticipate in your system that the bus will still be active but IOVDD may be powered down, would be to use an I2C bus repeater IC with its enable input connected to the IOVDD line. This way, the repeater will act to isolate the ADAU1701 if IOVDD is powered down when it is still on the bus.

Thank you Brett, for making the quick test and for the suggestions on the bus repeater setup! In my case, I'm actually working on modifying an existing design so it may not be possible to introduce a bus repeater. What to me is important now is to get a definitive answer to whether the ADAU1701 conforms to the I2C spec for fast-mode devices. If that is the case then I suppose the repeater would be superfluous, and it's always good to have lean and mean designs :-). I'll raise a formal support ticket with ADI on this issue and rest the discussion on the forum for now.

Thanks again.

Best,

-Ram

Hi Dr.Ram,

I was able to confirm with the designer today that since the I2C pins on the ADAU1701 are shared with the SPI bus functions, they require overvoltage protection diodes internally. If the signal on the SDA or SCL pin is held high but the IOVDD pin on the ADAU1701 is not powered up, the diodes will indeed turn on and a voltage drop will occur, interfering with the I2C bus. This will only occur when IOVDD is low; the state of AVDD and DVDD doesn't matter. AVDD and DVDD could both be powered down, but if IOVDD is high, then the device will not interfere with the I2C bus.

So, my suggestions would be to do one of the following:

-Isolate the ADAU1701 from the I2C bus with a bus repeater.

-Connect IOVDD to the IO supply for the rest of the system, ensuring that it will stay held high even if the ADAU1701's AVDD and DVDD pins are not powered.

Your support ticket on the SigmaDSP mailing address was assigned to me, and I will close it now. Sorry that it took me a few days to respond.

Dear Brett,

Thank you so much for checking with the designer on this important question, and for the two options you have confirmed based on the answer. For existing designs therefore it appears that the second option would work better because one could create a workaround by connecting just the IOVDD to, say, the EEPROM's VDD so that one could still program the EEPROM in-circuit with the DSP powered "off". It would be interesting to determine what the power consumption of the DSP chip would be with just IOVDD powered and I suppose that would be much less than when it is fully powered up - which would essentially serve the purpose of powering down the DSP.

I suppose one could still nitpick on that I2C spec statement and conclude that the ADAU1701 does not conform 100% to the I2C spec for fast-mode devices, though!

Thanks again.

Best - Ram

FWIW, this weekend I decided to try programming a 24LC32 EEPROM with its SCL and SDA pins connected to the ADAU1701 SCL and SDA pins with no power being applied to the DSP (not even to IOVDD). I connected pins 1, 37, 42, 12, 25, 33, 30, 38, 7 of the DSP to GND (as indicated in the self-boot schematic) of the EEPROM. Used an el-cheapo RS232 programmer as, e.g., at http://www.scienceprog.com/rs232-emulates-i2c-for-24cxx-memory-programming/ modified to work with 3V3 instead of 5V so as to not damage the DSP chip. Was able to program and read the EEPROM successfully, corroborating Brett's original observation that "the SDA / SCL pins appear to go high impedance when power to the device is disconnected." May possibly make it more reliable when IOVDD is powered up, but I thought it worthwhile to share this little nugget of info.

Best,

-Ram

Hi Brett,

Jumping into an old thread, because I'm debugging something similar. I can understand that having IOVDD powered down while SDA/SCL are high could lead to bus problems, but could it actually physically damage the device? I have a couple of dead ADAU1701's here that may have been subjected to this condition - I really see nothing else it could be as I try to debug the issue. But I don't have strong evidence (yet) that this is what caused the failure. I'm hoping I can either rule it out or blame the cause on this bad IOVDD state. Can you clarify?

(Failure mode: 1701 is disconnected from the microcontroller, but left connected to the EEPROM. EEPROM seems fine to read manually, but any attempt at "link/compile/download" with the USBi results in SCL stuck low.)

Thanks a bunch.

Hi,

I would like to clarify what you are seeing:

With the ADAU1701 powered up, you are seeing a 'stuck low' condition on the SCL line. If you swap the ADAU1701 for a new one, the problem is solved. What happens to the SCL line if you disconnect the USBi?

I would not expect that applying a reasonable voltage to the powered down part would damage the silicon.

Regards,
Coleman

Thanks, Coleman.

When powered up the clock is high, but any attempt to communicate results in SCL stuck low, meaning the USBi is waiting for a response. Disconnecting the USBi makes it go back to high. As far as I can tell this is consistent with the 1701 failing. (I haven't been able to communicate with it subsequently, even with third-party raw I2C tools, or my microcontroller).

I wouldn't expect that applying a reasonable voltage to the powered down part would damage anything either, but the evidence I have indicates that it does. However, I can't be certain at this point, which is why I'm asking about it in more detail here.

Hi Joe,

I am correcting the answer that I gave on March 7th:

Applying voltage to the I/O pins of any device that is powered down can cause damage. The I/O pins have ESD protection in the form of diodes that are intended to isolate the internal circuitry from any external voltage spikes - under normal operating conditions. If the IC does not have power applied to the IOVDD pins, and you apply voltage to the I/O pins - in the form of I2C pullups or even in the form of clocks and data - the ESD diodes will turn on and current will flow into the power section of the IC, effectively powering up the device...sort of. The protection diodes are intended for intermittent operation only; even though they can sink a voltage spike of a few kV, that spike cannot be longer than a millisecond. Long term, the diodes can suffer damage and cause very odd behavior from the damaged port.

If you are keeping the part powered down in order to save power between uses, I would recommend that you leave the IOVDD supply active and shut down AVDD and DVDD only. IOVDD powers the I/O pads, keeping the ESD diodes referenced to the voltage rail and protecting the internal device.

Best regards,

Coleman

Thank you very much for the thorough answer. I expect that this was my issue, and I'm glad it has an easy resolution.