ADV7441A failure mode: power-up/down

Hello:

We are observing a strange failure mode with the ADV7441A, described as follows:

GENERAL CONDITIONS

-          ADV7441A section implemented per ADI reference design suggestions and previous communication with factory (schematic available as needed)

-          analog signal with separate H- and V-sync inputs

-          part running in open air, not "hot" or "cold"

-          when failure occurs, mode persists across power down/up cycles (more below).

DETAILED DISCUSSION

The failure mode may be described as "the ADV7441A is unresponsive to all inputs including reset and produces no coherent outputs."

This failure only appears with analog signal applied and during a power down/up sequence - that is, the unit is operating normally and main power to the system is interrupted for a few seconds.  Once in this failure mode, subsequent power down/up cycles have no effect: the failure mode persists across power cycles.  The amount of time main power is removed from the system appears to be of no consequence: it may be left off for minutes and when re-applied, the failure remains.

The power-supply sequencing shown on pg. 311 of the “Hardware Database Manual” revision J (June 2010) seems to be saying that the 1.8V supplies should all be up near 1.8V before the 3.3V supply is powered-on.  We have verified that we meet this requirement by a large margin (~10ms).

Note that in a related thread on EngineerZone (topic 23376), a significantly different plan is described:

I believe that as long as the 3.3V does not exceed the 1.8V before the 1.8V is stable then everything should be OK.  The same applies in the power down sequence, the 1.8V should be stable until the 3.3V drops to 1.8V.  The reference schematic implements this functionality.

This seems to indicate that, during power-up/down of the 1.8V supplies, 3.3V should be at or less than the value of the 1.8V supply.

We have  ascertained via detailed analysis that the H- and V-sync signals from the external video source are charging the 3.3V rail through the ADI part when system power is off.  In other words, when an external (analog) video signal is applied to the device AND the device's power is OFF, the 3.3V rail remains at approximately 0.8V due to current supplied by the video source via from the H- and V-sync signals thru the ADI part.

We are concerned that this condition makes it impossible to meet the ADV7441A's unique power-down sequencing requirements and that we therefore violate that spec: that is, during power-down of the 1.8V supplies, the 3.3V rail does in fact become greater in amplitude than the 1.8V rail due to this strange “leak” of the H- and V-sync signals across the ADI part.

INITIAL QUESTIONS

Has the "Sync signal current leaking thru the ADV7441A to the 3.3V supply rail" behavior been characterized by the factory?  Is this expected behavior?

As mentioned above, it appears from the datasheet's power sequence section that 3.3V supply should never be above the 1.8V supply in amplitude while the 1.8V supply is in transition (up or down).  How is this achievable given the current "bleed-through" from H- and V-sync into the 3.3V rail?

What is the expected ADV7441A behavior when this condition - 3.3V rail at 0.8V and 1.8V rail at 0V - occurs?

Thanks for your help,

Randy

Parents
  • Dave,

    Here is an update.

    1. We have tried all suggested methods for reset with no change in behavior.

    2. Ditto power sequencing.

    3. We have verified that when an input video signal (analog OR digital) is present with the system power off, the ADV7741A "leaks" current/voltage to its 3.3V rail.  When the system is off, the value of this leaked voltage is ~ 0.8V.  There is sufficient current behind this leak to support a reasonable (1k-ohm) bleed resistor without affecting the voltage level.  ie., we have no way of knowing what sort of path the current takes through the '7441, so we cannot predict what type of load it presents to the incoming signals.  Since we would like to avoid destroying the '7441, we are hesistant to load up a high-current path from 3.3V to ground in an attempt to learn how much current is present.

    4. As shown on the previously-posted schematic, the source for the '7441's 3.3V rail is an ADP1713 regulator.  When the system is powered OFF, we see a reverse-biased leakage voltage on the INPUT to this regulator which measures ~ 0.4V.  That is, we are inputting sync signals which are limited to 3.3V which show up on the '7441's 3.3V rail as 0.8V, then which leak across the '1713 regulator and show up on its INPUT as 0.4V.

    5. The input to the 3.3V regulator is common to the other regulators powering the '7441 (specifically DVDD, AVDD, CVDD, and PVDD).  The failure mode is always accompanied, as mentioned previously, by the DVDD regulator (1.8V) appearing to be "OFF" - that is, it's input voltage (after a dropping resistor) is 5V, so no current thru the regulator.  Droppping this regulator's input V to ground ALWAYS clears the failure mode - that is, both the regulator and the '7441 come up after this event.

    6. Here is where supposition enters.  The model we are currently evaluating goes like this: with a video signal applied to a powered-down '7441, voltage is leaked across the part to its 3.3V rail, appearing as 0.8V.  This voltage is then leaked across the ADP1713 regulator, appearing on its input as 0.4V.  This node is common to all the regulators in this part of the system.  When power is applied to the system, the regulator supplying 1.8V to DVDD has a non-zero input voltage, which for some reason causes it to essentially shut down.  Reducing the regulator's input voltage to 0 and releasing it causes the system to work normally.  Here's where it gets a little screwy.  For whatever reason, when connected to a DIGITAL source and watching the '7441 3.3V rail at power down, we see a short blip down to 0V, then the "leaked" voltage actually climbs above that for an analog signal (probably due to average values at the input of the '7441 being greater for digital than for analog).  We believe this blip to 0V which occurs with digital but NOT with analog inputs is what prevents the system from failing when connected to a digital source.

    It appears that we may have stumbled upon a combination of parts which do not work as expected given the set of criteria we are working on, that is: video signal applied to the inputs while the power is off.

    I believe you indicated in an earlier post that you also saw voltage leaking into the '7441's 3.3V rail.  It would be helpful to know what sort of path this leak is taking, and how much current it can support in this mode.  It would also be helpful to know more about what parts of the '7441 the AVDD and DVDD rails are supplying.  From the names it seems like DVDD might be used to power the core that handles digital inputs and conversely for AVDD.  Is this close to correct?


    We are working with a local distribution FAE who is putting us in contact with the appropriate power group folks at ADI, so that hopefully they can Yea/Nay our suppositions regarding operation/failure of the regulators.

    Thanks as always,
    Randy

Reply
  • Dave,

    Here is an update.

    1. We have tried all suggested methods for reset with no change in behavior.

    2. Ditto power sequencing.

    3. We have verified that when an input video signal (analog OR digital) is present with the system power off, the ADV7741A "leaks" current/voltage to its 3.3V rail.  When the system is off, the value of this leaked voltage is ~ 0.8V.  There is sufficient current behind this leak to support a reasonable (1k-ohm) bleed resistor without affecting the voltage level.  ie., we have no way of knowing what sort of path the current takes through the '7441, so we cannot predict what type of load it presents to the incoming signals.  Since we would like to avoid destroying the '7441, we are hesistant to load up a high-current path from 3.3V to ground in an attempt to learn how much current is present.

    4. As shown on the previously-posted schematic, the source for the '7441's 3.3V rail is an ADP1713 regulator.  When the system is powered OFF, we see a reverse-biased leakage voltage on the INPUT to this regulator which measures ~ 0.4V.  That is, we are inputting sync signals which are limited to 3.3V which show up on the '7441's 3.3V rail as 0.8V, then which leak across the '1713 regulator and show up on its INPUT as 0.4V.

    5. The input to the 3.3V regulator is common to the other regulators powering the '7441 (specifically DVDD, AVDD, CVDD, and PVDD).  The failure mode is always accompanied, as mentioned previously, by the DVDD regulator (1.8V) appearing to be "OFF" - that is, it's input voltage (after a dropping resistor) is 5V, so no current thru the regulator.  Droppping this regulator's input V to ground ALWAYS clears the failure mode - that is, both the regulator and the '7441 come up after this event.

    6. Here is where supposition enters.  The model we are currently evaluating goes like this: with a video signal applied to a powered-down '7441, voltage is leaked across the part to its 3.3V rail, appearing as 0.8V.  This voltage is then leaked across the ADP1713 regulator, appearing on its input as 0.4V.  This node is common to all the regulators in this part of the system.  When power is applied to the system, the regulator supplying 1.8V to DVDD has a non-zero input voltage, which for some reason causes it to essentially shut down.  Reducing the regulator's input voltage to 0 and releasing it causes the system to work normally.  Here's where it gets a little screwy.  For whatever reason, when connected to a DIGITAL source and watching the '7441 3.3V rail at power down, we see a short blip down to 0V, then the "leaked" voltage actually climbs above that for an analog signal (probably due to average values at the input of the '7441 being greater for digital than for analog).  We believe this blip to 0V which occurs with digital but NOT with analog inputs is what prevents the system from failing when connected to a digital source.

    It appears that we may have stumbled upon a combination of parts which do not work as expected given the set of criteria we are working on, that is: video signal applied to the inputs while the power is off.

    I believe you indicated in an earlier post that you also saw voltage leaking into the '7441's 3.3V rail.  It would be helpful to know what sort of path this leak is taking, and how much current it can support in this mode.  It would also be helpful to know more about what parts of the '7441 the AVDD and DVDD rails are supplying.  From the names it seems like DVDD might be used to power the core that handles digital inputs and conversely for AVDD.  Is this close to correct?


    We are working with a local distribution FAE who is putting us in contact with the appropriate power group folks at ADI, so that hopefully they can Yea/Nay our suppositions regarding operation/failure of the regulators.

    Thanks as always,
    Randy

Children
No Data