• State Resolved
  • Locked Locked
  • Replies 5 replies
  • Answers 1 answer
  • Subscribers 16 subscribers
  • Views 360 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TS5A3166: Power Off behavior documentation

Trey German8
Expert 1930 points
Part Number: TS5A3166
Other Parts Discussed in Thread: SN74LVC1G3157

Mux Experts,

The power off behavior of this device is causing me considerable consternation. 

  • In the parametric search this device is listed as having power off protection.
  • The datasheet rev history says this feature has been removed from the features section.
  • The 0V leakage currents are still in the datasheet

Someone looking for a device to protect a system that has it's power removed from an external signal might think this device is just the ticket for their application and spend considerable resources both financial and otherwise to design it into their application...someone like me.  Then after hours of debugging they finally find this forum post which clarifies the behavior of this device:

https://e2e.ti.com/support/switches-multiplexers/f/388/t/660359?TS5A3166-Does-it-have-powered-off-isolation-feature-

Adam's last post in this thread is the succinct answer. If power is removed from this device, the internal FET can stay on until VCC = 0.  What ends up happening at least in my application is the signal passes through when VCC is removed and then passes through the ESD diodes of devices downstream which in turn holds up the VCC rail enough that the internal FET never turns off.  I think its clear that at least some TI'ers are aware of this issue with this device, so why isn't this documented better in the datasheet?  It would save engineers using this part valuable time and resources and would lead to a much better customer experience.

Trey

  • 0
    Expert 1930 points

    Follow up question:

    With a signal applied to one of the mux terminals, when power is removed is it possible for current to backfeed through the control or VCC pins on this device?

    I've isolated the downstream signal path and I'm still seeing about 850mV on my VCC rail.

    Trey

  • 0 in reply to Trey German8
    TI__Mastermind 23453 points

    Trey,

    Thanks for bringing this to our attention. The powered off protection here is valid since the isolation does occur when Vcc = 0 but it looks like it has been removed to keep consistent with branding of the term 'powered off protection'. Isolation at Vcc = 0V should still be noted though. In Adam's post it looks like he pulled Vcc to 0V and signal did not pass through the chip. But as Adam mentioned it looks like when Vcc is floating or above 0V the isolation is lost. 

    My understanding of your conditions are that the Vcc rail is powered down and then you see 850mV on the floating rail. This then wouldn't meet the conditions for Vcc = 0V. If you don't mind clarifying whethe Vcc is floating or not though, then I can help further with this and see if further testing is needed.
    Additionally, what are the signal conditions (DC? what voltage level?) are you applying to the signal path that you're worried about with regard to backfeeding? The back feeding to the Vcc shouldn't be possible, we would've seen a much higher level of current in the leakage when Vcc = 0. 

    Thanks,
    Rami

  • 0 in reply to Rami Mooti1
    Expert 1930 points

    Rami,

    Let me expand on the situation I'm facing.

    I'm using the several 3166s to isolate my system from external signals that are always present.  When I power down my system, I see the external signal that goes through the 3166 get clamped down due to it feeding power through some path in my system.  This power is ultimately keeping the VCC rail at a voltage high enough to prevent the 3166 from achieving its V=0 specification.  My VCC rail for these devices is 5V.

    The signal in question is DC and is specified as 0-5V.  When this signal is low (~1V) the system behaves better and switches off as expected.  When the signal is higher in voltage (3-4V) I start seeing the issue where when the device is depowered the external signal is pulled down due to it backfeeding somewhere in my system.

    I have about 50K across the VCC rail and I've tried putting lower value resistors across the rail to encourage the VCC rail to settle to 0V as quickly as possible.  This works, but isn't a clean solution.  I'd like to find the culprit so I can add an ideal diode in the next board revision to prevent the backfeed.

    There are 3x 3166s in my system.  The outputs of two of them go to voltage dividers connected to a MCU ADC on a 3.3V rail.  I don't see any power on this 3.3V rail so I don't think the issue is with either of these two 3166s. The last 3166 is connected to a SN74LVC1G3157 and then a DAC. 

    In order to isolate the problem, I pulled the 3157 so there was no where for the signal to go.  This improved the situation.  Before I pulled the 3157 I was seeing ~1.3V on the 5V rail.  Now with it gone I see the 850mV previously mentioned.  That's why I asked if there was potentially a way the 3166 could backfeed.

    Happy to share a schematic of what's going on, but can't do it on a public forum.  Really appreciate any advice or review you or other TI'ers could provide.

    Trey

  • 0 in reply to Trey German8
    TI__Mastermind 23453 points

    Trey,

    So you have 2x 3166 powered by 3.3V rails that work as intended just fine but 1 powered by a 5V rail that seems to give you problems.
    More specifically, when the 3157 was down stream you saw 1.3V on the powered down 5V rail and when the 3157 was removed 0.85V on the supply rail.

    Are the 3.3V devices going under that same conditions as the 5V one, and the only difference is the voltage being supplied? 
    And have you removed the 3166, apply the same 3-4V+ signal and make sure that the 5V rail and signal trace are isolated properly?

    In the meantime, I will take a look at the schematic off of the forum and see if anything stands out. I'll reach out via an E2E message.

    Thanks,
    Rami 

  • 0 in reply to Rami Mooti1
    Expert 1930 points

    Rami,

    I received the parts and was able to do some bench testing.  It looks to me like the 3157 was indeed forming a backfeed path for power which was keeping the 3166 in a weird state.  I added a diode and bleed resistor to the 3157 on my test setup and was able to almost get it to behave.  It looks to me like when the 3157 loses power, it tries to pull the signal down.

    I have some more parts on the way and I'm going to try rolling my own isolation solution to see if I can get this to behave nicely.

    Trey