• State Resolved
  • Locked Locked
  • Replies 4 replies
  • Answers 1 answer
  • Subscribers 62 subscribers
  • Views 1361 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.

TPS22916: Unexpected voltage drop and leakage current trying to design a power source switching through load switches

Adrian Cobos
Prodigy 90 points
Part Number: TPS22916
Other Parts Discussed in Thread: , TPS2121, TPS22917

Hello, I'm going to give some context first so you might understand our situation more easily

We want to design 2 different power management system with some diverse features as following:

Type A Expected features

  • Automatic Switching - system must change to back-up battery automatically when main supply fails. There is no decision level, so main supply is 5v when active, or none if fails. Power supply must return to major priority power supply if it returns.
  • Switchover time - As less as possible to prevent MCU reset, so seamless system is mandatory.
  • Output voltage - As less as possible to prevent MCU reset. Voltage must not fall below 3v any case.
  • Current Consumption – Systems works in two different stages, low current sleep mode and high-performance mode; because of that, we need good efficiency and very low Iq operating power management to not increase total sleep consumption from both power supply sources and achieve a maximum of 200 mA.

Type B. Expected features

  • Automatic or manual Switching - Both options are valid
  • Switchover time – A seamless system might be perfect, but breaks could be acceptable instead Instrumentation is able shutdown and restart with new power supply.
  • Output voltage - As less as possible to prevent instrumentation bad function. Voltage must not fall over 4.8Vdc any case.
  • Current Consumption – Power managements must achieve a maximum of 2A continuous consumptions.

In order to design a unique power management design for achieve both types (type A and B) features, we test load switches performance using TPS22916EVM following the next design (App report SLVA811A). We connect enable input to Vmain power supply with the following expected works:

When Vin1 is present: Enable is ‘1’ so Vout connect to Vin1

         When Vin1 fails: Enable is ‘0’ so Vout connect to Vin2.

We use TPS22916CL on the minor priority power supply load switch, and we test all the others on the major priority power supply. We use both Vin2 voltages (3.3Vdc back-up battery and 5Vdc regulated power supply). We make test without load, with 10ohm load and 1Mohm load.

We have same problems regardless circuit assembly:

Once Vin1 is connected, we connect Vin2, without output load added.

We have 600nA current consumption from Vin2 -> GOOD

We have ≈33mA current consumption from Vin1 -> BAD. Unexpected

If we add an output load of 10 ohm or 1Mohm (both current type needs) we have no change on output leakage show before (these values are added to normal consumption due to load).

What could be the problem?

Once Vin1 and Vin2 are connected and Vout=Vin1. We disconnect Vin1 (Vin1 fails), then Vout = 0.5v (so it doesn’t switch to Vin2), instead of that, we have a 0.5 Vdc present on Enable pin,

What could be the source of this voltage?

We have test using a 1Mohm pull-down resistor for enable pin, but performance is the same.

In order to design a circuit for type B requirements, we test load switches performance using TPS22916EVM following the next design (App report SLVA811A). We connect enable input to microcontroller digital output following expected works:

Enable is ‘1’ (3.3Vdc ) so Vout connect to Vin1

         Enable is ‘0’ (3.3Vdc ) so Vout connect to Vin2.


We have the same current leakage problems show before, in addition with an unexpected voltage drop when we use a 5ohm load (≈1A current output).

Enable is ‘1’ so Vout connect to Vin1, and Vout falls to 4.6 Vdc. BAD. Unexpected

Enable is ‘0’ so Vout connect to Vin2, and Vout falls to 4.9 Vdc. GOOD.

What could be causing this voltage drop?

Thank you in advance!

  • 0
    TI__Genius 10855 points

    Hi Departamento, 

    Thanks for reaching out on E2E! I see you also have a similar post regarding the TPS22115A.

    There are two potential issues that I see with this configuration. First, let me make sure my understanding of the configuration is correct. You have 5V on VIN1, and 3.3V on VIN2, and the enable pins for both enables are connected together AND connected to VIN1. When VIN1 is present, the enable signal is "1" so VIN1 is outputting to the downstream load. However, since you're using the TPS22916CL "active low" version, wouldn't this also cause VIN2 to be on? Is my understanding correct? 

    That's the first potential problem I see with this configuration. Second, if VIN1 is removed from the system (the input is left floating), the top device will not be powered. Because VIN1 is left floating, it will follow the voltage on VOUT and the switch will not detect VOUT > VIN and activate reverse current blocking. This could potentially damage upstream components. 

    The biggest problem that I see present is QOD (quick output discharge). Since you're using the TPS22916 versions with QOD, the output of the load switch will be pulled down to GND when disabled to discharge the rail. If this is being used in a power mux configuration, there will be a pulldown on whichever TPS22916 has the enable pin low. This will prevent the output rail from reaching it's proper voltage. This is probably why the output voltage rail is sagging in some of your configurations. 

    On another note, I feel this system can be easily simplified by using our newest power mux device, the TPS2121. The TPS2121 can be used in multiple configuration modes include priority operation, automatic/manual switchover, and 5us switchover time to minimize output voltage drop. In regards to Iq and Ron, the TPS2121 has 100uA Iq and 56mΩ typ. to meet low power applications. The device is also available in a small package (2.5mm * 2mm).  Would this device work instead?

    Thanks,

    Arthur Huang

    To find the latest information on Power Switches, visit www.ti.com/powerswitch

    Please click "This resolved my issue" button if this post answers your question.

     

  • 0 in reply to Alpha H
    Prodigy 90 points

    Hello Arthur,

    First of all, thank you so much for your clear and fast support. I going to answer some of your questions by quoting it in your post.

    Arthur_Huang said:

    There are two potential issues that I see with this configuration. First, let me make sure my understanding of the configuration is correct. You have 5V on VIN1, and 3.3V on VIN2, and the enable pins for both enables are connected together AND connected to VIN1. When VIN1 is present, the enable signal is "1" so VIN1 is outputting to the downstream load. However, since you're using the TPS22916CL "active low" version, wouldn't this also cause VIN2 to be on? Is my understanding correct? 

    You are right about our configuration and enable pin connection.

    Our intention was:

    When Vin1 is present, enable pins are connected to '1' so TPS "normal" is active and TPS "inverted" (TPS22916CL) is desabled then Vout=Vin1.

    When Vin1 fails, enable pins are floating (weak '0'... it could be a problem indeed, because it´s not a strong '0'), then TPS "normal" should be desabled and TPS "inverted" should be active, so Vout=Vin2.

    Arthur_Huang said:

    That's the first potential problem I see with this configuration. Second, if VIN1 is removed from the system (the input is left floating), the top device will not be powered. Because VIN1 is left floating, it will follow the voltage on VOUT and the switch will not detect VOUT > VIN and activate reverse current blocking. This could potentially damage upstream components. 

    Regarding the previous point, it means that we need to put a pull-down resistor close to Vin1 input, so, when Vin1 fails (power source comes floating") we would have strong '0' in Vin1 and Enable pin.

    Is that correct?

    Arthur_Huang said:

    The biggest problem that I see present is QOD (quick output discharge). Since you're using the TPS22916 versions with QOD, the output of the load switch will be pulled down to GND when disabled to discharge the rail. If this is being used in a power mux configuration, there will be a pulldown on whichever TPS22916 has the enable pin low. This will prevent the output rail from reaching it's proper voltage. This is probably why the output voltage rail is sagging in some of your configurations. 

    As we think, QoD is causing our unexpected dropout voltage and current leakage, so, we need to avoid "TPS" with QoD.

    Keeping it in mind, we thought on the following circuit, using TPS22917, in order to acomplish Type B features.

    With this circuit we might control power source selection throught using 2 pins. Due to TPS22917 have the external QoD pins that we leave unused because we want to avoid leakage currents through this resistor when this power source is disabled.

    When a source power is enable, quiescent current is 0.5uA, and 10nA when is in stand-by (because there is no more QoD).

    And, due to internal resistance, at maximum current (2A), output voltage should be 200mV

    Are those values correct?

    Arthur_Huang said:

    On another note, I feel this system can be easily simplified by using our newest power mux device, the TPS2121. The TPS2121 can be used in multiple configuration modes include priority operation, automatic/manual switchover, and 5us switchover time to minimize output voltage drop. In regards to Iq and Ron, the TPS2121 has 100uA Iq and 56mΩ typ. to meet low power applications. The device is also available in a small package (2.5mm * 2mm).  Would this device work instead?

    I prefer to discuss about this component on the other thread if you are agree, in order to have this more clear and focus on load switches circuits and out Type A design.

    Thank you so much.

     

  • 0 in reply to Adrian Cobos
    TI__Genius 10855 points

    Hi Departamento, 

    Thanks for the clarification! Please find the answers to your questions below.

    1. Regarding the previous point, it means that we need to put a pull-down resistor close to Vin1 input, so, when Vin1 fails (power source comes floating") we would have strong '0' in Vin1 and Enable pin. Is that correct?

               [AH]: That's correct, apologies I misread the circuit. You won't need an external pull-down resistor close to VIN1. The TPS22916 has a "smart ON pin pull down", which pulls down the ON pin if the pin is left floating. Therefore, the device will be off if the pin is left floating. This should work well for this configuration.

    2. As we think, QoD is causing our unexpected dropout voltage and current leakage, so, we need to avoid "TPS" with QoD.

    [AH]: Agreed, as you mentioned the TPS22917 would be a good solution here as well.When a source power is enable, quiescent current is 0.5uA, and 10nA when is in stand-by (because there is no more QoD).

    3.  When a source power is enable, quiescent current is 0.5uA, and 10nA when is in stand-by (because there is no more QoD).  And, due to internal resistance, at maximum current (2A), output voltage should be 200mv.  Are those values correct?

     

    [AH]: Yes, however please note that the Rds,on of the device can be slightly higher if the junction temperature is increased. Please refer to Figure 3 for a graph of Ron.

    Thanks,

    Arthur Huang

     

     

  • 0 in reply to Alpha H
    Prodigy 90 points

    Hello Arthur,

    In the end, we are going to use TPS22917 in order to avoid unexpected current consumptions due to QoD, this chip have the same "smart OM pin pulldown" features too.

    We are going to focus only on design Type B using 2 external control pins to change between power sources, because TPS22917 dont have "inverse enable" features (as TPS22916 has) so, we have to accept NO seamless system.

    Thanks for all.