Post Go back to editing

LT4416 - Single fault protection

Hello together,

we use the LTC4416 in our system to make an automatic switchover from one power source to another. As we want to have the primary supply path as master, we don't want to switch the V_BAT to Vs, when Vsys<VBat. The VSys should decide, when we want to restore. Here the schematics:

Now we need to put in a single fault protection to the system. This has to be done for the following case:

V_BAT active (VSys < 5.793V) --> V29 and V41 are closed, power on VS from V_Bat.

If we have a look on a the single fault, that V27 is shorted, Vsys could be on Vsys as well --> Power from VSYS on VBAT --> Battery is in a critical state.

We thought about to build in two extra MOSFETs, which are controlled by H1 and H2 to block each others' path.

Did anyone has experience with this?

Thanks a lot

Holger

Parents
  • Hi Holger,

    As we want to have the primary supply path as master, we don't want to switch the V_BAT to Vs, when Vsys<VBat. The VSys should decide, when we want to restore.

    Now we need to put in a single fault protection to the system. This has to be done for the following case:

    V_BAT active (VSys < 5.793V) --> V29 and V41 are closed, power on VS from V_Bat.

    Here is my understanding of your app:

    You DON’T want diode-OR behavior, where the higher supply is passed to the output (Vs).

    You want the output to be determined only by the voltage of VSYS.

    For VSYS < 5.793V, Vs = VBAT

    For VSYS ≥ 5.793, Vs = VSYS

    Is this correct?

    If we have a look on a the single fault, that V27 is shorted, Vsys could be on Vsys as well --> Power from VSYS on VBAT --> Battery is in a critical state.

    I believe you are concerned that if V27 fails short (recommend using “Q” prefix for transistors), current will flow from VSYS to VBAT, possibly damaging the battery. Is this what you are worried about?

    If so, I don’t think VBAT is in danger of getting charged by VSYS.

    Let’s explore 4 scenarios:

    1. VSYS > VBAT , VSYS > 5.793V
      1. Vs = VBAT . G1 ON, G2 OFF.
    2. VSYS > VBAT , VSYS < 5.793V  OR    5.793V > VSYS > VBAT
      1. Vs = VBAT – 0.7V. G1 OFF, G2 OFF.
        1. VBAT path is on, but the body diode of V26 conducts if VSYS exceeds VBAT by a diode drop. This is what causes VBAT path to turn off.
    3. VSYS < VBAT , VSYS > 5.793V
      1. Vs = VBAT . G1 ON, G2 OFF.
        1. VSYS path ON, VBAT path OFF.
    4. VSYS < VBAT , VSYS < 5.793V
      1. Vs = VBAT . VBAT path ON.
        1. Since VSYS < 5.793V, G1 OFF. V26 body diode does not conduct.

    Here is normal behavior of your circuit:

    Compare this to when Q27 is shorted from drain to source.

    The only thing that has changed is when 5.589V > VSYS > 4.5V, Vs = VSYS – diode drop.

    Simulation is attached.

    4416 EZ 12-8-21.asc

    -Aaron

  • Hello Aaron. We have made some simulations with your model as well. Here we could see a current flow, when we resistor to the battery soource path in series. Therefore we have manipulated the schmeatics in that way we could control some additional MOSFETs with the state of \H2 to overcome the problem current flowing into the battery.

    1. What is the # meaning inside your simulation on port E2 @R203/R204? If we take this for our simulation, it is not working as excepted anymore? (See SingleFault_E2#_NOTWORKING.asc)

    2. The 2nd simulation is working, but we have a bit a strange behavior when swithcing from BAT to EXT. Could you explain this behaivor (see SingleFault_E2_WORKING.asc)

    SingleFault_E2_WORKING.ascSingleFault_E2#_NOTWORKING.asc

  • 1. What is the # meaning inside your simulation on port E2 @R203/R204? If we take this for our simulation, it is not working as excepted anymore? (See SingleFault_E2#_NOTWORKING.asc)

    Both net labels should have been E2#.
    Here is the updated simulation:

    Notice that output, Vs, drops to 0V momentarily when VSYS falls below 5.589V.
    This is because C260 causes a lag in E2#, which delays Q20 and Q41 from turning on.

    2. The 2nd simulation is working, but we have a bit a strange behavior when swithcing from BAT to EXT. Could you explain this behaivor (see SingleFault_E2_WORKING.asc)

    Removing C260 shifts this lag period to when VSYS is ramping up (compared to when it was ramping down).

    You need to explain the purpose of the additional circuitry. More FETs in the power path introduce more loss.

    -Aaron

  • Hello Aaron,

    sorry about the late reply. I was on vacation.

    We need to be sure that we don't have any possibility current flowing inside the battery. If you put a resistor in series in VBAT path and Q27 is shorted, it could flow a current inside VBat, which must be prevented. We know hat we have more losses in teh path, but we need to keep the single fault safety.

    How could we overcome the problem with the lag of period? Did you have any idea?

     SingleFault_2022_01_12.asc

Reply
  • Hello Aaron,

    sorry about the late reply. I was on vacation.

    We need to be sure that we don't have any possibility current flowing inside the battery. If you put a resistor in series in VBAT path and Q27 is shorted, it could flow a current inside VBat, which must be prevented. We know hat we have more losses in teh path, but we need to keep the single fault safety.

    How could we overcome the problem with the lag of period? Did you have any idea?

     SingleFault_2022_01_12.asc

Children
  • The common way to overcome switchover delay is to change one of the channels from a load switch configuration (back-to-back FETs) to an ideal diode configuration (single FET with body diode pointing to load).

    By doing this, the supply with the single FET can conduct through the FET's body diode during the transition period, when both gate pins are OFF. See fig 3 from the datasheet below:



    If you insist on a load switch configuration for both supplies, you can cope by adding output capacitance to prevent Vs from drooping during switchover.

    -Aaron