• State Resolved
  • Locked Locked
  • Replies 15 replies
  • Answers 3 answers
  • Subscribers 62 subscribers
  • Views 479 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.

TPS2491: TPS249x Desing Help - Sirros IoT

Jeferson Pehls
Expert 1085 points
Part Number: TPS2491
Other Parts Discussed in Thread: TPS1H200A-Q1, , CSD19537Q3, TPS2490, TPS2663, LM5069

Hi.

I am designing the new industrial data collector of the company, it will also have 2 protected transistor outputs / switches. The nominal system voltage is 24V, but we have seen in the history that some power supplies of the machinery have overvoltage, resuming we need to consider that the voltage of the power supplies can reach 36V, so I use 36V TVS diodes on all pins of the connectors, I mean 36V of reverse standoff voltage.

For the transistor outputs I had selected the smart load switch TPS1H200A-Q1 before, which the VCC maximum absolute is 42V, but after a while I started to think that this IC could be damaged easily by transients in this application.

I searched so much in the internet for higher voltage load switches and did not found nothing enough. So I saw the hot swap controllers on TI website and found the TPS2491 interesting, because its ABS Max is 100V. The TVS diode used, SMAJ36CA, it can reach a maximum clamping voltage of 75V for a 8/20us pulse. The 2 series diodes of the schematic are rated for 3A continuous and 200V reverse voltage. The Mosfet N of the schematic is CSD19537Q3 and I chosed it because we already use it in another project.

Below are the calculations I did until now for the TPS2491.

---> Calculations of typical values only

Ilim Typ = 50mV / RS
Ilim Typ = 50mV / 22mR
Ilim Typ = 2.27A

2.27A ^2 * 0.022 = 113mW
Shunt = 22mR 1206 500mW or more

Vprog = Plim / (10 * Ilim)
Vprog = 60 / (10 * 2.27)
Vprog = 2.64V

VREF = 4V
Rdown = 2.64V, Rup = 1.36V
Relation = 2.64 / 1.36 = 1.941
Rup = 5.6K
Rdown = 5.6 * 1.941 = 10.86K
Rdown = 12K

---> Recalculating:
Vprog = VREF * Rdown / (Rdown + Rup)
Vprog = 4 * 12 / (12+5.6)
Vprog = 2.72V
Plim = Vprog * (10 * Ilim)
Plim = 2.72 * (10 * 2.27)
Plim = 61.7W

I Vref = 4V / (12K + 5.6K)
I Vref = 0.227mA

Ctimer = Tfault * 25uA / 4V
Tfault = Ctimer * 4V / 25uA
Tfault = 4.7uF * 4V / 25uA
Tfault = (4.7*10^-6) * 4 / (25*10^-6)
Tfault = 752ms
T re-enable ~= Tfault * 10
T re-enable ~= 7.5s

The PCB will have 2 times the circuit below, but I did it only one time until now, I want to check if it is all ok with it before copy this to a second identical circuit.

CURRENT SCHEMATIC:


PDF OF THE SCHEMATIC:

6406.SCH_01.pdf


CURRENT PCB (2D):


CURRENT PCB (3D):

The board space is limited, it has 10cm x 9cm.
OBS: The 2 TVS diodes of 36V are placed at the other side of the PCB, very close to the connector.


OBSERVATIONS:

(1) Capacitor C41, 100nF, is Low ESL type.

(2) The circuit is intended to drive loads like relay coils, contactor coils, led strips, solenoids, self-oscillating buzzers. I know that these are the kinds of loads, but I don't have their specs.

(3) Lets suppose that the positive and negative wires of the power supply are directly connected to the input and output of the circuit, respectively, and the Mosfet is turned on, a short. In this case, the series diodes must not get damaged, what you can say about this? OBS: The series diodes withstand 100A non-repetitive peaks according to the datasheet.

(4) For this circuit, it could be used another solution with some other IC of TI, like a load switch or something similar, but remember that the VCC ABS Max should be 80V or more, and that the IC should be small and the circuit also, you can have an idea of the available board space in the pictures.

(5) Would it be better to use TPS2490 instead? And in case of faults, the MCU monitor the PG pin? I wanted to know more about the operation of TPS2490... I imagine the following, please tell me if my reasoning is correct: for example, if Tfault is set to 1s through the capacitor on TIMER pin, then the MCU put level 1 on EN pin, the Mosfet conduces and if there is a short, the current would be limited and 1s after the switching ON of the Mosfet it will be turned off, and the PG pin will keep in level 0? (or will generate a pulse to 0?). Then the MCU put the EN pin to level 0, and after some time put the EN pin again to level 1, then the transistor will turn on again? It works like this? In TPS2490, is the signal of PG pin (going to low level) related to over-current and short circuit situations?

(6) Does TI engineers have observations or suggestions?


Thanks very much in advance,
Jeferson.

  • 0
    TI__Guru* 95911 points

    Hi Jeferson,

    TPS2663 eFuse is ideal fit and using large TVS diode like SMCJ36CA helps to clamp the transient voltages effectively. If you need controllers with higher VCC ABS Max, TPS249x or LM5069 are rated for 100V. 

    In your schematic, one series diode D4 should be good enough. D5 can be removed. During power-into-output short, these diodes needs to handle current of Plim/Vin for Tfault interval which can be managed easily. D17 should be schottky diode B260-13-F to clamp the -ve transients at the output effectively. rest all looks good to me.

     3323.schottky diode at Vout.pdf

    Best Regards,

    Rakesh

  • 0 in reply to Rakesh
    Expert 1085 points

    Hi Rakesh.

    Thanks for giving the options and for your support, but I still prefer TPS249x.

    Sorry, I forgot to explain other details, the module is intended to be "miswiring-proof", so I can't place the freewheeling diode directly at the output, in case the situation bellow occurs then the freewheeling diode would be biased directly and burn:
    Power supply + ---> GND of the board
    Power supply - ---> Output

    So I added another diode to the circuit (D19), with the same PN of D4 and D5 (I want a 80V+ diode), it would be used only for freewheeling situations, the 36V TVS diode at the output is to protect the diode D5 (MBRAF3200T3G). I have chosed MBRAF3200T3G for D4, D5 and D19 because this PN is already used for other components of the board. MBRAF3200T3G is 3A x 200V.

    In diode MBRAF3200T3G datasheet, there is the typical and the maximum foward voltage graphs in functions of the instantaneous current, below is the graph of the maximum foward voltage. It was only today, after see the need of this diode, that I saw that there is a -1V voltage limit on OUT pin of TPS249x, the datasheet also shows "OUT will withstand transients to –2 V for 1 ms or less".

    You said: "During power-into-output short, these diodes needs to handle current of Plim/Vin for Tfault interval which can be managed easily". I will consider a system power supply of 20V to 36V, the worst case is 20V. I plan to change Plim to 50W (instead of 60W), so 50W/20V = 2.5A (the diodes are for 3A), but 2.5A is higher than the 2.27A set for Ilim according to the original post. I think still there is something unclear for me.


    How about the power on the MOSFET if the output is in continuous short?

    Regards.

  • 0 in reply to Rakesh
    Expert 1085 points

    Hi.

    Please could you check this?

    What is more recommended? The second?

    This trace is the OUT pin.

    Regards.

  • 0 in reply to Jeferson Pehls
    TI__Guru* 95911 points

    Hi Jeferson,

    The device has ESD diode from GND to OUT pin. so, when miswiring is applied at the output, you will have two path as shown below.

    you can separate device gnd and power gnd via a signal diode to block the path.

    The controller will take care of FET under continuous short condition.

    Best Regards,

    Rakesh

  • 0 in reply to Rakesh
    TI__Guru* 95911 points

    Hi Jeferson,

    I would suggest to get an EVM and try the configuration.

    Thanks,

    Rakesh

  • 0 in reply to Rakesh
    Expert 1085 points

    Hi Rakesh, thanks!

    How about this? Using D19 this way? D19 being MBR1H100SFT3G (1A x 100V), the bias current of OUT pin is 40uA Max, there would be a very small voltage drop across D19, I believe.

    In practice, our setup will be like the picture below. The red diode is the internal diode of TPS2491 that goes from GND to OUT pin. The green diode is D19, the black diode is D5, the TVS is D17. The freewheeling diode would be installed directly across the wires of the load, externally to the board.

    Do you think its acceptable this way? See problem on this?

    Regards.

  • 0 in reply to Jeferson Pehls
    TI__Guru* 95911 points

    Hi Jeferson,

    Yes. That should be fine. I don't see any concern.

    Best Regards, 

    Rakesh

  • 0 in reply to Rakesh
    Expert 1085 points

    Thanks very much!

    Another question: if we consider that both series diodes are rated for 3A maximum continuous, that P lim is set to 60W currently, and that the minimum suppply voltage would be 20V in a worst case, do you see problem? To be safer, should I reduce P lim a little? To 55W or 50W? I mean, to guarantee that the series diodes won't be damaged in case of a constant short circuit at the output.

    Regards.

    [EDIT]

    For the series diodes, if needed I can also use the following diode below, it is rated for 4A x 100V:

    www.digikey.com/.../6024269

  • 0 in reply to Jeferson Pehls
    TI__Guru* 95911 points

    Hi,

    Under sustained short circuit, the TPS249x will cut-off the path after Tfault interval

    The selected diode dissipates 2W at 3A. Taking Tfault as 10ms, the temperature rise would be 3C/W * 2W =6C above the ambient. So should be fine.

    Best Regards,

    Rakesh

  • 0 in reply to Rakesh
    Expert 1085 points

    Hi Rakesh, thanks!

    On the last schematic I just changed the capacitor of TIMER pin from 4.7uF to 1uF to reduce Tfault.

    4.7uF: Tfault = (4.7*10^-6) * 4 / (25*10^-6) = 752ms
    1uF: Tfault = (1*10^-6) * 4 / (25*10^-6) = 160ms

    Plim is currently calculated to be 61.7W, and at page 16 of the datasheet is said "The TPS2491 automatically initiates a restart after a fault has caused it to turn off Q1. Internal control circuits use CT to count 16 cycles before re-enabling Q1". So for TPS2491 which features automatic retry, and in case of a sustained short-circuit, we have a duty of 1/17, 61.7W * 1/17 = 3.6W average power dissipated on the transistor. [I can be wrong with my statement].

    One of my concerns is about damaging the transistor.

    I'm thinking to switch to TPS2490...

    - At page 11 of the datasheet says about EN pin: "A TPS2490 that has latched off can be reset by cycling EN below its negative threshold and back high".

    - At page 13 says about PG pin: PG is false (low or low resistance to ground) whenever VDS of Q1 has not been less than 1.25 V, VDS of Q1 is above 2.7 V, or UVLO is active.

    So I think I just need to be able to detect the faults (like a short) with the microcontroller through the reading of the PG pin of TPS2490, for then after some time put EN pin to '0' (below the negative threshould) and back to '1' (above the positive threshould) again. Is this Ok?

    Regards.

  • +1 in reply to Jeferson Pehls
    TI__Guru* 95911 points

    Hi Jeferson,

    The TIMER has a 1:10 charge-to-discharge current ratio so, the duty cycle is 1% max and not 1/17. The average power dissipation is 0.617W for power limit of 61.7W

    Other option is to use latch-off variant.

    The approach you mentioned with microcontroller is fine.

    Best Regards,

    Rakesh

  • 0 in reply to Rakesh
    Expert 1085 points

    Thanks very much Rakesh.

    Could you please check if the calculations below are correct?

    Tfault = Ctimer * 4V / 25uA
    Tfault = 0.15uF * 4V / 25uA
    Tfault = (0.15*10^-6) * 4 / (25*10^-6)
    Tfault = 24ms

    D Retry - Fault retry duty cycle
    - 0.5% min (200x) - T retry Max = ~4.8s
    - 0.75% typ (133x) - T retry Typ = ~3.2s
    - 1% max (100x) - T retry Min = ~2.4s

    Regards.

  • 0 in reply to Jeferson Pehls
    TI__Guru* 95911 points

    Hi Jeferson,

    Your calculations are correct.

    Thanks,

    Rakesh

  • 0 in reply to Rakesh
    Expert 1085 points

    Hi Rakesh.

    I'm done, could you check below if you see any problem?

    FINAL SCHEMATIC:

    PCB LAYOUT:

    CALCULATIONS:

    ---> Calculations of typical values only

    Ilim Typ = 50mV / RS
    Ilim Typ = 50mV / 15mR
    Ilim Typ = 3.33A

    3.33A ^2 * 0.015 = 166mW
    Shunt = 15mR 1206 500mW or more

    Vprog = Plim / (10 * Ilim)
    Vprog = 80 / (10 * 3.33)
    Vprog = 2.4V

    VREF = 4V
    V-Rdown = 2.4V, V-Rup = 1.6V
    Relation = 2.4 / 1.6 = 1.5
    Rup = 10K
    Rdown = 10K * 1.5 = 15K

    ---> Recalculating:
    Vprog = 4V * (15/(15+10)) = 2.4V
    Plim = Vprog * (10 * Ilim)
    Plim = 2.4 * (10 * 3.33)
    Plim = 80W

    --->
    I Vref = 4V / (10K + 15K)
    I Vref = 0.16mA

    Ctimer = Tfault * 25uA / 4V
    Tfault = Ctimer * 4V / 25uA
    Tfault = 0.15uF * 4V / 25uA
    Tfault = (0.15*10^-6) * 4 / (25*10^-6)
    Tfault = 24ms

    D Retry - Fault retry duty cycle
    - 0.5% min (200x) - T retry Max = 4.8s
    - 0.75% typ (133x) - T retry Typ = 3.2s
    - 1% max (100x) - T retry Min = 2.4s

    P transistor Max @ sustained short-circuit:
    80W * 0.01 = 0.8W (800mW)

    Regards,
    Jeferson.

  • +1 in reply to Jeferson Pehls
    TI__Guru* 95911 points

    Hi Jeferson,

    The calculations, schematic and layout looks good to me.

    Have you verified the SOA margin of the design using design calculator? Please check once and proceed. The design tool is available at

    https://www.ti.com/lit/zip/slvc033 

    Thanks,

    Rakesh