• State
  • Locked Locked
  • Replies 17 replies
  • Subscribers 63 subscribers
  • Views 2616 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

TPS92512 minimum Vin

David Dean
Expert 2315 points
Other Parts Discussed in Thread: TPS92510, TPS92512, TPS92513, LM46002, LM3401, LM3414

The TPS92512 datasheet clearly states a minimum Vin of 4.5V, but also an undervoltage lockout of only 2.94V.  So - what happens if I attempt to use the part with Vin = 3.5V?  I only have one LED, with a Vf around 2.9V, so there's enough headroom otherwise.  I know that the TPS92510 is rated down to 3.5V, but I really like the Iadj feature of the TPS92512.

  • 0
    TI__Mastermind 25500 points

    The part will work down to 3.5V input. However, the electrical characteristics specs can no longer be guaranteed below 4.5V nor can any other guarantees be given when operating out of spec. I've tested it lower and it will still work pretty well, but the gate drive is limited and therefore the maximum output current will also be limited. It will not do 2A or 2.5A at 3.5V input, 1A to 1.5A maximum would be more likely (1A for sure, 1.5A maybe).

  • 0 in reply to Clinton Jensen
    Expert 2315 points
    I now have the TPS92513 eval board, which is designed to begin switching at about 6V. I'd like to use it down to 4.5V, so I changed the UVLO resistors so that 4.5V would trip the UVLO threshold.

    However, as a turn up the input voltage on my lab supply, it only turns on at about 5.2V. I can watch the UVLO pin on my oscilloscope - it passes 1.22V at Vin = 4.5V, but the part does not begin switching until 5.2V. At that point, I can turn the voltage DOWN, and it continues switching at lower voltages.

    Further, it seems that, if I set my supply to 4.5V and simply turn it on, or connect the wire while the supply is on, the TPS92513 will sometimes begin, but not always.

    It is going to be critical for me to power this from 4.5V, and also critical that it is guaranteed to start. What might be going on here?
  • 0 in reply to David Dean
    TI__Mastermind 25500 points
    Can you tell me what you changed the UVLO resistor values to? One thing to try in the mean time is to change the 10M resistor from Vin to the IADJ pin to 1M (or lower, just keep the maximum pin current below 100uA at your max input voltage). The 10M is probably too high a value to bias the IADJ clamp at such low input voltages (it was sized for much higher potential input voltages up to the maximum Vin of the device).
  • 0 in reply to Clinton Jensen
    Expert 2315 points
    First attempt was to change R3 to 75K (the closest value I had on hand). After that didn't do it, I removed both R3 and R7, but no change.

    Changing the 10M resistor to 1.5M had no affect.

    What about the BOOT-PH UVLO? It looks like that might be be threshold I'm passing when it starts switching at 5.2V, but hard to be sure.
  • 0 in reply to David Dean
    TI__Mastermind 25500 points
    Tricky. I can try this out in the lab if needed, but first lets just make sure you have enough hysteresis on the UVLO. Can you try changing R3 to 133k and R7 to 47.5k? Leave the lower resistor value to IADJ as well for this low of an input voltage.
  • 0 in reply to Clinton Jensen
    Expert 2315 points
    I have 49.9K and 120K. No effect. It would be great if you could recreate this and see if you can find a solution.
  • 0 in reply to David Dean
    TI__Mastermind 25500 points
    Sure, I happen to have one EVM handy. I'll try it out, probably tomorrow at this point. I have set the UVLO at 4.5V before without issues, but not this exact design. Could you give me the Vf of the LED load you are using so I can re-create as close as possible? Also, have you tried changing the IC in case there is some minor damage to the one on the EVM?
  • 0 in reply to Clinton Jensen
    Expert 2315 points

    My LED Vf is 2.9V.

    I only have the one EVM, and no loose parts, so I can't change the IC, at least not at the moment.

  • 0 in reply to David Dean
    TI__Mastermind 25500 points
    No problem, I actually checked it out and it is something I had seen some time ago and had forgotten about. You were correct in thinking about the BOOT UVLO. What happens is if you are sweeping Vin slowly (or directly connecting live wire and removing) the output capacitor will actually stay charged to some voltage because the LED leakage current is low. So when you sweep back up (or reconnect the wire) the voltage between BOOT and PH is not sufficient to overcome UVLO since PH is sitting at the output voltage. If you use a supply that pulls to 0V when it is disabled it will discharge the output cap through the schottky and you can then immediately turn it back on at the lower voltage. This obviously only happens at low input voltages that are close to the output voltage. So you need to make sure the output falls sufficiently to start up correctly again. If the input will not perform this function you can add a bleeder resistor in parallel with the LED to discharge the output capacitor much more quickly. You do still need to use a lower value for the VIN to IADJ resistor though. 10M is too much, I reduced mine to 499k, but I think 1.5M should be low enough.
  • 0 in reply to Clinton Jensen
    Expert 2315 points

    That makes sense AND seems to have done the trick.  Thanks so much!


    One concern - I was nervous that it was possible to get in state where the driver wouldn't start, depending on how the input voltage was applied.  With the bleeder resistor installed, is there any way to get stuck in a situation where the driver won't start when 5.0V is applied?

  • 0 in reply to David Dean
    TI__Mastermind 25500 points

    The only way you could have that happen is if the power is removed and reapplied before the bleeder resistor can discharge the output cap sufficiently. So you can tailor your resistor value to make that time constant as fast as you like. You can also use a smaller output cap to speed up the discharge.

  • 0 in reply to Clinton Jensen
    Expert 2315 points
    Got it.

    What would you say is the minimum Vin that would guarantee startup? Without the bleeder resistor I was seeing it start switching at 5.2V. Is there a combination of parameters to look at that would give me that value? UVLO thresholds, LED Vf, etc? Seems the safest thing would be to provide an input above that threshold, if possible.
  • 0 in reply to David Dean
    TI__Mastermind 25500 points
    If you want to guarantee startup without the bleeder you need to take into account a few things. The Vin to BOOT regulator will have some voltage drop, maybe around 1V. Then the boot uvlo level can be as high as 2.8V or so. So you will need to have a Vin of at least 3.8V above what the LED/cap discharges to immediately upon removing the input voltage. If you want to be really really safe you would want to have Vin at least 3.8V above the LED Vf during normal operation.
  • 0 in reply to Clinton Jensen
    Expert 2315 points
    The reason I'm a bit nervous about this is that my Vin is fairly unstable and a drop under under the UVLO and then back up can be expected. It's fine if the LED goes off during that dip, but I'd need it to definitely come back on when the voltage goes back above the minimum, and I might not be able to guarantee 6.8V. If my voltage rise is slow enough, even with the bleeder I could still get locked out. Is there anything I can do around the boot capacitor to absolutely guarantee that the driver can start at 4.5V?
  • 0 in reply to David Dean
    TI__Mastermind 25500 points
    It's not the prettiest thing in the world but yes, you can. You would need to use a small switched capacitor doubler to double the input voltage. Then connect a resistor from the switch cap output to the BOOT pin and connect a 6.2V zener across the boot cap. That will prevent ever entering boot uvlo while input voltage is applied.
  • 0 in reply to Clinton Jensen
    Expert 2315 points
    Right - not very pretty. I still like the TPS92513, but are there other devices that don't have a similar BOOT UVLO? TPS92510, maybe? Also, I've had luck modifying a standard voltage regulator to act like an LED driver, in particular LM46002. Do you know if I'd run into a similar situation with either device?
  • 0 in reply to David Dean
    TI__Mastermind 25500 points
    I am not familiar with the LM46002, but the TPS92510 operates in the same manner as the 513 so it will have the same issue. However you might want to check out the LM3414 depending on your maximum current, it can do up to 1A and has a low side switch so no bootstrapping is required. Analog dimming would need to be done using something like a digital pot with that part. Another option is the LM3401. It is a PFET controller so it also does not require bootstrapping. It does not have analog adjustment, but that can be accomplished by injecting a voltage through a resistor to the SNS pin.