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.

TPS92518HV: LED Driver for fast flash

Part Number: TPS92518HV
Other Parts Discussed in Thread: TPS92515, TPS92518, TPS92515HV,


We need to control a string of LED for fast flashes that might be a few seconds apart.

We have started to look at the TPS92518 as well as TPS92515.

The main parameters are :

DEL Strip curent  = 1 A
DEL Strip Voltage = 40 V 

Time ON = 1 us to 4 ms
Time OFF = 500 ms  to 10 seconds.

We will control the ON / OFF time of the pulses with an external signal.

We might want to use Shunt Fet Dimming for the fastest way to have short current pulses.

Is the TPS92518HV a good part selection ?  Or the TPS92515HV ?

Is there any potential problems or important design issues with the fact that there is a long OFF Time for the DEL Strip ?

We want a minimal power cunsumption.

Thanks !


  • Hello Stephane,

    I have a few questions and comments:

    What switching frequency range were you looking at?

    The output is 40V at 1A, what is the input voltage?

    How accurate do you want the pulse width to be?  At 1 us you would most likely need to use shunt fet dimming.

    Do you expect to use output capacitance?  If so the shunt fet has to be sized to discharge that capacitance.  It also cannot be a very large capacitance value due to your narrow pulse width requirement.

    Both the TPS92515 and the TPS92518 have boot capacitors.  Long off-times will discharage that capacitor.  On the TPS92515 the Roff resistor will help discharge the capacitance and pull the switch node to ground (which charges Cboot).  The TPS92518 may need an added load resistor to pull the switch node to ground.  The reason is you don't want a boot refresh to happen right when you input a narrow PWM pulse.  With shunt FET dimming this is not an issue, it's just not as efficient.  You may be able to use a combination of shunt FET and regular PWM for better efficiency, the regular dimming would have to go high before the shunt FET dimming to allow the current to ramp.

    Note that the TPS92518 is a dual channel and requires SPI bus to operate except for one operating point by pulling EN above 23.6V (see the datasheet).

    Best Regards,


  • Hello, 

    Thank you for the fast reply...

    Actually, the switching frequency is not a concern for me, I want the most fast rising / falling and stable current pulses.

    The Input voltage accessible will be what is required,  as it is yet to be designed..

    For the accuracy, it's not that much of an issue, maybe 5%.  But we want the pulse to be the same amplitude and duration from flash to flash...



  • Hi Stephane,

    Do you only need one channel?  Are you okay using SPI bus?  The input voltage will affect the design since the switching frequency depends on the off-time and on-time.  The off-time, once chosen, will be fixed and determine the current ripple.  The on-time will be longer since your output voltage is high.  For example if the input is 44V the on-time will be 10X the off-time.  At 50V input it will be 4X the off-time which will be over twice the switching frequency compared to 44V input.  The input voltage is an important parameter for calculting this design.  So can it be assumed that there will be little or no output capacitance?

    Another difficulty will be the narrow pwm pulses when shunt FET dimming.  The ripple current at low pwm pulses will affect the average current during the on-pulse.  I'm trying to gather most of the information needed to design a circuit to see what is possible.

    Another way to make the pwm pulse more consistent is to use standard pwm with a fast current rise time.  This may require the pulse sent is actually a little wider than needed.  The input voltage would have to also be fairly constant for this to work and will need to be known to create a design.  The output voltage range/consistency would also need to be known.  Higher input voltage would help for trying to do it this way.

    Best Regards,

  • Hello, 

    Actually, we need 2 channels,    I would prefer not using SPI.  So that is why I would prefer the TPS92515.

    TPS92515 vs TPS92518, except SPI, any important differences ?

    One is better than the other for what I want to do ?

  • Hello Stephane,

    The TPS92515 and TPS92518 operate very similar.  The TPS92515 has an internal MOSFET, the TPS92518 requires external MOSFETs.  This allows the TPS92518 to operate at higher currents.  Two TPS92515s are probably best for your design, no need for SPI, internal MOSFET, 1A of current fits the TPS92515.  The TPS92515 also uses an external  Roff resistor which will pull the switch node low when off for long periods, this will keep the boot capacitor charged (as long as nothing is pulling up on the output and there is minimal output capacitance).  That resistor can also be added to the TPS92518 to do the same thing.

    Best Regards,

  • I am working with the TPS9515 design calculator and the datasheet design procedure...

    They seem to be different in their approch and input parametes.  Which one is preferred ?

  • Hi Stephane,

    Those are just general tools to get a design.  It's best to design the circuit to your needs.  You can start with either method as well as look at the EVM design.  You can even get an EVM to test.  The portion of your design that will require looking at closer is the 1 us pulse width and long PWM low.  I don't use either tool because I just create each design as how I would create them using some of the equations in the datasheet.  This is a buck converter so the power stage design is similar to most buck converter LED drivers.  I can also review your design to see if it will work.

    Best Regards,

  • Hello, 

    I will have a 36V, 1A LED strip, with input voltage around 42-44 V...   Is this enought margin ?    

    For the Shunt Dimming Transistor, I had a look at the evaluation module of the 92518 and there is a bit FET...  600V, 4.6A MOSFET.

    Is there a reason for such a part ?  I was thinking more of 100V / 3-4 A for the Mosfet.

    Thanks !

  • Hi Stephane,

    Yes, that is enough margin.  A 100V part makes more sense, not sure why there is a 600V FET for the shunt FET.  Look at the RDSon since that will be conduction loss (I^2*R) and you're design can be near 100% conduction loss for the shut FET.  For the TPS92515 you'll also have to implement the second Roff to keep the current continuous when the Shunt FET is on, this is in the datasheet 8.3.4 starting on page 12 and figure 14 on page 13.

    Best Regards,