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.

TPS92515-Q1: On the position of SW of Shunt FET Dimming

Part Number: TPS92515-Q1

Hi all,

The data sheet has SW between Vled and GND.

Is it OK to place SW between diode and GND?

Is it necessary to connect as shown in Fig. 14 of the data sheet?

And, what is the allowable frequency of SW ON / OFF?

Best Regards
Koji Hayashi

  • Hello Koji,

    Yes, it is necessary to connect it like fig 14. The point is to keep the inductor/FET current slewed up to the same average value as the LED current so that the LED response time is very fast. Connecting it to the diode would not work correctly and may even damage the device and/or the FET.

    The switching frequency you use is really just limited by the slew rate and the linearity you need. It works well up to 30kHz with no problem as demonstrated by:

    You can go faster depending on linearity and contrast ratio requirements.

    Regards,

    Clint

  • Hi Clint,

    My Customer want to used at 20MHz.

    What kind of settings can you use?

    Please let me know if it can be handled with other IC.


    Also it is described as Time of Flight (TOF), in the data sheet.

    How much is assumed for the frequency?

    Best Regards
    Koji Hayashi
  • Hello Koji,

    20MHz is very fast. This type of converter with shunt FET dimming is as fast as it gets for high current LEDs. The parasitics alone, even if you have the FET directly across the LED(s), will result in 10s of nanoseconds rise and fall times.

    So the answer is yes, you could do it but there are things to consider. You will be very limited in your duty cycle range and you can't expect to get a square wave current waveform. It will look more sinusoidal. It will also be faster than the device switching frequency so that will be superimposed over the peak current each cycle. I have seen this be ok for some TOP applications.

    One other possibility is to use a large output capacitor and a resistor in series with the shunt FET. I have seen this done also though I've never tried it. Again it will result in more of a sine wave superimposed over the device switching frequency.

    Regards,

    Clint

  • Hi Clint,

    Thank you for quick response.

    It is described as Time of Flight (TOF), in the data sheet.
    How much is assumed for the frequency?

    Are there other ICs that can be used with TOF?

    Customers say they want to use square waves.
    Can I approach a square wave by changing R, L, C etc?

    Best Regards
    Koji Hayashi
  • Hello Koji,

    This device is as fast as switching regulators get regarding shunt FET dimming. When it was tested for TOF it was tested up to 300kHz PWM frequency. At that point it is still relatively a square wave (not completely, you can see the slope on the edges) and works well. But you do still have the switching frequency ripple superimposed on the LED current peaks.

    I doubt you will get much better regardless of component choice. It's simply physics. Even nHs of parasitic inductance will prevent you from getting a square wave at really high frequencies. The parasitics in the FET and the LED alone would prevent it.

    Regards,

    Clint

  • Hi Clint,

    Thank you your response.

    I understood that it can not be adjusted at 20 MHz.

    Best Regards
    Koji Hayashi
  • Hi Clint,

    You said,
    " Connecting it to the diode would not work correctly and may even damage the device and/or the FET."

    What is the damage to the device and the FET?
    Please tell me the mechanism.

    Best Regards
    Koji Hayashi
  • Hello Koji,

    Connecting a FET from the switch node to ground does not allow the inductor to control current. Ideally the device self current limits to the peak current you set, but without the inductor to control the slew the current could rise to 10s of amps before the converter could respond (sub ns, more l like A/ps current rise times) so you could blow the internal FET and/or the bond wires, or even the shunt FET, before anything could react. The shunt FET must come after the inductor to control current slew rates.

    Regards,

    Clint

  • Hi Clint,

    Thank you for your response.

    Sorry for the confusion

    The first question is "Is it OK to place FET between LED and GND? (LED-FET-GND)"
    (Is it OK to connect the FET in series, instead of connecting the FET in parallel with the LED?)
    The answer was "Connecting it to the diode would not work correctly and may even damage the device and/or the FET."

    What is the damage to the device and the FET?
    Please tell me the mechanism.

    Best Regards
    Koji Hayashi
  • Hello Koji,

    I see. I'm not sure I understand why you would want to put the FET in series with the LED. You could, but when the FET is off the output voltage will go to Vin unless the PWM pin is pulled low. It may cause some odd transients when the FET is turned on. It would be best to either just use the PWM pin with no FET or to use the FET in parallel with the LED for even faster rise/fall times.

    Is there a reason I do not understand that you would like to have the FET in series with the LED? Thanks.

    Regards,

    Clint

  • Hi Clint-san

    Thank you for response.

    In the case of parallel, turn on the FET when it do not want to light the LED.
    At this time high current flows.

    In the case of series, turn off the FET when it do not want to light the LED.
    As a result, no current flows.

    So, my customer want to connect FET and LED in series.

    ”when the FET is off the output voltage will go to Vin”
    Why is that?

    I think that Vout is determined by Vf of the LED.

    Best Regards
    Koji Hayashi

  • Hello Koji-san,

    If PWM is not pulled low and you open a series FET the output will be open, so the 515 FET will turn on and stay on trying to get the current sense to the peak value. The peak value cannot be reached however if the output is open so the output voltage goes to Vin (the LED is not in circuit with an open series FET). If PWM is pulled low it will stop switching and that will not happen.

    Parallel shunt FET dimming does continue to regulate current through the parallel FET. That is so that when the FET turns off the LED current will slew very fast since the inductor current is still regulated. The power loss is low with the FET on since it is just I^2*Rds(on), but that is the way to get the fastest slew rate. 

    A series FET will cause Vout to go to Vin if PWM is high, or to zero if PWM is low. In that case it is just best to not use a series FET and just use the PWM pin. It would be just as fast and you would not need an extra FET. I series FET in conjunction with the PWM pin could be used for a faster response if and only if you are using an output capacitor. But with a capacitor you would need to use the PWM pin or the capacitor would charge to Vin and likely blow the LED when the FET is closed.

    Regards,

    Clint

  • Hi Clint-san,

    I can be understood that switching is not performed when PWM is Low.
    Why that "A series FET will cause Vout to go to Vin if PWM is high, or to zero if PWM is low. "
    Please give an explanation.
    At this time FET is ON or OFF?


    Best Regards
    Koji Hayashi
  • Hello Koji-san,

    When the regulator is operating and there is an open output (series FET is OFF) the switcher will go to maximum duty cycle to try and get the current sense voltage to rise to its peak setting. So the internal FET is basically about 99.99% of the time so Vout will rise to Vin because there is no current flow to reach the peak current in the sense resistor.

    Regards,

    Clint

  • Hi Clint-san

    Thank you for response.

    I understood the answer to the previous question.

    You mean that if LEDs and FETs are used in series,  the LEDs is sometimes destroyed when FET is turned off without Cout ?

    Best Regards
    Koji Hayashi

  • Hello Koji-san,

    With the series FET off there is no path for current flow so the LEDs will be safe regardless of whether there is an output capacitor or not. But, if there is an output capacitor and PWM is high the output capacitor will charge up to Vin, so when the FET is turned ON the capacitor will discharge through the LEDs with a high current and could cause damage. If there is no output capacitor there is no energy storage so the LEDs will not get damaged when the FET is turned ON.

    Regards,

    Clint

  • Hi Clint,

    Thank you for response.

    " Connecting it to the diode would not work correctly and may even damage the device and/or the FET."
    The above refers to the case of using Cout,It is not damage when Cout is not used.
    Is this correct?

    Best Regards
    Koji Hayashi
  • Hello Koji-san,

    When I had said that I had thought that you meant placing the shunt FET across the rectifier schottky diode, not the LED diode. If it is across the LED (shunt FET dimming) then it could be damaged with Cout, but it will not be without Cout.

    The same is true with a FET in series with the LED unless you also pull PWM low so that the Cout voltage doesn't rise during the off time.

    Regards,

    Clint

  • Hi Clint-san

    Thank you for response.

    Sorry for the confusion..

    My customers use Vout - LED - FET - GND and do not use Cout, so they are expected not to damage.

    Best Regards
    Koji Hayashi