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.

TPS92641: A few TPS92641's intermittently shut down/don't start

Part Number: TPS92641

Hi All,

I have a couple designs that use the TPS92641 to drive a single high-current (15-20A) LED under the control of a microcontroller.  One of the designs uses a single TPS, the other uses 3.  The single channel design works well.  I perform a mix of manipulating the a filtered PWM voltage applied to IADJ and PWM chopping of SDIM to control the LEDs' output over its full range from around 1% to its full output at 100% (up to 20A).  For higher output levels, I leave SDIM high and vary the voltage applied to the IADJ input by controlling the PWM duty-cycle.  As LED desired light output drops, I hold IADJ at about 10% its nominal value and chop SDIM to further reduce light output.  I am working with a couple of different wavelength LEDs, which both show this issue.

On the three channel board, I am having some intermittent issues with two of the channels (one channel works fine) but the other two intermittently shut down in spite of the fact that their control inputs are being properly driven.  I am sure this isn't an overheating problem since the controller chips don't get warm to the touch.  When not operating the controllers holds all its gate output signals in their inactive state.  The 8.5V supply is working and the VREF is 3V as it should be.  The UDIM input is biased to the controller's supply with a pair of resistors that set the brownout threshold to about 10V.  I am running the boards from a 12V @ 20A or 12V@25A supply which always measures 12V.

When I increase the IADJ input to nearly 100% (1.78V), the chips begin working.  If instead I increase IADJ to 20% or 30%, the chips remain dormant, as if their internal clocks have stopped.  Once the IADJ input has gone high enough to get the chip "started", I can reduce it to 10% or 20% and the controller remains on.  It will however, intermittently, after minutes or longer shut off.  If I immediately increase IADJ to 100%, the controllers resume operation as before.

Two of the controllers on the three channel board do this.  I have been over the boards dozens of times with a microscope and oscilloscope looking for the intermittent solder connection or missing signal and don't find one.

Appreciate any thoughts you might have.

Thanks,

Scott

  • Hello Scott,

    " I am working with a couple of different wavelength LEDs, which both show this issue." Is this on the three channel board or do you see this issue on the one channel board which "The single channel design works well." Have you tried changing the LEDs to the other drivers to see if it is Vf related? Different wavelength LEDs will have different forward voltages. Are the three channels the same design as the single channel (values, etc)?

    The Brown out threshold, is that rising or falling input voltage? Are there capacitors next to the Vin pins of the IC and power stages of the three channels? Have you powered just one of the malfunctioning channels on without the others?

    BR,
  • Hi BR,
    Thanks for the reply.
    The one channel board works well with either LED type. I don't believe the issue is Vf related because the controller chip isn't outputting any signals to switch the inductor. It turns out that the forward voltage of both LED colors is nearly identical.
    There are no capacitors/inductors in the UDIM circuit, just two resistors (68K to Vin and 10K to ground) which set the operating threshold to about 10V as I understand.
    The controller is bypassed with 100nF, 1uF and 22uF ceramics on Vin and 1 uF on Vcc. There are three 22 uF capacitors bypassing the +12V supply next to the NMOS.
    All of the test results I discussed above are using only one channel at a time.

    Scott
  • Hello Scott,

    Yes, the UVLO is set to 9.95V with a high tolerance of 10.45V. Have you looked at all the pins on the IC that would prevent it from running, including UVLO in case something is installed incorrect? Comp would be another pin to look at. (Being that you increase PWM up to 100% and it comes on is strange also). What is your current sense resistor value? What is your PWM frequency and switching frequency?

    I assume each controller is bypassed with capacitors?

    Best Regards,
  • Hi Irwin,
    The current sense resistor is 0.01 ohm. The nominal PWM frequency is 60 KHz. I am using a 2-stage R-C filter for IADJ (10K/100 nF followed by another 10K/100nF). Yes, the controllers are bypassed with 100nF, 1uF and 22uF, all ceramic.
    I'll check out comp tomorrow.
    The design is based on TI's 20A LED driver example circuit. I haven't deviated from it much except the addition of a IADJ being driven by a filtered PWM.

    Thanks,
    Scott
  • Hello Scott,

    I have not heard back from you so I believe you may have solved the issue.
    The thread will close now but will reopen if you reply.

    Best Regards,
  • Hi Irwin,

    The issue is more or less under control.  I discovered that very odd things happen when IADJ is driven at small PWM values, under 10% and SDIM is chopped to control brightness.  This type of control is somewhat historical.  In our original design, I used a 1 uH inductor and saw noticeable LED flicker for IADJ values at and under 10%.  Since we needed lower light output levels, I began chopping SDIM to further reduce brightness.  This control strategy carried over into our new design that uses a 10uH inductor.

    The first thing that happens is the driver (TPS92641) won't start operation.  It just sits there with no outputs.  This mode of operation seems to induce a threshold.  The controller doesn't start until it's inputs are driven up to a threshold value then reduced. Depending on the initial drive levels, the threshold can be quite high, 80-100%.  By this I mean if I initially drive IADJ and SDIM at 10% the controller doesn't start.  Increasing IADJ to 100% then once the driver starts reducing it to the original 10% will get the chip going.  Sometimes after minutes or hours the chip will shutdown when operated at these low duty-cycles.  When it doesn't shut down, there is significant LED flicker (random light output fluctuations).

    I have discontinued use of the SDIM input (it just remains high while the controller is in operation) which has eliminated the flicker, and also the threshold.  With SDIM high all the time, the controller seems to reliably start when IADJ is as low as 0.1% (our resolution limit).  So all is good.

    Thanks,

    Scott