• State Resolved
  • Locked Locked
  • Replies 3 replies
  • Subscribers 34 subscribers
  • Views 170 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.

TCAL9539: Nature of Current Controlled Outputs?

James Olson
Prodigy 230 points

Part Number: TCAL9539

I am testing a PCB I designed.  It contains a TCAL9539 to drive qty=16 red surface-mounted LEDs of 0603 size.  The TCAL device is powered from 3.3VDC. 

Based on the part of the specification shown below plus various prose in the data sheets I concluded that the TCAL has constant current outputs.  In fact, I concluded that the currents are user selectable from 2.5mA to 10mA in four equal sized steps.

TCAL9539.jpg

I remember thinking that TI folks are extremely talented to be able to create constant current outputs with so little margin between the forward voltage of the LED and the 3.3VDC VDD rail.  Moreover, it seems that TI engineers were able to create a bi-polar constant current source for each pin.

In my PCB design I tied the anodes of my red LEDs to VDD and their cathodes to the P port pins.  I used no series resistors because I assumed that the on-chip constant current sources associated with the P ports would serve the purpose normally served by series dropping resistors.

For a brief moment I wondered about the power dissipation in the TCAL absent any series resistors, but the package I chose has a bottom side power pad.  

At boot up the tiny SMD LEDs emit a tiny puff of smoke and give a brightness level corresponding to currents far greater than 2.5mA to 10mADC. Not the behavior I expected. 

My first thought was "hey, where's my constant current source? Is there a hidden 5th constant current value at >100mADC awaiting for users that misconfigure the 40-43 registers?"

And then I did an intense re-read of the data sheet, finding appended to the description of the "fingers" a lengthy discussion of shoot-through currents of push/pull output drivers and the profound effect on the VDD rail's noise when 16 outputs change state.

Could it be that the user programmable fingers are not intended to facilitate direct, resistor-less drive of LEDs but to minimize shoot-through current effects on VDD?

Could it be that the table in the specifications mentioning 2.5/5.0/7.5/10mADC currents is provided merely to allow the user to chose his external series resistor for each LED by giving estimates of the ON resistances of the on-chip MOSFETs?

What say you on this?  How did I draw the erroneous conclusion that the P outputs were constant current?


  • 0
    TI__Guru 65698 points

    Hi James,

    I will respond soon. 

    Regards,

    Tyler

  • 0
    TI__Guru 65698 points

    Hi James,

    It sounds like the LED was shorted. 

    James Olson said:

    Could it be that the table in the specifications mentioning 2.5/5.0/7.5/10mADC currents is provided merely to allow the user to chose his external series resistor for each LED by giving estimates of the ON resistances of the on-chip MOSF

    TCAL9539.jpg

    The push-pull output structure of a P-port when sinking 10mA through the nFET at max drive strength 11b when VCC = 3.0V, the output voltage VOL(max) <= 0.1 V. 

    The IOL is varied depending on the load connected to the p-port. 

    For your LED, without any additional series resistance, the current is limited only by the push-pull driver RDS_ON resistance. 

    If we assume that the drivers are "strong" and have a small resistance, the LED is essentially seeing VCC at the anode and GND at the cathode. This would create a high-current path and short out the LED.

    Regards,

    Tyler

  • 0 in reply to Tyler Townsend
    Prodigy 230 points

    Tyler, I could not find a reply button underneath the answer you provided below.  However, I assume this reply location will reach you.

    Your hand-drawn schematic below is not representative of my circuit.  All 16 of my LEDs have their anodes attached to VDD.  Each of the 16 LEDs has its cathode tied to one P port output pin.  Said differently, each LED is driven by one or more N-channel "fingers" of a given P port pin based on how the 40-43 registers are configured.

    Not sure how you drew the conclusion that my LEDs are wired **between** P port pins. They are not.

    However, I think the gist of your response is that there is little if any current control designed into the P port outputs, and certainly none that can serve in lieu of an LED series resistor.  Even one active N-channel finger of a four-fingered N-channel output possesses sufficient sink current capability to overpower an 0603 red LED having no series resistor.

    In other words, the multi-fingered output architecture of the P port pins should not be viewed as a predictable current source/sink in the range of 2.5-10mADC, regardless of the VDD voltage.  If the chart I cited in my original questions seems to imply that the P port output currents are 2.5/5/7.5/10mA based register values, that conclusion is incorrect.

    And though you did not address this topic in your response, I've concluded that the motivation for the four-fingered architecture was simply to allow some P port pins to have very strong drive capability where needed while other P port pins only 1/4 as much drive. 

    The motivation for trimming back drive capability on some P port pins (where high drive capability is unnecessary) is to lower the total amount of shoot-through current into the VDD/VSS power planes at the time such push/pull outputs switch. 

    In other words, the four-fingered I/O pin architecture is a noise suppression strategy.
     
    Thanks for the quick response to my query, Tyler. Texas Instruments' E2E feature is a very useful design tool for an engineer like me.

    Best regards,

    Jim Olson
    Carmel, IN US