• State
  • Locked Locked
  • Replies 8 replies
  • Subscribers 63 subscribers
  • Views 887 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.

79533 Latch up in RF rich environment?

James Olson
Prodigy 210 points
Other Parts Discussed in Thread: TPS795

We're in the second engineering build of a 2.4GHz RF telemetry product that operates from a single 3.6V lithium battery.  The 79533 is used to down-regulate the battery voltage to 3.3VDC for the RF power amplifier IC.  The power amp IC can't tolerate 3.6V.

The first build of 50 units worked very well.  No 79533 linear regulator problems.  The second build is plagued with intermittent latch up of the 79533, likely RF-induced.  The latch-up occurs during an RF transmission, but not on every RF transmission.  It might go 50 transmissions before it latches up.

We noticed the latch up when the batteries in our telemetry transceivers started dying prematurely.  Upon further study we see that during latch up the 79533 draws 250mA but the output pin #4 and the enable pin #1 are zero.  The regulator's output being zero means that the RF power amp is turned off during latch up.

During latch up the chip is non-responsive to its enable pin #1.  A scope placed on the noise rejection pin #5 shows constant 250mVDC.

Latch up can be cured by disconnecting and reapplying the power applied to the 79533 on pin #2.

Per the datasheet the chip has input and output bypass caps.  Per the datasheet the noise rejection pin #5 is equipped with both 10nF and 10pF caps in parallel.  The 10nF cap is the recommended one for regulator noise rejection.  The 10pF cap is for RF suppression.  Both caps are 0402s anchored to a ground plane in a 4-layer board.  

The PCB layout is consistent with that shown in the datasheet: extremely tight.  It's in the middle of a well-designed 2.5GHz transceiver circuitry.

Any ideas? 

Jim Olson

Viewpoint Electronics

Indianapolis, IN US

  • 0
    TI__Guru 55940 points
    Hi Jim,

    Could you please provide the complete orderable number for the device you are referring to?

    Very Respectfully,
    Ryan
  • 0 in reply to Ryan Eslinger
    Prodigy 210 points

    Ryan,

    This is from the Digikey purchase order: TPS79533DCQR

    The "good" batch from the January 2016 build that did not latch were marked:

    PS79533

    56A076U

    The present batch that latches is marked:

    PS79533

    63AX1VU

    I notice that the engineer who designed this board used only a 100nF cap at the VIN pin.  The VIN cap is within 3mm of the VIN pin and is surrounded by stitches to the ground plane.  

    Is it possible that the 79533 regulator is not latching but instead is oscillating at a very high frequency--too high for this piss ant Tek 'scope to see?

    Jim Olson

    (317)590-0700 cell

  • 0 in reply to James Olson
    Prodigy 210 points
    Ryan,

    If this this bulletin board permits posting of photographs, I can send you one of the 79533 and the parts around it.

    Jim Olson
  • 0 in reply to James Olson
    Prodigy 210 points
    Upon further study I'd like to retract my previous statement that VOUT does not follow EN when the IC is hot. So, it appears this is not a latch up situation. Probably oscillating.

    When EN goes high while the IC is hot (drawing >250mA in phantom current) VOUT will go to 2.5VDC and remain there. But not 3.3VDC as it should .

    While EN is low the NR pin rests at 250mV. When EN goes high the NR pin voltage begins ramping up but does not quite reach 1VDC before our transmit interval ends and EN drops low. Our transmit interval is about 10msec long.

    This measurement of the NR pin is with a 10nF cap to ground on that pin--the recommended cap value in the data sheet.

    So with this new observation, do you agree that the problem is oscillation rather than latch up?

    Our output capacitor is 10uF ceramic. Our input cap is 100nF X7R ceramic. Any problem with these values? What changes in them would subdue the oscillation?

    Jim Olson
  • 0 in reply to James Olson
    Prodigy 210 points
    Additional data for your review:

    We have some 1uF X7R caps in the lab in the 0603 size. The length of an 0603 is approximately the spacing between two pins on the 79533.

    I soldered one of these 1uF X7R caps to the IC at the epoxy body so that the cap straddles pins 2 and 3 (VIN and GND). This effectively makes the CIN=1.1uF.

    There is no improvement in the problem. It occurs with the same intensity and with the same frequency of occurrence.
  • 0 in reply to James Olson
    TI__Guru 55940 points
    Hi Jim,

    Thank you for all of the information! Would it be possible to post a scope shot with Vin, Vout, EN, and Iout? This will help give a better indication of what is going on.

    Increasing the input capacitor will help the upstream device supply the inrush current to the 10uF capacitor. As long as your upstream supply to TPS795 is able to supply the inrush current, increasing the input capacitor will simply provide some prefiltering. Any scope shots that you can provide with Vin, Vout, EN and Iout will help us debug.

    Very Respectfully,
    Ryan
  • 0 in reply to Ryan Eslinger
    Prodigy 210 points

    Ryan,

    Let me pause a bit in this exchange and implement a fix on many units that was successful yesterday on one board.  

    If it works today on 20 units it will likely be the final solution.

    The fix:

    WAS: Cin=100nF, Cout=10uF

    Now: Cin=1uF, Cout=1uF

    Thanks.

    Jim Olson