• State
  • Replies 6 replies
  • Subscribers 48 subscribers
  • Views 112 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

LMH6321: EF Pin Unexpected Behavior

DiodeWhiz
Prodigy 10 points
Part Number: LMH6321
Other Parts Discussed in Thread: OPA192

Tool/software:

Hello!

I am using the LMH6321 in the following design:

LMH6321 In Circuit

I input a 0 - 2.5V signal from the left and at the output I get a -0.2 - 10.2V signal. Using the 53.6kΩ resistor I am setting an approximate 120mA current limit. Fortunately, these parts of the circuit seem to work as anticipated.

Unfortunately, what doesn't work is the EF pin, as you can see in the schematic diagram I have inserted in a 10kΩ resistor which pulls EF up to +5V. During normal operation, I get the expected +5V output at this pin. However, under an overheat condition (introduced by either a short circuit or heat gun) the EF pin begins to oscillate:

Note that the EF pin 1 is the blue trace, and the output pin 6 is the red trace. The IC is sourcing roughly 120mA into a short connected to the output through the 10Ω resistor. If I continue to heat the IC up with a heat gun the oscillation changes period and frequency:



By measuring the voltage across R55, the 10 Ω output resistor, we can measure output current. The blue trace is the EF pin, and the red trace is the output current at a scale of 50mA per division.


If the IC is heated without a load similar behavior on the EF pin occurs. If the IC is heated enough (does not occur during normal conditions, must use heat gun for an extended period of time) then the EF pin finally goes to ~0V, but as soon as the heat is removed it almost immediately goes back to +5V. Here are a few things I tried:

  • Replacing the IC to ensure it wasn’t damaged. No change in behavior.
  • Adding a capacitor (1nF or 100nF) to the EF pin. This stopped the oscillation but clearly was covering up the issue, the EF pin had an analog voltage on it that varied with how much heat the IC was exposed to.
  • Added larger 10uF ceramic decoupling capacitors and thick 24AWG solid core wire to connect all grounds together and to supply ground. No change in behavior.
  • Used +3.3V instead of +5V to pull up EF pin. No change in behavior.
  • Removed 1nF capacitor between OPA192 output and LMH6321 and replaced resistor with 5.1kΩ instead of 10kΩ. No change in behavior.
  • Tried a different variant of this IC. I noticed we had some older LMH6321 that did not have a pin 1 dimple and slightly different etching on the top, so I tried using one of these to see if there was a change that affected the behavior of the EF pin. No change in behavior.

I read the datasheet, and the EF pin has a 10°C hysteresis so I am surprised that I am seeing this oscillation and that it snaps back so quickly when heat is removed. Here is a snapshot of the layout for what it’s worth. Keep in mind that the EF pin pullup and the series 10kΩ/1nF resistor capacitor combo have been modded-in by me.

Red is the top layer and the vias all connect the exposed pad to a -15V plane. We make another device that uses the LMH6321 and I tried using the EF pin in this design (which had a different layout) and it still exhibited this oscillatory behavior but at a lower frequency.

I was wanted to use the EF pin to tell the main MCU that a fault has occurred but due to this phenomena that doesn’t seem like it will work. What is going on here? What is the expected behavior of the EF pin?

Thanks!

  • 0
    TI__Genius 10490 points

    Hi,

    Thank you for providing all the details to the issue you are experiencing. I do think the issue is due to the device going in and out of thermal shutdown. We have seen this with other devices where the device goes into a loop similar to the one you are seeing. Depending on the operating conditions, a device turning on could increase its junction temperature enough to put it into thermal shutdown. Does this occur when the device is running in its normal condition or only when you use a heat gun and force it into thermal shutdown?

    Best Regards,

    Ignacio

  • 0 in reply to Ignacio Vazquez Lam1
    Prodigy 10 points

    Hi Ignacio,

    Thanks for getting back to me! I suspected the LMH6321 may be going in and out of thermal shutdown but wasn't sure when I realized the frequency was around 600kHz. Is it possible for a thermal oscillation to occur at such a high frequency?

    "Does this occur when the device is running in its normal condition or only when you use a heat gun and force it into thermal shutdown?" Great question, it occurs in 2 conditions:

    1. There are four (4) of these circuits in close proximity, when all four circuits are simultaneously shorted and limiting current then this condition occurs. We believe our circuit needs to tolerate or protect itself from a shorted output.
    2. We apply a heat gun (with or without the device sourcing current.)

    In my above post, I also mentioned another device we designed that also uses the LMH6321. That particular layout doesn't have as good heatsinking and it will occur during 100mA IOUT operation. In summary, if the device gets too hot from sourcing a large current or being externally heated we observe the oscillating EF pin phenomena.

    Hope this helps!

  • 0 in reply to DiodeWhiz
    TI__Genius 10490 points

    Hi,

    The turn-on/turn-off frequency will vary depending on numerous factors however the frequency you are seeing is not out of the question. This loop is due to thermal shutdown which is why we advise customers to keep the device out of these adverse conditions. If only one device goes into this shorting condition, does it also go into this oscillation or is it only when all four circuits get shorted?

    Best Regards,

    Ignacio

  • 0 in reply to Ignacio Vazquez Lam1
    Prodigy 10 points

    Hi Ignacio,

    When only one device is shorted this condition does not occur because it doesn't get hot enough. When all four are shorted, the entire area heats up, and eventually (after about a minute) the LMH6321s enter this condition.

    Also, I noticed that the datasheet shows a 10°C hysteresis.



    I am speculating here, but it seems hard to imagine that the die temperature varies 10°C in less than 1μS. Am I missing something here?


    Thanks,

  • 0 in reply to DiodeWhiz
    TI__Genius 10490 points

    Hi,

    We do know it is from thermal shutdown and will vary depending on things like layout. We have seen devices that heat up 10C from just turning on due to the iq of the device so the jump of 10C when turned on is not unlikely. 

    Best Regards,

    Ignacio

  • 0 in reply to Ignacio Vazquez Lam1
    Prodigy 10 points

    Hi Ignacio,

    So just to be clear, here is a rundown of what is happening;

    1. Due to excessive power dissipation the die reaches 170°C and the device shuts down.

    2. With the output current nearly 0 the die rapidly cools down about 10°C in about 200-300 nanoseconds.

    3. The device temperature is now less than 160°C so the device turns back on and current flows again.

    4. With current flowing the die heats up rapidly and the cycle repeats.

    Does that sound right?



    Is there any litature on these high-frequency thermal flucuations that occur on IC die? I couldn't find a lot of infromation about them on the web.

    The EF pin decreases by about 1V during the 300ns time period. Since this is a 5V logic level, 4V is above the threshold of most typical logic gates. Do you have any suggestions on how this could be used to protect the IC? I was thinking maybe a comparator and a latch that disable the control signal driving the LMH6321?

    Thank you,