LMX2595: heating issue

Part Number: LMX2595
Other Parts Discussed in Thread: LM2595,

Hi Experts,

I am using LM2595 for generating a fixed frequency. I am attaching settings on TICS Pro and registers which i program in IC. 

I am getting correct frequency, correct power from LMX2595. The problem  is the IC heats up very much during operation and power starts decreasing over time as circuit heats up and finally after 20-25 minutes, it stops giving frequency. What can be problem? Here is my registers file : 

HexRegisterValues.txt 

LM2595 settings.jpg

Regards,

Marvin

  • Hi Marvin,

    With this configuration, the device on the eval board consumes 425mA current, the device should not be extremely hot and should work normally. Did the the exposed DAP have sufficient via connecting to the GND plane of the pcb? Did you check the power supply? 

  • Hi Noel,

    Thank you for waiting. Please see reference below:

    LMX2595.pdf

    Please ignore refrence designators on PCB. They dont match to schematic. When LMX2595 is programmed by microcontroller, it takes around 250mA current. But LMX2595 heats up.

  • Hi Marvn,

    your OSCin circuit is not correct. Shunt resistors for the HCSL clock are missing.

    It is a bad idea to use linear regulator to convert 12V to 5V. In this case, the LMX2595 consumes 450mA, the power wasted in the regular is 3.15W, it is going to be very hot. 

  • Hi Noel,

     I see that connection to osc and LMX2595 is problem. Do you think it should contribute to the excessive heating of LMX2595? Also, though it can be a problem, but LMX is outputting correct frequency with required power at beginning. With temperature rise, the power goes down slowly. after this fact (LMX correct frequency output),Do you still think that only this is cause of problem ?

    Secondly, I am using linear reguators to convert 12V to 5V because of space constraint on PCB. For switching regulator, I cant afford bulky inductors on PCB. And BTW, there is no heating in both regulators.

    Regards,

    Marvin

  • Hi Marvn,

    The heat is definitely came from the LDO, I forgot to count the PA, it consumes another 100mA, so the LDO must be hot.

    BTW, the PA is rated for Vcc=4V but you gave 5V to it.

    The unused differential pin should be 50Ω terminated.

    in the pcb layout, make sure you have sufficient via to distribute the heat of LMX2595 to the pcb.

  • Hi Noel,

    For the points you have mentioned, I have following response.

    1 - The regulator is on separate PCB so there is no way that LDO heat is transferring to LMX2595.
    2 - The PA has absolute max rating voltage of 6V. So its working fine and not blowing up. Its also not heating up as checked by me.
    3 - Unused pins are terminated on PCB but not visible in schematic. Please ignore this mistake in schematic.
    4 - I am attaching picture of PCB in which you can see that there are enough vias to distribute heat of LMX2595.
    5 - the problem of terminations on OSCin Shunt resistors for the HCSL clock are missing. I have soldered 2 resistors as shown by you previously but LMX2595 is still heating.

    Do you see any other point which can cause this heating?

  • Hi Marvn,

    How many layer does this board have? Do you have a big GND copper plane to dissipate the heat from LMX device?