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.

TLV760: RθJA or RθJB - which thermal resistance to consider

Part Number: TLV760
Other Parts Discussed in Thread: TLV755P

Hi there,

for powering two current loops, I need an LDO for Vin = 24 V and Vout = 12 V (i.d. Vdrop = Vin - Vout = 12V). I suggested the TLV76012 and checked power dissipation as follows:

I studied http://www.ti.com/lit/an/spra953c/spra953c.pdf "Semiconductor and IC Package Thermal Metrics" and the paper seems to support the use of Junction-To-Board resistance while it does not explicitly favor one metric over the other. I calculated the nominal load current to be I = 2 x 20 mA = 40mA. Worst case load would be Imax = 2 x 30 mA = 60 mA (short circuit current limited by current loop current limiter). So I calculated the temperature difference between board and junction to be:

dT= RθJB * Imax * Vdrop  = 56.8 °C/W * 0.06 A * 12 V = 40.1 °C

Assuming a board temperature of 30°C I found that the device would be operated within specification.

However, my colleague pointed out that I should rather use the much higher RθJA for this calculation as detailed in the specification note.  RθJA  is 275.2 °C/W and using that value results in dT= 200 °C which would be out of specification. So he'd suggested to rather use a 7812 with TO package.

Which way is the correct design procedure?

Thanks.

Dan

  • Hi Daniel,

    The best thermal resistance metric to use is dependent on which temperature you actually know. Using the junction to board thermal resistance, Rjb, is great to use when you have an actual PCB in your hands which you can physically measure the temperature (Tb) of the PCB right next to the IC, it is not the enough to know the average temperature across the entire PCB the measurement must be made right next to the heat source which is the IC. If you don't have an actual PCB yet this can be done during the design phase if you have the thermal modeling tools required to simulate the temperature of your system, however this is not very common due to few people having access to such tools. 

    If you do not have a good estimate of what the temperature will be right next to the IC then it is often best to use the junction to ambient thermal resistance, Rja, since you probably have a good idea of what the ambient temperature of your application will be. Rja is the most commonly used thermal metric since the ambient temperature is one of the few system level temperatures that most system designers have a good estimate for. 

    Now one thing to note is that Rja is very influenced by the actual PCB design. What you see listed in TI datasheets is the JEDEC standard for reporting thermal metrics using their standard High-K Board. The JEDEC standard is decent for dissipating heat however it is not optimized specifically for it. We actually published an app note showing how various PCB designs changed the effective Rja on those different PCBs and found a PCB optimized for thermals can decrease Rja listed in the datasheet by 30%-55% allowing for more power to be dissipated in real applications which design their board with thermals in mind. 

    Here are some additional links which may be helpful: A quick video on LDO thermals and a thermal calculator based off PCB copper area 

    I hope that is helpful, let me know if you have any other questions.

  • Hi Kyle,

    Thank you for your helpful answer and sorry for not getting back to you earlier.

     As an example the TLV755P (SOT-23 package) has been tested in the app note you mentioned (page 10). The Rja from the datasheet is 231 K/W. Using an optimized board layout the app note reports a reduced Rja of 108 K/W. This value is still quite far away from the Rjb of 64 K/W given in the datasheet.

    Thus my lesson to take home is: Rja (given in datasheets) is a rather conservative value (so one errs on the right side when using it). Rja could be reduced to about 50% of the datasheet value using a board layout with thermal considerations in mind. Generally. Further reduction of Rja is possible by using a different package (bigger, thermal pad). Rjb should not be used to estimate thermal performance of such a device.