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.

# LM3480: Problem with the LM3480 at 70°C

Part Number: LM3480

Hi,

I am using a LM3480 on a board to generate a 3.3 V power supply with a 12 V power supply.

While testing several of my boards at 70°C, I have problems with some of them for which I was able to see on a scope the power output of this LM3480 falling down below the minimum power supply of the components it is expected to power up.

For an ambient temperature of TA=70°C, I measured a TC(max)=88°C on the LM3480.

Since, from the datasheet RθJA = 269.6°C/W and RθJC = 141.1°C/W, RθCA = RθJA - RθJC = 128.5°C/W.

I am then expecting to have P = (TC – TA) / RθCA = 140 mW and TJ = TA   RθJA * P = TC   RθJC * P = 107,76 °C.

Since the recommended temperature of the component is from TJ = -40°C to  125°C, I don't understand why its output is falling down.

Its consumption doesn't seem to be excessive as well.

Could you tell me if I missed something?

Antoine.

• Hi Antoine,

You mention that you were able observe the Vout on a scope. In order to help debug, could you provide a scope shot of Vin and Vout? Also, what is the output current?

Very Respectfully,
Ryan
• Hi Ryan,

If I just provide a scope shot of Vin and Vout it won't tell everything, because my problem is that it is working at the beginning with Vin=12V and Vout=3.3V at 70°C.

But after a while, when the component takes temperature, I can observe that Vin=12V and Vout falls in some minutes from 3.3V to 1.7V.

Instead of a scope shot, I did a video where you will be able to see Vin and Vout.

Ch1 = Vin = 12V and Ch2 = Vout but with an x10 probe (i.e. 1V/tile).

Since video quickly take space, once I got the real one I made a second one by filming a quick version on my lap top.

Please find the fast version, the time on the scope get the real time.

I made computation to estimate the output current : 31.7 mA with an ISO7242 on the board or 14.2 mA with an ADuM1412 on the board.

Please find the Excel file with those computations.

I had one board with a good LM3480 (i.e. a LM3480 working fine at 70°C) and a bad LM3480 (i.e. a LM3480 with a falling Vout). I exchanged them on my board and the defect follows the bad LM3480 (i.e. it is not related a bad component powered by this LM3480).

My board manufacturer also took a bad LM3480 and only loaded it with a 100-ohm resistor (i.e. output current = 33 mA). He observed the same Vout falling with this resistor as a load.

Best Regards,

Antoine.

Alim_courant_OPG.xlsx

• Hi Antoine,

Thank you for the video! While this is not a traditional thermal shutdown which would have looked like a square wave on your output, I agree that the video clearly shows LM3480 limiting the thermal dissipation.

Due to the amount of power being dissipated in your application you are reaching the point where LM3480 actively limits its junction temperature around 150C. Please note that LM3480 is designed, characterized and tested for an operating junction temperature maximum of 125C.

Since you have ample headroom, you could reduce the power dissipated in the LDO. You can conserve power by using a dcdc before the LDO to provide a mid rail while still leaving enough headroom for the LDO to operate. Another common option is to place two LDOs in series. While this will not save you any power, it will spread the dissipated heat and provide your system higher PSRR. A third option is the simplest to implement; however, it will only spread the dissipated heat. You could place a series resistor before the LDO.

Very Respectfully,
Ryan
• Hi Ryan,

From the behavior of the LM3480 output that we can see on my video, this is what I was expecting: the LM3480 is actively limiting its junction temperature. However I still have two problems:

- The first one is why is the LM3480 actively limiting its junction temperature? As you explained and as I could see it in the datasheet of the LM3480, the LM3480 should be able to work fine up to 125°C junction temperature. If we make case temperature measurement on the LM3480 and try to compute its junction temperature, I found 107.76°C, as explained in my first post. (See the attached picture. From the datasheet, RθJA = 269.6°C/W and RθJC = 141.1°C/W, so RθCA = RθJA - RθJC = 128.5°C/W and P = (TC – TA) / RθCA = 140 mW and TJ = TA   RθJA * P = TC   RθJC * P = 107.76°C.) I am then expecting the LM3480 to still be more than 15°C below the junction temperature it is expecting to move into protection... I don't understand why it is actively limiting its junction temperature...

- The second one is that I agree with you than generating a 3.3V with a 12V and an LDO without a serial input resistor is not a good idea. However, this a design I inherited from a previous version of the design of my board. Up until now it worked fine (probably because since there are several sources for some of the components powered by the LM3480, I probably had the ones consuming less in the previous productions of the board). Since this board is in production, I am not expecting to redesign it right now. So I can't use a DC/DC or another LDO or a serial resistor before the LM3480. If I want to replace the LM3480 with a DC/DC (which seems more appropriate to generate a 3.3V with a 12V), this DC/DC should fit in the SOT23 foot print of the LM3480... I have not found such a component. Do you have any suggestion or any other idea to solve my problem?

Best Regards,

Antoine.

• Hi Antoine,

The Thermal Information numbers are typical and assume a JEDEC board as described in the following Application Report:
www.ti.com/.../spra953c.pdf

The typical LM3480 will limit its junction temperature to 150 C. It is possible that the devices you are using are from a lot that limits its junction temperature a bit lower.

Unfortunately I am unaware of a pin to pin replacement for LM3480 that will allow the amount of power dissipation your application requires while remaining within the recommended operating temperature.

Very Respectfully,
Ryan