OPA547: OPA547F/500

Hey Akshay, 

What are your supply rails? Are you soldering the thermal tab to V-? Ideally, this thermal tab should connect to a large copper pour to help dissipate heat. 

See figure 4 of the datasheet for more information:

There are some conditions in which an external heatsink is required, but in general it should be possible to operate the device like figure 4 shows. 

What is the VOUT voltage when you are sourcing 240mA? 

Remember, power dissipation is going to be dominated by the difference in voltage from V+ to Vout. 

If we cannot alter the copper pours, then you will need to place a makeshift heatsink on the top of the device, but the thermal performance will likely not improve significantly due to the higher thermal resistance from the top of the package. 

Feel free to send over the PCB layout pictures of the design. I can review these and tell you what I think we should do. 

Thanks,

Jacob

Hey Akshay, 

It is good that you are using single supply, this means that the thermal tab can be easily connected to the ground planes on your board. 

This is the most common method for removing heat form the device. A top mounted heatsink can help, so it is good to hear that you were able to install one in your application. 

Please let me know if you are able to obtain the design files. Pictures of the layout are okay as well. 

In general, I am looking to see how large the thermal plane is. Per my previously included figure 4, you can see just how impactful the plane is to the thermal performance of the device.

Hi Akshay, 

Thanks for sharing the pictures of the layout, this is very helpful for me to see.

We certainly are on the smaller side for pour dimensions, but the thermal vias also help to add surface area. 

Assuming a Vout of 15V with ~250mA current draw, we will have a power dissipation of about (30V-15V)*.25A = 3.75W. If Ta is 25C and we want to be less than 150C Tj, we need an effective theta Ja of (150C -25C) / 3.75W = 33.333W/C. Per figure 4, we need about 1.5 square inches of copper to have sufficient heat transfer. With some design headroom, I would want to be at least 2.5 or even 3 inches of copper pour. 

Elevated ambient temperature would make this even harder on the device if Ta was not 25C. 

A top mount heat sink can help improve the performance, but the best bet may be to rework the copper pours. 

Is the system working as desired with the top heatsink?

Thanks,

Jacob

Hi Akshay, I do not know of another device that fits this footprint. 

What heatsink are you using? a part number can help me figure out the thermal performance. 

If you have good thermal conductivity to your PCB, you could even mount another small heatsink on the bottom of the PCB. 

A HS like this may work for the bottom of your PCB:

A similar style but smaller XY dimension could work for the top side. 

What is this net circled in green?

Is this not a GND plane? 

Some customers utilize blower fan style heatsinks for our larger devices. This type of implementation could help lower the thermal resistivity to the junction. 

When the device enters thermal shutdown, is the PCB hot in the thermal pad area? IF it is not, it may be an indicator that the thermal tab is not sufficiently connected to the thermal pad. 

Thanks,

Jacob

Understood on the thermal performance, I see that these are V- pours now.

Admittedly, the OPA547 is probably one of the best devices to use if you want ~250mA out of the device. OPA551 is a similar device, but I would still likely take the OPA547 in my opinion. 

Alternatively, something like OPA593 may be possible, but we will not have much margin if you really need 240mA. 

Reworking the board, or going to the through hole variant with a heatsink may be the best option. 

If you were able to add larger ground pours, we would have acceptable levels of power dissipation. 

 Thanks,

Jacob

Hey Akshay, 

I would use the OPA547 and make the thermal pour larger to help dissipate heat. 

I can review the design before we send to fab.

Thanks,

Jacob

Hi Akshay, 

You are correct in your thinking, the high cap load will make this more suited to a DC to DC converter. 

Thanks,

Jacob