Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

High Current SMD Diode Cooling

Status
Not open for further replies.

xieliwei

Member
Hi all, just a quick question!

Imagine I have a design that requires passing about average 8A of current through this diode: https://www.vishay.com/docs/89057/ss10ph45.pdf

Given that this is within specifications, does it mean that I can simply use it? (Granted ambient temperature of 30degC, no forced air and proper ventilation). My guess is at about 7W power dissipation at full load, things will heat up pretty quickly, maybe even exceeding the 175degC limit. Then, how does one cool such a small diode?
 
8A .45V=3.6W from DC loss. I do not know if you will have AC loss.
If you use the recommended PCB pad size then, 60C/watt.
1w=60C+25=85C
2w=120C +25=145C
3w=180C+25=205C too hot!
I would use a large copper area to cool off the diode.
 
Hard to believe but its rated up to 175 deg C.
Add extra copper trace around it to help dissipate heat.
 
I agree. I think it is a poorly written spec. Even with a square inch of PCB heatsink 50C is about it. Maybe an inch under the pad and an inch off of the leads.
 
ronsimpson: Thanks! So that's how one uses thermal resistance! However, going by your method and using Fig 2 (Forward Power Loss), I see I'm in a lot of trouble if my waveform has a duty cycle of less than 50%. Even at 100%, the graph gives about 4.5W of dissipation at 8A. Which will heat things up to about 295degC?

4pyros: I'd think designing it to run anywhere near that would be a problem since most parts (that would be near it) aren't rated to take such a high temperature. One thing I'd like to ask is, how much copper is enough?

It seems quite weird to me, probably because of my lack of experience, to see such a part exist when it seems almost impossible to use it to its full potential without sacrificing large amount of PCB area for the heat dissipation. What advantage does this have over traditional TO-220 packaged diodes? It uses more land space and even costs more than the associated cost of an extra TH component.
 
Don't know what to tell you. But there must be some kind of general design guidelines.
 
The 7W reaches a junction temperature of 60*7+30=450 degrees Celcius, given the recommended pad size, which is a bit high for your diode. You can buy SMD heatsinks, or glue a heatsink onto your component & pcb (using heat-transfer glue). You can also just solder thick copper wires to the diode pads to act as heatsinks (and antennae) (only suitable for prototypes though). You can also get think silicone transfer sheet to move heat between the PCB and enclosure, or from the diode to the enclosure/heatsink.
 
using a pcb pad for a heatsink is only good up to about 1 sq in, plus it helps to remove the solder mask and leave the copper exposed to air. We use thermals vias and many layers with as much copper connected to the thermal vias as possible, I'm talking like 8 layers. Plus thermal pads connected to the bottom side and transferring the heat to the box, we're able to push 40-60W through our boards this way.

To get full use of these parts, you really need to add a heatsink. They give you the thermal resistance junction to Ambient, then you can decide how much power you can push through a part just as it is. If you want to go the full monty, you have to add heatsink/air flow. This is true for many parts, especially power diodes and regulators. The 7805 can do up to an amp, but needing 1.5V overhead x 1A = 1.5W, and you can't push that through a TO-220 without a heatsink + airflow. You're lucky to get 200mA through it w/o one.
 
using a pcb pad for a heatsink is only good up to about 1 sq in, plus it helps to remove the solder mask and leave the copper exposed to air. We use thermals vias and many layers with as much copper connected to the thermal vias as possible, I'm talking like 8 layers. Plus thermal pads connected to the bottom side and transferring the heat to the box, we're able to push 40-60W through our boards this way.

To get full use of these parts, you really need to add a heatsink. They give you the thermal resistance junction to Ambient, then you can decide how much power you can push through a part just as it is. If you want to go the full monty, you have to add heatsink/air flow. This is true for many parts, especially power diodes and regulators. The 7805 can do up to an amp, but needing 1.5V overhead x 1A = 1.5W, and you can't push that through a TO-220 without a heatsink + airflow. You're lucky to get 200mA through it w/o one.
Ok Mike How would you heat sink it?
 
Here's an app note that should help with PCB heatsinking.
 
Ok Mike How would you heat sink it?

I've used something as simple as a metal bracket that came down from the case so that it sandwiched the part when the board was put in the case, using a thermal pad of course because of the tolerances of making the brackets. But, like I said in my post, we use thermal vias to pull the heat to the bottom of the board, then use thermal transfer pads to send it to the case.

BTW, very good appnote Crutschow!!!
 
Hi all, thank you for the very great insight provided! =) Transferring the heat to the bottom of the board and then to the case via vias and thermal pads is a great idea!

I did look into SMD heatsinks but found it hard to believe that gluing or friction clamping on a heatsink provides sufficient thermal conductivity to transfer heat between the package-heatsink interface. I'm used to traditional packages that either require bolting the package to the heatsink or clamping it down hard with some retainer(s).

To elaborate on the background as to what I'm doing, I'm revising an old SMPS design that currently uses a TO-220 packaged diode for the 8A output rectifier (amongst other TH components), in the hopes of shrinking it down. The reason I asked the question in the original post is that I'm unsure if there would be any advantage to switching to a smaller sized, SMD diode if I still have to figure out a way to lose a similar amount of heat.

Mike's idea could be very useful though as it might be possible for me to do away with a separate heatsink by using the device chassis.
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top