so I have created a PCB with a lot of components on it, its a rather large PCB, roughly 18cm x 19 cm. A lot of the components' datasheets (things like Resistors, Switches, capacitors, relays, CAN controller/ transceiver) don't mention the thermal resistance of the components, nor does it include the power dissipation. Does anyone have suggestions to work around this? or generalizations / rules to estimate how much a PCB will heat up.
If it helps, some more information about my board,
Board lays horizontally, 1.6mm FR4 board
3oz/ft^2 thickness, 2 traces run 30A, 3 traces run 15A and 6 traces run 10A. They just run to a connector going off the board, only running through current sensor and relay.
But I am more interested for learning purposes, how thermal changed can be estimated with limited data?Also, do diodes heat up a lot typically and is it a concern for board temperature? because I have 9 small diodes it says their Thermal resistance is like 1000C/W 100mW dissipation
Best Answer
The power dissipation of a resistor is totally dependent on your circuit. Similarly with diodes.
Your question asks about "how much the PCB will heat up" which is a different question from "how hot will my semiconductor's junction get above the PCB temperature".
You can treat the PCB as a flat sheet of material and if you know the total power consumption and other sources of dissipation on the PCB (like those hot traces) you can estimate the temperature rise above ambient assuming natural convection or forced convection or whatever your actual situation is. To get a really accurate number you might have to prototype it or build a mockup, including the enclosure.
Your heat sources like a diode junction will heat above that PCB temperature based on the thermal resistance from junction to case. That may be worth calculating in some cases.
Practically speaking, usually only a few sources account for the bulk of the heat dissipation and only a few of those are worth looking very closely at.
One way to get a handle on the dissipation of individual components is to run simulations, using a program such as (free) LTspice, which will show you the typical dissipation of a part even with a complex waveform. For example, a power rectifier that is passing "spikey" current will have a higher power dissipation than you might estimate from the average current because there's a resistive component to the diode.
If it's just DC like a series resistor you can easily calculate it by hand, or even roughly estimate it in your head, of course. A 10K pullup on an input dissipates only 2.5mW so it can usually be ignored for thermal design under normal atmospheric conditions.