I have a 100x100mm board that is directly attached to a same sized heatsink on one side (touching heat producing IC packages). The board is 4 layer, with a large ground middle layer. Is there any reason I should NOT be connecting the heatsink screw hole mount pads to this middle ground layer?
-
\$\begingroup\$ Is it electrically in contact with any other part of the circuit or wiring to it? \$\endgroup\$– Andy akaCommented Jul 31, 2018 at 17:52
-
\$\begingroup\$ Whats sort of IC(s) are you heatsinking? Could you post a link to the datasheet(s)? \$\endgroup\$– Nick AlexeevCommented Jul 31, 2018 at 17:53
-
\$\begingroup\$ Is your circuit kept inside a box/enclosure? Is the enclosure conductive or nonconductive? Does the heatsink stick outside the enclosure or is it inside? Does the heatsink contact the enclosure? \$\endgroup\$– The PhotonCommented Jul 31, 2018 at 17:55
-
\$\begingroup\$ Its not in contact with any other part of the circuit/PCB, and its inside an enclosure...I can make it that the heatsink contacts or does not contact the enclosure. Im heat sinking FPGA type ICs. \$\endgroup\$– electronutsCommented Jul 31, 2018 at 18:23
3 Answers
Is there any reason I should NOT be connecting the heatsink screw hole mount pads to this middle ground layer?
One reason not to is that you may turn the heatsink into an antenna and unintentional radiator if your ground is especially noisy.
This is an unlikely scenario though and is very much dependent on the design. In most cases grounding the heatsink electrically will help cut back on noise.
The other reason is if the heatsink could bleed heat back into the PCB in an area you may not want it.
Something to think about and design for.
-
\$\begingroup\$ How do you see a heatsink, with a thick horizontal base plate, acting as an antenna? I would see it acting as an RF shield. With the components sandwiched between the internal ground plane and grounded heatsink I would expect very little RF interference being radiated or received. \$\endgroup\$ Commented Aug 13, 2018 at 12:43
-
\$\begingroup\$ This is the way I see heatsinks become an antenna. Less likely for an FPGA or micro, but not impossible since sometimes large parts of the whole chip can be switched on and off. Probably better to connect to ground, but if the ground is moving around then it will turn the heatsink into a radiator, this is very very unlikely but it is something to think about. \$\endgroup\$– Voltage Spike ♦Commented Aug 17, 2018 at 5:28
Being a large, conductive part that is not intended to carry current, if it is in proximity to line voltage conductors (in the same case and not mechanically separate for example), it should be grounded to the case, which should be bonded to the bonding(green conductor). If this is the case and it cannot be grounded (eg. being used as a conductor) it should be mechanically separated from the user (protected by casing) and possibly insulated.
Reason this matters is that if you're involving line voltage, the ground for your circuitboard is actually the white wire(neutral/identified conductor), not the green one, and it is important to use the white wire for all current that is intended to flow, and the green wire for all current that is not intended to flow.
If you're designing a device that does not come into contact with line voltage, I'm pretty sure you just need to worry about RF effects and if you want you can use it as a large ground bus.
The internal ground plane should be used as a heat spreader. This means there should be thermal vias from the heat source(s) to the ground plane.
There is no reason the heatsink should not be electrically as well as thermally connected to the ground plane.
Even without thermal vias the internal ground plane will spread the heat to the heatsink mounting holes.
In technical literature is very often recommended to use the ground plane as a heat spreader and the component's thermal pad be connected to the ground plane.