Floating copper on PCB will act like an antenna?

[treez]

Hello,
We had switching node copper on our PCB, which was mounted on a heatsink…the proximity of the switching node copper next to the heatsink caused EMC issues, so we literally cut this PCB copper away from the switching node on the PCBs. The piece of copper is now “floating” (not connected to any node. I trust this is OK? Its only about 5 cm squared.
The product is an offline LED driver. It is linear mode, but has a small 1W Buck SMPS bias supply.

 

[betwixt]

Yes it will make an antenna but if your switching frequency is below about 2GHz and you are using PTFE PCB substrate I wouldn't worry about it!

Brian.

 

[KlausST]

Hi,

I'd like to add my comment on brian's post.

For sure with a 5cm squared copper you won't build a good antenna (for RF communication).
Indeed it may be a bad antenna for your switching frequency.
But:
* it is an antenna.
* EMI is not only caused by your switching frequencies but additionally by all it's overtones. They are all higher in frequency ... and with higher frequency your "antenna" becomes more efficient.
* and EMI is caused by ringing of your circuit. When a Mosfet becomes high/low impedance, when a diode becomes high impedance. The riniging frequency is defined by your design and usually is way higher than switching frequency.

What I want to say:
My personal opinion is, that a 5cm squared copper may develop to a relatively big EMI problem.
This is one cause, why I don't like copper orphans on a PCB.
Even unconnected they get influenced from signals nearby and therefore act as an additional antenna.
Even (connected) copper pour may decrease EMI, because the "connection" often is high impedance for high frequency..means about non existent

My recommendation: (from best to worst solution)
* remove the copper at all (if possible)
* leave it floating
* connect it to GND, shield or case with low impedance (maybe use copper foil instead of a thin wire)

Klaus

 

[FvM]

The copper pour might increase radiated emissions if it's effectively coupled to a "hot" switching node. Is it?

 

[andre_luis]

The copper don't even need o be too close to have significant impact on the board; A long time ago I was assigned to solve a serious problem on an ECG signal acquisition board. Although it had a well distributed GND plan on both sides, and despite there were analog filters, and even other engineer have made an attempt to elliminate the artifact by digital filtering, it still had a strong presence of noise in the printed waveform. The first solution I presented was to place a copper plate between the equipment chassis and the PCB, but it did not work, even moreover worsening the signal quality. The 2nd solution was to just ground this plate to the GND of the board, and Bingo, the noise disappeared. The learning I gained from this experience was that any floating metal object in the nearness of a net can accumulate/reflect electrical charge from the nearby EM source, particularly if its circuit has high impedance paths; therefore, Yes, floating copper can destroy the signal integrity, sometimes it is more appropriate to remove them than to keep them there, specially if overall ground copper plane clearance is short.

 

[treez]

The copper pour might increase radiated emissions if it's effectively coupled to a "hot" switching node. Is it?

Thanks, this copper used to be connected to the switchinbg node of a 1W buck converter bias supply.
-But we cut it away from that because of EMC.
Its now just floating copper...not connected to anything....its not that well coupled to the switching node now....not much stray capacitance between the (now) floting copper and the switching node.

 

[marce]

Switching nodes should be kept small... some of these recent posts do worry me, all seem to refer to bad design practices. A switching node as said should be a large as required and NO larger, you use the thermal pad on most switcher ICs down to ground to get the heat out...