Electronics > Projects, Designs, and Technical Stuff

Ground loops inside an instrument

(1/1)

ezalys:
Hey all, I'm designing a piece of instrumentation, and to keep it understandable and easy to debug, I'm building it as a collection of boards I'll be connecting together using SMA cables. If a board accepts both power connections and an SMA cable, or two SMA cables, I've now made a ground loop. I've seen plenty of T&M equipment that features multiple SMA cables connecting boards, so what gives? Do we just not worry about ground loops with such tiny loop area? Should I decouple the connections in some way?

profdc9:
I think there's a number of ways to handle this situation depending on exactly what kind of signal you expect to radiate or be susceptible to, and the current or voltage drops in the supply.

There are no universal rules here, but in general, the low frequency loop (power loop) is laid out in a "star" ground with the currents for the power going to/from each board to the power supply which is at a central point to prevent ground loops for the power.  A star grounding pattern can be applied to high frequency signal loops as well as long as the currents for that loop can be separated from the power supply loop so that transients in the power supply current are isolated from the signal loops, which requires good local bypassing of the power supply on each board.

For example, you have one current return loop for the supply power pins and ground, and a second loop for the signal in the coaxial cables.  If the supply power pins are locally bypassed with low-impedance capacitors, then the loop for the power is shorted out for high frequencies and so that loop does not influence the signal carried on the coaxial cables.  Likewise, if the signal on the coaxial cables is AC coupled using capacitors, this reduces the current in that loop due to power supply transients.  This way you can isolate the two loops from influencing each other (power loop and signal loop).

If you have a metal chassis, the ground of each board can be capacitively coupled to it at each board.  Capacitive coupling prevents the power supply current from returning through the chassis but allows the high frequencies to return through the chassis.  The metal chassis then serves both as a shield to prevent pickup in the loop antennas formed by the ground connections and functions as a low impedance return.  The power supply ground can be tied directly to the chassis, but only at the power supply at one point.  If the chassis is a sufficiently low impedance connection and the wavelengths are large compared to the chassis, then you can probably just route all return currents through the chassis and just not worry about separate loops like it is just another layer of the board, but this really depends on the magnitude of the signal transients and how closely the chassis can be coupled to the circuit.

If you don't want a metal chassis, you can try to run the power and supply cables along side each other to minimize the loop area and connect the grounds of them both together at both ends.  You could also put ferrite beads over the cables to suppress the common-mode currents being picked up by the cables.  But in general a metal chassis is going to make preventing noise pickup and minimizing loop area a lot easier.

 
 

David Hess:
Ground loops are a major problem which needs to be considered with broadband equipment like oscilloscopes which operate down to DC and like you suspect, special provisions have to be made between assemblies and modules to suppress common mode currents.  But most equipment which uses RF internally does not operate down to low frequencies making this less of a problem.

Vovk_Z:
I can tell my experience about ground loops:
I actually use boxes about 20×20×9 cm (or a bit larger) and module-type internal design. And I can tell if modules itself are done right and made with 'low-noise design type', and signal power supply is low-noise - then there isn't a problem with small ground loops (which are made by signal and power ground wires).
I had problems in previous, but then I started to read noise bibles like "Noise reduction techniques in electronic systems" H.W.Ott,  etc (can't remember now all book list), and now it is ok.

Navigation

[0] Message Index

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod