A DC benchtop supply is usually ground free, i.e. the negative output is floating wrt earth ground. I used to use HP E3630A which has an additional "Earth Ground" pin and to which I shorted the "-" pin.

Now I have a power supply (Keysight E3648A) that has only four outlets: "+", "-" for each "Output 1" and "Output 2". I need +/-5V for testing a PCB so I put them into track mode and short "Output 1 -" with "Output 2 +", calling this "ground".

This "ground" and the supply voltages go to my test PCB which has its grounds (indirectly) connected to other devices over SMA cable shields (signal generators, signal analyzers).

Now I am not sure how to best connect "ground" to earth ground or if I should keep it floating in the first place?

Interestingly the manual does not discuss earth ground at all and how to properly connect it. Any recommendations?


2 Answers 2


"Ground" is defined as 0V reference and used for measurement purposes.

"Earth ground" is used primarily for safety and bypassing common mode noise currents.

When you want to connect "Earth ground" to your local "ground" then this point becomes your 0V reference. Generally this done at the point of power entry on the board but it may also become connected to coaxial chassis mounted connectors etc.

Just keep in mind any power line filter Y cap to earth ground and/or shield currents to earth ground may influence your local ground.

It is usually OK to connect earth ground at the PS Return by a jumper except when RF is involved. This external low impedance "earth ground" connection may serve to attenuate high impedance stray coupling from sources of EMI.

  • \$\begingroup\$ Thanks - the main question is still open: Is there any issue NOT connecting local "ground" to "earth ground" at all? Wouldn't I risk ESD events when touching local "ground"? (ESD strap and mat are connected to "earth ground") Can you elaborate on the "except when RF" part/maybe you have a reference? Because this is exactly the case for me. I am actually already a bit concerned about ground loops too since my board ground is connected via SMA shields to multiple measurement boxes as well as to the benchtop supplies. \$\endgroup\$
    – divB
    Jan 3, 2018 at 0:58
  • \$\begingroup\$ In order to avoid ground loops in RF shields yet make local ground safe for ESD discharges as well as provide a low impedance for RF, often a 1nF Cap is used for coupling the two grounds. \$\endgroup\$ Jan 3, 2018 at 1:53

The main issue with NOT connecting two grounds together is usually surge generators or lightning generators which can generate thousands of volts and/or thousands of amps. In which case it has its own grounding rod to Earth ground it. But to safely measure the surge voltage we had to use differential probes rated at 40 kV -3 dB at 75 MHZ. The earth ground only provided a 'static' reference to ground. During a surge event it could have as much as -1,800 VDC on it because it was a 15 feet run to the ground rod, which acted like an inductor for the 20 uS event.

The bottom line here is that your source of signals determines whether:

1) You should add a Earth ground to your local AC outlet.

2) Accept the signal ground on your equipment as being good enough.

3) Or build an isolated Earth ground and make all measurements with differential probes.

An isolated Earth ground is not needed unless you are building or using surge generators.

Most of the time you will be using signal generators or DIY circuits that have a floating ground, unless your power supply has grounded common terminals. Either way you have at least a valid signal ground so your instruments should give you valid readings.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.