8
\$\begingroup\$

I inherited a power supply circuit design from another engineer, and he used a ground connection method I am not familiar with:

enter image description here

The input comes from an external isolated 48 V AC/DC power brick. The outputs supply several boards within a single enclosure.

My question is about the diode connection between the negative input terminal and the load-side ground.

The effects I see from this are

  • Isolation provided by the DC/DC converters is negated.
  • Negative input terminal will be held within about +/- 0.5 V of the load-side ground.
  • Current might or might not be able to pass freely between the negative supply terminal and load-side ground, depending how the input common mode drifts relative to the circuit ground.

Some other notes about this system:

  • The load-side ground is connected to chassis.
  • Reverse-connection protection on the input is provided by other components not shown here.
  • Several of these systems might be connected in parallel to the same 48 V source.

The question is, what was the design intent of adding the diode connection?.

\$\endgroup\$
  • \$\begingroup\$ Are there any non isolated data lines going between the low voltage side and the 48V side? If so, I think the intent was also to keep the ground levels relatively close to each other. I have seen such design with a device that generated a very large electromagnetic pulse. Without the diodes, there were errors in the high side's received data. \$\endgroup\$ – Lior Bilia Jan 21 '14 at 8:33
  • \$\begingroup\$ @LiorBilia, nope. The 48 V side is just a dumb power brick. \$\endgroup\$ – The Photon Jan 21 '14 at 15:58
2
\$\begingroup\$

This is a trick for clipping away small voltage noise coming into a device via the ground line (e.g. ground loop noise) while still providing a ground connection.

As long as neither diode is forward-biased, the ground is effectively lifted. It's like a voltage-controlled ground lift switch. The switch stays closed against small noises which don't exceed the forward voltage in either polarity.

If there is a fault in the device which causes current to be dumped into the ground, then one of the diodes will spring into action and carry the current. The only problem there is that the diode needs to handle that current without melting.

The "Ebtech Hum X" is rumored to use the same trick.

\$\endgroup\$
  • \$\begingroup\$ That's kind of what I gathered from my research, and it seems to make sense for audio stuff. Does this make any sense for an RF system? The diodes are some 3 A beefy rectifiers (PN when I get to work tomorrow if you care). \$\endgroup\$ – The Photon Jan 21 '14 at 5:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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