AC to DC Power Supply Ground Connection

[j777]

Hello,

I opened up an AC to DC power supply today to see where they connected the Earth ground wire and I found that it's connected to the negative side (labeled V-) of the DC output. V- is also connected to one side of the transformer's output. Why would they connect the Earth ground wire in this way?


Thanks

 

[berkeman]

If you don't need to have an output voltage (+/-) supply that is floating with respect to Earth ground, then that is the simplest connection.

 

[j777]

What is the advantage of having an output voltage that is floating with respect to Earth ground? Is one more robust (less noise on output, less susceptible to surges, anything that might be more desirable) than the other?

For a supply with the output floating with respect to Earth ground how do the electrical connections differ?

 

[Reality_Patrol]

What is the advantage of having an output voltage that is floating with respect to Earth ground? Is one more robust (less noise on output, less susceptible to surges, anything that might be more desirable) than the other?

For a supply with the output floating with respect to Earth ground how do the electrical connections differ?

If the DC output of a power supply is "floating" then it can be connected in series with other "floating" supplies to get higher voltages. This can be a simple way to generate controlled higher voltages from "low voltage" supplies in the lab.

The Earth ground can be connected to the power supply chassis (metal frame) in either case for safety. So, the only real difference in connection between the 2 types is that the floater doesn't have an output terminal connected to Earth ground.

 

[j777]

If connecting floating supplies in series is not desired should the 0V rail always be connected to Earth ground at the supply? How does this connection effect return current?

 

[berkeman]

A disadvantage with making the Earth ground connection would be in the special case of multiple power supplies sharing a common ground output, and supplying power to different parts of a distributed thing (like a room-size machine, etc.). The distributed Earth ground connections would provide a parasitic path for Earth ground currents to flow, and could add noise to the +/- power supply seen at various points. In the case where significant DC or 50/60Hz Earth ground currents can flow, it is best to use floating supplies sharing a common floating ground connection.

 

[j777]

So for single supply systems is it best to make the Earth ground connection or are there still reasons to use a floating supply?

I'm trying to understand when and why you would use one style supply over the other...

 

[berkeman]

For a general lab power supply, or a packaged OEM power supply, you will generally want to give the end user the most flexibility, so you will float the outputs from Earth ground, and provide a banana jack or spade lug or whatever to make it easy to strap V- (or V+) to Earth ground.

For circuitry in situations where significant Earth ground currents can flow (like on electric-powered Subway trains in New York), you will want to isolate all logic grounds from Earth/chassis grounds, and maybe only connect the two occasionally in the devices with a charge bleeder resistor like 1Meg in parallel with a small high-frequency cap like 1000pF.

For devices that don't fit the above, you will generally get the best EMI performance and ESD immunity if you hard tie Logic ground and Earth/chassis ground with a low-inductance connection (like a short metal standoff between a PCB plated-through hole and the metal chassis).

Situations vary, though, so you need to look at the noise environment and end use of the product, in order to determine the best overall grounding scheme.

 

[j777]

Thanks, that information is very helpful.

Is it best to hard tie Logic ground to Earth/chassis ground at the power supply or at the PCB being powered? I've seen Logic ground connected directly to plated through holes on PCBs (TS-7400 SBC made by Technologic Systems) and maybe the reason they are designed that way is to make it easy to make the Logic ground to Earth/chassis ground connection. If the connection was made at a PCB like the TS-7400 I would think it would be a good idea to only connect one plated through hole to Earth ground to avoid ground loops or is there a scenario when multiple connections would be desirable?

How is EMI performance and ESD immunity improved by making the Logic ground to Earth ground connection?

I hope you don't mind all the questions; I learn best when I have all the angles covered in my mind. On that note can you recommend a good book or other resource on these types of things?

 

[berkeman]

It is generally best to use a "star ground" configuration, where there is a single tie point between logic and chassis ground, near where cables enter the device. See the section on "PC Board Layout Guidelines in chapter 4 of this network transceiver databook, for example:

http://www.echelon.com/support/documentation/manuals/transceivers/005-0139-01D.pdf

The classic textbook on EMI is Ott's "Noise Reduction Techniques in Electronic Systems". Your technical library should have multiple copies.

 

[j777]

If the device's logic ground is connected to this chassis ground, then connection should only be made at this
single point in the center of the star ground.

Is it assumed that the chassis is connected to Earth ground at the center of the star ground or is it acceptable to connect logic ground to the chassis at the center of the star ground and the chassis to Earth ground elsewhere?

 

[berkeman]

Is it assumed that the chassis is connected to Earth ground at the center of the star ground or is it acceptable to connect logic ground to the chassis at the center of the star ground and the chassis to Earth ground elsewhere?

In general, if the metal chassis is wide between those two points, it would be okay (and is actually done fairly often). It wouldn't be a good idea, though, if ESD/Surge currents clamped to chassis ground had to run a ways through the chassis right next to the bottom of the PCB, for example, in order to get to the Earth connection out of the device.

 

[j777]

I've spent some time now sifting through Google results for information regarding this topic and your answers make the most sense to me. Unfortunately what I'm finding is that applying it to a real-world application is not all that straight-forward. I'd like to explain the present application in which I'm trying to apply what I have learned to and see what you think of the solution that I've come up with.

The project involves a control enclosure and a power supply enclosure (both plastic) located about 1ft apart. The power supply enclosure contains an open frame AC to DC power supply with a floating output and the control enclosure contains 2 PCBs on a PC/104 style stack. The top PCB contains a 5V switching regulator, IOs, and a separate analog section which has an ADC used to read 2 load cells (the strain gauges I mentioned in a previous post). The bottom board has a microprocessor, Ethernet, and an SD card.

Based on what we have discussed I need to connect the logic ground of the 2 PCBs to Earth ground. The chassis is basically the aluminum plate in the bottom of the control enclosure that everything is mounted on. The best way that I can think of to make all of the required connections (logic ground to chassis ground to Earth ground) would be to use the PCB mounting holes to connect to the chassis and then run an Earth ground wire from the power supply enclosure to the control enclosure and connect it to the chassis. Since all of the mounting holes are connected to logic ground I can either connect all of them to the chassis or connect only one of them (use nylon standoffs for the others) to facilitate a star ground scheme. I'm somewhat undecided on the best way to go with the standoffs (connect all or only one to chassis?) after examining how it is done in a desktop PC. Using my DMM I found that in a desktop PC the chassis, logic ground, and Earth ground are all connected in the power supply. The motherboard's mounting holes are all connected to logic ground which are then connected to chassis ground by the mounting screws.


What do you think of grounding scheme I've described above?

What would you recommend I do with the standoffs (another potentially important thing to note is that connecting 1 standoff still means making the connection for both boards since they are stacked)?

 

[berkeman]

It would help if you could post a picture or sketch of the setup. Also, what DC voltage goes from the power supply to the PC/104 assembly? What ESD protection circuitry is on the PC/104 PCBs for any wire entries or other user-accessible (hit-able) buttons, etc.?

My initial recommendation would be to use just the single conductive standoff stack for the tie, and make that at the PCB hole closest to where the power supply cable enters the stack. If there are IO cables all around the periphery of the PCBs, though, then I'd vote to tie to chassis at all of the PCB holes, to better allow ESD clamp currents to get out to chassis without going through sensitive circuitry on the way out.

 

[j777]

Sorry for the lack of details.

The power supply has a 12V output. What follows is a list of connections that enter the top PCB on the stack and the ESD protection (if any) for each:

- 12V power supply input has a unidirectional TVS diode (PN: 1SMB12AT3G)
- 9 digital inputs used for inductive proximity sensors and photoelectric sensor have 10K series resistor followed by a diode clamp (PN: NUP4301MR6T1G)
- 1 digital input for user-accessible button has a 10K series resistor followed by a diode clamp (PN: NUP4301MR6T1G)
- Analog inputs for 2 load cells have 5.49K series resistors. These inputs are connected to 2 LT1167 instrumentation amplifiers and according to datasheet the series resistors bring protection to IEC 1000-4-2 level 4 spec.
- 10 Digital outputs which use ULN2003A Darlington Transistor Arrays to sink up to 400mA
- 14 pin header for character LCD which is sealed inside of enclosure and viewed through transparent enclosure lid

There are 4 mounting holes total but the only ones connected to logic ground are the ones labeled in the image.

The only connection entering the bottom PCB is Ethernet and no special ESD protection in addition to the transformer is provided.

The top PCB has a CPLD used to handle the digital inputs and outputs which is powered by a 3.3V linear regulator.

See the attached image to get an idea of what the layout of the top PCB looks like and where the ESD protection exists in relation to the mounting holes.

 

[berkeman]

First cut -- I'd use conductive standoffs through holes 1-3, and use non-conducting standoffs for the Analog section's mechanical support (its one hole).

 

[j777]

Thanks for looking at the details of the design. I plan on not having the mounting hole in the Analog section connected to logic ground so I'll just use conductive standoffs for all of the holes.

Back in post #5 I asked about the path that return currents will take when logic ground is connected to Earth ground; I'm still not very clear on this. I understand that this connection makes the logic ground Earth referenced but does the current sourced by the power supply return on neutral or Earth or both? If ESD would take the Earth path then wouldn't all return currents flow directly to Earth ground instead of taking neutral to Earth ground also?

 

[berkeman]

Back in post #5 I asked about the path that return currents will take when logic ground is connected to Earth ground; I'm still not very clear on this. I understand that this connection makes the logic ground Earth referenced but does the current sourced by the power supply return on neutral or Earth or both? If ESD would take the Earth path then wouldn't all return currents flow directly to Earth ground instead of taking neutral to Earth ground also?

I believe that you said the 12Vdc power supply's outputs were floating with respect to Earth ground, so the supply current for the logic will flow out the + terminal and back into the - terminal, with no inclination to use Earth ground as a return (even though Earth ground and logic ground are tied at your PC/104 device. ESD and Surge hits, however, are trying to get back to Earth, so those currents will flow out of your PC/104 device and return via the separate Earth ground wire in your connection cable, and through the chassis of the power supply.

 

[j777]

Thanks. That takes care of most of the holes in my understanding of this issue except one... Why does the supply current for the logic have no inclination to use Earth ground as a return while ESD does?

 

[berkeman]

Thanks. That takes care of most of the holes in my understanding of this issue except one... Why does the supply current for the logic have no inclination to use Earth ground as a return while ESD does?

Because the output of the power supply is floating with respect to Earth ground. There is no path back to the V- output of the power supply from Earth ground (at least in the system the way you have described it).