23
\$\begingroup\$

I just received a 120VAC to 5VDC (20A) switching power supply (the kind with a passively ventilated metal casing) for use with a microcontroller project I'm working on (driving long LED strips).

I've wired up Line-Neutral-Ground correctly to a 3-prong plug. It tests out fine (and the case IS bonded to ground internally) but I see the outputs are labeled DC- and DC+ and indeed DC- is not bonded to ground so the outputs are floating (though I'm not detecting a high voltage differential).

Is there any danger in bonding the DC- output to Ground with a short wire, to ensure my circuit ground is also referencing the home ground? I'm not sure simply letting the outputs float is wise, but I don't want any dangerous side-effects if I ground DC-. (The circuit sharing this power will at times be connected to my PC which is itself quite thoroughly grounded so I'm inclined to emulate that behavior.)

Side-note: I have an after-market laptop supply that's two-prong... it works either way, but if I plug it in one way the metal trim on the laptop has an interesting "buzz" to the touch. Not shocking, but definitely noticeable. I suspect that when it's plugged in the right way the output ground is weakly referencing neutral, and the "buzzy" way has the output ground weakly referencing line voltage (it's floating - it'd be quite a bit more than buzzy otherwise). The OEM supply with the 3-prong plug? Fully grounded to the shield of the DC power cord.

So I suspect that when it comes to AC to DV power, getting Line and Neutral wired up wrong is far more risky than letting the output float, and that grounding the output is preferable to letting it float if you're working in an environment where you'll interface directly to grounded equipment. I just want to be sure...

\$\endgroup\$
2
  • \$\begingroup\$ This question is incomplete and misleading. It cannot be answered properly in a vacuum without reference to the REST of the system here. Maybe the complete system should be grounded there at the power supply and maybe that is a terrible idea. We just don't know. \$\endgroup\$
    – Richard Crowley
    Jun 9, 2016 at 3:05
  • 2
    \$\begingroup\$ My question stands - the concern was whether the action of bonding DC- (or for that matter DC+, but obviously not both at once) to the case (which itself is bonded to earth) would cause a power supply like this to act up. Turns out this isn't very unusual for this particular type of supply (versus one where there is an explicitly marked GND and no "DC-"). \$\endgroup\$
    – MartyMacGyver
    Jun 9, 2016 at 6:06

5 Answers 5

Reset to default
12
\$\begingroup\$

Is there any danger in bonding the DC- output to Ground with a short wire, to ensure my circuit ground is also referencing the home ground? I'm not sure simply letting the outputs float is wise,

Assuming the power supply is properly built either is fine.

Side-note: I have an after-market laptop supply that's two-prong... it works either way, but if I plug it in one way the metal trim on the laptop has an interesting "buzz" to the touch. Not shocking, but definitely noticeable. I suspect that when it's plugged in the right way the output ground is weakly referencing neutral, and the "buzzy" way has the output ground weakly referencing line voltage (it's floating - it'd be quite a bit more than buzzy otherwise). The OEM supply with the 3-prong plug? Fully grounded to the shield of the DC power cord.

Power supplies without an earth connection often have problems like this, especially as they get larger.

Capacitors have to be placed between input and output to control EMI. In an unearthed design this ends up with the output being "weakly referenced" to the input. How weak that reference is (or to put it another way how big the "touch current" is) depends on the size of the capacitors. Unfortunately there is a tricky compromise here, bigger caps are better at suppressing EMI but produce larger "touch currents". Larger power supplies tend to suffer worse from this than smaller ones (this is why laptop PSUs from reputable brands are usually earthed while phone chargers are usually not).

In an earthed design this can be mitigated either by tying the output to ground (pretty much universal in in desktop PC power supplies, occasionally seen in laptop supplies) or by splitting the EMI suppression capacitors into two parts, one from output to earth and one from input to earth.

\$\endgroup\$
1
  • 1
    \$\begingroup\$ This "capacitors have to be placed between input and output to control EMI, creating weak reference" solves a few cases that I saw but never got a good understanding why. Thanks! \$\endgroup\$
    – minghua
    Dec 20, 2017 at 21:58
4
\$\begingroup\$

As you are only talking about a 5V supply, then there is little risk from it either way, but in principle; For a current to flow there needs to be a circuit and if the negative is connected to Earth on your mains system then it is also connected to anything else earthed, and to the earth itself, so if you hold the live a circuit will be made through you to the earth. Not a problem with a 5V system and you wearing shoes/ standing on flooring or otherwise quite well insulated, But were it a higher voltage system and you touched the live while leaning on a sink or touching your computer or barefoot outdoors, then a current (relative to the supplied Voltage and the resistance to earth) would run through you.

If you do not connect the floating negative to earth then you can only get a shock by making contact with Both the + and - connections as otherwise there will be no circuit.

Mains powered equipment (in the UK) used to always have a 3 pin connection so that the metal casing could be earthed and any Loose Live wires inside could not make the case Live, but would short to the earthed casing, blowing the fuse. Now most things have Plastic cases, and modern circuit breakers are much more sensitive than fuses, this is not needed and mains connections are often 2 pin.

Grounded metal cases also have a shielding effect against magnetic waves and cannot suffer from Static build-up, which would be a major reason for your computer's metal casing and earth connection.

\$\endgroup\$
4
  • \$\begingroup\$ I see what you mean, except that most common DC voltages are relatively low (24V and below). I'm not sure how a lack of a ground is safer in this low-voltage regime. And if there were a fault (mains voltage on the 5V line or the casing) then you'd be having a pretty bad day if the casing was ungrounded and you touched it (since mains is ground referenced to begin with). And while not everything has a grounded plug they seem prevalent on consumer electronics here in the States (e.g., PCs). \$\endgroup\$
    – MartyMacGyver
    Jan 23, 2013 at 18:18
  • \$\begingroup\$ As I mentioned, devices with exposed metal where there is a possibility of a loose live mains wire reaching it are usually grounded, but plastic casings are common as are floating dc supplies. In the UK, mains power tools are always run from an isolating transformer, specifically to float the power and avoid electrocution. \$\endgroup\$
    – SamB
    Jan 23, 2013 at 22:12
  • \$\begingroup\$ Got it, though I have never heard of isolation transformers being in common use in UK households for normal power tools and such. Double-insulated wiring for two-wire systems like powertools is another thing altogether safety-wise but that's beyond the scope of this question. I've concluded that grounding the DC- of my power supply (tested for leakage current) is safe & reasonable (also my next "bench" supply will be a re-purposed PSU, which will by default be grounded internally anyway). But in isolated projects for deployment I will consider leaving things floating. \$\endgroup\$
    – MartyMacGyver
    Jan 24, 2013 at 9:27
  • \$\begingroup\$ @SamB BS. firstly it's only power tools used on construction sites. Secondly the transformers used are center tapped earth not floating output. \$\endgroup\$
    – Peter Green
    Jun 9, 2016 at 0:19
1
\$\begingroup\$

'Ground' is a relative thing. In the case of home electrics it's either literally a stake in the mud (earth) or the neutral of the three phase to the local sub station (or sometimes both, called PME - Protective Multiple Earth).

Live is relative to the neutral (which is sometimes earth/ground - but not in all countries)

An 'isolated' DCDC, like yours has no connection to live, neutral or ground at the output.

In the real world this is impossible. There has to be a capacitance and resistance between the + and - of the DCDC output and all of the mains connections (live, neutral and earth), however this might (and should be) very insignificant, with resistances in 10's of mega Ohms and capacitances in picoFarads

In your DCDC where you get a buzz, I expect these 'parasitics' are not insignificant, something is breaking down, lifting the DCDCs output common mode voltage through a high impedance to the live. In effect it is still 5V, but it has a common mode AC component relative to ground, on which you are standing.

(BTW this 'earth leakage' is often caused by poorly designed power factor correction circuits)

A simple test for your 'isolated' DCDC is to connect a milliammeter between each of the outputs (+ and -) one at a time to ground (earth). You should see very little current if any. Any more than 1mA or so and there is a problem with earth leakage.

If you see little current then connecting the + or the - to ground is OK.

However I will add that not connecting ground is a really good idea!

A high resistance (or low capacitance) route to ground is a really good way of preventing electrical shock in the event of a fault.

\$\endgroup\$
4
  • \$\begingroup\$ Your statement about not grounding being safer surprises me. What sort of fault would be less dangerous because the ground reference for 0VDC was a poor (high resistance) or absent (floating)? I'd think the opposite - all other things being equal, a grounded device is safer. The only cons I've seen are that some hum could get introduced into the system (considering audio or ADC applications). \$\endgroup\$
    – MartyMacGyver
    Jan 17, 2013 at 10:36
  • \$\begingroup\$ And to add, I read no current between DC+ or DC- and the ground or case. \$\endgroup\$
    – MartyMacGyver
    Jan 17, 2013 at 10:55
  • 3
    \$\begingroup\$ @MartyMacGyver: if you are touching just the live half of an AC outlet, you get a little tingle, but if you also have a good connection to ground, you will get the full power of the outlet, limited by your body's resistance or the circuit breaker. Making a good connection to ground unavailable is one way to avoid this. A GFCI is another way. \$\endgroup\$
    – Phil Frost
    Jan 17, 2013 at 12:45
  • 1
    \$\begingroup\$ So... why the trend towards three prong plugs and grounded electronics (e.g., computers, laptops and monitors)? By this logic those things must be more dangerous by virtue of their "earth"-grounded chassis and circuitry. \$\endgroup\$
    – MartyMacGyver
    Jan 17, 2013 at 19:19
0
\$\begingroup\$

Metal enclosure or internal metal shielding "bonded" (electronics speak not electrician speak) to PE (protective earth conductor, green wire on a three prong US plug).

Long way of saying, float whatever you want but shield the PSU. Grab an EMF meter and poke around a bit, you will see why.

\$\endgroup\$
-1
\$\begingroup\$

I would not bond the dc ground to frame ground on smp,it can cause noise issues and some SMP can create even more EMI issues by doing this. The idea is the dc side of connect pcb floats,if it mount in a metal case the frame ground bonds to case as well. The idea is if lightening hits out case it will route thru the bonded ground and not around your circuit DC ground causing more damage. THe smp prob has a 6kV flash over point. You must also make sure that any jack sockets that are ref to DC ground do not touch the chassis ground via insulators. You may also have to fit external 22nF 310V AC 3-6kV (up 15V) or 680nF 310AC 3-6KV(up to 30V Dc) on both the DC pos to frame ground and DC neg to frame ground(mains earth) You can check with DVM on AC and see how much AC leakage there is on the DC output side you will be surprised! TDK Lambda fit these capacitor internally in there SMP but other don't due to cheapness and a general lack on information on this subject. These new Fully Isolated SMP are not documented very well generally and on internet and mis-information is plenty on internet. A lot of Industry electronics use these isolated ground system nowadays but its not very well known and documented,it took me ages to solve AC leakage on DC outputs and ensuring DC neg is not bonded in anyway to frame ground and caused all manner of problems with CPU crashing and ADC picking up all sorts of crap due to SMP increasing EMI issues,its isolated for a reason. If you still insist in grounding dc neg on a metal case,I would totally isolate the SMP so Earth goes straight to smp and mount SMP on nylon standoffs,you can then ground the DC neg to frame of enclosure. On a old linear PSU the frame ground was often commoned up to DC negative. I have checked and PAT tested include HV flash test and all passed fine with above system. I would also recommend a AC line filter to your SMP as well and a 2 turn ferrite ring on DC side,it good practise.

All the best

Rob

RP Comms & Design UK

\$\endgroup\$
2
  • \$\begingroup\$ Also copper shield box helps if its a open frame type to help reduce RFI emissions. \$\endgroup\$
    – rob
    Jun 9, 2016 at 0:14
  • 1
    \$\begingroup\$ As it is, your answer is really hard to read. Please break up the text so we can breathe and emphasize the main points. Also the greeting and signature at the end is superfluous. \$\endgroup\$
    – dim
    Jun 9, 2016 at 9:46

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.