PSU Manufacturer - Is this acceptable?

[ssashton]

I purchased 100 PSUs from 'HK Ying Hui Yuan Group'. I specified a 3 pin output connector with V+ pin1, V- pin2, GND pin3. I confirmed it 3 times.

They joined V- and GND inside the PSU and connected pins 2 & 3 to the same wire.

[Miyuki]

The question is if that ground is connected to input ground

[ssashton]

Yes, where else can it come from?

FYI in my specific use case it causes ground loop noise from an audio amplifier when external grounded equipment is connected to the amp input.

The company seem to think it's a tiny mistake, I think it's a ridiculous mistake.

[Psi]

I'm confused, did you want V+, V-  and EARTH?

Saying GND is likely to confuse people.

[ssashton]

Interesting, maybe this was some of the confusion! Still, they did connect V- to earth.

To me GND is earth. Equipment is floating or grounded.

Although I do accept that on schematics GND is often used in place of Com or V- .

[wraper]

Interesting, maybe this was some of the confusion! Still, they did connect V- to earth.

To me GND is earth. Equipment is floating or grounded.

Although I do accept that on schematics GND is often used in place of Com or V- .

GND is not earth. GND can be without earth in double insulated device or in battery powered device. Or both of them can be present in the same mains powered device but separate from each other. Multiple GND can be present within a single device too. GND is nothing more than a reference point which in most cases is a negative power rail.

[wraper]

Also I don't think that particular PSU design allows negative power rail to be separated from earth. It will void safety regulations, to do it safely, it would need to be double insulated.

[ssashton]

I had it as a floating output last time I ordered. C8 input.

I'm really surprised you guys equate ground with common or V-. To my mind, especially when talking about power supplies, ground is earth and it is called ground becasue it comes from a metal pole stuck in the ground. In an AC mains circuit I doubt you would confuse neutral ground, even though they are connected together?

Anyway lesson learned, I MUST start using the term earth.

I don't think this was actually their confusion though, since my last order was floating and by asking for pin3 to be GND they added a 3 pin mains inlet and connected V- to earth.

[wraper]

I had it as a floating output last time I ordered. C8 input.
I'm really surprised you guys equate ground with common or V-. To my mind, especially when talking about power supplies, ground is earth and it is called ground becasue it comes from a metal pole stuck in the ground. In an AC mains circuit I doubt you would confuse neutral ground, even though they are connected together?

Nobody equates them. Secondary of the transformer needs to be tied with earth unless it's a double insulated design. Otherwise all those safety certifications are BS.

[ssashton]

Wait.. you don't equate ground with common or V-, yet you also say ground is not earth? Then what is it?

[wraper]

Wait.. you don't equate ground with common or V-, yet you also say ground is not earth? Then what is it?

GND is nothing more than a reference point which in most cases is a negative power rail.

Edit: Ground is common or common ground. But common does not mean earth.
Edit2: at least in context of electronics. In context of electrical power distribution it's common to use earth, ground or earth ground to describe earth connection.

[ssashton]

Then why is it called ground?

EDIT: Okay, so you do equate GND with common. Fair enough. I find that a very imprecise usage but I accept it is a standard way. I even used the term 'virtual ground' in my circuit to describe a common bias point.

[ssashton]

On a side note, I've seen double insulated equipment with a metal chassis. Do they need to have insulating sheets under the PCBs or something to comply with standards?

[wraper]

Then why is it called ground?

Ask one who started using this term. Or ask linguists why a single word can have multiple meanings. BTW in electronic circuits in old Soviet magazines from 60's, most of the transistor circuits had positive rail as GND.

[wraper]

On a side note, I've seen double insulated equipment with a metal chassis. Do they need to have insulating sheets under the PCBs or something to comply with standards?

Depends on distance to PCB. It's more about transformer insulation, interference cap insulation, creepage distance and similar.

[brabus]

Unclear specification.

The connector issue is the last thing I would complain about: those markings look suspicious. Have you obtained proper documentation and a traceable CB report?

[Cerebus]

If you see a specification with three connections labelled +V, -V and GND you go back to to the specifier and ask what their intent was: "Is GND a 0V centre rail?", "Is GND intended to be mains earth?", ""Is the supply meant to be floating?" or "Is GND intended to be tied to one of the rails and if so which one?" and so on. Then you get it confirmed in writing. You do so precisely because we all know that the term "GND" is poorly defined and that it gets thrown about loosely with people using it to mean everything from "a floating return line" to "circuit protective conductor". 

You certainly don't assume that "They specified three connections but only really wanted two and we'll arbitrarily assume that we will tie GND and -V together". The requirement for three connections ought to give you pause for thought if you propose to change that to effectively two connections.

The specification is faulty because it is ambiguous. The manufacturer is also at fault because they failed to get that ambiguity resolved before moving to manufacture. Were I a judge deciding a dispute about this I'd rule 50/50 and make both parties eat half the costs unless one party was, or was supposed to be, significantly better qualified than the other - in which case I'd ladle more blame on the party that was, or ought to be, better qualified. The most I would depart from 50/50 would be 30/70 because everybody involved needs to learn a lesson here and a kick in the wallet is a superb learning tool.

[Miyuki]

In 99% of cases, you communicate with someone who barely knows English and have no idea about technical stuff

[ssashton]

In 99% of cases, you communicate with someone who barely knows English and have no idea about technical stuff

True. It is certainly a skill to do business in China. I remember there is a book called something like 'Engineering Chinese for Foreigners' or similar. I guess I need to get a copy!

[Psi]

yeah, I do agree that the word 'ground' is a bit misleading due to it's literal meaning.
But if you look at any schematic it will have the common voltage reference labeled as GND,DGND,PGND etc., which has nothing to do with the household EARTH wire.

I suspect the word 'ground' originated from very old circuit construction techniques before earth wires existed.
Electronics built inside a metal case which is wired to the common voltage rail. Where a 'ground' connection is a connection to the metal case at the bottom of the device. eg, with respect to the device, the floor of the metal case is the lowest vertical point i.e. ground.

[TimNJ]

GND is circuit ground (AKA "common" or "return"), i.e. the 0V reference node of your design, usually denoted by a triangle symbol on a schematic. It is somewhat unfortunate that most(?) engineers still use the word ground in this context, although I've met a few people who are very diligent about calling it "common" or "return", which I think most would agree is a much more informative name.

Earth is the actual planet Earth, in this context, refers to some conductor which is connected (at some point) back to a water pipe, earthing rod, etc. On a schematic, it is the upside down pitchfork symbol and (sometimes) the three parallel lines (in the shape of a triangle) symbol.

I've somewhat consciously trained myself to never use the word ground when talking about earth. It helps takes the ambiguity away, especially if you are trying to communicate with others.

[Phoenix]

Calling a large plane of copper "ground" in the context of electronics seems OK to me. The Oxford Dictionary definition has multiple meanings including:
      an area of land or sea used for a specified purpose.
      "shore dumping can pollute fishing grounds"
This feels similar to its use in electronics and is not related to the tangible dirt under our feet.

That being said I do use 0V instead of GND as a net label to avoid the ambiguity especially in designs that have some 0V planes floating and some at earth potential. In speech I usually say ground plane as that is the general vernacular.

[Siwastaja]

Simple: shitty specification which is just meaningless. Also any assumption about it is meaningless, so it can't be "wrong" either.

GND is an arbitrary net label in electronic design. I absolutely loathe power supplies that label something directly connected to PE as "GND". PE is PE, PE is not GND. PE may be GND in a particular design, though.

But here, PE is meaningless as well, it may or may not be connected to the output negative, both styles are widely seen in switch mode supplies, both are fine if it passes EMI and safety standards, and it's well possible to pass those with or without connecting the output negative to PE.

This specification just screams that they couldn't decide whether to call the output negative "-" or "GND" so they did both! Or, they reused both the connector, and the specification image from a bipolar supply, you know, one used to power audio opamps and similar with both positive and negative rail, without "virtual ground" tricks, but ended up just paralleling the "-" and "GND" labeled wires because their supply is unipolar.

[Benta]

Simple: shitty specification which is just meaningless. Also any assumption about it is meaningless, so it can't be "wrong" either.

GND is an arbitrary net label in electronic design. I absolutely loathe power supplies that label something directly connected to PE as "GND". PE is PE, PE is not GND. PE may be GND in a particular design, though.

But here, PE is meaningless as well, it may or may not be connected to the output negative, both styles are widely seen in switch mode supplies, both are fine if it passes EMI and safety standards, and it's well possible to pass those with or without connecting the output negative to PE.

This specification just screams that they couldn't decide whether to call the output negative "-" or "GND" so they did both! Or, they reused both the connector, and the specification image from a bipolar supply, you know, one used to power audio opamps and similar with both positive and negative rail, without "virtual ground" tricks, but ended up just paralleling the "-" and "GND" labeled wires because their supply is unipolar.

+1

"GND" just means "something", and every engineer has her/his own concept of what it is. In the early transistor days using Germanium PNPs it was the positive rail.

The only "GND" (which it isn't) that has a bulletproof definition is PE or "Protective Earth". It's also the only one that has a standardised schematic symbol.

[T3sl4co1l]

If this amplifier is custom, this is probably relevant:
https://www.eevblog.com/forum/projects/should-i-use-a-ground-plane-in-my-audio-frequency-pcb/msg3694912/#msg3694912
In short, amplifiers shouldn't be designed to draw ground currents through their inputs, through proper application of differential sensing.

Tim