Power supply questions.

[BillyO]

Okay.  I have a Mornsun LM15-23B05 SMPS.

https://www.mornsun-power.com/html/pdf/LM15-23B05.html

However, there is nothing unusual about this unit and there are a million units just like under different names (maybe lower quality though).

Internally the earth ground pin is connected to the chassis via a screw through the PCB.

I want the secondary to be earth referenced too. So...

A) Is it okay to connect the -V to the earth ground directly?

B) If the answer to A) is no, why is the answer no?

A tad more background before the final question.  I will be using this supply in a function generator.  The great number of cases where thie generator will be used, it will be feeding into an earth ground referenced device (like a scope, or something attached to a scope, etc..).  In this preponderance of cases the little PS will have it's -V side being connected to earth ground anyway.

C) Would it not be best to avoid doing that over test leads and connect directly to earth ground where the earth ground enters the function generator?

[TimFox]

Like most supplies in this format, the "GND" terminal connects only to the metal structure of the unit.
Normally, one connects the usual three wires of an AC cable to AC(L) and AC(N) and to the GND terminals.
You may connect either polarity of the DC output to GND at the supply terminals, if you wish.
Optionally, you can connect the DC output terminals to ungrounded load, within the insulation specification of the supply.
The data sheet wording ("1 to 4 any must be connected to the earth") is not ideal English: they do not want the internal circuitry to "float" completely with respect to ambient ground.
However, they do specify a high isolation voltage capability from input to output.
(Strictly, "any" could include connecting both 1 and 2 to Gnd, which is obviously wrong.)
Taking their wording literally, it appears they do not recommend connecting "+ Out" to ground.

[BillyO]

Thanks Tim, that's pretty much what I thought.

[Damianos]

In their application notes, they say to avoid a direct connection between output and ground!
Because the documents are a little vague, try to communicate with them to clarify.

[elianto]

In my view their reference for 1-4 is not to pin terminal but to screw position, connect any of the mounting screw to ground.

[TimFox]

That would make more sense.

[BillyO]

In their application notes, they say to avoid a direct connection between output and ground!
Because the documents are a little vague, try to communicate with them to clarify.

I see that, however that's kind of fanciful.  For instance, as mentioned above, if I connect my function generator to an oscilloscope, or other device that is ground referenced, then the -V out is going to be connected to ground.  Like it or lump it.

In the same note they mention what I am trying to avoid, and that is the leakage current (and voltage).  They don't, however, suggest any good ways to get rid of it other than adding a Y capacitor.  I will try that (I have some coming), but if it's not really effective I will connect -V to ground and make it switchable between earth ground and Y capacitor.

There is not a ton of leakage, but at ~20V AC and nearly 15uA AC, that is enough to toast a sensitive and hard to replace FET

[Terry Bites]

There is usaually a degree of butt covering in the blurb. They quote an isolation voltage of 4kV, once you ground the output side you have zero volts of isolation.  Any crap on the mains ground becomes crap on your output ground.

An aside. Lately I've been exploring the pros and cons of ground wire chokes.
They have saved from inbound mains ground borne crud- mostly from old UPSs lately. Pro
They (of course) increase the ground impedance.  Hmm. Con?

[Damianos]

In their application notes, they say to avoid a direct connection between output and ground!
Because the documents are a little vague, try to communicate with them to clarify.

I see that, however that's kind of fanciful.  For instance, as mentioned above, if I connect my function generator to an oscilloscope, or other device that is ground referenced, then the -V out is going to be connected to ground.  Like it or lump it.

In the same note they mention what I am trying to avoid, and that is the leakage current (and voltage).  They don't, however, suggest any good ways to get rid of it other than adding a Y capacitor.  I will try that (I have some coming), but if it's not really effective I will connect -V to ground and make it switchable between earth ground and Y capacitor.

There is not a ton of leakage, but at ~20V AC and nearly 15uA AC, that is enough to toast a sensitive and hard to replace FET

I agree that your device have to be referenced to the ground.
A solution may be to connect your circuit directly to ground (PE) and use a choke inductor to connect the power supply to it. This to reduce transferring of spikes to the output side, due to inrush current and the line filter capacitors. As referred in replay #7 by Terry Bites.
I think that a better solution is to find another power supply, that is more clearly "defined" for what it does or use a "classic" linear one ...

BTW: a function generator does not need a symmetric power supply?