How come negative power supply has negative voltage only

So I have seen power supplies that produce e.g. -48V DC (used e.g. in telecom but probably not only). For sure, that power supply is floating and in no way is negative terminal connected to earth. So, what doesn't make sense to me is why we need to specify minus sign on power supply? What does it actually means?

Current flows from positive to negative (or actually from higher potential to lower). So, if we have negatiive power supply - ok, current will flow from 0V to -48V, so it would be reversed. But then, if you have +48V PS - you swap terminals and got -48V. What's the actual difference? Is it sink vs source of current, or is it related with human safety? For me from engineering point of view it somewhat doesn't make sense unless you need both positive and negative voltage rails.

I do get this notion with ATX power supply, where we have +12V rail and -12V though

Upd 1 found this link which indcate why telecom uses negative voltage but that doesn't explain what it really is. https://deltaelectronicsindia.com/faq-items/why-does-telecom-use-negative-voltage/

Update 2 Another thing came to my mind as result of thought experiment... So we have polarized DC MCBs. Those should be connected correctly in terms of polarity, otherwise those can catch fire because arc won't be properly extinguished. So, how do you properly connect those? Something tells me that 0V should be connected to "+" terminal and -48V for "-"

• Your question really starts with a wrong assumption; unless it's stated that the supply is isolated, it's probably not floating. Sep 21 at 11:57
• @MarcusMüller So, what you're essentially telling is that from usage context in negative voltage power supply "positive" terminal is actually ground (0V) and negative is 48 volts below(-48V)? Did I get it right? I still don't understand why don't you just rename terminals... Might be I'm missing something Sep 21 at 12:19
• I read the information in the update link and I think parts of it are straight up wrong, like the lightning safety part. Sep 21 at 12:21
• WTF, really, aside from the arguably true "negative metal conductors see less electrolytic corrosion" (which just simply ignores there's still a positive pole…) they seem all totally wrong. Sep 21 at 12:28
• I looked at the site in Upd 1, and it has -- issues. Four of the points are flat-out wrong, and one is not supported by evidence. "Since negative voltage has a lack of electrons" - - - REALLY - !!! Sep 21 at 15:41

6 Answers

A voltage source is negative or positive according to how it is measured. It has no inherent polarity of its own.

Only the terminals can have a polarity relative to another reference.

The telecom industry uses the positive terminal of a 48-volt battery as the reference (0 volts) making the other terminal -48V.

ATX power supply, where we have +12V rail and -12V though

These voltages are relative to the 0 volt rail which happens to be connected to ground.

If the black wire of a voltmeter is connected to -12 rail with the red wire connected to ground. A value of +12V, a positive value, would be indicated.

Whenever voltage and its polarity is discussed or measured, it is always with respect to a reference node which is usually chosen to be sensible for the associated system.

• so let me ask this again - how is it different from having 48V power supply and connecting positive terminal of power supply to negative of the load and vice-versa? Is it just marking on power supply for convenience? Sep 21 at 12:45
• as four engineers have now told you, seeker: it's a question of convention. Nothing more, nothing less. Sep 21 at 13:43
• @MarcusMüller: If there's any electrical leakage between a buried cable and the surrounding soil, wires which are more positive than the soil will suffer much worse corrosion effects than those which are more negative. Sep 21 at 21:21
• @supercat: Agreed. The problem here is thinking a node terminal has an inherent absolute voltage associated with it. The polarity of the earth (node) that the cable is buried in is measured relative the the cable (node). Otherwise the determination of whether the cable will corrode cannot be made Sep 21 at 22:12
• @chux-ReinstateMonica: The principles involved are the same as with electroplating. If one has a positive and negative electrode with a fluid path between them, metal will migrate from the positive electrode to the negative one. Moving even a small amount of metal from thin phone wires onto a heavy grounding stake would be likely to make the phone wires fail. Moving even a much larger amount of metal from a heavy grounding stake onto phone wires, however, would be no problem. Sep 22 at 15:53

In telecom the positive terminal of the battery or power supply is grounded.

That makes everything powered by the -48V negative or 0 at most relative to ground, which is superior in preventing the more damaging electrochemical reactions should the circuits get wet and current leaks to ground.

• +1 for mentioning this, thank you! So if I get it right in telecom the current flow is reversed to be from ground to negative and this way you don't have a problem if smth leaks to ground, nice. Sep 21 at 13:57
• I'd add that in practice, many PSUs will be designed with the option of tying 0V to chassis ground. Sep 23 at 7:36

A terminal of a power supply is marked 'positive' or 'negative' depending on which terminal is taken as the reference terminal (marked 'zero').

Here's a pictorial (with batteries) for easier understanding.

• Thanks, I do get this, my point of confusion was rather that a) minus terminal is always ground and b) negative voltage is something rather special Sep 21 at 13:41
• Anytime! I'm glad you got it. Sep 21 at 13:48

The importance of +48V vs -48V in the telco context is entirely related to the complete solution deployment.

The whole idea of -48V comes from the +ve terminals on the battery banks being grounded to reduce galvanic corrosion on external phone lines, which are powered at 48V.

Why this then impacts supplies is tied to the two main grounding schemes:

• IBN (Isolated Bonding Network)
• MBN/CBN (Mesh or Common Bonding Network).

The IBN model has each rack of kit "floating" (literally in insulated pads), with + & - power terminals on equipment not connected to chassis, and rack ground wires all going back separately to a common grounding bar. This was designed to limit ground loops and mains hum on analog phone lines. It delivers on this goal, but maintaining galvanic isolation on all the interfaces can be a pain, and the long ground connections have a high inductance which makes them less useful for sinking high frequency EMC.

In the CBN/MBN model, the alternative approach is taken, and the goal is to form a dense mesh of ground connections, to reduce loop impedance for HF EMC. Critically, this generally includes equipment containing a jumper/link which connects chassis ground to the "0V"/"+ve" terminal. A jumper is used so the same kit can be deployed in an IBN environment.

(This is a bit like the AC PME setup, where N & E come from a common conductor at each supply head.)

It is this default grounding of the +ve/"0V" terminals in equipment, that enforces the concept of the -ve terminals being "-48V", because every element (equipment or supplies) has to either be floating or agree that it is the + terminal that is to be grounded.

It also impacts on colour coding of wires used in exchanges, where the US telephony convention is that 0V/gnd is red (as +ve) and -48V is black (for -ve), whilst in Europe the convention is to use blue for 0v/gnd, and grey for -48V (as opposed to brown that is the default colour for a positive DC wire).

Having said that, with the exception of where the grounding link sits, there is no real difference in the supplies.

• Thank You for providing historical and technical context on how and why it's done in telecom. It's interesting and totally makes sense to me. On the other hand telephone lines are long gone but that still explains everything. Funnily enough though, I've met another power supply which has red terinal of 0V and blue of -58 or something i.stack.imgur.com/j7cyx.png Sep 21 at 14:08
• I've also seen kit which is black of 0V and Red for -48V. Frankly I always check rather than rely of colours! Sep 21 at 14:41
• This link also makes interesting reading on the IBM vs MBN subject vdocument.in/… Sep 21 at 14:50

So, if we have negatiive power supply - ok, current will flow from 0V to -48V, so it would be reversed. But then, if you have +48V PS - you swap terminals and got -48V.

There's cases where you cannot swap terminals. For example, you have a power supply that plugs into a backplane where ground is a specific pin on a fixed connector. Here you'd have to specify whether you want a +48 or -48 V model. Internally, the manufacturer might have merely swapped the output, but the end-user can't do that.

Another case is in test and measurement. You might not want the labor of swapping the connectors when measuring the I-V curves of a batch of diodes, for example. Four-quadrant and bipolar power supplies allow you to "swap the terminals" electronically by simply programming in a positive or negative voltage.

If you only have a black box with two terminals and nothing else, whether you mark something 0V/+48V or -48V/0V is arbitrary.

A car battery is like that. But already an AA battery has an enclosing shell that somewhat naturally (because it might touch other things) is dubbed 0V, with an isolated tip being the +1.5V electrode.

A mains-operated power supply will tend to have one electrode that is better suited for sinking and sourcing disturbances that are referenced towards the main voltage.

And once you have more than two terminals, there will be one terminal that will be used for sourcing and sinking disturbance currents via "blocking capacitors" from arbitrary other locations. A simple 2-terminal unregulated DC power supply may be symmetric; but once you stabilize voltages, you will use one terminal as the reference (0V) and the other as the regulated terminal.

Whenever the inherent symmetry of the two-terminal black-box is breached (like when a mobile phone goes off and injects RF into it, or the capacity to the ground may play a role) and the sourcing and sinking capabilities of the two terminals stop being treatable as a compulsory matched pair, the difference between 0V/+48V and -48V/0V may surface.