78xx in Negative Supply Mode

[Boondock Journeyman]

First post here.  If I am in the wrong place or am violating the culture, please excuse me and point me accordingly.  No offense is intended.  I took the advice to new people seriously and dumped my whole story rather than just the core question.  Hence, the length.

I have a +5vdc, +/-15vdc PSU for breadboard work.  I obtained a used Tek 2232 scope as my first bit of test gear beyond my multimeter.

Naturally, I started taking everything apart and learning how to probe without sparks flying.  When I opened up the subject PSU and put the protective earth ground connection to the ground plane of the PSU on a switch and started poking around (thank you Dave for the YouTube on that point), I found a pair of 7815s rather than the 7815/7915 pair I expected on the +/-15vdc.

I pulled a bunch of data sheets and found the "negative output voltage circuit" figures wherein pin 2 seams to drive the output rather than pin 3.

My question is, how does this work and is it safe long term?  How is current getting passed function blocks like the starting circuit, reference voltage, thermal protection and SOA protection?  Is the use of 78xxs in "negative" mode common?

I guess I maybe looking for a reason to rebuild this thing with a 7815/7915 pair.  If the "negative" use of the 7815 by-passes internal protections, I would have my excuse.

Thank you.

[IanB]

Well, firstly I wouldn't change a working power supply just for the sake of it. As the saying goes, "If it ain't broke, don't fix it!"

As far as the 7815 is concerned, it has no way of knowing whether you think the regulated output voltage is "positive" or "negative". The circuitry inside cannot tell, so all of the protections and other functionality will work just the same.

You can tell this by taking any power supply, for example the +5 V DC one. If you measure the output voltage of this supply and swap the leads of your multimeter around, black to red terminal and red to black terminal, you will suddenly find it is a "-5 V DC" supply. Positive and negative voltage is purely in the eye of the beholder.

Some power supplies have two independent 15 V outputs. You can put them in series to get 30 V, or put them in parallel to get 15 V at twice the current. If the supplies are independent like this, then the use of two 7815 regulators will be fine.

However, the "negative" 7915 is possibly a better choice if the supply has a shared common rail and +/- 15 V supplies referenced to this. In this case the 7915 may be more stable as it won't be "piggybacked" on top of another regulator.

[Boondock Journeyman]

IanB,

Thank you. 

This PSU turns out to have three separate "ground planes." All three stages come from separate transformer windings and rectifiers, no center taps.  Then they tied one pole of each plane together and referenced them all to the protective equipment ground.

So you are right. They just connected the "+15 vdc" of one source to the "ground" of the second.

I separated all the internal referencing and put in a set of switches. Now I can get 5, 10, 15, 20 and 30 in either +/-.  I can only run three channels at one time, but I can power op amps and logic. 

I will fool around and build a +/-12 vdc board in the same box. Lots of room.

Thanks again. I have not thought about multiple reference planes before.

[macboy]

The 7915 is not as good as the 7815. The 79xx use a NPN pass transistor since they are easier and cheaper to produce (or at least were, when 79xx was hot snot), but doing it this way produces an extra pole into the system, exactly like an LDO regulator. With 270 degrees of phase shift due to three poles, it then requires careful compensation in order to be stable. Unfortunately, the extra pole is dependent on both load and output capacitance. 79xx regulators can end up oscillating more easily with various combinations of input and output capacitors and loads. 78xx regulators, with one fewer poles, are much easier to tame and get right. On a bench supply, where you can't control what the user connects as a load, this is much preferred.

[amyk]

I suppose another way of looking at it is to consider the 79xx a "low side regulator" and the 78xx a "high side regulator".

[codeboy2k]

I suppose another way of looking at it is to consider the 79xx a "low side regulator" and the 78xx a "high side regulator".

This is a very good way to look at it, because once you are thinking of them like this you realize you can use a 79xx with an inexpensive and easy to find NPN like the 2N3055, and use it in the negative rail.  This eliminates the need for a maybe hard to find PNP in the positive rail to bypass a 78xx positive regulator.  In this case, the positive rail is the common rail, and we are regulating the low side.

The attached (simple!) schematic is a 12V 2A positive PSU from a negative regulator (with no protection! -- it's supposed to be simple).

The collector of the 2N3055 is connected to the case of the TO-3 and you can directly mount the the TO-3 to the metal project case.  You do not need to insulate the collector from the case.  The case becomes the output GND reference.

Result:  You get a positive voltage regulator out of a negative 79xx series and can use a TO-3 attached to the heatsink without the need to insulate it from the heatsink or case.

EDIT:
(1) I know TO-3's are pretty much obsolete these days in favor of the cheaper plastic power packages.  A lot of old hardware still has 2N3055's so they're still easy to find. But that's not the point here. The point is that there are more choices for NPN vs PNP so I'm showing how it can be done with an NPN and a low side monolithic regulator;
(2) In fact, there is no such thing as a positive or negative PSU, there's just convention, and it's all relative anyway. A voltage is always referred to some other reference point in the circuit. As IanB said, turn your probes around and the polarity changes.  This circuit just happens to regulate on the low side rail, and it's called ground (GND) here.  But you could just as easily call that "-12V Regulated Output" and call the other side GND if you wanted to, and that would be the standard way to boost a 79xx when making a "negative PSU"
(3) it's a trick of nomenclature

[Boondock Journeyman]

Thank all of you.

I understand now that my whole world view on ground planes and reference points came from my work on 3 phase motor control circuits.  I had no concept at all about these rails (if you will) being relative to nothing at all until they are referenced to something by a designer.

This is a big learning point. When your utility company establishes your reference plane, the world is simple.

I pulled out Horowitz and Hill and will try to understand the scope and limitations of the NPN/PNP structures. This is fascinating and might give me some direction in part selection.

This is a good place to learn.  What a nice change from the usual virtual mosh pits out there.