Defining an Isolated Power Supply

[MoltenVoltage]

I just picked up a Voodoo Lab Pedal Power Plus 2 and opened it up before even turning it on because I wanted to see what made the outputs "isolated".

Expecting to see separate transformers for each output, to my surprise there was one huge transformer that dropped the wall voltage down to 13.3 VAC.  That power then goes into 8 different bridge rectifiers then to each separate voltage regulator.  Each output also has a 1000uF 25V cap and a 10uF 25V cap.  I haven't traced it, but assume that it is similar to RG's geofex design.

My question is what are the minimum requirements for the output to be electronically "isolated".

1) Can two outputs share a transfomer and still be considered "isolated"?
- apparently

2) Can two outputs share a bridge rectifier?
- they don't in the Voodoo lab, so do the separate rectifiers create the isolation?

3) Can two outputs share a 1000 uF filter cap between the rectifier and the voltage regulator?
- they don't in the Voodoo lab, so i am assuming those need to be separate as well

4)  Can two outputs share a voltage regulator?
- it seems obvious that they cannot - perhaps this is the magic point of "isolation"?

5) Are EACH of these parts necessary to create isolation?

6) Is there any design that will ensure complete isolation of outputs that are all using the same wall plug, or is isolation simply a relative term?
- my gut tells me that there are just degrees of isolation

7) Are 2 unregulated wall warts that are on the same power strip "isolated"?
- seems like they should be since they have separate transformers, but maybe not if they are missing the regulator

8) Are 2 regulated switch mode power supplies on the same power strip "isolated"?
- seems very likely

Thanks!

[darron]

Isolated means that they do not have a connecting common electric path in my view. More below.

So the answer to MOST of your questions is no, they can not share common parts.

The Geofex spider is isolated. If you measured the outputs they shouldn't have continuity from any point to another. They should also not 'touch' the primary mains either.

there was one huge transformer that dropped the wall voltage down to 13.3 VAC.

Do you mean it has 1x 13.3VAC output or 8?


By not having a common electrical path I mostly mean that they do not electrically touch. There could be exceptions to the simpler 'no touching' explanation though. For example a power transformer on an amplifier does isolate the supply, so the amp isn't really directly connected to the wall mains. However, one of the wires off the transformer could still be connected to the metal chassis Earth/framework, which would connect back to neutral at the switch box. So technically the mains could still be touching the isolated output. The current would not be flowing through the chassis Earth though so there's no ground loop noise problems. That's still isolated, in my view.

edit: Interested to know if someone has a textbook answer.

[Kesh]

Are the grounds connected after the bridge rectifiers?

[defaced]

1) Can two outputs share a transfomer and still be considered "isolated"?
- apparently

Yes, that is one of the principal reasons devices have a transformer.  The other is to convert to the voltage/current needed for the device.

For questions 2 - 5, the answer is no.  To concur with what darron said above, isolation means there is no connected path between the sections of the circuit.  Sharing parts means things must be connected, thus not isolated. A transformer is just windings of insulated wire with no apparent electrical connection.  Transfer of energy is done via magnetic fields (induction). Thus different windings of transformers are isolated.  

6) Is there any design that will ensure complete isolation of outputs that are all using the same wall plug, or is isolation simply a relative term?
- my gut tells me that there are just degrees of isolation

Partial ground lift, but this is sketchy because details matter.

7) Are 2 unregulated wall warts that are on the same power strip "isolated"?
- seems like they should be since they have separate transformers, but maybe not if they are missing the regulator

Are 2 regulated switch mode power supplies on the same power strip "isolated"?
- seems very likely

Depends if they both reference one side of the output to the input ground.  If so, then no, the two power supplies are not isolated.

[R.G.]

My question is what are the minimum requirements for the output to be electronically "isolated".

1) Can two outputs share a transfomer and still be considered "isolated"?
- apparently

Sharing a transformer is not an issue. Isolation is dependent on whether the outputs share a single secondary winding. As noted, isolation depends on whether there is a DC conductive path from one to the other. If the transformer has multiple secondaries that are isolated, then the outputs are isolated. See the article at geofex on winding a "Spyder" transformer. In that, I took a two-secondary transformer and rewound it to have eight isolated secondaries. This could produce eight isolated DC outputs.

So the answer is "Yes, if they are on different secondary windings." You might want to check and see if there is a unique pair of wires from the transformer to each bridge rectifier. If there is, they're probably isolated. I've not been inside this thing, but I assumed that it was a Spyder on steroids. I've found that if I really need an isolated output for some pedal setup, another $20 1Spot is cheaper than buying a many-output brick, but that's just me and I'm cheap that way.

2) Can two outputs share a bridge rectifier?
- they don't in the Voodoo lab, so do the separate rectifiers create the isolation?

No. Two outputs with separate rectifiers still share a DC path between them. That is not isolated.

3) Can two outputs share a 1000 uF filter cap between the rectifier and the voltage regulator?
- they don't in the Voodoo lab, so i am assuming those need to be separate as well

No. Two isolated outputs must not share a filter cap and rectifiers.

4)  Can two outputs share a voltage regulator?
- it seems obvious that they cannot - perhaps this is the magic point of "isolation"?

No.

5) Are EACH of these parts necessary to create isolation?

You must have separate secondaries. That means you must also have a unique rectifier, filter and regulator, so yes. But the isolation is created by the transformer windings being separate.

6) Is there any design that will ensure complete isolation of outputs that are all using the same wall plug, or is isolation simply a relative term?
- my gut tells me that there are just degrees of isolation

"Isolation" is a technical term which has a meaning that is beyond what you're asking. For your purpose, "isolation" means "comes from a different transformer winding", pure and simple. There are degrees of isolation, but I'd have to type for a long time to get that across and it would lose most of the people who read it here.

7) Are 2 unregulated wall warts that are on the same power strip "isolated"?
- seems like they should be since they have separate transformers, but maybe not if they are missing the regulator

Outputs are isolated (for the limited purposes mentioned here) if they come from different transformer secondaries, or different transformers. The issues of rectifiers, filters, and regulation are irrelevant to that discussion.

Are 2 regulated switch mode power supplies on the same power strip "isolated"?
- seems very likely

Yes. The outputs come from two different transformer windings and share no conceivable DC path between them. They are isolated (again, for the limited purposes being discussed here.)

[totes]

Thanks for such informative posts! I have a question of my own regarding the "degrees of isolation" in power supplies, and how it affects hum.

If I were to build my own power supply, for example, with six 9v outputs, how would the hum change if each secondary from a transformer is to be used for 1, 2, 3, or 6 outputs?

Additionally, how much worse does the hum get from zero shared secondary's to shared secondary's, to shared rectifiers, and even to shared filter caps or shared voltage regulators?
If I'd like +9v and -9v outputs, can a pair of those share a secondary?
I understand that to achieve isolation, separate secondary's/transformers are needed, but I'd just like to get an idea of how bad it gets without that condition.

Oh and a second question: Would the filter caps need to be rated anything above 25v?

Sorry for the terrible formatting.

[R.G.]

If I were to build my own power supply, for example, with six 9v outputs, how would the hum change if each secondary from a transformer is to be used for 1, 2, 3, or 6 outputs?

There is no clear way to answer that, mostly because hum comes from so many places. Isolating the power supplies of pedals is one way that helps hum in certain situations, but by no means all situations. So to answer your question, I'd have to actually have the specific pedal setup you use and measure it.

In my professional capacity, I have experience with literally thousands of setups where all power comes from one single DC source, and there is no perceptible hum at all. In a lot of these cases, the setup is used by a professional session or touring guitarist who can have anything on his board that he wants, and has no patience for hum at all. So the answer is "that's the wrong question."

Additionally, how much worse does the hum get from zero shared secondary's to shared secondary's, to shared rectifiers, and even to shared filter caps or shared voltage regulators?

As above. It depends on where the hum comes from. If and only if leakage of AC into the ground or power lines is where the hum comes from in the first place, then you could construct some setups to test that. Even then, it might well depend on exactly how the individual wires were run. Hum is a many-headed Hydra.

If I'd like +9v and -9v outputs, can a pair of those share a secondary?

Yes. The important point here is that there is one and only one ground.

I understand that to achieve isolation, separate secondary's/transformers are needed, but I'd just like to get an idea of how bad it gets without that condition.

There is no clear way to answer that without lots more information.

Oh and a second question: Would the filter caps need to be rated anything above 25v?

If the transformers and rectifiers can possibly make a peak voltage on the caps above 25V, yes.
If the transformers and rectifiers CAN'T possibly make a peak voltage on the caps above 25V, no.

In both instances, "possibly" has to include maximum no-load voltage from the transformer secondary even under high-AC-line voltage.

[MoltenVoltage]

Thanks for all the help RG!

I opened the Voodoo Lab back up and traced the circuit.  There are in fact 8 separate 2-pin outputs on the transformer (16 pins total).  Each 13.3 VAC outputs go into a bridge rectifier (16.4 VDC) which supplies an adjustible voltage regulator (LM317L) that outputs 9.2 VDC.

I would have thought that the 2 unregulated wall warts would share noise, but am glad to know they are isolated.

Hum is not my concern, I am trying to ensure that MIDI Clock noise does not leak from one pedal to another.  Since MIDI controllers are opto-isolated from MIDI-controlled pedals, if the MIDI controllers are on an isolated power circuit, then any digital noise they might generate should not appear in the audio circuit.

[R.G.]

I opened the Voodoo Lab back up and traced the circuit.  There are in fact 8 separate 2-pin outputs on the transformer (16 pins total).  Each 13.3 VAC outputs go into a bridge rectifier (16.4 VDC) which supplies an adjustible voltage regulator (LM317L) that outputs 9.2 VDC.

That's not a huge surprise; I'd expected something like that. I think the pedal power appeared after I posted the "Spyder" article. If I had been making a commercial product, I'd probably have done it in a similar way.

I would have thought that the 2 unregulated wall warts would share noise, but am glad to know they are isolated.

Hum is not my concern, I am trying to ensure that MIDI Clock noise does not leak from one pedal to another.  Since MIDI controllers are opto-isolated from MIDI-controlled pedals, if the MIDI controllers are on an isolated power circuit, then any digital noise they might generate should not appear in the audio circuit.

Maybe. Probably. That gets into some of those other semi-esoteric "isolation" facets I mentioned. Generally a 60Hz transformer is a pretty good trash bucket for digital noise, with its high (by digital circuit standards) inductance and self capacitance. Sometimes you get it passing through common mode digital hash, though.

As far as MIDI signals leaking, this is much more a characteristic of the MIDI receive/send/pass circuit on the board. If that's locally decoupled well and some attention paid to small loops and NOT shared ground lines, it's not much trouble. Or at least I haven't had issues with any of the stuff I've done.  Keep the power down and the decoupling local, the grounds not shared on the PCB in the pedal and you're probably OK.

'Course, it's always possible to make a wireless MIDI transmitter ...   

[bukas]

if i may add something:

if you plan using

1000uF 25V cap and a 10uF 25V cap.

be aware that you can expect 300mA of current. even though lm317 can deliver 1A a 1000uF cap just can't feed it so then you can expect current variations if supplies are sharing secondary coil.

[defaced]

"Isolation" is a technical term which has a meaning that is beyond what you're asking. For your purpose, "isolation" means "comes from a different transformer winding", pure and simple. There are degrees of isolation, but I'd have to type for a long time to get that across and it would lose most of the people who read it here.

Can I get "Keywords for Searching" for 200, R.G.?

[R.G.]

 
Good one.

Let me talk around it for a while. Maybe we can make up some key words.

Isolation is one of those slippery concepts. It really has almost no meaning without a context of what you're trying to isolate, and what manner/path of communication you're trying to sever.

Yeah! That's a good way to start thinking about it. The path of communication. "Isolation" can mean something like "way, far away" in some instances, like an isolated island. That means as a practical matter, that it's distant, and you can't simply walk out there to the island. You can't really have much of an effect on the island or people/animals/plants on it without expending a lot of effort to do it. It's isolated from sight, sound, and other means by distance. It may not be isolated by radio communication at all. There's an island in the (?) Indian Ocean that only recently got a satellite phone link with the rest of the world, but a ham radio operator could have gossiped there all the time.

Circuits communicate by transferring electric currents, electric voltages, or electromagnetic fields between them. They can do it all three ways. In addition, there is the concept of the frequency of the communications medium mattering. The communication between two circuits can be from (literally) DC to daylight and beyond perhaps. If your circuit can't "perceive" any signals above 10kHz, then it probably won't get interfered with by anything up in the GHz range. If it is self contained and powered by batteries, no wires between it and any other circuit, then DC and low frequency signals that can't couple across the intervening free space gap won't interfere with it.

Coupling and isolation are near-opposites. High isolation and low coupling are much the same.

I read a book on shielding techniques the other night, for fun. Yes, I have a somewhat odd idea of fun. The book talked about the shielding effectiveness of various shielding techniques, and compared that against the strength of the possibly-interfering signal, and the sensitivity of the receiving circuit at that frequency band, and made suggestions on how to think about how those worked. If, for instance, a received signal of 100uV is the threshold below which you don't care about or can't detect interference at some frequency, and the circuit couples signals in that frequency band in with a loss of 32db from the free air, but there is a 100kW radio transmitting tower putting out +85db of signal at the frequency, then you need to shield enough to cut the signal by 53db with reference to 100uV at that freuqency. Riveting reading, really. 

Coupling can be capacitive: literally every conductor in the universe is capacitively coupled to every other one by the intervening electrical field. But this drops off as the square of the distance between the two conductors, so moving things apart is a very cheap way to decrease their capacitive coupling. Capacitive coupling drops with the drop in surface area of the two conductors, so making them smaller *and* further apart is good too.

Inductive coupling works on magnetic fields. If the sensitive stuff shares a conductor loop area with another conductor loop, any current in one loop is transferred to a voltage in the other loop by the change of the magnetic field in the shared area. You reduce magnetic coupling by making the loops small, so they don't emit well or receive well, by moving them further apart (inverse square law again) or making them orthogonal so the flux couplings don't cross the conductors well.

Interestingly, it's almost impossible to shield out a DC (i.e. permanent magnet or static electromagnet) magnetic field. About all you can do is try to attenuate it by shunting it away from sensitive stuff with high-mu iron and such.

Electromagnetic radiation as RF is coped with by the laws of antennas. For good radio reception, you want good antennas. For shielding and RF immunity, you do everything you can to make BAD antennas that don't pick up. A solid metallic enclosure is a very bad antenna indeed for stuff inside. Slots, holes and gaps actually make for TUNED reception antennas at frequencies dependent on their dimensions.

I'm blathering, trying to hit something that may be useful for you. The principles are the same: identify what your circuit is sensitive to, and to what degree. Identify what the ambient interference is in terms of frequency and amplitude. Then figure out how much you need to knock down that ambient soup of interference, then shield, filter, and attenuate till you get there.

See why it's hard to enter search terms?

Actually, my old standby "Noise Reduction Techniques in Electronic Systems" by Ott is a great packaged introduction and reference to the issues. But it's a deep pool to be tossed into.

[defaced]

Thanks R.G.  Thankfully I'm at least familiar with those concepts so that made sense.  Funny you bring up the Ott's Noise book, I bought it about a year ago after you mentioned it  Last time I opened it I got as far as looking up a few topics (grounding of the shield on shielded wire was one I distinctly remember), but that was about it.   I think I'll take it home and crack it open again.

I know all about riveting reading.  My stainless steel welding metallurgy book used to be my go-to book when I needed something to read.  Nothing like reading about alloying elements and phase morphology to kick a night off.