That is not floating, that is isolated, a totally different concept. None of the circuits presented will work in that case, as the only way to get an isolated supply is with a transformer.

Terminology is very important with electronics. Ground is another issue altogether, which several people did try to explain.

Ground is one of those concepts that can be confusing since it's roots come from several different sources (this is what David Bridgen was trying to point out). That one word means several different things.

When you blow people off you might try to understand they are trying to tell you something important. A switching mode power regulator is fundamentally the same in function as an analog power supply, but it converts with higher efficiencies. I did try to mention transformers several times, if you read the earlier posts. To get isolation, if that is what you truly need, you are going to have to build the equivalent of a AC/AC inverter followed by a DC power supply.

 

That is not floating, that is isolated, a totally different concept...

Sigh!

Isolated is when there is no electrical connection whatsoever between the input and output. Floating is where neither output is tied to either input. Subtle difference. The latter is what I wanted, as I stated in my OP.

All sorted now, without a transformer

 

Just out of curiosity, what did you wind up doing?

Here's what I was referring to in my prior post (see attached schematic/simulation; note that I revised the flyback transformer and added a bypassed Zener and optocoupler to provide the feedback to the 555's control voltage input. C5 is key to providing a soft-start function; without it, overshoot at start-up is rather severe.)

The simulation shows a ground on the output side; but that's just so that it didn't take forever to run, as if it were truly isolated in the simulation, some very large numbers are generated.

pssst.... Bill.... this is a good one for you to experiment with...

 

This is a floating supply, I can short one of either outputs to the source and nothing will happen. Not so with a SMPS or analog regulator, since they do have a common reference somewhere (as well as DC connections throughout the circuit), which can cause fire and smoke.

What kind of transformer are you thinking of Wookie? Winding something on a torrid?

Thanks for the idea, it is a keeper. I have bookmarked it.

 

This is a floating supply, I can short one of either outputs to the source and nothing will happen. Not so with a SMPS or analog regulator, since they do have a common reference somewhere (as well as DC connections throughout the circuit), which can cause fire and smoke.

Well, this DOES fall into the rather broad category of SMPS; it's a flyback converter. Remember those dozen SMT UC2842D IC's I sent you? (#18 on the packing list.) They're one of many switch-mode supply IC's that a pro designer would use in a design like this, but I thought you would have fun with a SMPS that had a 555 at it's heart.

What kind of transformer are you thinking of Wookie? Winding something on a toroid?

Yep.
Grab one of those item #39 T50 flat blue toroids (Al=1200, tests #24, 36), and use 9 or 10 turns on the primary, and 6 turns on the secondary using any magnet wire from AWG18 to AWG26. That should work OK for low power stuff. Best if you wrap a little tape on the toroid before you wind; that helps to delay the onset of saturation. I don't know all of the characteristics of those particular toroids; just the AL values derived from a few informal tests. I'm betting that they will saturate fairly quickly unless you tape them. I use a special 3M (tm) tape that's designed specifically for things like this; Permacel and #1350.

I should send you some more of the Micrometals toroids to experiment with; I've picked up a few more since I sent you that box o' stuff. I can send you some of the Permacel and #1350, too - but even masking tape or Kapton would work fine for just a test. The Micrometals Kool-Mu toroids don't need to be wound with tape; they have distributed gaps (voids) in the toroid itself, which serves the same purpose, along with eliminating eddy currents.

Thanks for the idea, it is a keeper. I have bookmarked it.

My pleasure. It's an interesting use of the control voltage pin to achieve a voltage-controlled PWM using a single 555.

I took Ronald Dekker's original work from his website and simply changed a few things around. His original schematic didn't have a snubber on the primary side of the transformer; that would result in very high voltages when the MOSFET turned off, and would've blasted the IRLR7821 off the board.

Also, I modded the feedback so that it could be completely isolated from the primary side. If you ever build a flyback that runs off the mains, you'll have to do something similar to this. As shown, the supply would probably work OK up to around 300-350mA. [eta] Increasing C3 to 470uF or higher will help keep the output ripple voltage down. If you wanted to go higher than that, you'd have to change the turns ratio on the transformer closer to 1:1 rather than 3/2 or 10:6.

BTW, the IRLR7807's I sent would work great for this - sure, they're overkill, but the price is right. They have a super-low gate charge (~5nC's), so they'll "snap" on and off very quickly, and don't worry about heat sinking them; just use some nice fat pads on the board and large traces going to wherever.

If you want more than a couple hundred mA output, you may need to go to a larger, better toroid. Like I said, I haven't performed extensive tests on those particular toroids. If it seems to be getting warm, and/or the power out starts dropping unexpectedly after a period of time, that's what I'd look at.

Oh, for the diodes - you don't have to use exactly what I used - but don't use slow rectifiers like 1N400x or 1N540x series; use fast-recovery or Schottky diodes; at least 1A rating. The slow diodes will kill any hope for performance.