Ideas for DC step down design

[akis]

For my existing bench PSU I have wound a few turns to get a 5V DC independent of the main sources to be used for the voltage and current meters (they are old style and pull considerable power). I imagine I could unwind half of it, tap it, and wind it back on so as to get a mid-range tap on the secondary. The wire is thick however and it would require a lot of patience.

The switched caps scenario need not be as bad as it sounds. The caps will not be discharged fully on every cycle. After the initial charge cycles the state of the switching caps will depend on the behaviour of the load. And like half-way/full wave rectification, we can add more switching capacitors to charge on the opposite side of the wave. I am working on a schematic, I will post it once it works.

[dannyf]

The switched caps scenario

A switched cap design has low current capability. smps has low efficiency under light load. So if your application is one where a switched cap power supply can potentially work and you care about simplicity, a linear regulator is the way and maybe the only way to go.

You are confusing two cases here: you used the battery charger example to justify the use of a smps when you are really designing for a low load situation.

[akis]

works quite well on breadboard.

The 2nd one wouldn't work - it is plain wrong.

Both lack any meaningful current capabilities.

It works really well. Both circuits are academic shown here just to discuss ideas. It can be used to power op-amps, I only need 2-5mA standby and 5-20mA at full load. If you want a quick -Ve from a single supply, eg say a 9V battery, the second circuit is quite handy. Depending on your current requirements you can power it straight out of the 555 IC, otherwise I have added a simple pass transistor or a push pull pass. The diodes and caps are there to shift the base voltages away from each other (the PNP base sees a double voltage and the NPN sees a negative voltage) so that the overlap between the two transistors is minimal otherwise they conduct at the same time and draw a lot of power.

[dannyf]

It works really well.

In the 2nd circuit, the transistors aren't even biased.

[dannyf]

I only need 2-5mA standby and 5-20mA at full load.

A to92 linear regulator will do more than fine in such a case. You are wasting brain cells trying to get a smps to work within your constraints here.

[WarSim]

There is not a best topology for every possible power supply. 
Apparently you need an ever increasing number of power supplies. 
Pick one of them, make it the go onto the next. 
Asking about several at the same time will confuse things. 
At least name the supplies so others can keep the issues straight. 

[silent]

Hello.

What is an acceptable efficiency for You?
I have faced a similar problem some time ago. I needed a 12V to 5V converter with 3A output. In my previous project I used LM2576 converter for 1,5A load which did the job, but required the radiator (4x4x1.2cm) and when the load reached 2,5-3A, it was able to fry an egg. The second problem was the required inductor - 100uH/3A - hard to find in SMD.

I have found a nice alternative - ST1S10. It is a good and cheap (4,5PLN in Poland - about $1,5) step-down switching regulator with 3A output. IC is available in tiny DFN8 and SO8 package with heat pad on the bottom, so the radiator is not required - the PCB is a radiator, but 2-sided board is required (recommended layout is available in datasheet). The high current-rated inductor is no longer a problem, because due to high switching frequency, 2-5uH is enough. Output voltage is set by two resistors, but maybe it could be set dynamically, i.e by a microcontroller and a DAC. Heat problem is also gone - efficiency reaches 90%. The circuit has also a thermal protection (I have tested it with thermal pad not soldered - it automatically switches off) and survives 20V input voltage (I have accidentally plugged the device to the second bench power supply (wrong hole! ;-) ) set to 20V).

Over the past 12 months, I have assembled about 100 devices with this converter, which are running mostly 24/7. There was no problems with it so far.

I think the switching converter is a very good option.

[akis]

You've gone from a battery charger to a varable bench-top regulator to a regulator with 4-5A capability now you are down to 2-5mA?

No, I imagined we could take ideas, schematics, and then scale appropriately. For example, if had a circuit to do DC 20V to DC 5V at some low current, what changes would I then need to make to increase the current capacity? Or if someone looked at DC Inverter circuit I posted above, what changes would need to be made to increase the output current capacity?

Acceptable efficiency, off the top of my head, would be no more than a couple of watts on the switching transistors, when the linear regulator wastes over 20W to do the same job, ie 10% of the equivalent loss on a linear regulator.

[Zbig]

No, I imagined we could take ideas, schematics, and then scale appropriately. For example, if had a circuit to do DC 20V to DC 5V at some low current, what changes would I then need to make to increase the current capacity? Or if someone looked at DC Inverter circuit I posted above, what changes would need to be made to increase the output current capacity?

Acceptable efficiency, off the top of my head, would be no more than a couple of watts on the switching transistors, when the linear regulator wastes over 20W to do the same job, ie 10% of the equivalent loss on a linear regulator.

I can't help but think you somehow feel that few of your random ideas followed by some quick brainstorming on the forum could lead to some new, groundbreaking, 100%-efficient power conversion method, perhaps by smoothing PWM with just right configuration of some capacitors. Well, it's not going to happen so you can stop now. Ever heard of NASA improving the performance of their rocket engine after receiving an e-mail from some midnight hacker? Yeah, me neither. There's decades of engineering gone into modern, highly efficient switching power converters, there are people doing it for a living. Your time spent trying to reinvent the wheel will be the time wasted, a time you could spend by building something genuinely new and useful. But who am I to judge.