For a project I am working on, I need 5 Volts @ 10 mA maximum - to be supplied from 230 VAC 50 Hz mains.
The output needs to be galvanically isolated, but not "safety" isolated.
The output can have up to 100 mV ripple.
The solution needs to be small and cheap.

So far, my idea is to use a microchip SR087Adjustable Offline Inductorless Switching Regulator to power a small isolated 24V to 5V DC to DC converter.

This looks like a BOM cost of around $5.50

But it feels like thundering massive overkill?
This could output 1 Watt, but I only need 50 Milliwatts.

Ideas invited.

 

 

Thanks for your design input.

I cannot use a "Capacitor Dropper" configuration because the "GND" of the power supply needs to be connected to a floating node in another mains-connected circuit.
The actual project involves adding DALI dimming capabilliy to existing low-cost mains-powered LED lights that are typically not dimmable.
The idea is working, but I need to power the Micro and the switching MOSFET with a floating 5V supply.
For my prototype, I am using the guts of a cheap USB phone charger, which is essentially an isolated 5V SMPS.

The DALI interface provides isolation through optocouplers both sending and receiving data from the fixture.
The optocoupler on the receiving side is powered by the DALI bus itself.

 

Is there some reason why you can't use a Transformer ?
Does the Dali-Bus run over the mains wiring, or does it use separate Low-Voltage Control Wiring ?
What topology do you intend to use to Dim the cheap "non-Dimmable" LED Fixtures ?
Are you intending to create a "One-Way" communication Node on the Dali-Bus ?
Will the Dali-Bus continue to attempt 2-way Communication with the new One-Way Node ?
Or does it have an option for a One-Way-Comm Link Output ?

Some "cheap" LED Supplies will produce noise that may
interfere with the data integrity of the Dali-Bus Network.

A Micro Controller by its self may draw more than 10ma.
Even a tiny dip in Power Supply Voltage may cause your controller to re-set.
If you are PWM-ing a FET for dimming,
pumping the Gate Capacitance of the FET will probably draw more than 10ma., ( certainly higher peaks ).
So, your proposed Power Supply may not be over-kill at all,
( not withstanding its possibly questionable application as a solution to your problem ).

Providing documentation for your current ideas regarding the solving your problem is commendable,
( and unfortunately unusual in these forums ),
but you have not provided the needed details for getting from point-A to point-B.
So, as always, Please clearly state the exact, underlying problem, that you are trying to solve,
rather than asking for a solution to a problem that may help an odd,
and possibly very impractical, workaround do the job.

It sounds like you don't want to have to pay for a "Dali-Bus Compatible" Ballast/Power Supply,
are they really that expensive ?
If this is the case, you are adding quite a bit of complexity, and potential points of failure, to the system,
and possibly spending more time and money than a
"drop-in compatible", or external, Dali-Bus Compatible, LED Power Supply would cost.
.
.

 

Thanks for your continued interest.

The photo below shows what I have accomplished so far.

I have developed a piece of hardware that accomplishes the basic goal.
My hardware connects to an existing LED ballast with 3 wires and dims the LED output by shunting the current regulator with an FET.
When the control FET turns on, the current continues to flow in the ballast's switching inductor, but the voltage drop over the inductor is reduced drastically so the current decays very slowly.
The current regulator continues it's task without caring about the output voltage.
PWM controls the brightness at around 500 Hz, which is also smoothed a bit by the existing capacitor across the LED's- so there is no visible modulation.
The tricky part was eliminating the flicker caused by the asynchronous switching of the regulator and the PWM, I solved this by synching-up the PWM edges with the current regulator, since they are drastically different frequencies, it all works well

The whole setup draws 3.5 Ma from the 5 volt supply, the 3.3V PIC16LF1575 is very power thrifty.
The 5V supply is referenced to the MOSFET drain, which is at a voltage potential that floats relative to the mains. (the tricky problem)
I want to boil this all down to a small board that can fit with the stock ballast in a small box.

As I write this, it occurs to me that I might investigate ways to bootstrap the gate drive, so the PSU could reference the NEUTRAL?

The big picture idea is to outfit my home with inexpensive and elegantly controllable lighting. DALI is a nice simple standard that is widely used in architectural lighting, but the fixtures are usually very expensive. I believe they are artificially expensive as the market is generally institutional and not terribly price sensitive. These standard consumer lamps are so ridiculously cheap that adding a $15.00 board that results in a $20.00 DALI dimmable lamp seems worth the effort.

DALI has not caught on in the consumer market, people would rather use IoT stuff which I personally dislike, I want a simple and robust wired solution that I can understand and maintain myself.

 

hi Sensacell,
I used the NMA versions of the potted DC>DC converters, usually 12V > 5V or 5V > +/-12V.

On lighter loads the Vout can rise to about 10% above rated value, also the conversion efficiency drops off a low loads.
For any project with low signal levels, I had to add additional +/-12Vdc filtering.

E

 

Well, to each his own ...... Go for it man ....
But I think you are making the whole arrangement way too complex.

If I were to consider doing something like this, I would skip the Dali Hardware all together
and just design my own complete, centralized system.

I'm not directly familiar with the Dali System, but I'd be willing-to-bet that you can get a
Remote Controlled LED Power Supply that will run what ever number of fixtures that you'd like,
with zero electronics in any of the fixtures.

Measure to find out the average Voltage drop per fixture, and find the maximum Current Rating,
then wire your fixtures in series with each other,
if you get too close to your desired maximum Voltage, when calculating the number of fixtures that you want,
then create 2 equal length strings of fixtures, and then wire the 2 strings in parallel.
Each String of fixtures will always require the same regulated Current,
but the Voltage Drop will vary with the number of fixtures in a string.
2 equal length strings, when wired in parallel, will need twice the Current.
.
.

 

You cannot find fixtures that contain just LED's- a "Fixture" is typically a self-contained unit with a driver and LED's - unless you want raw LED strips or other unpackaged lights engines.
Half the battle is the housing, it's gotta look good and be easy to install and not cost too much.

To control a whole room full of lighting, you really need a master-slave configuration with many drivers or "ballasts" as they are sometimes called controlled from a network.
This seems to be why the IoT stuff seems to have taken off in the home lighting world.

The nightmarish complexity of the IoT idea makes DALI seem brutally simple by comparison!


https://en.wikipedia.org/wiki/Digital_Addressable_Lighting_Interface

 

I hear ya,
but those built-in boards are super easy to remove.
And there's no advantage to having an individual Power Supply inside each fixture,
unless you have absolutely no access to the space above the fixture to run new wire.

The fixtures probably require less than 1 amp, ( I'm guessing ),
so a single ~22-gauge Wire is really all you need between each fixture in a string.
~50 Volts is probably enough to power ~10 Fixtures wired in series,
depending, of course, on how the multiple LEDs are wired on the circuit board inside the Fixture,
and it's really easy to make a ~50-V - 1-A Power Supply with a single SMPS-Chip,
or you could probably buy ~10 of them pre-made from Ali-Express for ~$15.oo.
The entire SMPS circuit board might even fit inside a standard Electrical Junction Box.
.
.

 

A USB phone charger might be the ticket.