A member in another board wants to replace some 20w 12v halogen spots with LED retrofits. I asked him to check the power supply type, and it looks like the standard dimmable electronic HF AC device

Now as i know those emit quite high peak voltage, if that is rectified it won't make the 17V DC you get from a magnetic transformer's 12V AC sine, but way more

What are the actual power requirements of the 12V LED retrofits ? Do they rectify into a capacitor ? Do they use PWM then or just linear ? Do they work well on 12V or 17V DC input ?

He said he tested one LED lamp on the electronic supply and it worked ok, but i dont think it means a lot - If he exceeded the max v of the electrolitic or such there will be smoke, maybe not right now but later

He also asks about converting to 240V LED retrofits, i think its bad (less reliable than the 12V due to more circuit, with high voltage and with higher heat output being crammed into same space), any experience of yours ?

The LED retrofits use a diode (usually Schottky, due to low dropout) bridge on the input, then quite undersized capacitor (so it does not filter any of the 100Hz ripple, only the high frequency). Downstream is quite regular buck converter controlled for constant current output.
The voltage rating for all the semiconductors is at least 40V, but the electrolytic capacitor is rated only at 25V. So that mean it could well tolerate operation voltage of 25V (include all the tolerance contributors) with up to 40Vpeak occasional surges.

I never met any "electronic halogen transformer" doing overshoots, all of them generate a rectangular wave at few 10's of kHz, amplitude modulated by the rectified 50Hz sinewave. Given the basic circuits, there is no place to cause such overshoots...
So after the diode bridge (it have to be fast diodes, what Schottkey's are) the output matches the output with regular 50Hz transformer.

What would have problems are not the LED lamps, but the "electronic transformer" by itself. In order to work correctly, it have to be loaded by a resistive load of some minimum wattage, usually about 20..50% of the wattage rating. Without that their transistors may overheat and the "transformer" die...
With the LED's, neither the minimum load, nor it's resistive character requirement would be met, so I would expect the "transformer" to fail quite soon.

If the transformer overshoots (and i do expect some 20-50V peak from it), i see a problem if this is charged into a 25V electrolitic

Now how about 12V DC input ? Do they allow it, or they count on the capacitor having at least 17V in it ?

If the transformer overshoots (and i do expect some 20-50V peak from it), i see a problem if this is charged into a 25V electrolitic

What overshoot is the classical "core and coil" bulk transformer, mainly when large part of it's load disappear (e.g. the halogen lamp burn out).
The aluminum electrolytic rating stand for long term exposure (maximum tolerance with regular ripple on it, what may appear for longer than 1second), for occassional short time surges they allow way higher peak voltages (the 25V ones allow at least 40V overshoots with 1 second time limit - the component start to dissipate a lot of heat during that overvoltage event, cumulated time for such exposure over the lifetime in hours).
The electrolytic capacitors do exhibit a self recovery on an overvoltage breakdown (the damaged oxide layer is grown up back by the electrolysis), but during the recovery process dissipate heat, therefore the time of one event is limited.
And unlike in the film capacitors (where the damaged part is disconnected), the recovery in the electrolytics is a real 100% recovery of the dielectric oxide layer. But as the recovery generate heat and even gases, it put the rubber seal under larger stress, so with the recovery process happening too often, the seal would fail sooner.

Now how about 12V DC input ? Do they allow it, or they count on the capacitor having at least 17V in it ?

It is the real minimum: LED's are usually in series triplets, what mean about 9..10V drop on them. Additional 0.5..1V drop across the input diode bridge (two diodes in series). In some literature is depicted the use of two MOSFET's as synchronous rectifier instead of the bottom side diodes, what save voltage drop of one diode. But in real products I've seen only the diodes.
So the regulated 12V supply (so an AC/DC adapter,...) gives only 1V minimum headroom for decent current regulation, tolerances and wiring drops, what is really a minimum for decent operation.
Some cheepeese models use even regular 1N4002, what mean voltage drop in the 1.5..2V range, what mean the 12DC won't work correctly.
The switch element in the buck could be moreless neglected (given the other drops and their tolerances).

It is not suitable for 12V battery operation, as the "empty" voltage of 10V is way too low. Only the types with the synchronous rectifier would work reasonably (so you would have to replace the bottom two diodes by MOSFET's)

Operating at higher voltage is OK, all of them would work correctly in the range between 13..24V. It is actually what they are designed for...
So virtually any notebook supply (15..22V) with at least twice the power rating would work well, only the halogen capsules could not be used anymore.

That's an interesting idea for alternative uses for these power supplies .. they are cheap to buy and have a good supply voltage range!


BG

I am not about the single event spike when switching off the magnetic ballast. It normally outputs sine voltage, with a peak at 17V

I am about the electronic ballast, which can emit series of spikes (with 12V RMS), subjecting the capacitor to permanent overvoltage (as it charge tothe peak value not the RMS)

I am about the electronic ballast, which can emit series of spikes (with 12V RMS), subjecting the capacitor to permanent overvoltage (as it charge tothe peak value not the RMS)

The electronic halogen ballast does not generate any overvoltage. On it's output is the 50Hz sinewave chopped in polarity by the few 10's of kHz. When you rectify that using simple diode bridge (well, you have to use fast recovery rectifier, otherwise the diodes would burn out; but in the MR16 are mostly used Schottky, what have virtually zero recovery time, so in that matter ideal), you get exactly the same as when rectifying the secondary of a regular transformer. And with "exactly" I mean really exactly, including the 17V peak value of 12Vrms output (when neglecting diode drop).

And as the electronic have way lower output impedance, this is way less affected by any eventual switching events on the secondary side than the regular transformer. So if something could generate overshoots, it would be the "classic" bulky 50Hz transformer and not the electronic ballast...

How is it going so far for that member? Any smoke or other problems? I’m thinking of retrofitting LEDs to some of my halogen lamps too but I’m not quite sure how to go about it…

have not heard from him since but maybe he just left it as halogens

if it would be me i'd get one, open it, power up and measure the voltage on the capacitor

How is it going so far for that member? Any smoke or other problems? I’m thinking of retrofitting LEDs to some of my halogen lamps too but I’m not quite sure how to go about it…

With regular bulk iron/copper wire transformer you would have no problems at all.
With the "electronic" transformers you have to keep at least one bulb as the halogen, preferably close to the "transformer" in order to provide the necessary load for the "transformer" to operate. Otherwise I would not expect any problem either, given the LED's ballast electronic is reasonably made...

Quote from: lesliemorris85 on February 07, 2013, 09:16:39 PM

How is it going so far for that member? Any smoke or other problems? I’m thinking of retrofitting LEDs to some of my halogen lamps too but I’m not quite sure how to go about it…

With regular bulk iron/copper wire transformer you would have no problems at all.
With the "electronic" transformers you have to keep at least one bulb as the halogen, preferably close to the "transformer" in order to provide the necessary load for the "transformer" to operate. Otherwise I would not expect any problem either, given the LED's ballast electronic is reasonably made...

I see, thanks a lot for this information.

today i replaced a lot of burned halogen with led lamps.
despite they are cheap (8€/pc) they run very well on electronic halogen transformer.
however, one or two halogen lamps are required to operate others lamps (otherwise, they tend flicker)

would have done some testing but i recycled the "electronic transformers" i had here.
they are all rf polluters.
i remember a friend modding a cheap desk lamp that had one and it overshot badly enough to kill the 4 luxeon k2 tffc neutral whites he had installed.
that sure sucked as these parts were in short supply and high price at the time.it got a radio shack iron transformer installed.he had me wire it up.didnt want to poof $50 of led's again!

kill the 4 luxeon k2 tffc neutral whites he had installed.

I'm just curious, how the LED's were connected to the "transformer", what kind of ballast the setup used.
Because all the "electronic" types I have ever seen differ from the bulk transformer only by the low output impedance and no overshoots when the load is switched on the secondary side.

At this point I don't count their reliability and robustness, as that would kill only the "transformers" and not the LED's...


I have some idea about what had happened to the poor LED's, but I would like to know the arrangement first, before posting that idea here...

If that lamp was well isolated around the 12v output, they migh have used something nasty like a choke or electronic "dimmer" to make the "12V"