Seems they are targeting business esp on AC to DC conversion possibly for office buildings.
Seems they are targeting business esp on AC to DC conversion possibly for office buildings.
I had no idea that the Acrich products were that important for Soeul. A quote from the article:
"It has trademarked the light "packages," which can be placed into many different kinds of final goods, from table lamps to car headlights, that run on AC with the name Acrich."
I also had no idea that car headlight were powered by AC! I'll just assume that this is just a techo typo by the writer.
So.... what is the Acrich technology? What are they doing to power the LEDs that no one else has figured out? I just assumed that they included a bridge rectifier, but I haven't looked into it at all to see if they've included a current regulator or what.
acriche is just a big die with a bunch of little LEDs etched into it, basically. You only need a resistor and a heatsink to hook it up to mains.
years and years ago I saw some acriche-powered bulbs at walmart. I waited several years for them to make their way to the reduced section. They reduced them by $0.40, still not cheap enough for me to bother with. They were cheap enough for someone to bother with, though, someone bought them all up I guess.
They were GE branded, and marked as 4W GU10 bulbs, I think.
that does sound familiar. Hard to believe that SSC would be that excited about it, though. Using a resistor to control the current is fine for low power lights, but to run high power will require a proper switching power supply, and their big claim to fame (according to the WSJ) is that the Acrich avoids the poor reliability of power supplies.
I'd be more impressed if the Acrich included an on-board linear current regulator, but there is still the issue of a proper power factor that would need to be addressed if SSC has ambitions to being a big player in mains powered lighting.
Put enough LED diodes in series and you don't need a driver circuit to run them off mains power, just a bridge rectifier. I have build LED lights using this principle. And perhaps you could even fashion the rectifier out of the LEDs themselves. Yes it takes a lot of diodes, but in a multi-emitter LED with 50 or more emitters, this should not be a problem. However, since everything is in series, if one diode fails, they all go out....
But what you save in a driver you more than lose by the need of having every part of the leds being high voltage - complete with both security and insulation requirements (you cannot really trust the "to ground" insulation of typical mountings for 100s of V...)
Also, this still leaves potential runaway do to sinking Vf with increased temperature.
There is also sometimes a slight decrease in Vf as the LEDs "burn-in" or age. Both this and thermal run-away are easily planned for by including a fuse in the circuit. On one fixture I built, about 3 months after putting it in service the 300ma fuse blew and I had to add about 4 more emitters to get the current back down where I wanted it
Building a light like this is not for the faint of heart, nor for anyone who does not understand how to properly ground and insulate the high voltage bits. I had access to a number of obsolete power LEDs and set out to make a few low cost, reasonably efficient fixtures as work lights
The forward voltage of the LEDs themselves will limit current, so long as the combined forward voltages of the series string of LEDs is greater than the supply voltage. A low ohm resistor is not a bad idea, it can protect the LEDs from small power spikes, but a resistor will waste power. For efficiency my target current was around 150-200 ma current through the LEDs, the lumens/watt of first and second generation luxeons is not too bad at that current and waste heat is less of a problem. To get that current I use about 36-40 white LEDs, and 15 amber/red for a 120 volt AC source fed through a full wave bridge rectifier. The calculated Vf of this string is about 166 volts but is in line with the fact that peak voltage of a 120 volt RMS AC supply will be higher than 120 volts. I put what I thought would be more than enough LEDs in series and gradually reduced the number until I hit the target current. Each lamp has a different number of emitters because of so many individual variations in Vf/LED brand/color combined. Once the heatsink was at a stable temperature, and again after some hours of "burn-in", I made small adjustments to the number of emitters. The lights will win no beauty contests, but they serve well as worklights and are more reliable than the halogen fixtures they replaced.
But if I was building a multi-emitter LED light engine, with all the emitters well matched, I could use this method to eliminate almost all of the need for a driver when operating off mains voltage. but yes, you probably would need at least a resistor as "insurance"
well, designing the forward voltage of a string of LEDs so that it barely turns on at the peak of the AC voltage is one way to do things. The disadvantages of this approach are:
1. that the LEDs only conduct for a very small portion of the AC waveform, increasing the flicker of the light.
2. that any minor variation in the AC voltage (due to local load variation, poor wiring, etc) will either produce noticable dimming or increase the LED current sufficiently to cause damage.
3. that the LEDs will need to be oversized and underutilized in order to handle the large variation in current that results from item #2.
If the goal is the highest efficiency in terms of lumens per watt, then this approach might be the best. There are certainly no losses in a regulator circuit, and the need to run the LEDs at lower than rated currents means that they will be more efficient. The price that is paid is higher cost, increased flicker, and lower reliability (...possibly...).
Adding a resistor is certainly a simple and very reliable way to limit the LED current, although at a cost to efficiency. Linear and switching regulators improve regulation and should have minimal impact to reliability, assuming that they are designed with suitable derating of components. The more complicated/sophisticated designs will add cost relative to the resistor, but could be worth it in terms of higher efficiency, etc.
I have to admit that this is the first time I've heard of someone building an AC powered LED light w/o current limiting! I wouldn't recommend it, but as long as it works for you and makes you happy, then it's a great thing! (well, and it also doesn't seem like it should set your home on fire either... you do have a fuse wired in series with it, don't you?). Can't be too careful!
Funny you should mention #2, part of the inspiration of the "almost all LED design" was to see if I could come up with a way of making an LED light that could be dimmed with a common household dimmer. It works, and is pretty surprising the range of dimming you can get. Most drivers are not dimmer friendly, or require different dimming protocols.
Flicker is not as bad as I thought it might be, the full wave bridge means it should flicker at 120 Hz, but it really isn't that noticeable. I was prepared to add a capacitor to the circuit if I needed to.
Yes there is a fuse, and if one was to build a light like this with anything but "surplussed" LEDs the $ per lumen would be very high, unless it was designed for typical drive currents using modern efficient emitters.
I found a little more info on the latest Acrich modules from Seoul... it's from Design News magazine, in a little advertising/info column by Cary Eskow of Avnet Electronics:
borrowing from the article:
"The black IC noticeable in Figure 2 performs most of the work. Internally, it rectifies the 110 or 220 VAC source, measures the instantaneous DC voltage, and selectively drives several independent strings of series-connected LEDs using built-in mosfets. As the AC cycle advances its phase, the IC turns on more strings on the module. Think of it as an "automatic transmission". Since only the fewest number of strings needed are on at any moment, efficiency is high; "
It does sound like it is a matter of switching LEDs in series as the voltage rises and falls. It does seem a bit complex in principle, but if built as a single IC, then it should be quite reliable. I wonder if they are available in low quantity??
Interesting! I don't think their first acriche arrays had this IC.
Agreed! The first gen Acrich didn't impress me that much, but this is much more intriguing! This design is conceptually complex, at least to me right now. The lack of large components is very attractive, and especially the lack of low reliability caps. The ability to maintain a good power factor is also impressive.
I did find that Mouser carries these, and the prices don't seem bad. The luminous efficacy didn't seem great.... around 70 lumens per watt, at least for the 8W module that I looked at. The lack of losses in an external power supply may compensate somewhat.
All in all, I'm very interested in trying some of these out for a nice table lamp or similar!
just an update.... Mouser has the 8w version for about $11. I'm thinking about getting four of them to make a replacement for a bathroom light that currently uses four 40W incandescents. Just need to add a 22 ohm series resistor and MOV for each. Much nicer than needing to provide a power supply! Honestly, the ability to get rid of a switching power supply and all of the heat and poor/unknown reliability parts is a big attraction.
I'm just wondering if anyone has experience with these yet. Any concerns? Flicker? Color? Failure rate?
One thing that I can't find an answer to is whether the circuitry is isolated from the MCPCB... I'm assuming that it is.
thanks -- Steve K.
I'm thinking you will probably end up being the guinea pig. I do like not having to find a spot for a bulky PSU, and might consider these for a project of mine, a floor standing lamp that shines a 400W halogen at the ceiling. Not that I plan on matching the output with acriches..., but something that provides light without heating up the place.
For something driven directly by AC, the Acriche 2 seems to be the best thing available. A shame that they aren't a bit more efficient, but the price is reasonable... $11 for a 8W module. That might be a bit pricey for a 400W light, though. I think that 3 or 4 of the 8W modules will replace my four 40W bulbs, so you might need approximately 10 8W modules (or just use the larger modules). Shouldn't cost more than $110 for the Acriches. The bigger question might be the cost of the heatsinks! Might be worth checking on the final cost if you were to use the new Crees that are happy to bake at 85C.
I have a bunch of active heatsinks, just not sure how well they'd do in passive mode. The most complex part of this light is starting to look like the circuitry to drive the fan lol