Current mirrors are transistor based that controls the current acurately between two parallel strings. The method that Meanwell uses is simple, but doesn't save you from total disaster. Plus it adds a ton of wiring. A current mirror is placed at the end of the strings.

The ELN-60-48 can be turned down to just under 1000mA.

Dag Nabbit Evil, now you've got me worried!
So I've swapped one of these( ELN-60-27,non-dimmable) for a for the buckpucks. Two // strings of 8 LEDs (a mix of blues and whites to a color temp of my taste). I realize no dimming - but its not important.

I'll have to recheck my numbers tonight, but I'm at ~28V and ~800mA per strand. So far, this puppy is working great! but what about, as someone alluded to above, when an LED goes out down the road? What do I need to "protect" myself?

Thanks,
Nick

I have been doing more reading into it, and current mirrors can be made really easily with two transistors.



Article here.

Interesting article. The last section really describes how the current mirror protects things.

1st: an led magazine!!! Sweet!
2nd: sweet article, I can't believe the variation that can form across two strings. A must read for all!!! Thanks fors sharing.
3rd: This does seem very simple. Can I get these tranistors and resistors at RS, any suggestions for protecting these from the "elements" of SW?

Thanks, Nick

You should be able to get similar transistors at RS, but you won't find the right resistor values. Might as well get it all online.

When it says the two transistors need to be thermally connected, could this be as simple as them mounted to the same piece of aluminum (scrap piece) with some AS5? or do they need need something with more mass, like a heatsink?

I guess what I'm asking is will these things generate a lot of heat?

Thanks, Nick

They could generate heat, but it's more to make sure they drift at the same rate because of temperature changes.

evil you make my head hurt brother !!!

How the hell do you retain all of that ?

Thats a good question

Hi Sherman,

any finding ?

Await for result !

Hey all,
saw this thread from another one (evil referenced this one).
1. Did anyone actually get their hands on a dimmable version of the power supply? There's a much cheaper solution that would probably work, and
2. You should be able to use that PWM circuit evil posted on your existing non-dimmable power supplies. Just connect the source of the transistor (the part with the arrow) to ground and make sure your LEDs are above that. You have to make sure the FET (transistor) you're using can handle the current and it's a 'logic level' FET. From a cursory glance the schematic evil posted looked good but if anyone wants to try it i'll look at it closer. Worst thing that happens is that you let the magic smoke out.
3. When you're building current mirrors, for the two matched transistors, use ones that are packaged together. Here is a quad-packaged set of BJTs (another type of transistor, different from a FET) that you can use for current mirrors. In addition to being thermally coupled, you can be sure that you'll have the same process characteristics -> good mirroring. However, you'd probably have to look a while to find a set of BJTs that would support 1A for the LEDs folks are using on this forum.

edit: that quad pack of BJTs would work fine for a 350mA current mirror as above
edit2: I now see there's a dc-dimmable version and a pwm dimmable version, #1 should work for the dc-dimmable versions

zingtaw,
I'm new to this thread (and to the site too). I have something simple you can try:
If you're using the 48V version, connect a 36K resistor to the +48V output. Connect a 10k resistor to the 36k resistor and then to ground. Get a cheap BJT (transistor, 2n222 works), connect the collector to the 10k/36k node, the emitter to ground, and your arduino output to the base. The DIM+ is connected to the 10k/36k/collector node and dim- to ground. You also have to make sure your arduino is connected to your PS ground.

The 36k/10k resistor combo is a 'resistor divider' and creates a ~+10V reference that only draws 1mA of current. The transistor switches that node between the 10v reference and ground.

This and other resistor divider solutions (like a 36k resistor in series with a 10k pot) that don't use a separate 10v power supply are based on a couple of facts I'm guessing on that are missing from the power supply datasheet:
1. When the current is set to 0 by dim+/-, it's not really 0 (leakage from the power supply)
2. Dim+ is a hi-z input (1Mohm+) and doesn't require much current to work.

#2 I'm pretty sure about but #1 needs some experiments.

I would be careful suggesting pwm'ing the output on a buck driver. If it is not designed correctly (the driver) the pwm could cause damage with a virtual switch to ground. Many times when a buck driver uses a pwm external switch, it's before the inductor, not after.

I think you're right on that... I'll ask someone that knows a bit more about power supplies than I do (I'm a digital design/electromagnetics guy). Are you sure the meanwells are buck converters? They might just be choppers.

Also, if you have one of the DC-controllable supplies can you test the resistor divider circuit out when you have time? 36k-10kpot-1k (to keep the min voltage to 1v, give the pot more range). I'm curious to see if there's enough leakage to make it work correctly.

I'm pretty sure they are bucks. I do have one, and I have been meaning to get around to play with it, but time hasn't been on my side

Today I try to power the LED rung(6 in series) using the Meanwell ELN-60-24-P after getting a PWM for motor speed control

Light up the rung .

Was able to dim the LED rung but the dimming range was very small.
Less than 5 degree of the potientiometer range.

Details on the PWM unit?

This is the Circuit of the PWM

Bigger please?

Also, what voltage are you supplying this?

I use 9 VDC first then after that I use 12 vdc

I would suggest building the circuit I posted earlier, just due to the fact that you know what the output frequency is based on the resistor values. It's possible that the circuit you used may be out of the usable frequency range.

It say the usable frequency is 3.6kHz

Well, thats a potential problem. Max input frequency on the driver is 3khz.

So that is the culprit. What can I do to change the frequency?
Or maybe it easier to build one myself

Don't know how that circuit works. The labeling of the inputs is not very clear. It could be hard coded in the PIC. I would try and build the circuit I posted. The default values of the resistors will get you a frequency of about 140Hz which is inside the tollerance.

Evil,
I am sorry. Which post? The one from LED Magizine on thermal runaway?

Ben

Start here and work your way down a few posts for more details.

For those of you who have used the meanwell power supplies so far what are your opinions? Would you recommend them over the buckpucks? I would like to power my design at 1A and then dial it back from there so that I am not wishing for more power. For what I want I would need 8 meanwells to power the leds at 1A or 15 buckpucks. Its a difference of 40$ with the meanwells being cheaper. Doesn't include pots to dim the buckpucks or the pwm circuit to dim the meanwells. The bucks would be easier to wire because I don't have to build a circuit. Who thinks meanwell and who thinks buckpucks?

Thanks

Performance wise, they won't be any different. It's all down to convenience and cost for you. I'll try and get a Meanwell driver hooked up in the next day or two and tell you the results.

that would be great. I am really interested in the meanwell, and to see how the pwm, ps, and led are all hooked up. I have no experience with building circuits but this seems like a really interesting project. Please post pics of meanwell setup when completed. Thank you evil!! cant wait.

Evil- do you have an estimation of what par ratings would be at 20-24" with 40 degree optics if the LEDs were running at 650mA?

Complete off the wall ballpark guess, about 120-150 PAR. You lose a lot of light droping to 650mA compared to 1000mA. I have hit 200 PAR at 1000mA before at 24"

and if I remember correctly the LEDs MUST be in either quantities of 13 if ran at 1A or 26 if ran at 650mA? They wouldn't just burn off the extra energy like the bucks do, right?

They work exactly the same way that Buckpucks do. You can run anywhere from 1-13 LEDs in series without issue. Not sure where people are getting the idea that you can only run 13 LEDs on these drivers.

Test results are in. Looks like the "P" versions are for sure analog dimmable. It's not quite as easy as just hooking up a pot and a 10v supply, but not much harder. All it takes is an LM317 voltage regulator, 5k ohm pot and a 680 ohm resistor to get you 1.25v to 10v adjustment. You can use a 12-32v power supply to run this, so that gives you a bit of flexibility.

I did try the pwm circuit, but didn't see any real results. Not sure if Meanwell was fibbing (or was just plain clueless) when they told me this was a pwm dimmable version, or if something is not working correctly. I don't have the tools here to check, but I do at work. It's possible that I just need to raise the frequency to a range that is a little more suitable for the driver. The frequency I'm running now is only a little higher than the minimum frequency they require. Stay tuned on that one.

Did you try different pwm frequencies? I was going to get a couple of meanwells from the group buy, but if they're not pwm dimmable I'm not sure I can use them in my microcontroller project...

Maybe doing pwm fast enough (1-4mhz?) might yield a signal that's analog-ish enough for dimming?

I might try tomorrow at work or in class. I just don't want to screw with things without watching what I'm doing.

I have been using the non-dimmable version with success, and have recently been working to use the Arduino to dim the P version. I have it working as of tonight. (Work has been busy, so only an hour here and there) I am using a the following component. Arduino to drive a Norton Op Amp (TI LM3900n) powered by a 20volt supply hooked through a variable voltage regulator (DE-SWADJ) adjustable so I could make sure when Arduino is pushing max the voltage out of the Norton Op Amp is pushing 10v exactly. This hooked up to the Meanwell P version, and it dims as expected. Very please at this point. I now have to combine my Arduino code that dims with the Clock code and electronic relay code.

Do you see the LED increase in brightness suddenly when you cut the power to the driver? The rise time on it seems to change with the input impedence. Using a 100K pot took forever, but using an LM317 shortened the rise considerably.

I just had the arduino looping from min to max and back to min over and over for testing (~7.5 seconds between min to max). It would be completely off for about .5 seconds then spring on with what appears to be at about 10-15 percent of full brightness. (I dialed my max current to 1.5 amps or 750ma to each string of 13 LEDs) Then it would evenly increase and decrease according to the Arduino signal. I hope this helps,

I was talking more about when you actually kill the AC wall power.

I have an RKL controller coming in. Since I do not really have any experience with building circuits, is there any way I can use the controller to dim the leds through the meanwells?

Not with that version. It doesn't support the moonlight module that has 0-10v outputs.

The circuit you need for dimming the driver is super simple and a lot easier than the pwm circuit I showed before. It is literally three parts that are readily accesable at Radioshack.

Thank you very much for the reply!

The 3 parts you are talking about are the lm317, 680 ohm resistor, and 5000 ohm pot? Im assuming these are all put together on a breadboard? Can you post a simple diagram(like a ms paint diagram haha) with all the components included (i.e. meanwell, lm317, resistor, pot, ps, leds).


Is it... meanwell ps -> lm317 circuit ->led?


Also would this module get the job done?
http://www.digitalaquatics.com/saltwater/RKM-ALC


P.S. I figure Id post in this thread rather than the GB thread...makes more sense Sorry for being a pain evil, I really do appreciate the help!

I'll try and get a diagram up later, but it will connect to two seperate wires on the driver specifically for dimming. You will also need a 12v supply. It can be a small little wall wart type, as the circuit draws very little power.

That module will work. I thought the moonlight module was the only one that would support 0-10v outputs. either that, or this is the same module I'm thinking of and they have changed the compatability of the RKL. That module wasn't originally supported when it was released IIRC.

icic, well I was just looking in to it and thought maybe that would work. It is really expensive tho at $95... Id rather learn electronics and build a $10ish circuit.



okay awesome. Ill probably head over to radioshack today to pic up 2 sets of parts. (one for each meanwell) and await your diagram for assembly . Thanks again!

For $95 though, you now have the lights fully automated. Sounds worth it to me. You can always get it later.

my thoughts exactly after thinking about it... but for now Ill stick with the lm317, unless i get itchy fingers for the module.

With the meanwell 60-48...which is what is available with the GB... I know I can run a max of 13 per string in parallel...now would I run into any problems in these 2 scenarios... Sorry its too early in the morning for me to be capable of thinking...

1. Run 1 string of 6 straight off of the meanwell ps...(b.c. ps output is 48v and total voltage for 6 leds is ~24v), whats needed to do this?...

2. Run 2 strings of 6 on one meanwell, one string of white and the other of blue. (This is probably an obvious answer that I am gonna just need to confirm) Is it possible to wire 2 pots or other dimming device, one on each string... to dim each string?

Thank you for the help again evil!


P.S. I got itchy fingers...I have the ALC ordered...