Constant Current Power needed... I'm confused...

[tedoro]

What would I search for when seeking the unit?

this looks close:

https://www.amazon.com/dp/B005T6K78K/?tag=cpf0b6-20

 

[tedoro]

I think I found one!!! A little spendy, but it looks right.

https://www.amazon.com/dp/B06XFJB28D/?tag=cpf0b6-20

• MEAN WELL HLG-320H-48A constant voltage and constant current LED power supply
• Full range AC input: 90~305Vac
• Operating case temperature: -40~+90°C
• UL8750 type HL LED power supply
• Meet 4kV surge immunity level (IEC61000-4-5)
• Built-in active PFC
• IP65 durable aluminium enclosure
• High efficiency
• Output voltage and constant current level can be adjusted through internal potentiometers
• Protections: short circuit / over current / over voltage / over temperature
• Applications: LED electric display, lighting, LED street lighting and indoor/outdoor lighting
• MEAN WELL 7 year warranty

 

[JustAnOldFashionedLEDGuy]

So if I run 3 in parallel and get 36vdc, then run a bunch of those in series, perhaps 5 times, I would get 1.4 amps... is that your thinking? (my apologies if my nomenclature is improper.

I was kinda hoping for the cleanest strand of 8 leds I can do... (this is an art project using these as grow lights for plants on a grid)

You are getting series and parallel mixed up. Parallel is side by side. Parallel increases current rating. Series is stacked on top of each other. Series increases the voltage.

 

[JustAnOldFashionedLEDGuy]

You would have 25 in parallel, and 4 in series .... or 25 parallel strings of 4 in series. The former, putting 25 in parallel at a time, then stacking those 4 high in series, is safer. That gives you 270mA for each LED, but with mismatching from paralleling what are cheap LEDs, is probably a bit much for those. It has a pot though that allows you to adjust it down to 3.35A, so you can tune the output. It also has dimming wires that you can hook a 0-10V dimmer up to and reduce it to a much lower current if needed.


I suggest buying it from a reputable distributor, such as Mouser or Digikey. It's a bit cheaper as well: https://www.mouser.com/ProductDetai...OaCslb6T0uJ%2B4qvaR/SmMi35/eVjOw7rr%2Bpz25A==


You may pay more for shipping, but worth it.

 

[tedoro]

I would much rather you wire that electrically that as two strings of 4 in parallel (which could still be physically arranged as one string). 12 strings of 8 would become 24 strings of 4. To run them near full power, but not quite, let's say 250 mA per string. That adds up to 6A.


Okay, whoops, I know better.... let me try again.

okay, I take all 100 LED's and put them in strands of 4 in parallel (12vdc 1000ma) I take two of those strands and put them in series to make my 8 LED strand I need for the design (24vdc 1000ma) then I take these 12 strings and group two together in series (48vdc 1000ma) and then lastly, take those 6 groups and wire the whole thing together (48vdc 6000ma)

Is that what you were thinking?

 

[tedoro]

Okay, if I have it right, and this is an okay and safe design that I can run these bulbs a bit lower than rated... what is the procedure on fabricating and testing it? I fear doing one wrong move and blow everything up in a nanosecond. its a fair bit of labor in soldering alone...

 

[DIWdiver]

Yes, it looks like you have it.

 

[DIWdiver]

That Mean Well power supply looks perfect. It's a solid brand, and I agree, buy it from a distributor like Mouser, DigiKey, or others. I plugged it into Octopart, and it looks like Newark has the best price at $72.32. Check their shipping, though, as it could swallow most of that savings.

The first powerup is always scary. I'm coming up with no simple way to do it in stages.

 

[JustAnOldFashionedLEDGuy]

Spend the extra and get the https://www.mouser.com/ProductDetail/MEAN-WELL/HLG-320H-48B?qs=DNaZHaGatO3I17c08YAcZw== ... this is the B model. It has 0-10V dimming wires. Unfortunately the data sheet does not show what happens when the dimming wires are 0V (i.e. you short the two dimming wires together), but normally they would go to the lowest level or 10%. What is not clear is whether that is 10% of the current set with the potentiometer but it should be. That would give you with the dimming leads shorted, 10% of 3.35A or 335mA. That is going to do a lot less damage than 3.35 or 6.7A. You will need 2-3 units in series to get up to 24V before the driver will work properly. But even two in parallel should not be damaged with 335mA.

 

[Dave_H]

Parallelling LEDs needs some consideration. You could try it and hope you get lucky. With small number of parallel
strings, if one fails open (due to LEDs or merely a bad connection), all current is forced through remaining
string(s) which could cause damage, or a least run brighter and hotter. With large number of strings the effect might be less dramatic,
but brightness could vary across strings.

Inside a LED ceiling fixture I found 24v CC driver with series connection of 8 sets of 4 paralleled SMT LEDs, with no sign
of resistors or any obvious load balancing. They can get away with this by close match of forward voltage
of LEDS (vf binning) and that they operate at approximately the same temperature. Not sure if either is happening
with design in this thread.

With smaller SMT white LEDs I found vf range typically 2.7v to 3.2v at same current and temperature; typically divided into
4-5 ranges ("bins"). Not sure what LEDs being used here have, looking at spec. now.

LEDs have negative voltage coefficient with temperature, which in some cases can lead to thermal runaway.

Osram has (or had) a bunch of good application notes including this...I might be able to dig up a copy; or someone
could search this out.

Not trying to discourage DIY; but the current, voltage and power levels are high and needs caution, especially
if new to the game.

Dave

Interesting article (though a bit older) dealing with 2 parallel case:

https://www.ledsmagazine.com/leds-s...ay-when-driving-multiple-led-strings-magazine

 

[DIWdiver]

Spend the extra and get the https://www.mouser.com/ProductDetail/MEAN-WELL/HLG-320H-48B?qs=DNaZHaGatO3I17c08YAcZw== ... this is the B model. It has 0-10V dimming wires. Unfortunately the data sheet does not show what happens when the dimming wires are 0V (i.e. you short the two dimming wires together), but normally they would go to the lowest level or 10%. What is not clear is whether that is 10% of the current set with the potentiometer but it should be. That would give you with the dimming leads shorted, 10% of 3.35A or 335mA. That is going to do a lot less damage than 3.35 or 6.7A. You will need 2-3 units in series to get up to 24V before the driver will work properly. But even two in parallel should not be damaged with 335mA.

The B model does not have the internal adjustments. If you want both the internal and external adjustments, you want the AB model. There does not appear to be stock of these at american distributors at the moment.

 

[Dave_H]

Interesting article (though a bit older) dealing with 2 parallel case:

https://www.ledsmagazine.com/leds-s...ay-when-driving-multiple-led-strings-magazine

BTW not saying the method in this article is what the OP should necessarily use.

Not sure the probability of LEDs failing short but that could throw parallel balancing off quite a bit. Different modes of degraded operation are possible but self-destruct should not be one of them!

Likely there is an OTS parallel drive board/module which could balance currents or at least sense loss of string and reduce current or shut off; perhaps used in combination with multiple low-current drivers.

An aside, I designed a discrete 48vdc/200mA linear regulator circuit for some LED strips with 14 series (2 parallel) times 3 (84 LED total) This could be driven by regulated 48v switching supply, common in telecom industry for IP phones. Measured vf total came in at 45-46v which left comfortable overhead, not much power loss (3v*0.2A = 0.6W). Considering the adapter is probably 90% efficient, overall efficiency of 80+% is not bad for this power level. Single-chip dc drivers are attractive but I wanted to customize some functions. For example, if the input voltage got too high (e.g. from using unregulated 48v "brick" adapter), it would switch to lower current to keep dissipation down, avoiding large heatsink on the MOSFET; or shut off completely.

Obviously the designers of those strips thought they could get away with that series/parallel arrangement; matched LEDs on the same PCB. Or maybe that's why I got them surplus for such a good price...

Dave

 

[tedoro]

That Mean Well power supply looks perfect. It's a solid brand, and I agree, buy it from a distributor like Mouser, DigiKey, or others. I plugged it into Octopart, and it looks like Newark has the best price at $72.32. Check their shipping, though, as it could swallow most of that savings.

The first powerup is always scary. I'm coming up with no simple way to do it in stages.

Well, I guess I'll move ahead with this plan. I hearing its a big project, and has some juice running through it. I will take care. Many thanks for the help gentlemen. I will post with my result/further questions. T

 

[JustAnOldFashionedLEDGuy]

Dave_H,


The majority of commercial and even residential fixtures are made with paralleled low power LEDs. We don't have the wide Vf variations we once had, though effective resistance is lower which makes things worse. COBs are already matched internally for Vf.


The reason why you parallel the LEDs first, then stack them in strings is it averages out differences better and mismatches are per parallel section, not on whole strings.

 

[arek98]

You could use something like this https://www.onsemi.com/pub/Collateral/NSI45060JD-D.PDF
They are $0.89 at DigiKey for example.
Look thru the spec (page 7 in particular)
You can connect it with LEDs directly to AC. Well thru the bridge .
You can use one bridge per connection point (all LEDs if they are plugged in one socket)
Then make a strings of LEDs in series with this device in each.
One goes to 100mA (you can adjust it with resistor between 60-100mA) but you can connect couple in parallel with one LED string to get more output.
If you connect 3 those devices in parallel (see page 6) you can set any current between 180-300mA
Don't be scared by 45V rating at the beginning of spec, it is max voltage between anode and cathode of this part. It means that input voltage can't be higher than 45V from voltage dropped by LEDs.

For example, if you use 120V AC input, your LEDs are 9-12V I would put 10 of them in one string. Measure voltage, see how close it is to input, remove LED if is to high (LES are really droping 12V or more each) or add LED if is low and there is a space (e.g., you get 90V on 10 LED)

Edit: Adjustment resistor goes between pin 3 (Radj) and cathode (pin 4), value you get from Fig 6 on page 4 (0.125W resistor). Not connected pin or anything above 1000ohm gives 60mA output, about 7ohm for 100mA. Looks like 33ohm should give 80mA and 68ohm 70mA (these are common resistor values)

 

[Dave_H]

You could use something like this https://www.onsemi.com/pub/Collateral/NSI45060JD-D.PDF
They are $0.89 at DigiKey for example.
Look thru the spec (page 7 in particular)
You can connect it with LEDs directly to AC. Well thru the bridge .

.....

Perhaps the OP can comment on whether he wants to take on driver design at the component level. Looks like a
small (or maybe not so small) PCB to mount these.

This is a linear device which brings a number of design issues including thermal management; and dealing with
line voltage requires correct component selection and safety issues e.g. fusing.

In my electronic design and component engineering experience over time, I find you really need to read
the datasheets and application notes carefully. For example maximum power per device is as low as 1.77W at
25C with minimal heatsinking (~1/2 sq. in. of copper), which for many cases is unrealistic and needs to be derated. At
100mA, means limiting voltage drop across each device to below 17v including variations in line voltage. Minimum voltage
across device is about 4v, below which current drops off. This adds up to a narrow window for number of LEDs in series.
Actual vf of the LEDs needs to be accurately known (or at least measured) for given current; 9-12v is a wide range.

Not sure the device has thermal and short-circuit protection.

I don't want to recommend too many specifics but looks to me like pre-made properly-certified ac-dc CC drivers are
a strong approach. They offer flexibility, simplicity, efficiency, safety, and overall outweigh any benefits of linear
drivers for this level of power; and may cost less in the end.

And again, not objecting to linear drivers in some circumstances as noted from my 48v/200mA discrete design above. BTW
Infineon also has some linear CC devices for this type of use.

Dave

 

[Dave_H]

You could use something like this https://www.onsemi.com/pub/Collateral/NSI45060JD-D.PDF One goes to 100mA (you can adjust it with resistor between 60-100mA) but you can connect couple in parallel with one LED string to get more output. If you connect 3 those devices in parallel (see page 6) you can set any current between 180-300mA Don't be scared by 45V rating at the beginning of spec, it is max voltage between anode and cathode of this part. It means that input voltage can't be higher than 45V from voltage dropped by LEDs. Edit: Adjustment resistor goes between pin 3 (Radj) and cathode (pin 4), value you get from Fig 6 on page 4 (0.125W resistor). Not connected pin or anything above 1000ohm gives 60mA output, about 7ohm for 100mA. Looks like 33ohm should give 80mA and 68ohm 70mA (these are common resistor values)

Setting each CCR to 100mA (its max.) using 7 ohms will result in much lower RMS current (looks like 50% or less) because 100mA limit is peak current, and full-wave rectified ac results in pulsed output. Application note AND8433/D referenced from datasheet details this. Higher-output e.g. 160mA CCRs are available but still leaves a lot of limitations (and calculations); still need at least 3 in parallel. This will also result in 120Hz flicker which may or may not be an issue. This configuration is similar to LED Christmas lights, 60Hz flicker is visible (and sometimes disturbing), but 120Hz is better (my experiment by adding a full-wave bridge). Looks like using switching-type CC driver module is still a better approach. Dave