please explain the meanwell 60-48D startup/surge voltage

[110galreef]

It states these draw 60a on startup.....

I would have 10-12 of these given a 120-148 LED setup. Now i have heard rumors of issues possibly from this many drivers & startup. However, only half at most would ever start up same time.

Thanks,

[kcress]

That is a lot. Too many for me to ever consider.

You need to move up to power factor corrected drives. Using that many 60-48D can cause a lot of electrical mischief - none of it good..

Alternatively use a PFC DC supply(s) and a bunch of BuckPucks.

What you propose is buying a bunch of non-PFC 60W supplies to provide about 32W each.

Consider instead, buying one or two 200W PFC supplies and running everything with no harmonics disasters and no nuisance popping breakers.

[kcress]

BTW that current is the inrush to charge the capacitor front ends. Stack a bunch up and they will trip your circuit breakers.

[110galreef]

what's a bunch. If only 5 start at same time any issues? Are these quesses or a way to find exact. I've seen many of 8-10 driver setups w/o issues?

[110galreef]

any specific reading or threads where people ran into these problems? Just trying to wrap my head around this.
thanks

[XSiVE]

First, what's listed there is Inrush Current (Max) meaning it may not ever hit that much inrush current on startup.
Second, it's only for a few milliseconds as they power on.

Sure, if you've got 20 of these units all powering on at once you *might* run the risk of popping a gfci or your circuit breaker. If you're really that worried about it just stagger their startup.

I start 6 at once with no issues.

edit: the other mean-well power supply that most everyone uses with buckpucks still has a 40A inrush current, so it's not too much different.

[Rancherlee]

yep, a standard 20 amp breaker takes a few second to trip with a high amp draw, you only need .10 of a second or less to power up the drivers. One way to find out, try it!

[110galreef]

First, what's listed there is Inrush Current (Max) meaning it may not ever hit that much inrush current on startup.
Second, it's only for a few milliseconds as they power on.

Sure, if you've got 20 of these units all powering on at once you *might* run the risk of popping a gfci or your circuit breaker. If you're really that worried about it just stagger their startup.

I start 6 at once with no issues.

edit: the other mean-well power supply that most everyone uses with buckpucks still has a 40A inrush current, so it's not too much different.

It's not some much of a worry, just wont to be sure i understand what I am getting into. Also trying to simplify this w/o doing to much DIY.

for 120/144 LED's it looks like:
10/12 meanwell 60-48D (6 for RB, 4 for CW/ 7 RB/ 5CW)
At most would be 7 drivers starting at the same time for the RB's.
I can easily have 3 start, then the other 4 for RB's.....then the 3 cool whites & then the last 2 cool whites....

or
20/24 buck pucks & like 2-4 powersupply's depending on How i do the heatsinks & arrangements. Also here only 2 power supply's at most starting at the same time.

Also both options seem to cost about the same as well.

[110galreef]

Also what is PFC..

And what is the difference then between non PFC :hmm5:

[therman]

I've got a 156 array with half starting at a time. I went a little crazy from all the warnings and ran 2 new circuits so they were split up (along with other tank hardware). I have 20 meanwell ELN-60-48 running in my house split across 3 circuits with no perceptible issues and will soon be adding another 6. Wiring does not get hot, harmonic gremlins have not taken over my house, breakers and GFIs do not pop, and the sky has not fallen. The inrush current is not detectable by a multimeter or Kill-a-watt meter either.

I asked a LOT of people, including Meanwell engineers about this, and they did not think it would be a real issue. I am no EE, but at least one person told me that any large motor, some power tools, and some other home appliances all have massive inrush currents, but they do not cause problems.

So as a disclaimer, I can't guarantee you won't have ANY problem, but I sure haven't. That number has been running for 5 months so far in my house.

FWIW you might want to check out the Meanwell HLG (http://www.meanwell.com/search/hlg-240/hlg-240-spec.pdf) series. Can drive much larger arrays, but does require parallel wiring, and has PFC.

Good luck, and feel free to PM me if you have any specific questions.
-Tim

[110galreef]

what about this meanwell.

It is the HLG line....appears the 185H & the 320H have the 0-10v dimming option, which I need/want. For some reason the 240H line doesn't. Also these are all active built in PFC....about 94% effecient.

So would either the HLG-185H http://www.meanwell.com/search/hlg-185h/default.htm

or HLG-320H
http://www.meanwell.com/search/hlg-320h/default.htm

work to power 3w cree LED's. If so how many would they power? Do I still want the 48v versions?

I assume I would basically have to wire say 36 LED's in parrallel w/ the 185H-48 and 60 LED's with the HLG-320H-48 ....:worried:

[kcress]

To answer your PFC question. The Meanwell 60-48D has just diodes on the front-end that charge a capacitor. Capacitors store energy with voltage. What happens is the capacitors charge up to some level and operation begins. Every power line cycle which varies from 170V to -170V can only add power back to the capacitor, (that is continuously drained by running the LEDs), when the absolute value of the voltage exceeds whatever the drained down voltage of the capacitor is at that moment.

This means the power can only be gotten from the outlet in tiny sporadic gulps. Those gulps are large sudden current spikes because all the energy needed by the system has to come in those tiny gulps.

Those gulps can be 10A even though the name plate states 1A or 0.5A. That's because the name plate shows the average current. When you stack up a pile of these things you may have 300A pulses being drawn from your outlet even though the average current is only 10A.

The heating in your breakers and wiring and outlets is only a function of I-squared R. Which means the heating is very sensitive to the current I. Your wiring is set up for "average" current not these repetitive pulses.

In most cases you won't have a problem - you'll get away with it. Your wiring will be running hotter than predicted by the average current values listed on the Meanwells. If you have no weaknesses (today) like a bad connection in your house wiring you probably won't have an issue.

As for the startup problem with the inrush the inrush listed only occurs when you power-on the MW and the power line voltage which cycles, as stated above, is at the maximum values. Any other time it's less. This means you are rolling the dice every time you energize the MWs. You may turn them ON and the first time the breaker trips. Or the 300th time the breaker blows as it all depends on where in the power line cycle the switch gets thrown.

If you always leave the MWs ON but just dim them then your breaker tripping opportunities are obviously reduced.

The aforementioned problems that diode-front-end power supplies cause is so onerous to the power companies and to places where a lot of switching supplies are used, like offices with a bunch of personal computers, that the laws are changing on the subject of switching supply front ends.

Enter the Power Factor Controlled (PFC) front ends. Instead of diodes hooked to capacitors the diodes are replaced with a controlled network of transistors and inductors. The control directs the current drawn from the power source to be drawn smoothly over the entire power line cycle not just in gulps - no gulping allowed. This avoids ALL the current pulsing and wire heating. This same network can provide the bonus of eliminating the inrush too. As the initial energy drawn on start-up can be actively limited.(Note that these PFC Meanwells don't limit inrush much.)

The HLG line is Meanwell's PFC or "active front end" current drivers. They would be the choice for those wanting a "bunch" of drivers.

On a side note:
Running 10, 15, or 20 ELN60s is a gigantic waste of money. Those units are only about 88% efficient at FULL LOAD. Loading them at something like 700mA means you are only running them at about 60% load which is also something like 60% efficient. It's so bad they don't even show you a graph. This means you're paying another %40 percent for your lighting power and dumping %40 more heat into the space. You'd be better off using resistors and a fixed voltage power supply, unless you want dimming.

As for the HLG-185H or HLG-320H since they are more powerful than the 60s I would use them in parallel series setups. There's additional work and a few more parts like resistors and possibly fuses when you parallel. You can read about that in the BIG LED thread. That would reduce the huge pile of under utilized MW 60s too.

[110galreef]

Wow great explanation. that helps alot! Thanks for taking the time to lay it out!:thumbsup:

Ok so my option other than running 10 MW 60-48D was to run:

20 of the dimmable buck puck drivers and then like 4 small power supplies. 2 for the cool white and 2 for the RB..... Probably will have 2 pendants of 60 LED's

Or go with like 4-6 of the HLG-185H and wire the strings of LED's in parallel. Have to calcualte the exact numbers of drivers, but a lot less than the ELN 60-48D's

Aside from these options, do you have any others to simplify a build of this matter. I do WANT dimming capabilities via 0-10v or 0-5v like the buckpucks can do.
Thanks again! Gathering info and ideas to build this in a few weeks!

[sfsuphysics]

so is that Meanwell SNAFU with the power supply only restricted to the dimming versions? Or both dimming and fixed output?

[kcress]

Wow great explanation. that helps alot! Thanks for taking the time to lay it out!:thumbsup:

Ok so my option other than running 10 MW 60-48D was to run:

20 of the dimmable buck puck drivers and then like 4 small power supplies. 2 for the cool white and 2 for the RB..... Probably will have 2 pendants of 60 LED's

Or go with like 4-6 of the HLG-185H and wire the strings of LED's in parallel. Have to calcualte the exact numbers of drivers, but a lot less than the ELN 60-48D's

Aside from these options, do you have any others to simplify a build of this matter. I do WANT dimming capabilities via 0-10v or 0-5v like the buckpucks can do.
Thanks again! Gathering info and ideas to build this in a few weeks!

Thanks 110galreef. I appreciate the feedback as these long posts really take time.

I'd probably consider the: http://www.jameco.com/webapp/wcs/stores/servlet/Product_10001_10001_2115151_-1

Please note I haven't used one of these but that's how I'm reading MW's data sheets.

42V@5.7A

5.7 / .7A = 8 strings.
42V / 3.5V = 11 LEDs

One of those drivers = 88LEDs. 240W

Note that they do not dim completely. They dim plenty to adjust colors and shift from cloudy to sunny but not all the way down. It would be sort of pointless anyway as a little bit of 150+ LEDs is still going to be a lot of light. If you need that use one string on a 60 or something like that.

Two of those units represent 176 LEDs and should provide all your bulk LEDing.





so is that Meanwell SNAFU with the power supply only restricted to the dimming versions? Or both dimming and fixed output?


Both. SNAFU is kinda harsh. :) It's only an issue when you start stacking them up. The efficiency issue comes with not fully loading the drivers.

[110galreef]

kcrees.....Ok I think I understand all this math Jive now! I hope!!:wildone:

Ok so lets say I plan 2 pendants of 60 LED's each. i want about a 60/40 split between RB:CW...(36 XRE RB's : 24 XPG CW's)
XRE RB --3.9 fV max, 1000mA max....~3.55fV @ 750mA planned driven at
XPG CW--3.75 fV max, 1500mA max....... ~ 3.3fV @ 1000mA planned driven at

So the HGL 120H-48 for the XPG's:
-2.5A/ 1000mA = 2 strings
48v / 3.3V = 12 LED's-14LED's what should I use as fV...the max or the number given at the driven mA??
THUS my 24 LEDS effeciently light w PFC and only 1 driver!

Then also the HGL 120H-48 for the XRE's
-2.5A/750mA = 3 strings total 2.25A is this enuf margin of error...could I go up to 800mA w/o issues??
-48v / 3.55fV @ 750mA = 12 LED's...13 max what should I use as fV...the max or the number given at the driven mA??
Thus my 36 XRE's effeciently lit w/ PFC & only one driver.

So in total for the 2 pendants I will have 4 drivers. I really like this if I am correct. This really is the minimum drivers I can do as well as I want one for each color and each pendant independant!

[kcress]

kcrees.....Ok I think I understand all this math Jive now! I hope!!:wildone:

Ok so lets say I plan 2 pendants of 60 LED's each. i want about a 60/40 split between RB:CW...(36 XRE RB's : 24 XPG CW's)
XRE RB --3.9 fV max, 1000mA max....~3.55fV @ 750mA planned driven at
XPG CW--3.75 fV max, 1500mA max....... ~ 3.3fV @ 1000mA planned driven at

So the HGL 120H-48 for the XPG's:
-2.5A/ 1000mA = 2 strings
48v / 3.3V = 12 LED's-14LED's what should I use as fV...the max or the number given at the driven mA??
THUS my 24 LEDS effeciently light w PFC and only 1 driver!

Use the voltage number at the desired current - but leave some on the table! Don't try to squeeze every LED in you can.

Yes one driver.

Then also the HGL 120H-48 for the XRE's
-2.5A/750mA = 3 strings total 2.25A is this enuf margin of error...could I go up to 800mA w/o issues??

-48v / 3.55fV @ 750mA = 12 LED's...13 max what should I use as fV...the max or the number given at the driven mA??
Thus my 36 XRE's effeciently lit w/ PFC & only one driver.

Looks right to me.

So in total for the 2 pendants I will have 4 drivers. I really like this if I am correct. This really is the minimum drivers I can do as well as I want one for each color and each pendant independant!

That was always my point. ;)

Now since you're going to go with parallel strings there is a little more work to do.

When you string a bunch of LEDs together they add up to the SUM of their INDIVIDUAL Vfs. They aren't all the same.

You will need to match them. If you don't, one string can get the lion's share of the current.

Here read this: http://www.reefcentral.com/forums/showpost.php?p=17723160&postcount=21

[DFason]

KCress - the driver that you posted. Would the 88 LED's be ran in series or parallel?

-Dave

[kcress]

Both. You have to run (several chains of serial wired LEDs) in parallel.

In the example above, 8 parallel strings each with 11 LEDs.

[wld1783]

To answer your PFC question. The Meanwell 60-48D has just diodes on the front-end that charge a capacitor. Capacitors store energy with voltage. What happens is the capacitors charge up to some level and operation begins. Every power line cycle which varies from 170V to -170V can only add power back to the capacitor, (that is continuously drained by running the LEDs), when the absolute value of the voltage exceeds whatever the drained down voltage of the capacitor is at that moment.

This means the power can only be gotten from the outlet in tiny sporadic gulps. Those gulps are large sudden current spikes because all the energy needed by the system has to come in those tiny gulps.

Those gulps can be 10A even though the name plate states 1A or 0.5A. That's because the name plate shows the average current. When you stack up a pile of these things you may have 300A pulses being drawn from your outlet even though the average current is only 10A.

The heating in your breakers and wiring and outlets is only a function of I-squared R. Which means the heating is very sensitive to the current I. Your wiring is set up for "average" current not these repetitive pulses.

In most cases you won't have a problem - you'll get away with it. Your wiring will be running hotter than predicted by the average current values listed on the Meanwells. If you have no weaknesses (today) like a bad connection in your house wiring you probably won't have an issue.

As for the startup problem with the inrush the inrush listed only occurs when you power-on the MW and the power line voltage which cycles, as stated above, is at the maximum values. Any other time it's less. This means you are rolling the dice every time you energize the MWs. You may turn them ON and the first time the breaker trips. Or the 300th time the breaker blows as it all depends on where in the power line cycle the switch gets thrown.

If you always leave the MWs ON but just dim them then your breaker tripping opportunities are obviously reduced.

The aforementioned problems that diode-front-end power supplies cause is so onerous to the power companies and to places where a lot of switching supplies are used, like offices with a bunch of personal computers, that the laws are changing on the subject of switching supply front ends.

Enter the Power Factor Controlled (PFC) front ends. Instead of diodes hooked to capacitors the diodes are replaced with a controlled network of transistors and inductors. The control directs the current drawn from the power source to be drawn smoothly over the entire power line cycle not just in gulps - no gulping allowed. This avoids ALL the current pulsing and wire heating. This same network can provide the bonus of eliminating the inrush too. As the initial energy drawn on start-up can be actively limited.(Note that these PFC Meanwells don't limit inrush much.)

The HLG line is Meanwell's PFC or "active front end" current drivers. They would be the choice for those wanting a "bunch" of drivers.

On a side note:
Running 10, 15, or 20 ELN60s is a gigantic waste of money. Those units are only about 88% efficient at FULL LOAD. Loading them at something like 700mA means you are only running them at about 60% load which is also something like 60% efficient. It's so bad they don't even show you a graph. This means you're paying another %40 percent for your lighting power and dumping %40 more heat into the space. You'd be better off using resistors and a fixed voltage power supply, unless you want dimming.

As for the HLG-185H or HLG-320H since they are more powerful than the 60s I would use them in parallel series setups. There's additional work and a few more parts like resistors and possibly fuses when you parallel. You can read about that in the BIG LED thread. That would reduce the huge pile of under utilized MW 60s too.

kcress

Outstanding explanation of harmonics. Two thumbs Up:D

Just Have two questions:

I'm running over 20 drivers with out a problem. Most are on fused surge protectors. Do you feel this would be a fix to this theoretical problem and have there been any documented cases of fuses being blown due to this phenomena?

Also I did some PAR/Watt testing using multiple ELN 60-48Ds and found the fixture to be more efficient under driven. I set three drivers to 750mA at nine volts control input driving 12 Royal Blue/Cool White mix, and one at full driving two parallel strings of Royal Blue at appx 650mA. The readings were 330 PAR at 24" and 186 watts or 1.77 PAR per Watt value. At 1.5 volts control input the readings were 56 PAR at 24" and 21 watts or 2.66 PAR per Watt value...I know the LEDS are more efficient when driven more gently but your 40% claim of reduced efficiency confuses me. Could you explain the loss of efficiency better so a pilot like my self can understand it.

Thanks

Bill

[kcress]

kcress

Outstanding explanation of harmonics. Two thumbs Up:D

Thanks!



Just Have two questions:

I'm running over 20 drivers with out a problem. Most are on fused surge protectors. Do you feel this would be a fix to this theoretical problem and have there been any documented cases of fuses being blown due to this phenomena?

Um... First it's not a theoretical problem, it's an actual real problem. :)

There is no way to stop or mitigate the harmonic problem short of preventing it from ever occurring in the first place.

About the inrush problem you can mitigate it a little bit. In fact it's already being partly mitigated by all your wiring. The wiring simply can't supply the huge inrush so what happens is the voltage sags - a lot. This will reduce the inrush because there is less voltage available to push current. Sometimes this works against you as now the really large short surge turns into just a large surge but now it has to last longer! It all depends on your circuit breaker's response curve. Since you are not seeming to have any problems with your setup I wouldn't change anything.

If you aren't having nuisance trips - do nothing.

Power strips and other things will lessen the inrush but increase its duration.

With that many nonPFC supplies I would invest in a temp gun. ~ $25 I'd do a serious audit every other month and keep a log. After the lights have been on at their brightest for a few hours 'gun' the outlets and if possible the wire somewhere and the face of the breaker in the panel. If any of them are really warm you should monitor them more regularly. If you note a significant rise in temp then you may be having some increased resistance from oxidizing connections. These are what lead to house fires so you'd want to investigate what's happening. Screws working loose due to heat cycling or bass springs in the breaker or the outlet are relaxing for the same reason.



Also I did some PAR/Watt testing using multiple ELN 60-48Ds and found the fixture to be more efficient under driven. I set three drivers to 750mA at nine volts control input driving 12 Royal Blue/Cool White mix, and one at full driving two parallel strings of Royal Blue at appx 650mA. The readings were 330 PAR at 24" and 186 watts or 1.77 PAR per Watt value. At 1.5 volts control input the readings were 56 PAR at 24" and 21 watts or 2.66 PAR per Watt value...I know the LEDS are more efficient when driven more gently but your 40% claim of reduced efficiency confuses me. Could you explain the loss of efficiency better so a pilot like my self can understand it.

Thanks

Bill

You're not measuring the electrical efficiency as related to under-loading the driver. The efficiency is the amount of current supplied verse the current drawn from the outlet. You've already shaved off this efficency by setting your driver's maximum current to something less (750mA) than maximum nameplate (1.3A).

Dimming once the current has been limited will not change the efficiency a lot.

Also using a PAR meter to wash the whole measurements thru probaly adds more confusion.:lmao:

[der_wille_zur_macht]

Also I did some PAR/Watt testing using multiple ELN 60-48Ds and found the fixture to be more efficient under driven.



You're not measuring the electrical efficiency as related to under-loading the driver.

Bill, to reinforce what kcress is stating, it's important to understand exactly WHAT efficiency you're measuring. kcress is talking about efficiency of the DRIVER itself.

The driver takes "X" amount of power from your wall socket. It delivers power to your LEDs. Because of losses within the driver itself, the amount of power delivered to the LEDs will always be less than X. This difference lets us calculate the efficiency of the driver itself. This is what kcress is referring to - when run at anything less than full bore (SVR2 set to max, AND output loaded to max voltage), the ELN will be fairly inefficient.

On the other hand, if you're measuring watts on the input of the driver and comparing to PAR, you are effectively measuring a (heavily weighted, skewed, and hard to control accuracy/repeatability of) efficiency of the whole system, including the LEDs themselves.

Let's look at a few theoretical examples:

1) A Meanwell ELN 60-48 (a 60w supply) run under typical conditions. 1000mA drive current set by svr2, 12 HP LEDs in series.

2) A Meanwell ELN 60-48 run at a lower than typical drive current, say, 600mA drive current, 12 HP LEDs.

3) A theoretical "little brother" to the Meanwell ELN. Imagine if they made a version that ran at 48v but was only a 30w supply, i.e. it had a max current of ~600mA. Run this with 12 HP LEDs at full blast, i.e. 600mA, 12 LEDs.

#2 will create more PAR per watt consumed by the LEDs, at the output side of the driver than #1. It may create more PAR per watt consumed at the wall than #1 but it's hard to guess, because you're losing efficiency in the driver but gaining it at the LEDs. The driver itself is operating more efficiently in situation #1, but the LEDs are more efficient in situation #2.

#3 will be the most efficient in pretty much all possible measurements. The driver will be operating more efficiently, the LEDs will be operating efficiently, AND you'll be getting more PAR per watt consumed at the wall plug.

This is yet another reason why I'm not personally a huge fan of the ELN 60-48 drivers, at least run in the way reefkeepers tend to run them. You're paying for capacity you're not going to use when you buy the unit, and then you're losing money in the longterm due to poor driver efficiency. The only way to run ELN60-48s at their best efficiency is to run parallel strings and/or peg the limits on typical LEDs, both of which create other issues that probably aren't worth the tradeoff.

[kcress]

The only way to run ELN60-48s at their best efficiency is to run parallel strings and/or peg the limits on typical LEDs, both of which create other issues that probably aren't worth the tradeoff.

DWZM; I'm starting to think it probably is worth the trade off. Since an ELN60-48 is 1.3A if you wanted to run two 600mA strings you would be running the LEDs at good efficiency and the ELN at good efficiency. Toss in the 1 ohm resistors for easy current measurement, and a little load balancing, and you'd have it. Use the ELN's voltage limit to limit the current during a burnout and you'd have a good system.


Oh, and nice addition above.

[der_wille_zur_macht]

I suppose that is a good argument. In strings of 12 it's fairly likely that the loads will balance well anyways.

The ultimate solution would be if Meanwell would make a high-wattage supply that had a low enough nominal current that you could get good results running it wide open on series LEDs.

Either way it's not really something I'm losing sleep over as ELNs are expensive and have poor dimming performance compared to other options so they're not really on my radar anyways.

[wld1783]

DWZM; I'm starting to think it probably is worth the trade off. Since an ELN60-48 is 1.3A if you wanted to run two 600mA strings you would be running the LEDs at good efficiency and the ELN at good efficiency. Toss in the 1 ohm resistors for easy current measurement, and a little load balancing, and you'd have it. Use the ELN's voltage limit to limit the current during a burnout and you'd have a good system.


Oh, and nice addition above.

kcress

+1

This is what I recommend to other Reef Keepers. Two parallel strings of twelve LEDs. I like the idea of adding a 1 ohm resistor and will test that soon. In theory 650mA each but in un-lab tested reality maybe 700mA through one string and 600mA through the other string or more... Not a big deal as these LEDs have been driven with good results from 350mA to 1000mA.

Also

If its an actual problem are there any documented cases (Myth Busters included):confused:

Is not best PAR per Watt our ultimate goal:confused:

DWZM

I'm a results type of guy and although my measurements are not scientific...Control, numerous drivers, ect... I still find this loss in efficiency contrary to my measurements. In fact to a point of significant disagreement to raise the whats up flag. I've read the Mean Well data sheets and talked to their engineers and I'm still confused about what your saying and I do understand your examples.

Driver efficiency is one thing and overall efficiency is another. I've done similar measurements from drivers from the LPC 35-700 to the CLG 150-48 and found the results consistent and at best only a 10% difference in efficiency. When your saying your paying for extra/ unnecessary capacity, the street price of the ELN 60-48D is only 10 bucks more than the LPC 35-700. With a dimming function and option to drive 24 LEDs at a lower mA and at best a 5% lower efficiency.:hmm2:

For me its about how many watts and whats the PAR :beachbum:

[der_wille_zur_macht]

kcress

+1

This is what I recommend to other Reef Keepers. Two parallel strings of twelve LEDs. I like the idea of adding a 1 ohm resistor and will test that soon. In theory 650mA each but in un-lab tested reality maybe 700mA through one string and 600mA through the other string or more... Not a big deal as these LEDs have been driven with good results from 350mA to 1000mA.

Despite the fact that variations will likely be within SAFE ranges, I would still never run (or recommend someone else run) HP LEDs in parallel strings without at least checking for current balance between the strings, and correcting it if it's out of balance. A VERY VERY small difference in Vf can make a HUGE difference in current, and even if it might be "safe," that can translate into a large deviation from "expected" light levels over part or all of the fish tank.

Also, I'd never run any of the typical LEDs in a pair of series strings without some form of protection if a single string were to fail.

Finally, I think running two strings in parallel has some potential disadvantages even once you've covered those two issues. Mainly, 24 LEDs is just a LOT to be in one circuit for most of our tanks. Unless you're building for a very large tank, that's a big "chunk size" to work with for dimming, on/off times, and getting color ratios dialed in.

If its an actual problem are there any documented cases (Myth Busters included):confused:

If you're talking about imbalance, then yes, it's a known, documented thing.

Is not best PAR per Watt our ultimate goal:confused:

Within reason. See below.

I'm a results type of guy and although my measurements are not scientific...Control, numerous drivers, ect... I still find this loss in efficiency contrary to my measurements.

That's because your experiments are obscuring the difference in efficiency of the driver and efficiency of the LEDs. You've designed an experiment where you cannot measure these separately. kcress and I are not suggesting that your data or conclusions are wrong, we're just pointing out that it (and typical behavior of reef keepers) doesn't really point to the overall "best case" in terms of efficiency because in nearly all cases that an ELN60-48 is used, it's being used inefficiently. It's like taking a bunch of cars to the drag strip but putting junky tires on all of them. Yes, you can determine which car is the fastest, but you won't know really how fast that car can be.

Driver efficiency is one thing and overall efficiency is another.

Agreed - but to maximize overall efficiency, we must understand how to maximize efficiency of every component in the system.

For me its about how many watts and whats the PAR :beachbum:

If that's the case, then you should probably run 4 strings, or 6, in parallel - at lower currents - not just two. Six strings run on a Meanwell ELN60-48 would mean that the LEDs were seeing ±215mA each. WAY more efficient than two strings at ±650mA each. Of course this means you're buying a lot more LEDs, so upfront cost skyrockets. Or, you'd be running a carefully selected power resistor and a carefully selected DC supply. WAY more efficient because you can get current regulation losses to nearly zero, but not controllable. Or, you'd be running DIY drivers - controllable, more efficient, but difficult to work with if you're not in to that sort of thing. Which brings up my comment above - most people are trying to balance ultimate efficiency with other goals (reasonable upfront cost, controllability, ease of implementation). Because of the current and power levels available from Meanwell's off the shelf drivers, it's very hard to maximize ALL of these goals at the same time.

Not to mention that it's incredibly hard to use PAR as a measure of output across different lighting rigs, at least in the case of measuring efficiency - things like the angle or placement of the components or the sensor, ambient lighting, and so on can have a huge impact. Plus, of course, PAR measures output at a single point, not total output of a given rig. So it's going to be incredibly hard to get consistent results, much less normalize them in a way that you can correlate the results to a specific intended design parameter.

[wld1783]

DWZM

Thanks for the feed back.

Could you explain more your concerns with a slight difference in forward voltage being a big problem in strings of 12 from the same lot and manufacturer. Yes mixing a bunch if different LEDs would not be good.

I've read most of your threads and would like to know why your so opposed to parallel strings... Please be specific. I know the big downside of an open and all the current rushing through the other string but with a driver thats turned down I find this inconsequential to the benefits of parallel strings.


Bill

[wld1783]

Not to mention that it's incredibly hard to use PAR as a measure of output across different lighting rigs, at least in the case of measuring efficiency - things like the angle or placement of the components or the sensor, ambient lighting, and so on can have a huge impact. Plus, of course, PAR measures output at a single point, not total output of a given rig. So it's going to be incredibly hard to get consistent results, much less normalize them in a way that you can correlate the results to a specific intended design parameter.

Also whats so hard about putting a PAR meter at a measured distance above a fixture and reading several measured points (average them along the way) and having a KilloWatt meter read the total Watts.

Your kidding about the ambient indoor lighting which is 11 PAR pointed directly at the ceiling not at the fixture...Right???

Are we in the class room or in the field?

Bill

[laverda]

3) A theoretical "little brother" to the Meanwell ELN. Imagine if they made a version that ran at 48v but was only a 30w supply, i.e. it had a max current of ~600mA. Run this with 12 HP LEDs at full blast, i.e. 600mA, 12 LEDs.


Mean well does make a ELN-30-48

[der_wille_zur_macht]

DWZM
Could you explain more your concerns with a slight difference in forward voltage being a big problem in strings of 12 from the same lot and manufacturer. Yes mixing a bunch if different LEDs would not be good.

Firstly, unless you're buying straight from a major distributor (or Cree directly I guess) in bulk lots, it's hard to know if the LEDs you're getting are from the same order or not, and most of us don't have that luxury. Secondly, I have indeed measured Vf of individual LEDs confirmed from the same batch and have indeed seen Vf differences I consider significant (more below).

I've read most of your threads and would like to know why your so opposed to parallel strings...

I'm not opposed to parallel strings - I'm just opposed to using them haphazardly or without understanding the implications - hence I wouldn't personally implement parallel strings, or suggest someone else does, without understanding/explaining these implications.

Please be specific.

The most common parallel arrangement I've seen people doing or asking about is putting two strings on an ELN-60-48 and running it at full blast (1.3A). If you're using XP-E or XR-E, this is a critically dangerous situation - if one string shuts down, the other string will get a current level above the maximum specified by the manufacturer. Without fuses or voltage limiting, this situation basically means you're risking maybe $75 of LEDs if there's an open failure in one of your strings.

Plus, there's the potential for voltage imbalances. IMHO it's easy for these to seem insignificant but to impact output in a way I would consider significant. I've measured nearly 5% Vf differences in strings of just 6 LEDs from the same batch. A 5% Vf difference doesn't sound significant, but depending on the LED, it can translate to a 60 - 80% difference in current between the strings. That is ABSOLUTELY nontrivial. Even if both strings fall within safe limits, you're going to get a very skewed result in terms of efficiency, much less intensity and distribution on your tank.


Also whats so hard about putting a PAR meter at a measured distance above a fixture and reading several measured points (average them along the way) and having a KilloWatt meter read the total Watts.

Nothing is hard about it, but what's hard is for anyone to judge the confidence in such measurements without understanding the exact conditions under which they were made. Check out Sanjay Joshi's articles available on the web - he's mostly testing MH rigs, but he goes in to great detail in describing the methods and conditions for his experiments.

Plus, at a very basic level, the very concept of distilling efficiency into par/watt is flawed without actually describing the experiment. PAR is a measure of intensity at a specific physical location - it's not a measure of the unit's total output. Meanwhile, watts is indeed a measure of the entire unit's input. So you're trying to calculate efficiency by comparing whole-system input with single-point output.

Again though, all of this is somewhat moot. Going back to the original discussion, I wasn't posting in an effort to discredit your results, just trying to point out that they didn't segregate between driver efficiency and overall system efficiency.

[wld1783]

DWZM

I too share your concern about using parallel strings haphazardly or without understanding the implications. I'm a fan on using quick blow fuses just in case.

Thanks for all your input they are well received as it helps the collective knowledge of this forum.

Please see If I got this right...I'm still confused on the naked downside of parallel strings. 80% current difference is bad but I'm having difficulty finding it.

Please walk me through this as others will follow along.

Worse case each LED has a forward voltage range between 3.2 and 3.5. (I know this is outside the specs of the Cree XR-E at 700mA) But I choose the extreme. With two parallel strings of 12 LEDs driven by a Mean Well ELN 60-48D set to 1300mA total out put. In an absolute worse case 3.5-3.2=.3 * 12 = 3.6 difference between each string or an absolute worse case of 3.6 forward voltage difference between two strings in this example. Given a constant current at 48vDC (assumed for this question) That gives me two strings that work.

42 vs 38.4 is less than 10%

So with a 10% difference one string has a worse case of being driven at 715mA and the other at 585mA. Or a 130mA difference between the two strings. This should provide very close to the PAR of two strings at exactly at 650 mA and similar longevity results.

Am I looking at this the right way?

Bill

[kcress]

DWZM

I too share your concern about using parallel strings haphazardly or without understanding the implications. I'm a fan on using quick blow fuses just in case.

Thanks for all your input they are well received as it helps the collective knowledge of this forum.

Please see If I got this right...I'm still confused on the naked downside of parallel strings. 80% current difference is bad but I'm having difficulty finding it.

Please walk me through this as others will follow along.

Worse case each LED has a forward voltage range between 3.2 and 3.5. (I know this is outside the specs of the Cree XR-E at 700mA) But I choose the extreme. With two parallel strings of 12 LEDs driven by a Mean Well ELN 60-48D set to 1300mA total out put. In an absolute worse case 3.5-3.2=.3 * 12 = 3.6 difference between each string or an absolute worse case of 3.6 forward voltage difference between two strings in this example. Given a constant current at 48vDC (assumed for this question) That gives me two strings that work.

42 vs 38.4 is less than 10%

So with a 10% difference one string has a worse case of being driven at 715mA and the other at 585mA. Or a 130mA difference between the two strings. This should provide very close to the PAR of two strings at exactly at 650 mA and similar longevity results.

Am I looking at this the right way?

Bill

Yes you have it pretty well. The reality is that statistically you will never have those worse cases anyway.

Commercially pretty much no one would run only single series strings. The cost is prohibitive and the lack of efficiency would drive the rest of the nails in that coffin.

Most commercial apps wouldn't bother with the fuses I recommend either. I think the general consensus is that if a string has a problem one of the LEDs will be the fuse or listed fixtures would seem to have fuses in them.

I do recommend both fuses and resistors presently but with time I may drop the fuses. I will always recommend a resistor (1 ohm) because they will always allow you to instantly measure any string's current with out disturbing anything or the hazards and hassle of opening a string for inserting an ammeter.

In our DIY apps, using the resistor, we can tune up all our strings easily to a quality level higher than commercial fixtures usually are.

I was probably the first person to fly the caution flag on paralleling but as our builds are progressing and growing in size my recommendation is having to evolve.

[wld1783]

kcress

Thanks for the feed back.

Bill

[AdamHI]

. . . Enter the Power Factor Controlled (PFC) front ends. Instead of diodes hooked to capacitors the diodes are replaced with a controlled network of transistors and inductors. The control directs the current drawn from the power source to be drawn smoothly over the entire power line cycle not just in gulps - no gulping allowed. This avoids ALL the current pulsing and wire heating. This same network can provide the bonus of eliminating the inrush too. As the initial energy drawn on start-up can be actively limited.(Note that these PFC Meanwells don't limit inrush much.)


kcress - thanks for the explanation - i didn't know anything about PFC front ends. You mention that the PFC Meanwells don't limit inrush current much - and the spec sheets confirm that. So I don't understand what advantage is left for the MW PFC supplies vs the non-PFC. Could you explain? Are the peak instantaneous currents different than what the data sheets are showing? Or am i missing something else?

--adam

[kcress]

Inrush and harmonic distortion are two separate problems. PFC fixes the latter. It can fix the former too but MW didn't bother.

So the advantage is that your house and other electrical devices won't suffer from the harmonic distortion that a bucket full of non-PFC Meanwells will be spewing. That's the advantage.

As for inrush.. If you use the larger PFC MWs you won't have a problem with it as one or two drivers "in-rushing" shouldn't trip any circuit breakers. And, if you somehow used a truck-load of them, they wouldn't overheat your house wiring or breaker panels.

[Skeptic_07]

if you do parallel strings, listen to what these guys say about checking the actual vf of each of your LED's and use fuses as well. these things can drop 200ma just by blowing on them to cool them off when they are really hot. even if you check everything and have all your strings vfs matched, you can over-current a string if a couple of the LEDs in one string are not properly heat-sinked, or if one area is not cooled properly, or if you have another overlapping string that is making one of the parallel strings hotter than the others. this happens when you have lights coming on and going off at different times. parallel strings can be very tricky with these LED's.