DIY LED driver for reef lighting

[der_wille_zur_macht]

Here's the circuit... 4101 circuit

Simon

Ditching the 1uF cap? Trusting your PS is stable enough I guess?

Eagerly awaiting feedback. I noticed in the datasheet that it's not very accurate (current for VLED = 1 V, RSET = 1.47 kΩ can vary from 315 to 385; Rset pin voltage range is 1.17 to 1.23) so it might take some real-world measuring and adjusting of Rset to get the exact value desired in a particular application.

It's just so dirt simple though; very attractive. If you had a well regulated power supply and adjusted it so it was at the minimum of only .5v over the Vf of the LEDs it would maximize efficiency and minimize heat problems.

Perhaps the delay from the board house I was going to use was a blessing in disguise. . .

 

[TheFishMan65]

I think very few are running at 1 amp or really know exactly where we want to be. Just calculate RSET with a voltage of 1.23 and you won't ever overdrive the LEDs. I was thinking of putting a potentiometer (10k) in series with RSET this gives me manual dimming since I don't have a PWM circuit yet. Later I can take the pot to 0 (or whatever I want my peak to be) and use a PWM.

 

[dc3]

How many Leds are you guys thinking of driving with this CAT4101 regulator. ?

6 Leds at 700 mA would generate over 17 W which needs to be dissipated
by the device.


Gisli

 

[TheFishMan65]

I am not sure where you got your number dc3. Are you using 24 volts at .7 amps for 16.8 watts?

The chip does not have to get rid of all that. Assume the LEDs have a voltage of 3.5 and 700 ma. So 6 times 3.5 is 21 volts. So the chip only dissipates (24-21)*.7 = 2.1 watts.

As DWZM says if we can get that power supply at to 21.5 volts then we are only wasting .5 volts times .7 amps or .35 watts and using 21 * .7 = 14.7 watts so the efficiency is 14.7 / (14.7 + .35) = 97.6%.

[EDIT]
To really calculate efficiency I should have added the loss from RSET.
1.2 *1.2/549 = 2.6 milliwatts
14.7 / (14.7 + .35 + .0026) = 14.7 / 15.0526 = 97.67% (guess my rounding was off before)

 

[dc3]

Well I was reading page 8 of the datasheet. Pd=Vin x Iin + Vled x Iled.


Gisli

 

[TheFishMan65]

Well I missed the power for efficiency from the actual chip (the Vin * Iin). Vled is the voltage on the PIN label LED. This is not 24 volts, but is 24 volts minus whatever is dropped by LED (in the example above 21 volts).

power used by the chip is 5 volts * 8 ma = .04 watts
Efficiency = 14.7 / (15.0526 + .04) = 97.40%

 

[TheFishMan65]

Help

Help

Could one of you who has number please help these folks
how efficient are LEDs really? - http://www.reefcentral.com/forums/showthread.php?t=1794967
I would hate for them to give us a bad rap. Yes, I suggested they read some of the threads.

Thanks in advance

 

[der_wille_zur_macht]

Well I missed the power for efficiency from the actual chip (the Vin * Iin). Vled is the voltage on the PIN label LED. This is not 24 volts, but is 24 volts minus whatever is dropped by LED (in the example above 21 volts).

power used by the chip is 5 volts * 8 ma = .04 watts
Efficiency = 14.7 / (15.0526 + .04) = 97.40%

Plus arguably a tiny fraction of a watt lost in the other components, but still, it's very good. As with the other linear chips mentioned earlier in this thread, I'd assume that you would REALLY want to watch your power supply voltage and try to keep it as close to Vf + 0.5 as possible, because anything above that just turns into heat.

Could one of you who has number please help these folks
how efficient are LEDs really?

I did a brain dump in there. Hopefully it helps. Though I'm worried it'll just convince people that I'm loony.

 

[TheFishMan65]

Aren't we all. Thanks for trying to straighten them out.

[EDIT]
DWZM I just read the other post. Great, anyone considering LEDs might want to read it since it is a great summary.

 

[SpacedCowboy]

Ditching the 1uF cap? Trusting your PS is stable enough I guess?

For my tests, I'm using a bench PSU. It ought to be stable to the Nth degree [grin].

Eagerly awaiting feedback. I noticed in the datasheet that it's not very accurate (current for VLED = 1 V, RSET = 1.47 kΩ can vary from 315 to 385; Rset pin voltage range is 1.17 to 1.23) so it might take some real-world measuring and adjusting of Rset to get the exact value desired in a particular application.

If you're trying to obtain a specific current value, I guess that's an issue, but then you'd also want to look into temperature variance, component accuracy etc. My POV is that I want it operating roughly around 70% maximum (for longevity), which gives me plenty of headroom for variability. Given that I can PWM it down from there, I'm happy with that level of ambiguity [grin]

It's just so dirt simple though; very attractive. If you had a well regulated power supply and adjusted it so it was at the minimum of only .5v over the Vf of the LEDs it would maximize efficiency and minimize heat problems.

Perhaps the delay from the board house I was going to use was a blessing in disguise. . .

The 8A/24V MPJA PSU can be trimmed down to ~21v if I remember correctly At least that's my plan, if this device actually works as advertised.

Simon.

 

[der_wille_zur_macht]

The 8A/24V MPJA PSU can be trimmed down to ~21v if I remember correctly

Yes, I can confirm that their 6.5 and 8.3 (or whatever the exact figures are) 24v supplies can be trimmed to 21v.

At least that's my plan, if this device actually works as advertised.

Eagerly awaiting results!

 

[OTCHU]

so der_wille_zur_macht what kind of leds to you plan to use ? I am just trying to figure out for Vf.

I just did order the CAT4101...

 

[der_wille_zur_macht]

My 360g will likely have a mix of R4 or R5 cool white XP-G and royal blue XP-E.

My various nano builds, planted tank builds, and algae turf scrubber build - just about everything under the sun. Which is why a power supply with a trimpot is really nice - you could tune it for as small a drop as possible regardless of the LEDs.

I bet a CAT4101 could drive 7 XP-G on a 24v supply since they have a lower Vf.

 

[OTCHU]

The XP-G run at 1500mA ... so we can't run them at the full power, right ?

for the XP-E ... Vf at 1000mA= 3.5v so 24v/3.5= 6.8 ... so we can drive only 6 XP-E led, am I right ?

 

[TheFishMan65]

OTCHU
Yes they run at 1500 ma. Some drivers could run them in that mode. The CAT4101 says that two can be run in parallel.

Yes only 6 XP-Es.

Just to qualify after reading the discussion on efficiency. I should point out that the efficiency will also be affect by the power supply. Anyone know how efficient the MPJA are?

Say 97% for a DYI driver and a 95% power supply we are down to 92%.

 

[der_wille_zur_macht]

The XP-G run at 1500mA ... so we can't run them at the full power, right ?

My personal approach is to run these LEDs well under their "maximum" ratings anyways - usually between 500 - 700mA.

TheFishMan, I haven't found exact datasheets for the cheap MPJA power supplies but similar cheap supplies are usually in the 80 - 85% range, so not as good as you're assuming.

 

[TheFishMan65]

That is what I was afraid of. So really we should be saying these are closer to 76% to 82% efficient since the power supply is an integral part of the usage. We don't have a lot of control over the power supply but to be fair to T5, T8, MH, etc. we are using their over all efficiency.

Not trying to start an argument really, but just trying to share info. I measured my fluorescent light. Ice cap (660?) running four T12s. From what I understand these are driven at 110 watts - even for regular bulbs. If this is not true of the Ice Cap I was badly misinformed. So we have a total power of 440 watts at 120 volts so about 4 amps right? Wrong only 2 amps - measured on a kilowatt meter. They are supposed to be very efficient. They must somehow use the discharge of one lamp to light the next.

Anyway I guess what I am trying say it may be harder to state the effiency than we think.

 

[der_wille_zur_macht]

Fishman,

Commercial AC LED drivers (meanwells) can hit 90 - 95% efficiency. I'm sure we could make or buy a more efficient DC supply if we wanted to. At any rate, I try to use efficiency as a relative measure - i.e. comparing one DC-DC LED driver to another (my NCP3066 design to this new chip, for instance). There are way too many variables to compare efficiency of one system to another, IMHO the better approach there is to just compare whole systems on like tanks rather than trying to work it up based on math - and having done that, LEDs typically require 30 - 50% the wattage of MH or fluorescent.

I'd check the output of your Icecap if you can - if it's running four T12 VHO lamps on only 2A it's either vastly under-driving them or something else is wrong. Or your house has 240v AC.

 

[TheFishMan65]

That is sort of what I thought. I am hoping someone else has the same setup and can measure it.

I agree that LEDs are more efficient that other lighting (especially when concerned with light in the tank), I am just wondering if our comparisons have been completely fair is all. Also it may pay to find a more efficient power supply

 

[der_wille_zur_macht]

FWIW years ago I put a watt-meter on several VHO fluorescent ballasts I was running at the time, and the results were in the 70 - 90% efficiency range. Didn't have any IceCaps though.

Efficiency has several components - one is getting electrical power from the wall to the device that actually emits light (the lamp or LED). In that sense, LEDs are probably not much different from other forms of lighting, so I only try to address efficiency in a relative sense as I noted above. In the other sense - converting electrical energy to light energy usable by a coral a few feet away in a fish tank - I think LEDs have pretty clearly won, but I'm sure others would debate that.