Aquastyleonline.com LED questions.

[richiero]

i was just wondering how many leds i should get?

i was just wondering how many leds i should get?

i am up grading from fresh water to salt and i have a 75 gallon tank at the start it is going to be live rock and fish with the intent of adding some easy to moderate sps and lps coral. i called reefled and he told be i would need 90 led he told me to get 2 Phoenix kits over 1,000 what do you guys think? i am looking at the aquastyle site and they have a diy kit for $189 for 72 and $ 229 for 90.is 90 leds over kill?

thanks for your help

 

[dread240]

90 would be overkill. I run 76 LED's over my 75g and have them dimmed way down so I don't bake the corals... Hell I've actually reduced my photo-period some too

 

[zachts]

you do not want to have too many drivers on one circuit starting up at the same time, not sure how much load they can take, but from what i have read forward voltage is an issue when using too many drivers, you can overload the wires. the options are to run the leds in parallel or have them on timers, so not all of them turn on at the same time.

I believe the reason you don't want more than 4ish drivers on a circut is that when they first turn on they draw somthing like 60-70 amps each for a split second to charge up the capacitors in the driver. if you have a lot of them on the same breaker it will trip every time they start up. It doesn't have anything to do with the forward voltage of the leds.

runnning parralel strings on bigger drivers or having them turn on at different times as nemo suggests is the typical work around. Course you could just put in a dedicated breaker for you lighting too........

 

[gbru316]

I believe the reason you don't want more than 4ish drivers on a circut is that when they first turn on they draw somthing like 60-70 amps each for a split second to charge up the capacitors in the driver. if you have a lot of them on the same breaker it will trip every time they start up. It doesn't have anything to do with the forward voltage of the leds.

runnning parralel strings on bigger drivers or having them turn on at different times as nemo suggests is the typical work around. Course you could just put in a dedicated breaker for you lighting too........

Aka inrush current.

A negative temperature coefficient thermistor can be placed in series with the circuit whose inrush current you wish to limit in order to reduce the current spike.

However, all decent drivers SHOULD incorporate some type of inrush limiting circuit to begin with. Without schematics or experimentation (or someone willing to open up their drivers), it's best to play it safe.

 

[Snausy]

to be honest with you, on a tank that big, i would use a mixture of xml warm white and cool whites and cree xte's for the royal blues.

if money is going to be a issue, then the 2x 120 kits would be the way to go, don't use channels, instead go with heatsinks, this way you can cluster the leds closer together and get better par out of them. also try to bix the inner rows of leds with 40 degree optics and then the outer with 60's. this is for the aquastyle setup. for a cree build, i think 60's would be good.

if you read back a couple of post, you will see that i started adding cree's to my build because my tank is so wide and deep. so to reduce the amount of leds and drivers i use, i went with cree's.

you do not want to have too many drivers on one circuit starting up at the same time, not sure how much load they can take, but from what i have read forward voltage is an issue when using too many drivers, you can overload the wires. the options are to run the leds in parallel or have them on timers, so not all of them turn on at the same time.

another option you can do is to get a 120 kit and then get 4 heatsinks from rapids site along with xml's and xte's to give you a high noon effect, then you can ramp them down or turn them off after a certain time.
so you would have cree's on heatsinks in the middle and aquastyle on the outsides via channels.

hope this helps.

thanks for that felix it was very helpful. seems like no matter what I'm going to do it's gonna be more complicated than what everybody else gets to do because of the size of my tank

 

[nemosworld]

that's what we are here fo, to help each other out. glad i could help.

 

[gammer_100]

I have 108 over my 125 and I have them running full blast. The coral growth is pretty stunning compared to my previous lights. I am keeping everything from sps, lps, softies, and clams underneath the LEDs. They have been up for about a month and a half now? Here's some eye candy pics with the Bridgelux LEDs

 

[jayinh]

does anyone know what type of pot i can use instead of the supplied one? i think mine that came with the kit are junk so i got one of these from the shack http://www.radioshack.com/product/index.jsp?productId=2062354 and it does dim alittle but wont bring the light for bore. what pot spec should i be looking for? has anyone used a different pot than the one that is supplied. thanks

 

[Honks69]

I have 108 over my 125 and I have them running full blast. The coral growth is pretty stunning compared to my previous lights. I am keeping everything from sps, lps, softies, and clams underneath the LEDs. They have been up for about a month and a half now? Here's some eye candy pics with the Bridgelux LEDs

Did you go with the heatsinks or did you use the C channels?

Looks sweet

 

[dread240]

does anyone know what type of pot i can use instead of the supplied one? i think mine that came with the kit are junk so i got one of these from the shack http://www.radioshack.com/product/index.jsp?productId=2062354 and it does dim alittle but wont bring the light for bore. what pot spec should i be looking for? has anyone used a different pot than the one that is supplied. thanks

you would want a 25ohm potentiometer, not a 10,000ohm

 

[gbru316]

Dread, what about using the lowest value digital pot available, then adding a parallel resistor or 2 to drop it to 25 ohms?

 

[dread240]

we're talking the range of .1 ohm to 5 ohm realistically here.

so lets use a 500 ohm digital pot just for reference, and for ease of use it has 50 steps, so 10 ohms resistance per step. We will use a 5 ohm resistor in parallel

500ohm + 5ohm in parallel = 4.9505 ohms

lets go halfway
250 ohm + 5 ohm in parallel = 4.902 ohms (you've already used up 25 steps of the pot and have seen no change

Lets go 98% which would be 10 ohms
10 ohm + 5 ohm in parallel = 3.333 olhms... that's 140ma current from the driver

Our last step should theoretically be no resistance and a straight switch
0 ohm + 5 ohm in parallel = 0 ohms... and that's 620ma current from the driver

It works, technically, but the resolution is crap, hence why you have to use the resistor array and tap different points through a mux (well you could just tap different points themselves and call it a day, but it will use up alot of pins that way, to where as you get 16 positions with only 4 control wires

 

[dread240]

I literally already went that way btw, It's still sitting here next to me... a 6 channel spdi digital pot, all code written and working, just no real resolution when it came to dimming. If you go back further in this thread you'll see the prototype board for and arduino sketch already made up

Realistically I should have been done this ages ago, I'm probably the worst person for everybody to be waiting on considering how busy I am with work and travel right now. Nobody else really stepped up to make it though.

I am planning on testing a few things to see if I can save myself some headaches on these boards and using the arduino to trigger the mosfet to let the dimmer circuit pass through (basically passing 2 dc circuits through one conductor). The actual electronics assembly isn't my specialty, control systems are, but I have enough knowledge in this at least to get myself in trouble

 

[gbru316]

I literally already went that way btw, It's still sitting here next to me... a 6 channel spdi digital pot, all code written and working, just no real resolution when it came to dimming. If you go back further in this thread you'll see the prototype board for and arduino sketch already made up

Realistically I should have been done this ages ago, I'm probably the worst person for everybody to be waiting on considering how busy I am with work and travel right now. Nobody else really stepped up to make it though.

I am planning on testing a few things to see if I can save myself some headaches on these boards and using the arduino to trigger the mosfet to let the dimmer circuit pass through (basically passing 2 dc circuits through one conductor). The actual electronics assembly isn't my specialty, control systems are, but I have enough knowledge in this at least to get myself in trouble

I figured you'd thought of it already, you seem pretty knowledgeable.

No worries about lack of time. I'm juggling a full time job as an RF technician and college for my BSEE, and a wife, 4 dogs and home upkeep.

It's a shame you aren't local, I'd be more than willing to donate my electrical assembly skills towards the project. I spend 8 hours a day troubleshooting/soldering smd parts via microscope (mixed in with copious amounts of RC through the iPhone).

So basically, you're going to try to control dimmer current by changing the size of the MOSFET channel by varying gate voltage?

 

[dread240]

Nope..

I don't remember max current on the dimmer circuit atm but it was north of 350ma. The only 16 channel multiplex chips I could find were surface mount and only capable of handling 150ma-200ma. I'm going off the top of my head, but I think long term we would fry a multiplex without isolating it.

N channel mosfets are dirt cheap, and can handle the current. By building the gate with voltage from the arduino, I can then conduct the dimmer circuit through the mosfet. One mosfet per channel for the 16 channels of the multiplex ic.

You would basically have 2 lines for the circuit... Both +VDC (straight line from the LED Driver and an inline resistor on the arduino supply before they join, otherwise you would need a resistor on every base leg) will be on a shared bus so to speak. The collector of the mosfets would all be off this bus. The base of the mosfet would go through the mux to arduino ground, and the emitter of the mosfet would go to a tap position of the resistor array.

The resistor array will simply be .1 ohm resistors in series. The -VDC of the driver will go to the front of the array, and you will tap the emitter's to various points within it. Once you command a channel on the mux to energize the gate of the mosfet, the driver circuit will flow from the common bus, through that mosfet to that tap in the resistor array, giving you the variable resistance you'll need to control the dimming.

The reason I typed it all out. That's what the guy that does not do electronics for a living has come up with. I thought of relays as well too but they're too expensive for this project imho. Once you reach a certain point you may as well opt for a pwm driver as you're not saving any money if there's too much cost in components. You on the other hand, may have a few people around work or yourself for that matter that might find some better components for this... like mainly a controllable chip (I went with the multiplex setup as it gave me 16 steps for only 4 wires, which was pretty smooth transitions). I haven't been able to dig too much on all the various mux chips. Ideally if we could find one that could handle 350ma then no mosfets are even needed whatsoever and it simplifies it greatly. Or if we could find cheap chips that may only be 8 channels but are 2 wire capable with higher current they'd work too.

That's just some of the few things other people could look into and try. I know my plan will work and give the end results I want, but I don't know all the various semiconductors and IC's out there, if it's the most effecient cost wise, or if there are just flat out better options for this low and tight of a resistance range.

 

[Jamesjkl]

can this be used as three separate chips ? I cant tell from this description but the n-channel fets are still needed looking at this less soldering

http://www.sparkfun.com/products/9832

 

[dread240]

That may very well simplify the circuit alot, and give you 3 dimmer channels to work with.

I was originally looking at http://www.sparkfun.com/products/9056 for each channel but I think that shield you posted there would make for a much cleaner installation, plus give me and extra mux if I needed it.

 

[Jamesjkl]

have you thought about this http://vimeo.com/27660337 http://www.sparkfun.com/products/733 for maybe fewer pin use? I'm not a programer since the did a way with FORTRAN IV

 

[dread240]

Those could work too but may make coding a bit more difficult. Don't confuse me with a programmer either LOL, I'm just working within my boundaries

 

[gbru316]

Those could work too but may make coding a bit more difficult. Don't confuse me with a programmer either LOL, I'm just working within my boundaries

If you're not trying to expand your boundaries, you're doin it wrong!



What about using a switch IC? Tie all the inputs together (1 side of the dimmer wire) and route each output through a different resistor value to the other dimmer wire. Use arduino to control the control pins.

The only downside is that it might be a little bulky when controlling several drivers. Maybe limit each driver to 3-5 intensities, selecting desired resistor values beforehand to make up for the low resolution.