DIY LED driver for reef lighting

Steedy,

This thread is about DIY drivers. If you're looking for Arduino programming help, you might be better off asking in a different thread or starting a new thread.
 
Steedy,

This thread is about DIY drivers. If you're looking for Arduino programming help, you might be better off asking in a different thread or starting a new thread.

Yes but yours DIY drivers need SW too.. or no? in other thread dont used a CAT4101 , here yes .. My LED lamps is DIY too not buy.
 
Steedy,

No, there is no software on the drivers we're discussing in this thread. Many of us are using Arduinos to control the drivers, but this thread is really intended to focus on the hardware itself. If you want software help, it would be best if you started a new thread or posted in one of the many Arduino threads in the forum.
 
Hi All,

Just wondering who has used the PT4115 driver. What are your likes and dislikes? On paper it seems like a nice solution.

Tom

Boards are being made now so waiting for them to come in. Then I'll build them up, start testing, and write arduino libraries for interfacing with the temp chip and dimmer chip.

Benefits are it's pretty cheap of you're only needing 700ma per driver. You can get the DX drivers for $2.10 each and in bulk they go down even more. Just got 15 more and they came down to $1.60 each. With those you can use the diode, PT4115 chip, current sense resistor, and inductor. This is much cheaper than buying these parts from digikey/mouser individually.

Another benefit is it should run a lot cooler than the linear drivers as long as you keep the input voltage close to the total voltage being used. For example, multiply your LED forward voltage by the number of LED's and add about 10-15% and you'll have the total input voltage you need. It can take up to 30v input but if you're only running 5 LED's per string you'll only need 18v so the chip will generate more heat if you use 24v or 30v.

I also have a variable resistor for each driver to adjust the total output amperage so if you're running 350ma LED's you can build out the boards with 700ma, 1000ma, or 1200ma parts and then turn them down to your desired output.
 
Thanks for the info. Its the heat that I am most concern with. I have different color LED's in the fixture - for example one color (I think its red) has a forward voltage of like 2.1 volts vs the white has a voltage of 3.8. Have a 27 volt input (24volt powersupply turned up all the way). Works great for the White, but the red only needs a little over 12 volt, so that a pretty big difference (27 - 13volt). I know the CAT4101 driver gets too hot, been using the MAX 16820 chip, and that works but just looking at other options.

Tom
 
hey, probably a stupid question, but how does one properly solder a CAT4101 chip to the PCB without burning the thing out? I just tried replacing some on a pre-built driver I ordered, and either fried them from overheating the chip, or from static caused by my sliding the PCB a few inches on my kitchen table. I originally used a cheap stick soldering iron, and tested the chip after soldering it on and it was good, then a few minutes later, tested it again and found it to be faulty but am ordering a hot air/ soldering iron station (has both). I am thinking the hot air would be useful for soldering the ground pad onto the PCB.
 
Thanks for the info. Its the heat that I am most concern with. I have different color LED's in the fixture - for example one color (I think its red) has a forward voltage of like 2.1 volts vs the white has a voltage of 3.8. Have a 27 volt input (24volt powersupply turned up all the way). Works great for the White, but the red only needs a little over 12 volt, so that a pretty big difference (27 - 13volt). I know the CAT4101 driver gets too hot, been using the MAX 16820 chip, and that works but just looking at other options.

Tom

The CAT4101 driver is only rated for 25V, not 27V I don't know about the MAX 16820, but that may be the source of your heat problem
 
Well, my phosphorus is back up to 95:

Code:
14-Jan-12	364	1.116024
5-Feb-12	300	0.9198
8-Feb-12	105	0.32193
12-Feb-12	60	0.18396
22-Feb-12	139	0.426174
25-Feb-12	134	0.410844
26-Feb-12	34	0.104244
27-Feb-12	30	0.09198
 8-Mar-12	95	0.29127

The big drops are when I've dosed - Guess I'll dose again. Some of the base rock in the tank is over 15 years old, and I skimmed much worse way back when. I have not put the phosphate reactor on yet, I'm worried I'll exaust the media.

You think the increase is due to poor maintenance OR is it leaching from the rocks?

During the last 2 weeks, I've had 2 skimmers running continously and a small water change (only 30 gallons) on the 120. I feed a pinch or two of flake food, and it's all consumed.

I don't get much precipitate but the filters do clog.

Advice?

== John ==​
 
Cat 4101 Troubleshooting

First off, a gigantic thank you to DZWM and everyone that has contributed to this thread as this work/knowledge has been a monumental help to me.

So I've been using this design for over a year now on my own LED array. I got a batch of Luxeon's stupid cheap so I just ordered more than I needed and started working :). My system is setup with 72 LED's (36 white, 36 blue) and each group of 36 on their own 24V power supply (the affordable adjustable switchers). I have a 6S6P arrangement for both whites and blues. Drivers are very simple CAT4101 prototypes with one of DZWM's earlier schematics (CAT, resistor, capacitor, 5V control, done). I have no dimming, just tie the 5V to PWM pin for instant on. I did dial in the supply voltage until the drivers were barely working but I did still need just a little fan on the heatsinks for all of them as they would occasionally thermal-overload and shut down (sealed in a plastic box with no airflow and small heatsinks).

Everything was fine for a year+ until about 3 weeks ago, RIGHT before I left for a long vacation one of my banks of 6 blue LED's decided to stop working. Lame, but I had no time to troubleshoot it before leaving so I just disconnected the negative lead from LED string and left. The blues are really more for color anyways and I wasn't going to look at it on vacation :). I come back from vacation early this week and everything's same as I left it, figure I'll troubleshoot over this weekend. So yesterday (sat) When the lights power on, I have no blues at all. Disconnect power supply, start it, works fine unloaded, also OK with a small load of fans. Pin out +/- and find a short. Great. My first instinct is blown driver, so I remove all the ground connections for the CAT4101's but still shorted.

The only other option is the 1uf caps on the + side to ground. Remove all 6 and find one of them is dead. So I'm wondering why/how the hell a ceramic 1uf cap like that goes bad (it's a Vishay 0805 size) and I'm now wondering if I'm defying the laws of physics. The original diagram for the CAT4101 has a 1uf cap from +24v to ground for the driver. Yet since I'm running 6 parallel I installed 6 caps, but that seems dumb to me now. I've essentially summed the capacitance to 6uf by doing so right? Should I instead have only used a single 1uf cap for all 6 strings of LED's since all the +24 are in parallel same with a single ground? Or was I right and this stupid freaking solar flare killed my poor capacitor. More importantly if I SHOULD have had 6 caps, will it run/will I screw something up if I use 5 until I can get a replacement?

BTW, I may be good at soldering SMD components, but servicing them is a B :(
 
Wrong thread! :eek2: No wonder my response didn't show up, LOL.

== John ==​

I hate that. You think someone removed your post because 'it's gone' the next time you look. You get all paranoid and angry. Then a day later as you're browsing you find it misplaced where YOU put it. :bdaysmile:
 
A little update in case anyone does read my troubles. Upon closer inspection of the LED array, I noticed that one of the focusers I had on them was missing, as was the protective cap on the diodes themselves. There was a nice burnt look/smell coming from it. So it appears that the LED was running un-protected in a seawater environment for a year. Not bad All things considered and it failed open instead of closed which I thought was interesting. Either way, that's probably my point of failure but my question remains about the caps. If I have 6 parallel strings, does each string require a 1uf cap from +24v to ground, or just a single cap per string? Right now It's back up and running (I smartly bought spare LEDs :)) with just a single cap.
 
Hey guys I have had a bunch of parts sitting around for a few years. I fired up my CAT4101s, they work, they dim and I'm happy.
But for the PSU, can I use a typical wallwart? For voltage, I knowunderload voltage can shift for unregulated PSUs. Can I somehow size them properly?
Or am I better off buying a regulated PSU(what voltage for 4 White XPG and 5 Royal Blue XPE?) at 2 amps?

Once I finish up my tiny 9 CREE light I'll if my coral adapts to it do a light for a 48" tank.
Electricity($.50+ per kilowatt) and shipping for new T5 bulbs is sky high.
 
I have just finished building my LEDs and i have noticed that the heatsinks get to around 50 to 53C and the enclosure where the drivers are is sitting at 45 - 50C. Is this temperature OK for the components?

Thanks
 
Thanks for the info. Its the heat that I am most concern with. I have different color LED's in the fixture - for example one color (I think its red) has a forward voltage of like 2.1 volts vs the white has a voltage of 3.8. Have a 27 volt input (24volt powersupply turned up all the way). Works great for the White, but the red only needs a little over 12 volt, so that a pretty big difference (27 - 13volt). I know the CAT4101 driver gets too hot, been using the MAX 16820 chip, and that works but just looking at other options.

Tom

You need a switching regulator. No linear reg will handle a drop like that. I'd suggest the LM3409 we've been talking about for the last several pages.

hey, probably a stupid question, but how does one properly solder a CAT4101 chip to the PCB without burning the thing out?

Very carefully? :D IMHO forget about the hot air part of the station you just got. I posted this a while back, but this is what I do:

1) Tin the ground pad on the PCB.
2) Place the part and solder a single pin to hold it in place.
3) This part will depend on the shape of the tip you're using, but crank the heat and position the tip along the top of the exposed part of the ground pad on the IC such that your iron is touching the exposed metal AND the pad on the PCB. Hold it for several seconds. As soon as it's hot enough, introduce solder. Use more solder than you think, you want to get it to wick all the way under the entire ground pad on the part.
4) Pull the solder wire away but leave the iron there. Have a tool handy - tweezers, chopstick, whatever. Grab it quick and push down a bit on the chip to ensure good contact with the PCB. Pull the iron off and you're done with the ground pad.
5) Solder the other pins on the chip as usual. They're pretty big and far apart so it is easy. If you make any solder bridges, clean them up with wick and you're done.

I just tried replacing some on a pre-built driver I ordered, and either fried them from overheating the chip, or from static caused by my sliding the PCB a few inches on my kitchen table.

My guess would be accidental solder bridges or a short somewhere else on the board. The chips are pretty durable.

I originally used a cheap stick soldering iron, and tested the chip after soldering it on and it was good, then a few minutes later, tested it again and found it to be faulty

Faulty as in it didn't work? IME these chips don't just stop working when they die, they go up in smoke. If it's simply stopped working, check for bad solder joints, a dead LED, or other conditions. Make sure the whole circuit is appropriate for your design, i.e. voltage drop is in a reasonable range, and so on.

If I have 6 parallel strings, does each string require a 1uf cap from +24v to ground, or just a single cap per string? Right now It's back up and running (I smartly bought spare LEDs :)) with just a single cap.

Call the manufacturer and ask. It's an interesting question. I asked waaaaaaay earlier in this thread and we didn't really come to an obvious consensus so I designed it the way I did, with a cap on "each" string of LEDs even though they're effectively in parallel.

Putting caps in parallel does in theory sum their capacitance, but (one of the EEs can correct me if I'm wrong here) that doesn't tell the whole story since different caps will react in different ways. So it may not exactly be the 1uF capacitance we're looking for, but some other property common to caps around that size.

Hey guys I have had a bunch of parts sitting around for a few years. I fired up my CAT4101s, they work, they dim and I'm happy.
But for the PSU, can I use a typical wallwart? For voltage, I knowunderload voltage can shift for unregulated PSUs. Can I somehow size them properly?
Or am I better off buying a regulated PSU(what voltage for 4 White XPG and 5 Royal Blue XPE?) at 2 amps?

You can always try it and see. With the CAT4101 design the biggest enemy is difference between input and output voltage. It might not be an issue that the supply is unregulated but it may be an issue that it's not adjustable, allowing for fine-tuning. Measure what voltage it puts out under load and then design an LED string with enough LEDs to get as close to that voltage as you can while staying at least half a volt under it.

And don't count on the wall warts lasting if you load them more than about 50 - 60% of their capacity...

I have just finished building my LEDs and i have noticed that the heatsinks get to around 50 to 53C and the enclosure where the drivers are is sitting at 45 - 50C. Is this temperature OK for the components?

Thanks

The LED heatsinks are at 50C or the driver heatsinks? Which driver design are you using? Is the enclosure air tight?

If you're using the CAT4101 it'll go into thermal shutdown if it gets too hot (the LEDs will start to blink on and off every few seconds as it heats up and cools down).
 
Back
Top