DIY LED driver for reef lighting

[shackscs]

Efficiency.

From the first page:

Regulators for LED use like this chip can generally operate in three ways. The difference in operation is in the relationship between input voltage and output voltage. The difference in implementation is where the inductor is in the circuit in relation to Vin and the switch.

1) Buck: The regulator chops down the input voltage to keep the LEDs at the desired current. In a buck regulator, the output voltage must always be LOWER than the input voltage. Think of it this way: The input voltage is trying to spike the LED array too high, so the regulator chops it down to keep it correct.

2) Boost: The regulator boosts up the input voltage. The desired output voltage must always be LOWER than the input voltage. Think of it this way: The input voltage is not high enough, so the regulator switches it higher to keep the LEDs happy.

3) SEPIC: It doesn't matter. SEPIC topology uses two inductors, one in each position, so the voltage can be anything (in range of the chips min/max) compared to the desired number.

Buck and boost are much more efficient, but SEPIC is much more flexible. In a situation where input voltage might be all over the map (think: LED lighting in automobiles, where system voltage is pretty wild) SEPIC is good. But in our builds, we generally have DC power supplies that are nearly dead-stable, so we don't need the flexibility. Hence that leaves buck and boost. Boost is a little more efficient and allows for higher LED count on a given input voltage, but buck allows for more resolution when dimming. I picked boost for this design, mostly due to the efficiency.

 

[swimmer4uus]

Thanks. I don't know if this should be in this thread or the LED thread, but here goes anyways. Last couple days I've been researching DIY drivers. I've been sort of following this thread for about a month, and finally sat down to read through all the pages. Next day I get off I'll do it again and take some notes. My PS is the 24V @ 6.5A. I have strings of 5 LEDs because I first went with the super simple LM317 regulator with resistor to set my current. Then started reading, and I'm wanting to switch to the NCP3066. Now, my string of LEDs doesn't let me use boost at the moment, but if I connected 2 strings in series, the voltage requirement would then set me into the boost topology yes? Instead of having 10 strings of 5, I'd have 5 strings of 10. Did any of you guys create a single driver per board? I understand why you guys did 2 or 3 on a PCB, but for me I'd still want it 1 per board until I'm more familiar with circuit boards.

 

[ackerman]

I might have missed this somewere in the last 29 pages but how many of the CAT4101x3 pcbs can i run off the 24v 6.5a ps safely?

 

[der_wille_zur_macht]

Efficiency.

Also, I didn't really say this, but it's more "efficient" in terms of cost because you get 8 LEDs per circuit on a 24v supply instead of 6 (which you'd get in buck mode). So your "cost to drive each LED" is lower.

Thanks. I don't know if this should be in this thread or the LED thread, but here goes anyways. Last couple days I've been researching DIY drivers. I've been sort of following this thread for about a month, and finally sat down to read through all the pages. Next day I get off I'll do it again and take some notes. My PS is the 24V @ 6.5A. I have strings of 5 LEDs because I first went with the super simple LM317 regulator with resistor to set my current. Then started reading, and I'm wanting to switch to the NCP3066. Now, my string of LEDs doesn't let me use boost at the moment, but if I connected 2 strings in series, the voltage requirement would then set me into the boost topology yes? Instead of having 10 strings of 5, I'd have 5 strings of 10. Did any of you guys create a single driver per board? I understand why you guys did 2 or 3 on a PCB, but for me I'd still want it 1 per board until I'm more familiar with circuit boards.

You can lay out the driver on perfboard or on those little "DIP PCBs" Radio Shack sells. That's how I started. Just look at the schematics and put pieces in place to get the simplest (i.e. fewest jumpers) layout you can.

Boosting a 24v supply to run 10 LEDs might be tricky - the datasheet claims 8 LEDs max for the NCP3066 chip iirc. Also any time you're running a config different than what's detailed here, you'll want to check limits on all your parts so you don't fry anything!

I might have missed this somewere in the last 29 pages but how many of the CAT4101x3 pcbs can i run off the 24v 6.5a ps safely?

Just figure out the current you'll pull from each circuit and add them up, with maybe a 30% overhead for safety. So if you're running at 700mA, you'll want to "budget" 1A for each circuit. That means a 6.5A power supply can run two of the 3x PCBs.

 

[swimmer4uus]

Ok, so I want to try to make one buck prototype with PWM dimming capabilites for an Arduino. I have the basic buck configuration from the data files on the NCP3066, and I could make that no problem. As far as the PWM I have no clue. How would I go about designing a new circuit based off of this one?

Also, couple questions on the chip and circuit. The load labeled D1, is that actually a connected wire, or just a reference voltage? Is pin 8 the PWM pin? Would I be using that pin with input voltage from the arduino to dim? I'm assuming it's the same pin you guys are using since it's the same exact chip you used in ur 3066 based design. Their "ON/OFF" label is confusing me I guess

 

[der_wille_zur_macht]

"D1" in that drawing is a schottky diode you must have in the circuit for it to operate correctly.

To get PWM working, you just connect a 10k resistor to the ON/OFF pin and connect your PWM signal (a pin on the Arduino) to the other end of that (i.e. put the resistor in series - it's there to limit current on the ON/OFF pin). Then of course make sure the GND from the driver is common with the GND from the Arduino and you're all set to go.

 

[swimmer4uus]

Is there different values for the diode? Is that going to be dependant on the design and output goals? Is there a reason that's not in the design spreadsheet? I just gives a value for duty cycle. Unfortuneately for what I'm doing, I'm a mechanical engineer, not electrical.

 

[der_wille_zur_macht]

The values for that diode aren't critical. Just make sure it's current rating is above what your circuit will be running at. If you want a through hole part, you can just use the one I spec'd. If you want an SMT part, you can use the one called for in the demoboard BOM from the datasheet.

 

[swimmer4uus]

Thanks DWZM, that cleared up SOOOO much.

Yeah, everything's through hole. I don't posses the juevos yet to have boards printed up yet. Mind if I post values and you take a look see at it? Maybe point out what's wrong?

 

[funkmuffin]

CAT4104 Triple and Independent Dimming

CAT4104 Triple and Independent Dimming

My lighting plans are somewhat ambitious.

In the end I'll need a couple boards where I can dim one or two of the drivers separate of the others. Looking at the circuit, it looks like simply cutting a trace from the common dimming connections to one of the drivers, then connecting the independent dimming signal directly to that 4101 would work, no?

I guess I could also design an alternate version with independent dimming. That would be a much prettier solution.

Was there a reason for common dimming, I don't seem to recall that coming up in the thread, but I very well may have missed it...

 

[swimmer4uus]

DWZM, when coming from the Arduino, how are you wiring it up to all the boards? There's one wire coming from the Arduino, but multiple boards, and with voltage/current division, how would you make sure the right signals are being recieved by the drivers?

 

[der_wille_zur_macht]

My lighting plans are somewhat ambitious.

In the end I'll need a couple boards where I can dim one or two of the drivers separate of the others. Looking at the circuit, it looks like simply cutting a trace from the common dimming connections to one of the drivers, then connecting the independent dimming signal directly to that 4101 would work, no?

I guess I could also design an alternate version with independent dimming. That would be a much prettier solution.

Was there a reason for common dimming, I don't seem to recall that coming up in the thread, but I very well may have missed it...

Any of your proposed solutions would work fine. Or, since most board houses are going to want you to order a lot of the PCBs, you could just build some "partial" PCBs with only one or two drivers populated.

The common dimming was a "keep it simple" thing on my part. On my entire gigantic tank I'm only going to want 3 or 4 different "channels" of LED control, so I saw no reason to break out each individual IC's dimming pin separately.

Basically, I designed the PCB for myself, so it's "stripped" of some functions other people might want (individual dimming, onboard 5v reg, etc.)

DWZM, when coming from the Arduino, how are you wiring it up to all the boards? There's one wire coming from the Arduino, but multiple boards, and with voltage/current division, how would you make sure the right signals are being recieved by the drivers?

For the sake of argument, imagine you had 72 LEDs you wanted to control "together." With 6 on a driver circuit, that's 12 drivers. With three drivers on a board, that's 4 PCBs. I'll just put those 4 PCBs next to eachother in a project box and run a +5V, GND, and PWM signal from the Arduino to the box, then split the wires inside the box to run to each PCB.

The CAT4101 EN/PWM pin has a pulldown resistance of 200k. At 5v it should draw something like 30uA which means you can stack a crapton of them on a single Arduino pin and not worry about overloading the Arduino.

 

[der_wille_zur_macht]

and with voltage/current division, how would you make sure the right signals are being recieved by the drivers?

I think I answered this, but to be clear, the PWM pins of multiple driver PCBs would just be in parallel across the Arduino PWM pin/gnd - so they'd all see 5v and the current they drew would be additive. As I described above, the current the IC draws on the EN/PWM pin is extremely small, so there's really no worry.

 

[stugray]

DWZM,

I have been paying attention to these drivers and I am baffled by one thing.

Why doesnt anyone offer a PWM based controller with the following features:

1 - PWM FET Driver output ( designed to drive an external Power FET N or P type )
2 - Current monitor input ( Op amp that reads differential voltage across a small shunt res. )
3 - Current setpoint input ( sets max allowable current )
4 - enable/disable input
5 - Temperature monitor with max temp setpoint ( for measuring temps of the LED string )
6 - PWM AND 0-10 V dim inputs.


The advantage of this design is that we could drive huge strings using external power MOSFETs.
Also, when driving the FETs in full PWM mode ( as opposed to linear region ), the FETS do not get hot.
This design would be the most efficient CC driver with all the options we need.

I have seen SOME that do most of the above, but they typically must put the string current through the part & I want an external shunt res.

I did some work in my senior design project on D-class amplifiers, so I know even a few more tricks to increase efficiency if we REALLY want that extra 2-3%.

Thoughts?

Stu

 

[der_wille_zur_macht]

I have seen SOME that do most of the above

Me too, and I'm right with you as far as having a certain level of frustration for the parts we've talked about to date.

It seems like either you get a simple-to-use part like the NCP3066 driver that has the limitations you don't want (but can drive an external switch fwiw) or you get something outrageously massive where the chip manufacturer puts EVERYTHING externally and forces you to practically design every major component from scratch. While I agree it would be cool to come up with a single "superdriver" that met everyone's needs and got those last few percent of efficiency, I don't have the bandwidth to do that myself and I probably wouldn't use it - assuming it was based on any IC I know of - because it would probably be more complex/expensive than any of the solutions we've come up with so far.

Another limitation is that if you're trying to work from a higher voltage level to get longer strings, you're probably increasing overall cost per LED, since once you get away from the standard-issue 24v power supply, DC supplies seem to get more expensive (in a pure $/watt sense). So you're either spending more per watt, or you're building your own AC/DC power supply, which is also something I'm not quite motivated to do.

FWIW the NCP3066 has most of the features you're listing and is still relatively easy to use - it has a separate current limit input and current sense input, can drive an external switch, and has a PWM/EN pin. No temp probe (but you could rig one on PWM/EN or one of the sense resistors). It can do PWM dimming via the EN pin, or analog dimming (through a voltage divider) on the COMP pin. And so on.

 

[shackscs]

Quote:
Originally Posted by swimmer4uus
DWZM, when coming from the Arduino, how are you wiring it up to all the boards? There's one wire coming from the Arduino, but multiple boards, and with voltage/current division, how would you make sure the right signals are being recieved by the drivers?

For the sake of argument, imagine you had 72 LEDs you wanted to control "together." With 6 on a driver circuit, that's 12 drivers. With three drivers on a board, that's 4 PCBs. I'll just put those 4 PCBs next to eachother in a project box and run a +5V, GND, and PWM signal from the Arduino to the box, then split the wires inside the box to run to each PCB.

The CAT4101 EN/PWM pin has a pulldown resistance of 200k. At 5v it should draw something like 30uA which means you can stack a crapton of them on a single Arduino pin and not worry about overloading the Arduino.

I assume this will also apply to the dual board that I made?

 

[der_wille_zur_macht]

Yep, more or less. The datasheet lists the EN pin current for the NCP3066 at 15uA for a 5v signal so in theory you could parallel ~2600 of them on a single Arduino pin. :lol:

 

[perikaruppan]

Just a small soldering Question on the CAT.
I got a few sample boards done up.
Got a question on the soldering Bottom of the CAT that is to be soldered to the ground plane:
1. Does only the projected portion of the CAT (metallic) being soldered to the ground plane is enough? or
2. I need to have solder below the CAT as well as solder the Metallic projection portion of the CAT to the Ground plane?

Cuz option 2 i think we need to be an expert to do that and it requires a high soldering wattage to solder that volume in one shot!!!. Do help in this regard.

Got all the 84 LEDs wired up and it is just waiting to be fired up with the driver!!!.

 

[perikaruppan]

Also, DWZM from the image of your CAT in page 23, it appears like you have the CAT slightly elevated so that only the Rear-protruding metallic base of the CAT is soldered to the Ground plane? or am i wrong?

 

[der_wille_zur_macht]

As long as there is an electrical connection it really comes down to quality of thermal transfer. So, technically, as long as there is electrical conductivity, the circuit will work. But to get the best thermal performance, you'll want the whole metallic underside of the part soldered down.

Typically you'd have a grounded pad on the PCB large enough such that the entire exposed surface of the underside of the part was contacting the grounded pad, and then you'd typically create a solder joint across that entire interface.

It IS a little tough to do but not impossible. Here's what I do:

1) Put the part on my work surface upside-down and tin the entire exposed ground pad. Just a THIN coat of solder.
2) Tin the entire exposed pad on the PCB in the same manner.
3) Place the part on the pad and hold the soldering iron such that it has as much contact as possible with both the exposed edge of the tab on the part and the pad on the PCB
4) Wait several seconds, then apply solder. As soon as it gets hot enough to melt the solder, just keep feeding solder for a few seconds. This should allow the solder to fill the entire void between the part and the PCB even though you're only applying it from the edge - it'll wick along the surfaces you tinned in the steps above.

This is probably the "hardest" part of assembling these drivers but on a relative scale of difficulty I'd say it's no harder than soldering a wire to a pad on a LED's star - in both cases you need to pump a lot of heat into the part in as short a time as possible.

I know that some people will solder one of the five leads on the front of the part to hold it in place while they're soldering the ground tab, so maybe try that, too - otherwise it can feel like you need three hands - one for the solder, one for the iron, and one to hold the part in place.