DIY LED driver for reef lighting

Ok I started over building the driver. I think i got it working right. My only concern is the output voltage is around 47v with no load on it. I dont want to hook it up to the LEDS without some input if this is correct
 
That doesn't mean a lot... Since you have nothing hooked up current drivers go to full output.

How many LEDs of what Vf are you expecting to drive with your controller? Maybe I can think of a tough/cheap load you can test with.
 
Take a standard 100W incandescent bulb. Tape leads to it and hook it across the output. It should just glow.
 
Since this is a boost driver, it will go to full output with no load - actually, it'll blow the chip apart with no load. (Ask me how I know!) :lol: That's why there's the zener diode in the circuit that you don't see in a buck circuit. If the voltage gets over the breakdown voltage for the zener diode, it starts conducting (in the reverse direction) and essentially acts as a load to keep the chip from running away.

Interesting that you're seeing that much voltage though, since it's a 36v zener, so it should start conducting at 36v. Then again I don't know enough about how they work to know how the voltage would behave above that point.

If you do try to connect an LED load, keep in mind that the forward voltage of the LEDs you connect MUST be above that of the power supply, or else you'll insta-fry the LEDs (ask me how I know that, too.) :lol:
 
Slight change of direction. I've got several prototypes of this NCP3066 design up and running, and will use them on my big tank or the ATS when it's up, but I'm not convinced I want to use this design for ALL of the drivers I'll need, so I'm on the fence about ordering dozens from a board house.

For the people who have contacted me to get on the list for the "extras" when I order - if you're still interested, please contact me. I'll send out individual PMs in case people miss this thread. If people still want the NCP3066 design, I'll go ahead and order a bunch - I don't want to leave people hanging who were counting on getting some of these. The NCP3066 design does have the advantage of being through-hole for people that want that. Plus it's far more tested - I've actually been running it on tanks for a few months.

Otherwise, here's a different approach to driving LEDs, based on the ZXLD1366 chip someone mentioned a page or two back. This design is all SMT, and the chip can handle a lot more voltage, which means more LEDs per string. I'm aiming for 12 LEDs per string at 48v. (48v DC power supplies are available from digikey and other places for only a little more $$/watt than the 24v supplies everyone uses.) I put three of these circuits on a single 5cmx5cm PCB, so you'll be able to drive 36 LEDs per PCB. This design has some pricey components but cost per LED driven should be about the same. Since it's driving more LEDs per PCB, wiring on a large tank will be simpler. And, it's SMT, for people who like that.

This ZXLD1366 chip is a step down (buck) switching converter. The ADJ pin allows PWM or analog DC dimming, though it's pretty funky to use, since it needs a pretty narrow range (.3 - 2.5v or something) as analog, and wants the PWM signal to be in that range, too. If you leave it disconnected it floats to the internal reference voltage (1.25v) which results in 100% output - bringing it above 1.25v results in up to 200% output. To allow for simple dimming, as outlined in the datasheet, I've designed this circuit using a small NPN transistor to switch it to GND, so it'll work with the (5v) PWM signal from an Arduino or other microcontroller.

Otherwise it's really straightforward. There's only 4 external components besides the two required for my chosen dimming configuration.

Also, I put 3.5mm screw headers on this instead of the .1" pin headers I used on the other design, because there was plenty of room.

There's a separate PWM dimming pin for each circuit, though it would be easy to just use one PWM signal to control all three drivers, which is what I'll probably do for my tank. With 36 LEDs per PCB, I'll only need 6 PCBs for my whole tank! This simplifies wiring compared to using THIRTY SIX buckpucks, or 18 meanwells!

This is still in draft stage. I'm planning on getting a prototype board made soon, and I'm ordering components later today, so I'm hoping to have a prototype up and running soon.

Here's the sloppy schematic:

ZXLD1362_3_SCH.gif


And the board:

ZXLD1362_3_BRD.gif


I tried to follow the layout guidelines in the datasheet regarding which components to put close to everything else, but it's just so dirt simple it wasn't very hard. I used a ground plane on both sides with lots of vias near the ICs to help with cooling. Chose the trace widths to allow for 1A per driver with a little headroom. I will probably run these at 500 - 600mA, since I may end up with some LEDs that have a 700mA max (Cree MC-Es).

As always, comments or suggestions to improve are welcome. I'm working out the BOM now and will post soon.
 
im still down with taking a couple of what ever the final product is, i dont know that any of this will get used in my project if it ever gets going, but i like trinkets..
 
So, in the configuration I want to run this new driver - 48v in, 12 LEDs - the datasheet calls for a 4.7uF input cap, though that's borderline - since above 50v they want you to have a 10uF cap.

I can find a 4.7uF cap with a 50v rating for a fair price (like 50 cents.) But trying to find a 10uF cap is crazy - the few out there are VERY VERY expensive.

That leaves me with a few options:

1) Use the 4.7uF 50v cap. Makes me nervous because it's barely adequate capacitance, and I'd be within a few percent of the 50v max rating.

2) Use two of the 4.7uF caps in parallel. More capacitance, but still have the voltage problem.

3) The datasheet mentions using a 10uF electrolytic with a .1uF ceramic if you can't find a suitable 10uF cap. That would be relatively cheap and straightforward.

I'm leaning towards #3 but wanted the input of some of the EE types that have participated in the thread thus far.
 
How set are you on the 48 volts. I started a design for 24 volts that allowed a few more options for dimming. It could be selected (with a little blue jumper - maybe 2) between fixed (comes on and stays on at a manual adjustabe value), 0 - 10 volts dimming, 0 - 5 volt dimming, I2C dimming, and was adding PWM (this might mean another jumper). Sorry no schematic, but I was going to try and learn Eagle tonight.

I added an LM7805 ($.50) to get 5 volts, and an MCP4725 ($.75) for I2C interface. Some resistors and a pot. I think these are the right prices.

I tried to design for people (like me) that don't have the dimming yet and want to add it later, but don't know what option they will take. Or already have something that does not have PWM.

Anyone interested?
 
Good thoughts.

I'm designing from the perspective of:

1) I have a BIG tank, which means LOTS of LEDs, which means anything that can reduce parts count and wiring hassle is important (hence the 48v input, allowing 12 LEDs per driver).
2) I will have an Arduino, so PWM dimming is the only functionality I really care about

If people want a 24v driver, we should probably look at a different chip in the ZXLD13xx family, since it seems this one is targeted at high voltage.
 
Possible parts

Possible parts

dwzm did you have a speific part in mind. I would love for some one to tell me where I am wrong (if I am). I think for 24 volt in and a current of 1 amp you have to use either 1360, 1362, or 1366. I don't see a lot of difference in the 3:
1) Voltage - all take 24 volts though
2) Accuracy - 5% is probably good why pay for 3%
3) package - lay out for what is needed, but TSOT23-5 is common to all 3
4) frequency - I think this is the ripple frequency on the output can anyone clarify
Did I miss something?
 
See this comment in the datasheet:

Operation at low supply voltage
Below the under-voltage lockout threshold (VSD) the drive to the output transistor is turned off to prevent device operation with excessive on-resistance of the output transistor. The output transistor is not full enhanced until the supply voltage exceeds approximately 17V. At supply voltages between VSD and 17V care must be taken to avoid excessive power dissipation due to the on-resistance.

If the supply voltage is always less than 30V continuous (or less than 40V for less than 0.5s) an alternative device is available, the ZXLD1360. Note that when driving loads of two or more LEDs, the forward drop will normally be sufficient to prevent the device from switching below approximately 6V. This will minimize the risk of damage to the device.

You're probably right that the 1366 would be fine at 24v, but that note seems to point you towards the 1360 if you're below 30v. If you look at the graphs on page 18 and 19 of the 1366 datasheet it looks like it would be a little less efficient at lower supply voltages. Though the 1360 is actually more expensive.

It would be a worthwhile exercise to actually do the math and see what the cost per driven LED would be at 24v and 48v.
 
Well I am confused the VSD for 1362 and 1366 is 4.75 and for the 1360 5.5, seems backwards doesn't it. So I agree 24 should work with either.

It appears the the best way to run these is as many LEDs as you can pack on for the input voltage. With 7 LEDs the 1360 and 1366 appear the same in limited overlap range. For few LEDs the 1360 appears to be better. So i would agree either will work at 24 volts.

I repeat do you know what the frequency is? or did I have that right?
 
I think it has to do with the resistance of the internal switch at lower voltages - it's higher for the 1366 since it's targeted at higher voltages.

But in general you're right, IME the most efficient way to run a switching regulator is to keep the input voltage and the output voltage close together, and run it at the high end of it's voltage range.

You're right about the frequency - it's the base operating frequency of the switch in the IC, which translates into the ripple frequency on the output. Not really that important for our considerations.
 
Yeah 12 LEDs per string is a bit too much for me. It only allows for steps of 24 LEDs (with 1:1 blue to white ratio and dedicated dimming per color). This, however, will work quite nicely for big tanks.
 
So, in the configuration I want to run this new driver - 48v in, 12 LEDs - the datasheet calls for a 4.7uF input cap, though that's borderline - since above 50v they want you to have a 10uF cap.

I can find a 4.7uF cap with a 50v rating for a fair price (like 50 cents.) But trying to find a 10uF cap is crazy - the few out there are VERY VERY expensive.

That leaves me with a few options:

1) Use the 4.7uF 50v cap. Makes me nervous because it's barely adequate capacitance, and I'd be within a few percent of the 50v max rating.

2) Use two of the 4.7uF caps in parallel. More capacitance, but still have the voltage problem.

3) The datasheet mentions using a 10uF electrolytic with a .1uF ceramic if you can't find a suitable 10uF cap. That would be relatively cheap and straightforward.

I'm leaning towards #3 but wanted the input of some of the EE types that have participated in the thread thus far.


http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=PCC2308CT-ND
 
ZXLD1366 24 Volt Driver

ZXLD1366 24 Volt Driver

I think this will work. Analog is not my specialty. It uses simple resistor divisor networks to drop different voltages to the 1.25 needed by the ZXLD1366. If you are interested I will explain this, but I am looking for comments if this will work - so I am assuming you understand this.


I set it up to allow dimming from a different options:
- 0-5 for those with 5 volt dimming controls (currently connected to the IIC D2A)
- 0-10 for those with 10 volt dimming controls
- For simplicity the light level can be set using the 5 or 24 volt as a reference. For those that want to use a regular timer.
Which mode is selectable by a jumper.

A final set of jumpers determines if you are in the above mode or in PWM mode.

The IIC D2A only supported 2 channels. I think we needed more so I added a port expander so you could control 8 levels with an IIC bus.

I was going to check with an EE, but he did not make in because of the snow. But will post see if people like and check with him later.

So -what do you think?
dwzm - thanks for the library
 

Attachments

  • LED Driver.jpg
    LED Driver.jpg
    72.3 KB · Views: 10
Back
Top