DIY LED driver for reef lighting

[Krakonosh]

few notes on the LM3409 v5 boards

few notes on the LM3409 v5 boards

Hi all,

I just wanted to say thanks for all the help on this tread and especially to TeraHz for the board design.
I have now build 10 of his version v5 boards. All working well except few that will not turn off on analog dimming.
i should start with a description of my intended use.
I have Apex controller, so 1 - 10 V analog dimming is what I will use, for now I just hooked up few 10k pots. I intend to use one driver per string as much as possible.
The drivers so far have been set up for 1 A (whites), 800 ma (blues) and 600 mA for greens and reds. The power supply is Meanwell SP-300-48. This turns out too hot and noisy, so I have SDR-240-48 coming in. Will report on those later.
Strings are 12 long for now and PS is set to 39 V. (yes it does go that low).
There is hardly any heat from the driver boards, just slightly warm to the touch. The whites are at 36 V and blues at 38 V.
So far so good. I'm still building my array, so I have only few strings to test it on as I'm taking phased in approach to the change out.
I also had to build an interface board, because the v5 boards will pull about 30 mA from the adjustment pin and Apex will not provide that much (not even close). I used an op-amp in voltage follower mode, had hard time finding cheap enough one that will provide at least 60 mA, I settled on OP279 dual op-amp. I can drive two of the driver boards with one output .
I have not tried it with the Apex yet, but it works on the pots and the input load is minimal.
Few pictures:

Cheers.

 

[terahz]

SkiFletch, I would be really surprised if the LM3404 can be a soft of a drop-in replacement (sans the transistor). I would treat it as a completely different chip unless you verify between the two data sheets that they are indeed similar.

Krakonosh, thanks for following up. Glad things are (mostly) working for you . For the problematic drivers, check the voltage between R6 and R7 (the dimming signal voltage divider) when you're at the lowest setting and compare it with the one between R8 and R9 (the comparator threshold voltage divider). If the first one is higher, you can either increase the value of R8 or decrease R9, either way you need to increase that voltage to be higher than what you get between R6 and R7. I assume you've adjusted R6 and R7 accordingly for your 10V signal (the ones in the data sheet are for 5V signal)? Maybe that's why you're not getting under the minimum voltage. It might also mean that you're loosing half of your dimming range if you haven't adjusted these resistors.

Let us know.

 

[Krakonosh]

Hi TeraHz,
I use 0 - 5V into the boards for dimming, I should have said that, later when I hookup Apex I will add a 10k resistor in line with the pots. Sorry for the confusion.
The shutting down issue is little more complicated.
Only number one and seven have this issue, but they work OK if the number two is on.
It's some kind of feed back i have to troubleshoot later, right now I'm turning the power supply off, so it's no problem. I did try to replace the whole board with spare, same thing happens. When I get more parts in I will try to add some diodes to separate the 5V supplies.
Thanks.

 

[kcress]

Krakonosh; If you're tying those board inputs together you probably have unequal sharing. Resistor dividers don't share worth a dang.

It would be way better to have an op-amp controlling each board. That way the don't share at all. Your Apex could drive handfuls of OPAMP inputs easily and then each OPAMP output could run its own board.

What is that second pictured board? Is that your "interface" board?

 

[der_wille_zur_macht]

A quick edit: I'm gonna start RTMFDS' to see if I can answer the last question about components myself, but hey, you may be able to save me some time

+1 to what terahz said. It's similar in theory of operation but the design will be totally unique (much simpler actually). The only reason I'm not really planning on using this chip in custom work is the 1A limit.

 

[SkiFletch]

terahz, thanks man, getting elbow deep in the datasheets and they appear similar but different. I might steal your board design but just use different components. Of course I'll probably just proto one up first to make sure I'm not doing anything too stupid

 

[der_wille_zur_macht]

The AI board:

The fixture is comprised of 24 LEDs on 8 3-up MCPCBs. This board sits right in the fixture and basically "plugs" on to male header pins on each MCPCB. The 8 sets of black female headers have a through hole under them to allow the male headers to plug up through the bottom of the board (pretty neat).

The left end of the board has three barrel jacks. The middle jack supplies a DC voltage source for the LED drivers and onboard electronics. The left and right jacks are data ports for connection to the AI controller, which dims the LEDs for various effects.

Near the left end of the board there appears there appears to be two logic chips/processors to interpret the signals from the controller. The larger chip with more pins has an unknown logo and the following markings: MSH4 CTJ 5HUP. The smaller chip has these markings: EZ1Q ST2028 MRC027. It's obviously from ST.

In the very middle of the board there's a pair of transistors and a few passives that apparently drive the cooling fan in the unit, which plugs in to the white JST jack in the middle.

There are 4 driver circuits on the board. Each has an LM3404 as previously mentioned. Each LM pretty much has the "typical" application circuit as shown on the product page:

http://www.national.com/pf/LM/LM3404.html#Overview

It looks pretty straightforward. Inductor appears to be 8.2uH. Diode has markings B16 13 and an unknown logo. Three ceramic caps, unmarked so we'll probably not be able to determine values. Some of the resistors are too small to be read and I haven't yet figured out which is which yet, but as soon as I figure out the ones I can read and what they are in the design I'll post back. They're definitely not all the same from driver to driver; I'm expecting that the different drivers are running at different nominal currents or something like that.

In the end there's nothing groundbreaking here but it is kinda interesting to look at anyways.

 

[der_wille_zur_macht]

The driver all jammed up with the jacks at the left end of the board is way too crowded to read values from. Of the other three drivers, two have Rsns marked R25 and one has R27. Assuming these are .25 and .27 ohm respectively. Can't read the other two resistors. If someone cares enough to know what current the LEDs are running at you could probably go plug these Rsns values and 8.2uH for the inductor into the calcs in the datasheet and get pretty close.

 

[terahz]

You sure that ST doesn't say 202B? cause it could be their RS-232 chip:
http://www.datasheetcatalog.org/datasheet/stmicroelectronics/7313.pdf

EDIT: and if that's the case I assume the other chip is some form of micro? does this fixture have any brains (as in does it support dimming ?)

 

[der_wille_zur_macht]

It might say 202B. There's no crystal or resonator near either of them, so if one of them is a processor, it has onboard timing.

The fixture is dimmable - there is a separate control unit that AI supplies, and it communicates to this LED driver board via a barrel jack - I have no idea what protocol or format the communication is and I have no idea what this board does with the signal, other than that the various colors of LEDs are individually dimmable via the head unit, so whatever signal it's sending is somehow interpreted by those two ICs, and then provided to the LM3404s, presumably via PWM.

 

[terahz]

the unknown looks like an ADC?:
http://goo.gl/p9Gz4

My chinese is rusty Maybe someone can read the original page and help us out.

 

[der_wille_zur_macht]

Of note - there are four bare solder pads (test points?) under the other mystery chip. Could these be some sort of ISP port to flash the mystery IC?

The ST chip is definitely ST202B. The caps around it are all on the "right" pins and there is a trace from one of the barrel jacks that goes straight to the R1in pin. It is hard to tell where the other Tx/Rx pins connect because some of the traces are going to vias underneath the IC - don't know which goes where.

Either there is some sort of voodoo magic going on with this board or I'm half blind, because it LOOKS like the DIM pin from each LM3404 goes through a tiny resistor to GND. I don't see how it connects back to the mystery ICs we're assuming control dimming.

 

[der_wille_zur_macht]

You're a wizard. That is definitely the other chip. It even looks like there's a little RC filter near one of it's output pins.

Now, just to discover what the heck is going on here that allows those two ICs to dim the LM3404's.

 

[jgwinner]

Go Willie!

== John ==​

 

[kcress]

Hey DWMZ can you get a closeup of the back of one of those 8 terminals you mentioned? I'm curious as to what you described but can't make it out.

 

[Krakonosh]

Krakonosh; If you're tying those board inputs together you probably have unequal sharing. Resistor dividers don't share worth a dang.

It would be way better to have an op-amp controlling each board. That way the don't share at all. Your Apex could drive handfuls of OPAMP inputs easily and then each OPAMP output could run its own board.

What is that second pictured board? Is that your "interface" board?

Hi Kcress,
That's exactly what the interface board is, it has dual op-amps in voltage follower mode that are fed via the pots and each output feeds a driver board.
Apex interface will come later.
Cheers.

 

[kcress]

Hi Kcress,
That's exactly what the interface board is, it has dual op-amps in voltage follower mode that are fed via the pots and each output feeds a driver board.
Apex interface will come later.
Cheers.

Oh! And that setup is what's leaving a few hung dim?

 

[TeslaReef]

Hello LED Driver group,

I'd like to take the dive into a DIY LED Setup and was wondering how or if this quick description is possible.

I would like to emulate an Ecotech Radion with the few Reds and Greens mixed in.

After much reading through this thread over the past few months, I have noticed the issue of voltage drop (regards to efficiency) is a major factor. I like the concern placed on efficiency, I think these systems sound much more efficient than any commercial option.

Another concern I've read are multiple powers upplies due to Power Factor and Harmonics issues.

What I would like to do is tackle both of these concerns before jumping into this huge but awesome project.

Can I get a huge PS like a 500W 48vDC (or one 24V and one 48V) and run some voltage regulator ICs on it to get the desired voltages per string?

The setup I have in mind would be 6 modules total with each module consisting of all XP(E/G) with about 10xRB, 8xBU 8xCW, 4xGreen and 4xRed.

I realize the different strings will require different supply voltages due to the varied Vf of each LED string.

XPG_ CW @ 700ma, Vf = 3.05 *8 = 24.5V (17.2W)
XPE_ RB @ 700ma, Vf = 3.4 * 10 = 30.4V (21.3W)
XPE_ BU @ 700ma, Vf = 3.4 * 8 = 27.2V (19.1W)
XPE_ Red @ 700ma, Vf = 2.3 * 10 = 9.2V (6.5W)
XPE_ Green @ 700ma, Vf = 3.7 * 10 = 14.8V (10.4W) ((74.5Watts total))


Of course, these numbers are not set in stone in any way. I just want to try and find out if there is a good way to get the desired voltages and reduce the # of PS. I will be playing around wit hteh mix but I think many are doing somewhere around 2:1 mix of Blue:White.

If step down regualtors are a good option, what family of ICs would be the way to go with? Are there any other options?

Thank you very much and AWESOME THREAD!
Phillip

 

[der_wille_zur_macht]

Hey DWMZ can you get a closeup of the back of one of those 8 terminals you mentioned? I'm curious as to what you described but can't make it out.

I will next time I'm at home with a camera, but if you look in the photos you can see how it works. The terminals LOOK like standard through hole female headers. BUT they don't have a "pin" underneath them to go down through the hole. Instead, the pin sticks out the SIDE of the strip, at the base, and alternates from one side to another. There is a typical plated through hole directly under the header. This allows a male pin to plug up through the hole, from the bottom of the board, into the female header strip.

From the bottom, it looks like a row of plated through holes - you can see them in the photo of the bottom. In the photos of the top, you can see the "pins" bent out to each side of the header, where they're soldered to pads on the board.

Hello LED Driver group,

I'd like to take the dive into a DIY LED Setup and was wondering how or if this quick description is possible.

I would like to emulate an Ecotech Radion with the few Reds and Greens mixed in.

After much reading through this thread over the past few months, I have noticed the issue of voltage drop (regards to efficiency) is a major factor. I like the concern placed on efficiency, I think these systems sound much more efficient than any commercial option.

Essentially, voltage drop is only a (serious) issue with linear regulators, like the CAT4101. If you use a well-designed switching regulator, like the LM3409, there is nowhere near as large an efficiency drop when operating with a big delta (Vout-Vin).

So if you want to drive a bunch of "different" LED strings off a single power supply, the clear answer is to use a switching regulator instead of a linear regulator. Stepping down a "big" voltage to different intermediate voltages for each driver would be more complicated and (likely) less efficient than just using well designed switching drivers straight off that "big" voltage.

Another thought - if your desired string voltages are WAY different from each other, consider adjusting the number of LEDs in each string. You can do this by literally changing the number of LEDs you use for a given type/color (and compensating by choosing a different drive current to keep the same overall intensity) and/or by running multiple shorter strings (in parallel or on different drivers) for LEDs you want more of.

 

[TeslaReef]

Another thought - if your desired string voltages are WAY different from each other, consider adjusting the number of LEDs in each string. You can do this by literally changing the number of LEDs you use for a given type/color (and compensating by choosing a different drive current to keep the same overall intensity) and/or by running multiple shorter strings (in parallel or on different drivers) for LEDs you want more of.

DWZM,

Thank you! I like the idea of minimal work. I will definitely look at LM3409. Off to the drawing board and much more reading on this thread to catch up. I was mostly stuck on the CAT board. I do like the LM3409's flexibility and options. Great idea on the compensations with LEDS/Color, parralleling, shorter strings etc..

Phillip