DIY LED driver for reef lighting

[kcress]

Ok kcress this is probably your question but anyone with insight help would be appreciated. I have six string of LEDs using the same driver (rewiring to measure on the same CAT4101). I have 5 strings drawing 890 and the sixth upper 400s. I also hooked up 2 sting to a power supply and the low current one draws very close to 50% of what I think of as a good string. I figured I had a bad solder joint, but measured the voltage from wire-LED-wire (to make sure I included solder joints) and they all had the same drop 2.46 plus/minus .01. Any ideas what is rong or what else I can measure.

:eek1: You must sit up late trying to baffle me...

I have no idea what could be rong. However what's wrong should show itself with a measurement from neighboring wire to the other neighboring wire, while running them with an adjustable supply. Hook up an ammeter, (or the ol' one ohm resistor), and the LED string to a variable voltage supply. Adjust the voltage up until the current is the same. Pick one, say, 500mA. Measure as described. Write them all down. That scheme takes in both wires and solder joints of the in-between LED.

Then put in a happy string and set the voltage so you end up with exactly the same 500mA. Measure them.

One or many have to have lower or higher Vfs. See what you get. I'm interested.

Another one. I have ten CAT4101 drivers nine supply 850 ma the tenth is in the 500 somewhere all with the same string of LEDs. All RSET measure 602-609. Any thoughts one this one.

Here you probably have a solder issue or you have a misplaced part which is common for SMT parts.

I always grab for high magnification at this point. Compare resistor and capacitor actual detailed looks between 'good' boards and the questionable one. That helps find a scrambled part.

Did you properly wash these boards?

 

[imagex]

I got another question..is there any way to drive these drivers to 1a??

 

[der_wille_zur_macht]

Pretty moot point though because there aren't many bargain supplies around that are more than 24V anyway and once you go much over 36V you enter much more hazardous territory.

Agreed in general but we are really "on the verge" with this design's 25v limit. It would be nice to squeeze 7 LEDs on one of these drivers, and the typical common power supplies are adjustable, so it's easy to get over 25v...

A few times I've toyed with the idea of doing a higher-voltage, higher-current driver for large, parallel-string arrays (i.e. make a Meanwell HLG from scratch) but it's just plain hard to justify given the cheapness of doing lots of these 24v circuits, when you consider the cost of DC power supplies (24v seems to be the sweet spot from a cost/watt perspective).

Youbetcha - just with a little noise included. (noisy smoke)

If anyone remembers waaaay back when I started the thread, I let the smoke out of the first official prototype of the dual NCP3066 board because I had 24v shorted to some signal line or something. It was more of a BAM than a pfffftzzzpp and sadly it fried the AVR in the Arduino I was trying to use to dim the driver... Though since that's been my only fatality in this adventure I'd say I'm doing pretty well.

 

[der_wille_zur_macht]

I got another question..is there any way to drive these drivers to 1a??

Drive current is managed by the Rset value you choose when you build the board. The datasheet explains how to pick Rset for a given nominal current and includes a chart of suggested values. For 1A current you would want a 549 ohm resistor.

Choose a resistor in the same package size as the one in the original BOM and you'll be all set. Something like Mouser part 660-RK73H2BLTD5490F would be perfect.

 

[TheFishMan65]

Thanks kcress those are the things I intend to try, but just wanted to make sure there was not a better path through the woods.

I have not cleaned them yet, I thought I would make sure they worked first. I will then clean them with isopropyl alcohol and a tooth brush.

 

[kcress]

Find my cleaning dissertation and use it instead.

 

[TheFishMan65]

Why is it better? The board designer I work with did not think water and soap works as well. He that it does not dissolve (IIRC) the flux but isopropyl does.

 

[imagex]

Drive current is managed by the Rset value you choose when you build the board. The datasheet explains how to pick Rset for a given nominal current and includes a chart of suggested values. For 1A current you would want a 549 ohm resistor.

Choose a resistor in the same package size as the one in the original BOM and you'll be all set. Something like Mouser part 660-RK73H2BLTD5490F would be perfect.

ahhh cool damn i shouldve ordered the 549 ones then... i'm going to run bridgelux 402's off these drivers... i should be able to run 2 off each cat4101 correct?

 

[DustinB]

Do you guys really think it's necessary to clean the flux from your boards when using standard rosin core solders? I've never cleaned one before and had no issues from it. The only time I've heard of a need for cleaning boards is when using an acidic water soluble flux.

 

[TheFishMan65]

From my understanding yes it must be cleaned. As you mentioned some solder used acid to eat the oil and dirt off the solder joint. Rosin is still cleaning oils and if left on does continue to eat the board apart.

Now the best way to clean may depend on the solder. What I had forgotten to mention is that I am using ROHS complaint solder (no lead). So it may require a different cleaning method.

 

[der_wille_zur_macht]

I only clean boards that I intend to last longer than "let's see if this works." Lately, most of my boards are in that category so I haven't done much cleaning.

imagex - I have no experience with or knowledge of that particular LED so I can't really say. Suffice that as long as your LEDs' total Vf is within about half a volt of your supply, and your target current (determined by Rset) is an amp or below, this design should run any LED out there. The good news is these resistors are only six cents each so ordering the wrong ones isn't a big deal. :lol: That said you'll get hit with postage again. You might want to look for a similar part at digikey as I think they'll do plain old first class mail which will be cheaper than priority mail, which is the cheapest that mouser does.

 

[OTCHU]

I am working on electronic since 25 years and never clean my boards .... never had a problem ... The only problem you can have with the flux it's when you are working with very high frequency ... that's it

 

[TheFishMan65]

At 00 ma I could not see the contact points. So I put 15 volts across a normal white string and the low current blue string. I then measured the voltage drop across all 12 LEDs unused contact to unused contact.. They all registered between 2.47 and 2.52 the supply measured 15.08.

In order to get 1 amp through the string it took 20.4 volts for the 'normal' string and 22.4 for the 'low current string'.

Any thoughts?

I took the low driver into the office and looked at it under a microscope and touched up a few solder joints that looked off. As expected that fixed the problem.

 

[OTCHU]

cold solder ..... lol good job

 

[kcress]

Washing boards.
I've seen many problems with unwashed boards. In some cases washing them was all it took. Only the cheapest Chinese junk comes with unwashed boards. Having one board control to a different current value would be a typical problem that a dirty board could cause.

These drivers have high frequency switching nodes. You get flux left across them and today it might work fine. Tomorrow, when it's humid out it might not work.

There are "NO CLEAN" solders that are specially formulated to not corrode things over long periods of time. They usually stink and leave a board looking horrific. The board had better be new and have no oxidation or parts more than a year old on them, if you want to use that rotgut. I use it only when I absolutely must. An example would be when you have to solder an LCD display onto a new, previously cleaned, control board. You can't generally wash LCD displays so that's why it's necessary.

As for alcohol.. You should only be using solder with "WATER CLEANABLE" flux core. If you're using non-water cleanable flux on a control board - you're being foolish. '44' is an example. Use it to repair an old board or use it to solder wires to something, but you should no longer be using it on control boards. It cannot be cleaned properly any longer. The chemicals needed are all banned and have been for about a decade. There is no point in using alcohol because hot water with some surfactant,(detergent), is superior.

 

[kcress]

Any thoughts?

Doing it that way. Nope.

 

[OTCHU]

I had problem with flux when the frequency was 2 GHz .... the cat4101 is working at fews Khz ...

 

[TheFishMan65]

Good white string at .32 amps 3.07, 3.10, 3.06, 3.12, 3.12, 3.14
Good Blue string at .32 amps 3.21, 3.18, 3.18, 3.27, 3.18, 3.19
Low Blue string at .32 amps 3.25, 3.16, 3.20, 3.48, 3.35, 3.43
Are those the number you are looking for. I looked at the last three LEDs and the joint look good, but I will try redoing them tomorow night.

 

[kcress]

Much better info - that.
The constant current is the key.

Wow. A 0.66V difference. Over half a volt. 9.87V at only 320mA All these voltages will go up with higher drive current and almost surely your driver will run out of 'compliance voltage'. That means the current driver has no more voltage to use in the effort to reach the set current. Hook it back up to your current driver and short one LED. I would expect the current to jump up to 850mA or whatever your other strings are running at.

This is exactly the case you get paralleling strings. You would change out the highest one 3.48V with the other string's 3.18V. That might get this string down enough for the available compliance voltage to drive it at 850mA.

 

[imagex]

I only clean boards that I intend to last longer than "let's see if this works." Lately, most of my boards are in that category so I haven't done much cleaning.

imagex - I have no experience with or knowledge of that particular LED so I can't really say. Suffice that as long as your LEDs' total Vf is within about half a volt of your supply, and your target current (determined by Rset) is an amp or below, this design should run any LED out there. The good news is these resistors are only six cents each so ordering the wrong ones isn't a big deal. :lol: That said you'll get hit with postage again. You might want to look for a similar part at digikey as I think they'll do plain old first class mail which will be cheaper than priority mail, which is the cheapest that mouser does.

ok cool... think frys or radioshack would carry smd ?