DIY LED driver for reef lighting

[ingoh]

CAT4101 is chirping

CAT4101 is chirping

Hi,

my 4101s (PCB layout taken from here) are chirping if PWM is active. With 100% PWM chirping is gone. PWM is generated by a hydra. Sounds like it is ~1kHz - the frequency the PWM is generated with (port 5/6).

Any ideas?

 

[kcress]

Is it the CAT or one of the capacitors? I don't see how the CAT would be doing it but the caps sure could. Perhaps changing caps or changing the cap mounting would fix it. You are using the 3 channel CAT board right?

 

[ingoh]

thanks kcress. Until 5 minutes ago I thought it is the CAT. Then I put a plastic pen on my ear and the other end on the CAT: silence. Then I tried the output cap (C4/5/6): bingo - that seems to be the source. Caps are the Mouser parts from the BOM.

I'll have a further look/ear at it tomorrow.

 

[daplatapus]

Ok, I hear lots are going with X-pg's. Is there any reason why not go with the XML's. I read you use 10-20% less of them because they are so much brighter. FYI, I found this local guy up here in Vancouver that does bulk buys. It's slow to come, but cheap for Cree LED's. Check out the more info tab for lense choosing and such.
www.ledroupbuy.com

 

[Tesla]

terahz,

Although all the success with the CATs is promising, I share your concern with PWM simming. Might not be a big deal though but I am still only researching and haven't pulled the trigger yet.

I do however have some questions about your driver board.

How big are the boards?
How many drivers are you looking at per board?
Will your boards have individual power inputs and/or individual dimmer control inputs?
Will you be posting your PCB layout files for placing orders from seeedsrudio?

Sorry about all the questions but this evolution of the project sounds interesting and definitely worth doing. I'll be sure to follow the progress. :thumbsup::thumbsup:

 

[kcress]

thanks kcress. Until 5 minutes ago I thought it is the CAT. Then I put a plastic pen on my ear and the other end on the CAT: silence. Then I tried the output cap (C4/5/6): bingo - that seems to be the source. Caps are the Mouser parts from the BOM.

I'll have a further look/ear at it tomorrow.

Often the cap and the board work together to make a speaker like those gift cards. You could try tomb-stoning the cap and then running a tiny wire from the top down to the board. That would completely decouple the cap and the board. Often the cap by itself is toooo small to make much noise.

 

[terahz]

How big are the boards?

The board will be quite small. I have only thrown the components on a pcb for now, but it looks like everything will be able to fit in a roughly 30mm x 30mm area. I'm not sure yet if that will be the final size, but it will be within the small seed/tead studio size for sure.

How many drivers are you looking at per board?

One. Given the price per board and the fact that this IC can easily drive 12 LEDs with 48V psu (or 2 x 24V), I don't see a reason to go with more.

Will your boards have individual power inputs and/or individual dimmer control inputs?

Current plan is to have Vin, GND, DIM (either PWM or Analog, jumper selectable), LED+ and LED-. So yes and yes.

Even though with this board you don't need to control the input voltage like you do with the CATs, I might add a resistor, as kcress has been suggesting, for current measurement.

Will you be posting your PCB layout files for placing orders from seeedsrudio?

Of course. Just need to sit down and finish it .

 

[Tesla]

Terahz,

Sounds really cool, looking forward to it.

I was thinking of driving around 4 leds per driver. Can this board go that low if I turn up a 12v psu or down a 15v psu?

 

[terahz]

First draft of the PCB (34mm x 28mm):

Needs renaming of components to make them in order top left to bottom right. Needs adjustment of the labels of the various components. And I also need to make a prototype to see if it actually works

Note to PCB gurus: I have no idea of PCB design rules so if something looks really bad, say so, but be gentle

Anyway, here is a 3d of it:

I will be ordering components this week to build a prototype and make sure it works. After that I'll fix the names and labels, and will order a batch from seeedstudio.

Any feedback is welcome.


p.s. happy 4th of July!

 

[kcress]

Um... Am I seeing an exposed pad MSOP package for the controller chip?

Have you soldered 12mil pads with 8mil space? Do you realize you can't even buy solder that small? The solder is as wide as two pads. It's pretty sketchy.

If you're going to reflow this are you aware of the cost of solder paste? Or how hard it is to screen the stuff on?

That chip needs to have that center pad connected to the back of the board. Are those plated-thru holes big enough for your solder to fit down so you can feed it through from the back? If not make one big one instead of 4 little ones.

 

[kcress]

Only other comment is that angles less than 90 degrees where traces connect is considered 'bad form'. You might want to fix those.

 

[terahz]

Um... Am I seeing an exposed pad MSOP package for the controller chip?

Unfortunately, it is the only package available

Have you soldered 12mil pads with 8mil space? Do you realize you can't even buy solder that small? The solder is as wide as two pads. It's pretty sketchy.

Yes, I've worked with SSOP28 and the wicking method seems to work well for small packages like that. Also using plenty of flux and just introducing a little bit of solder with the tip of the iron works. That's my experience anyway.

If you're going to reflow this are you aware of the cost of solder paste? Or how hard it is to screen the stuff on?

Not reflowing. No clue how to do it anyway

That chip needs to have that center pad connected to the back of the board. Are those plated-thru holes big enough for your solder to fit down so you can feed it through from the back? If not make one big one instead of 4 little ones.

Do they really need to be connected? I realize that will help with cooling, but there is a dedicated GND pin. I any case I'll make sure they are plated through. The data sheet recommends 4 holes under it, but maybe that's for reflowing. I'll take a second look at that area.

Only other comment is that angles less than 90 degrees where traces connect is considered 'bad form'. You might want to fix those.

Will do. Thanks a bunch!

 

[kcress]

Yes, I've worked with SSOP28 and the wicking method seems to work well for small packages like that. Also using plenty of flux and just introducing a little bit of solder with the tip of the iron works. That's my experience anyway.

OK. Just so you know, often the easiest way is to just run a a huge blob of solder along each side. Solder all the pins together like they were one huge pin. Then come back in with solder wick. The wick will draw almost all of it back out leaving just the solder under the individual pins.

Do they really need to be connected? I realize that will help with cooling, but there is a dedicated GND pin. I any case I'll make sure they are plated through. The data sheet recommends 4 holes under it, but maybe that's for reflowing. I'll take a second look at that area.

Yes they need to be connected. That's what the plated thru holes do under the pad to the solder side plane. You don't need to do anything different if your solder can fit down those four holes as is. If you can't poke your solder down them then you need a bigger hole. Those were sized for re-flow. The designer wasn't thinking some foo would be hand soldering it.

Let us know how it goes. That's a good chip 'cept for soldering.

 

[terahz]

OK. Just so you know, often the easiest way is to just run a a huge blob of solder along each side. Solder all the pins together like they were one huge pin. Then come back in with solder wick. The wick will draw almost all of it back out leaving just the solder under the individual pins.

Yeah, that's what I meant by the 'the wicking method'. Was lazy describing it .

Yes they need to be connected. That's what the plated thru holes do under the pad to the solder side plane. You don't need to do anything different if your solder can fit down those four holes as is. If you can't poke your solder down them then you need a bigger hole. Those were sized for re-flow. The designer wasn't thinking some foo would be hand soldering it.

Let us know how it goes. That's a good chip 'cept for soldering.

Wouldn't flux from under the chip and in the holes 'suck' the solder in just like it happens with through hole components? I admit, I've never soldered a pad like that, so have no clue, just guessing.

I suppose it is just easier to make one big hole there and be done with it .

 

[kcress]

I would want to be able to feed the solder into a hole and have it fill up a little slow so I know it's filling in the pad below or see it coming back up a neighbor hole also indicating the same thing. If your solder will fit in those four holes 'be done with it'.

 

[ingoh]

Often the cap and the board work together to make a speaker like those gift cards. You could try tomb-stoning the cap and then running a tiny wire from the top down to the board. That would completely decouple the cap and the board. Often the cap by itself is toooo small to make much noise.

I really don't like to disturb your LM3409 discussion - looks really great - but I think I've got it silent now ;-)

Moved PWM from D5/D6 to D9/D10 and decreased PWM frequency down to 122Hz. I had to move it to ports not contolled by timer0 because a modification at timer0 also affects the delay functions.

greetings,
Ingo

 

[terahz]

Here is a quick update on the LM3409 driver board. It turns out that after I started checking component availability with the proper specs, quite a few of them are in different packages. This is mostly good news, because it will make soldering a bit easier.

In summary:
All small components are no 0603.
Diode, Inductor, sense resistor and one cap are now much bigger.
I've made a few copper pours instead of traced (again got the idea from the demo board)
Components are renamed now in order and some lables were added.

I'm thinking of added one more piece of circuitry - POT for setting the max current. Not 100% set on that, because we can now do analog dimming anyway, but might be useful for people without controllers. Stay tuned.

 

[TheFishMan65]

I would add the pot if you have room. You did your order backwards from me. I am still working on the controller. If I had used fixed resistors with the CAT4101 I would have replaced them 3 times by now. Also IMHO it makes debugging easier. And finally in the future I will be able to use all steps of the PWM for dawn dusk rather than using some to limit the maximum level - will I notice probably not.

 

[der_wille_zur_macht]

Not intimately familiar with your design, but if you wanted a pot to set max intensity, couldn't you just wire the pot to the analog dimming channel?

I need to just read the thread's last few pages, but how do you plan on feeding an analog signal to the driver? Do you have a DtoA chip picked out that you can drive from an Arduino?

Also, it's hard to see what the pads are like for that inductor. It bugs me that there don't seem to be packaging standards for inductors. I am glad to see that several of the components grew. Should make it much more flexible if people need to adjust values. It's easier to hack a small part on a large set of pads but hard to put a big part on small pads.

 

[terahz]

You can use analog signal. Currently the design calls for 0-5V analog signal (arduino friendly) which is coverted to 0-1.24(from memory) range by two resistors. This is what the chip requires.

The reason for the pot is if you just want to have an adjustable driver, without a controller.

The package for the inductor is 12.5x12.5mm with pads about 6mm wide. It seems like this package has the most variety of values.


EDIT: yes, you can wire the pot. That is all I was going to do anyway .