DIY LED driver for reef lighting

[kcress]

Why so? The design is within limits for the IC and all components will be within their ratings.

My experience is that pushing those types of switchers results in amazingly HOT controller ICs and inductors. If the board is mounted vertically in a good room-ambient position and all data sheet details are followed then you can get the advertized power out of the design and it will only lightly char your fingers. That's why I'd suggest a fan. Probably one of the itsy bitzy ones, like you find on heat sinks. I'd include a header to plug one into. Then, if you need it you have it. If you run the design down around <75% then it wouldn't be an issue.

would you have an idea, in terms of efficiency, how do these two driver designs compare to the simple driver you posted kcress?

One I posted?

Sent from my phone, isn't mobile technology wonderful?

Much better!:spin3:

 

[DJ in WV]

ive been following this thread on and off but saw this last post on my email and thought that id add that it is always a good rule of thumb to go over what you need by 25% so you dont have to max it out to get your output you are looking for as that will drastically shorten the life of any electronic

 

[firechild]

Change the voltage divider (R6) to make 10V -> 1.24V (Something like 212KOhms). You will also either have to find a different opamp that works with 10V supply, or put a couple of same value resistors (say 212KOhm, if that's what you use on the first divider) on the 5V input (one in series with the 10V source, another in series with ground) to bring it down to ~5V. That should do it.

Ok, I've been doing some reading about this chip and I tried to make my own circuit (not dissimilar to yours, but more similar to the circuit on page 21 of the datasheet) but I just can't figure out these PCB design programs. I understand the voltage divider (R6 and R7) but I don't understand where the opamp comes into it. If the reference voltage is being reduced to 1.25V as soon as it hits the PCB, why does the opamp get affected?

 

[firechild]

Actually, I've just gone over your schematic and mine and the only real difference I can see is in mine (as in the datasheet), the EN pin loops around to the VIN pin. By the looks of it, yours goes nowhere. The datasheet says that the EN pin requires >1.74V to switch on.

Yours:

Mine:

 

[ingoh]

Actually, I've just gone over your schematic and mine and the only real difference I can see is in mine (as in the datasheet), the EN pin loops around to the VIN pin. By the looks of it, yours goes nowhere. The datasheet says that the EN pin requires >1.74V to switch on.

EN is wired to JP1 to allow external control.

 

[terahz]

If the reference voltage is being reduced to 1.25V as soon as it hits the PCB, why does the opamp get affected?

The datasheet says that the EN pin requires >1.74V to switch on.

The reason for the opamp can be found a couple of pages back in the thread. When I built the first board, the IC wasn't switching off the LEDs at 0V on Iadj. At least on my boards there was always about 8mA current going through them. Now the opamp is acting like a switch, when the dimming signal goes under ~ 20mV.

The EN pin is out on a jumper. You can PWM that pin to do PWM dimming, or just put a jumper to the pin next to it, which connects it to the VIN. PWM dimming is somewhat simpler and requires less components on the board.

With the opamp, in the latest revision, a reference voltage is required from the dimming controller. That voltage is used to set the low voltage at which the opamp sends the EN pin to GND. It is also used to enable the IC when the dimming voltage is above said low voltage. And is also used to power the opamp.

 

[firechild]

The reason for the opamp can be found a couple of pages back in the thread. When I built the first board, the IC wasn't switching off the LEDs at 0V on Iadj. At least on my boards there was always about 8mA current going through them. Now the opamp is acting like a switch, when the dimming signal goes under ~ 20mV.

The EN pin is out on a jumper. You can PWM that pin to do PWM dimming, or just put a jumper to the pin next to it, which connects it to the VIN. PWM dimming is somewhat simpler and requires less components on the board.

With the opamp, in the latest revision, a reference voltage is required from the dimming controller. That voltage is used to set the low voltage at which the opamp sends the EN pin to GND. It is also used to enable the IC when the dimming voltage is above said low voltage. And is also used to power the opamp.

Sorry, I've been working with version 3. I don't need to worry about the opamp since I will be switching my light unit off overnight because my controller starts its program when power is supplied, therefore needs to be reset to start the program again. So, version 4 is extra components which are unnecessary. Unfortunately I don't have eagle so I can't make the slight modifications I need to make. Otherwise, I'm good to go with the version 3 design and just need to change the R6 and R7 values to achieve 0-10V dimming.

 

[reefocean]

Reeefocean, i would not suggest using the original design from this thread. It is limited in its dimming range and some people had downright poor stability from it.

I am struggling to understand what you are paying an engineer by the hour to do when pretty much any DIY LED driving problem has been or can be solved for free in this thread.

hehe, I am actually going over this thread again. On page 20 right now. I am banking on the LM3409HV seems to do everything I want. I have no background in engineering at all, and my physics stopped in grade 11. But I am starting to understand more and more. I don't think I will get to the point of being able to design gerber files but at least I am starting to understand the functionality of each piece involved in construction of these drivers.

 

[der_wille_zur_macht]

My experience is that pushing those types of switchers results in amazingly HOT controller ICs and inductors. If the board is mounted vertically in a good room-ambient position and all data sheet details are followed then you can get the advertized power out of the design and it will only lightly char your fingers. That's why I'd suggest a fan. Probably one of the itsy bitzy ones, like you find on heat sinks. I'd include a header to plug one into. Then, if you need it you have it. If you run the design down around <75% then it wouldn't be an

Thanks for the info. FWIW, I will have a fan in the driver box already thanks to the poor thermal flexibiliy of the 4101's.

 

[der_wille_zur_macht]

hehe, I am actually going over this thread again. On page 20 right now. I am banking on the LM3409HV seems to do everything I want. I have no background in engineering at all, and my physics stopped in grade 11. But I am starting to understand more and more. I don't think I will get to the point of being able to design gerber files but at least I am starting to understand the functionality of each piece involved in construction of these drivers.

Hey, we all started somewhere. I have no formal training and didn't even know what a gerber file was a few years ago. Let us help solve your design issues and save your money for corals!

 

[theatrus]

Here is a design I made awhile back, which does work successfully, based on the LM3404HV integrated driver with on board microcontroller (AVR) and RS485 bus (hint: checkout the silly overcomplicated moonlight thread - it talks to it). Here is the slightly convoluted schematic (page 1 in picture, two page in PDF):

(PDF)

Board:

It can't hit the peak currents of the external FET LM3409, but that was one of the design tradeoffs. Its been used to run LEDs for my overlit desktop nano-cube as my main tank still is running T-5s.

Yes, its all surface mount parts (but those are easy). I'll setup the Wiki page and download later once I remember the actual BOM parts.

 

[Ivan Rose]

Hindering the topic a bit, I need some quick help ...
I have a somewhat silly question for the experts in PCB ... How do the boards in EAGLE fill with this track, where only the contours?

 

[ingoh]

Hindering the topic a bit, I need some quick help ...
I have a somewhat silly question for the experts in PCB ... How do the boards in EAGLE fill with this track, where only the contours?

Try the command "ratsnest"

 

[Ivan Rose]

Thank you so much INGOH !!!!

 

[Gorgok]

Shipment finally arrived. Everything looks just fine. I didn't get 10 boards though... =/. I got 12 =D. So i could really really blind someone. Getting to the point where i actually need to start a build thread, but a few more steps need to get done so the posts come together faster.


Stack of boards, with one having been cut into the final pieces. I need to figure out a better way to do that if i need to cut more.


Top side, will have a heat sink occupying the space at the top edge (CATs under there).


Bottom, not much to look at.

 

[snorkeler]

One I posted?

Yep, this one: http://www.reefcentral.com/forums/showthread.php?t=1751598 which inspired me to try a dimmable version with transistors and you ended up guiding me to MOSFETs (which are working great) in this thread: http://www.reefcentral.com/forums/showthread.php?t=1914090

 

[kcress]

On topic, would you have an idea, in terms of efficiency, how do these two driver designs compare to the simple driver you posted kcress?

Snork

Yep, this one: http://www.reefcentral.com/forums/showthread.php?t=1751598 which inspired me to try a dimmable version with transistors and you ended up guiding me to MOSFETs (which are working great) in this thread: http://www.reefcentral.com/forums/showthread.php?t=1914090

Hi Snork.

Well, I didn't specifically post a "driver". I helped you with yours and my 'cheap' thread was not really a driver situation.

Efficiency on any LED driver comes down to how much voltage is being dropped somewhere. "Dropped voltage" is wasted energy.

PWM systems that include resistance have higher losses than those that use inductors because in the inductor case there's no resistor. Some inductor designs need a diode. Diodes are very lossy and cause efficiency loss. Higher complexity designs use synchronous switching of more transistors to remove the diode from the picture.



To sum it up.

PWM with resistor limiting:
1) Are generally the least efficient. 40%~98%
2) Can be the most efficient if the load voltage very closely matches the source voltage as then there would be little to no voltage to 'drop'.

Switched inductor with a diode
1) Have good efficiency. 75%~86%
2) Fairly simple.

Switched inductor with synchronously controlled FETs.

1) Most complex and most expensive.
2) Highest general efficiency. 89%~95%

 

[shark boy]

Extra CAT4101 pcb

Extra CAT4101 pcb

Does anyone have 1 or 2 of the CAT4101 triple boards they would part with?
I am in need of at least 1.Please let me know as my leds are in the mail.
I live in the Atlanta area and would love to get those babies hooked up ASAP to my Typhon controller.

Shark boy

 

[der_wille_zur_macht]

If you cannot find any, iteadstudio has their 10-count 5cm board on sale for liike $9 this month with $4 shipping. Dirt dirt cheap for a small count order of custom pcbs.

 

[ingoh]

v0.4 of the LM3409 board is ready.
http://joro.geodar.com/code/LM3409 v0.4/
http://joro.geodar.com/code/LM3409 v0.4.zip

If I open this files eagle warns about inconsistencies between scheme and board. Is it OK to create gerbers from this file?

I'll try out the LM3409 altough CATs are ready and running. They have to drive only 0.7A at ~19V (6*XP[E|G]) but with only 1V too much input voltage they get really hot. No problem with a lab power supply but hard to handle with a "normal psu".

regards
Ingo