DIY LED driver for reef lighting

You CANNOT dim on the EN/PWM pin using an analog signal (i.e. that generated by a pot). However, you CAN dim on the RSET pin. This pin is normally used with a resistor to set the nominal current for the circuit. However, if you choose a resistor (to limit max current) and a pot (to provide adjustability up to the point) you can manually dim on that pin. I'm pretty sure TheFishMan had a design that did this, you might want to check with him.
 
Yes, I did it that way. I have had no trouble with the prototype boards (just went through a new round). I used a fix resistor for 900 ma (604 ohms IIRC) and then used a 10k (also used a 100k, but like the 10k better) in series for dimming.
 
also where is the rset pin?? happen to have a pic of how you did this?? also this is without using a arduino correct... i'm wanting to manually dim a driver with a pot
 
RSET is the pin connected to the only resistor in the circuit. http://www.onsemi.com/pub_link/Collateral/CAT4101-D.PDF

Basically, I think he is saying to remove the resistor on the board and use a 600ohm with a 10k pot in series. You can solder 2 lengths of wire to the pads where you removed the surface mount resistor. Then connect the resistor and pot in series between the 2 lengths of wire.
 
RSET is the pin connected to the only resistor in the circuit. http://www.onsemi.com/pub_link/Collateral/CAT4101-D.PDF

Basically, I think he is saying to remove the resistor on the board and use a 600ohm with a 10k pot in series. You can solder 2 lengths of wire to the pads where you removed the surface mount resistor. Then connect the resistor and pot in series between the 2 lengths of wire.

ahhh I see.. do i use any watt?? or just as long as it's 600 ohm?? so each cat4101 i have to use one pot to each?? or can i link them all to one pot with just one 600 ohm resistor..
 
The power rating on the resistor does not matter. You just want to make sure you pick the correct resistance to get the right "base" current.

Regarding your earlier question about pot tapers - there's no accuracy difference. The difference is how much of a change in resistance you get vs. how far you turn the knob. For a linear taper, it doesn't change. For an audio taper, the relationship gets much more steep near one end of the pot. You probably want a linear taper for this application.
 
Pretty much what every one else has said above. But here is a clear schematic of what I did.
picture.php


I would not recommend using the same pot for all CAT4101. It might work, but IMHO pots are cheap enough it is not worth the risk.
 
I would not recommend using the same pot for all CAT4101. It might work, but IMHO pots are cheap enough it is not worth the risk.

I would strongly reinforce this recommendation. If you were to parallel one pot across several RSET pins, then you're effectively tying together the feedback mechanism for each of the drivers. In other words, they can no longer react independently to their own driven string. So you could get into a situation where a change or condition on one string creates a huge problem for the whole array, for instance.

This is another reason why I'm nervous to use RSET for dimming, vs. the EN/PWM pin which is generally more suitable for the purpose. The problem there is you need more than just a pot to feed a signal to EN/PWM.
 
I think we discussed this back about ten or twenty pages - some time early last spring. :)

It's not really a big deal, but I'm just nervous about using the feedback pin for external control. I'm projecting a bit but it's just not designed for that. Look in the datasheet at how specific they are about things like trace length and layout for the RSET resistor. Putting a signal that sensitive through a pot (which are generally unstable and inaccurate) makes me nervous. Compare that to the EN/PWM pin, which they expect to be fed from an unknown-quality external signal, and is pretty wide open design-wise.

I'm sure that using RSET for dimming can work just fine, and in an implementation where it's working I wouldn't change anything - but, if designing from scratch, I'd try my best to use EN/PWM instead.
 
ahhh well it seem the steves led driver tied all the ones together.. wish i still had the driver i would take a pic of it.. but it was pretty cool to see all of them on one pot thou it would limit the pots used.. maybe i'm going to have to look at how he did it again..
 
"coolness" aside I would consider it a serious design flaw if he tied the RSET pins of multiple chips together.

That said there are plenty of other ways to get a knob to dim one or more of these drivers. The most straightforward way to me would be to do a simple 555 circuit to generate a PWM signal based on the knob's position. This would be like $2 in parts and would let you gang up as many of the EN/PWM pins as you wanted, and would avoid all the potential problems we've talked about in the last few posts.
 
"coolness" aside I would consider it a serious design flaw if he tied the RSET pins of multiple chips together.

That said there are plenty of other ways to get a knob to dim one or more of these drivers. The most straightforward way to me would be to do a simple 555 circuit to generate a PWM signal based on the knob's position. This would be like $2 in parts and would let you gang up as many of the EN/PWM pins as you wanted, and would avoid all the potential problems we've talked about in the last few posts.

the pwm signal on these drivers would only be a 5v correct?? so if i took a 7805 off the powersupply and run it to a 555 then run it into the pot then the drivers then it would work fine?
 
I am having a problem with duty cycle on Buck mode.

I am considering an ON-Semi NCP3065 driver, like early in this thread, but in Buck mode.
I want to run it close to the input voltage. (24 V supply, ~21V to LEDs), 1.2A
But in doing so, it violates the duty cycle.
Anyone hit this on other drivers?
Seems like it is a fundamental problem for everyone, the way Cout must
be charged in buck mode.

I can dial back the number of LEDs per driver of course, but annoying.

I am reluctant to go to boost mode, since a mistake on Vf calculations
can result in a blown string. (You cannot go below Vin)

I know the linear one is a big hit here, but I am reluctant to use that since
I have everything on one supply, so hard to get efficiency.
 
You could easily handle that by turning your 24v supply up a bit. Most of the DC supplies we're using have a voltage trimpot.

If you're really stuck I'd suggest emailing tech support at OnSemi. They've been good at responding to others who have contacted them.
 
After much searching, the latest plan is to run in Boost mode, but use an external switch, and run much higher voltage.
The theory is to run ~49V output, on a 24V power supply, 1.2A.
At that voltage, no way I will accidentally miscalculate Vf sum below 24V.
Could go higher, but safety and mosfet become an issue.
The external switch even increases efficiency. (95.8%)
Plus, while an external switch is more work, I need to build half as many.
And of course, it seems there is a great schematic on this thread to start with!!

Downside - I need to find a high voltage, high current, low RDS, mosfet.
That is also cheap, and through-hole.
 
i got about half way through this to try to figure out how to make my own drivers and wow i am at a TOTAL loss after reading half of this thread.

where are u guys getting all ur parts?

what are u changing to get higher mah values?

brain is fried at the moment, maybe ill just stick with eln's haha
 
Back
Top