DIY LED driver for reef lighting

Perhaps you meant "linear or audio" pot? referring to the taper? In that case, it kind of doesn't matter, but you probably want linear.

And yeah, pretty much any 555-based circuit will work - just search google, there's lots of info out there on the subject.
 
Question about hooking up the driver to the Typhon and Meanwell.

Four input lines:
24V - from the Meanwell SP-320-24
Gnd - common ground
PWM - 5V PWM from the Typhon
5V - 5V output from the Typhon? (This is the one I'm really unsure about)
 
Pretty much yes. The PWM pin on the driver should be fed from the PWM pin for the channel you want to control that driver with on the Typhon. The 5v can be ANY 5v. You can use the 5v pin on the FTDI header, for instance.
 
Wow FishMan! You really did it.

Thanks.

abt92e.jpg
 
Two interesting tid bits. Do not reverse 24 V and ground for some reason the CAT4101s don't like it.:blown::furious::sad2:

However another good reason to turn down you voltage supply is that if for some reason (hmm what could that be) the CAT4101 stop regulating you won't blow up the whole string of LEDs because there is not enough voltage.

kcress - Yeah I did, but did you read it? You didn't answer my question. What do they mean by kelvin connection in the data sheet?
 
Fishman, check this out:

http://www.cypress.com/?rID=37694

As far as I can tell, it has to do with separating traces used for sensing from traces used for power transmission. i.e. if you have a resistor that's in the power transmission trace of your circuit, and you need to connect it to a sense pin on an IC, do so with an independent trace, not by tapping off the power trace far away from the resistor.

However I'm as confused as you - I can't tell how this would or wouldn't apply to the CAT4101 other than as a general admonishment that you should be careful to protect the sense trace from noise.

Though I have to think that if you're putting a potentiometer inline with the sense resistor, inaccuracy due to noise might not be quite an issue, since you will just naturally "tune it out" with the pot - in other words, if you want 500mA current, you're going to turn the pot until you read it - you aren't going to calculate where in the pot's travel you should theoretically get 500mA, and then get ****ed off because turning the pot to that spot gives you 550mA because of noise.
 
I think that was the one I found initially and said HUH? Do you remember where you saw your information on RSET limitiation (trace length IIRC)?
 
To summarize the datasheet's instructions for this pin:

RSET pin is connected to an external resistor to set the LED current. The ground side of the external resistor should be star connected to the GND of the PCB. The pin source current mirrors the current to the LED sink. The voltage at this pin is regulated to 1.2 V.

The RSET resistor should have a Kelvin connection to the GND pin of the CAT4101.

My interpretation of this is that you don't want current traveling in the same path as the connection between the RSET resistor and the GND pin on the IC, and you want to make sure it's not a noisy layout.

Here's the suggested layout:

cat4101_layout.png
 
And where is their "star connection"? And the white spot is that a via? I had trouble with that picture too.

I looked at the 1.2 volts they used for mirroring. Most of the other drivers used less than an ohm in series with the LEDs to sense the current (so really low voltage). I figured at 1.2 volts there would be a lot less noise to worry about.
 
I have light! (on my first of twelve heat sinks :lol:)

But my question is: How do I adjust the current on a Meanwell SP-320-24?

I can adjust the voltage. Adjusting the PWM through the Typhon works great, 1% up to 100%.

Also, where do I hook up the multimeter to tell how many amps I'm pushing through? Do I unhook the positive line from the driver to the LEDs and put the multimeter in-line there? (+ lead on the driver and - lead to the LED?)
 
The Meanwell is just a constant-voltage DC power supply. It supplies current according to the circuit it's attached to - in other words, there really is no applicable concept of adjusting current on the Meanwell in this case.

If you want to measure current, you put your multimeter (configured to measure current!) in series with the LEDs. So, exactly as you described.

To adjust current, you swap out the Rset resistor on the board. Rset controls the nominal current that the driver operates at.
 
kcress - Yeah I did, but did you read it? You didn't answer my question. What do they mean by kelvin connection in the data sheet?

Um.. I scanned it. I didn't catch that.

Kelvin correction is as guessed. A direct connection, used only for sensing, no current carrying.

You can buy/search Digikey for current sensing resistors and see resistors with 4 leads. That's so the two inner leads can be used as sensing leads while the other two do the heavy lifting.

For something like the CAT chip you would not actually need the four lead resistors, you just need to route them as directly as possible between the pins of interest. Note that, "suggested layout scheme".
 
Um.. I scanned it. I didn't catch that.

Kelvin correction is as guessed. A direct connection, used only for sensing, no current carrying.

You can buy/search Digikey for current sensing resistors and see resistors with 4 leads. That's so the two inner leads can be used as sensing leads while the other two do the heavy lifting.

For something like the CAT chip you would not actually need the four lead resistors, you just need to route them as directly as possible between the pins of interest. Note that, "suggested layout scheme".

Just as a note - I have had one of these drivers on now for several (6? more?) months. It runs at normal heat (warm to touch, but not hot) and lights a string of 6 LEDs without any fuss. It's using all through-hole parts. I don't think it's quite as sensitive as they're making out - in fact it's the one pictured waaay back on page 12 of this thread...

CAT4101.JPG


Your mileage may vary, of course, but that's what I've found. I'm also running it with a voltage-matched PSU, so there's very little extra power to dissipate within the chip itself. That helps :)

Simon
 
Last edited:
The Meanwell is just a constant-voltage DC power supply. It supplies current according to the circuit it's attached to - in other words, there really is no applicable concept of adjusting current on the Meanwell in this case.

If you want to measure current, you put your multimeter (configured to measure current!) in series with the LEDs. So, exactly as you described.

To adjust current, you swap out the Rset resistor on the board. Rset controls the nominal current that the driver operates at.
Hmm...in testing.

Running two strings of 6 LEDs at 20V it's about 540mA at 100% on the PWM and about 2mA at 1%...

Not sure if I should go any higher on the voltage...That's running a bit over 80% of the 24V capacity...turning the pot all the way down seems to only drop it to 19V or so...
 
You can use 80% of the watts. So full voltage and 80% current. Or as you are trying 80% voltage and full current. Or a mix. Remember watts = voltage * current.. The 19V sounds about right.
 
Chris,

There are basically three ways the power in the LEDs can be adjusted. It's very important to understand the implications of each to make sure you're measuring and adjusting correctly.

1) Rset. This resistor sets the nominal current for the LEDs, i.e. the current they will run at when the IC is getting an "ON" signal on the EN/PWM pin, and Vin is higher than Vout + .5v. Basically what happens is that the IC regulates output voltage until it sees the correct current in the LED string.

2) Vin, aka the voltage supplied by the DC power supply. As long as Vin is greater than Vout (the LED voltage) plus half a volt, the IC will operate in it's normal constant current state, i.e. output voltage will be varied to maintain the target current.

3) EN/PWM pin voltage - this is a digital input, i.e. it has a high state and a low state. When you supply a PWM signal, the output voltage is rapidly switched on and off with a duty cycle matching the PWM signal. It's important to note that supplying a PWM signal does not simply shift the output to a lower current - instead, it alternates output between the nominal current set by Rset and zero. It's possible to use a simple multimeter on the output to measure an average current when feeding EN/PWM a PWM signal, but it's important to note that the LEDs ARE NOT seeing that measured current - they're seeing full current for a portion of the time, and zero current for a portion of the time. In other words, you DO NOT want to use a PWM signal to "set the current" through the LEDs.

Effectively, this is what I see as the best practice for operation:

1) Choose an Rset value to give the desired MAXIMUM current you want in your LEDs.

2) Build the rig, set your DC power supply to the lowest voltage it's trimpot will allow, and put a multimeter set to amps in series with the LEDs. Turn the rig on, and check the current you're getting. It'll likely be significantly lower than your target current. Turn the voltage on the DC power supply up, while watching the current. It'll climb as you turn voltage up. Somewhere very close to your desired nominal current, it'll plateau - adding more voltage will no longer result in more current. At this very point, the IC is operating in the most efficient constant current mode, which is the desired state of operation. You'll find that this point occurs right around half a volt higher than the Vf of the LEDs. Once you've found this point, you should no longer fiddle with the voltage adjustment on the power supply!

3) Take the multimeter out of the circuit, and operate as desired. Use the EN/PWM pin to "dim" the LEDs. Forget all about Rset and the voltage adjustment trimpot on the power supply, as you should never need to touch either again.
 
Back
Top