Hi,

I'm designing a circuit to control chains of 3 LEDs (3 LEDs in series). I'm planning to use the CAT4016 constant current driver (data sheet here).

My voltage source for the LEDs is 8.2V, each LED has a 2.4V drop at 50mA, so the voltage presented to each sink pin is 1V (8.2 - 2.4*3). I've attached a diagram of this setup (one circuit shown for simplicity).

The sink pins are 5.5V tolerant. There are other ICs which have higher voltage tolerances on the sink pins. I don't understand why that would be needed. Hence wanted to ask if there is reason for this? And more importantly, I wanted to check if my circuit reasoning is sound (i.e. if there are any conditions, such as startup inrush, where 8.2V might be applied to the sink pins).

Many thanks.

 

The sink pins are 5.5V tolerant. There are other ICs which have higher voltage tolerances on the sink pins. I don't understand why that would be needed. Hence wanted to ask if there is reason for this? And more importantly, I wanted to check if my circuit reasoning is sound (i.e. if there are any conditions, such as startup inrush, where 8.2V might be applied to the sink pins).

As 8.2V > 5V when the sink is off you'll put your eye out... or damage the sink pin

Also it's rather risky to rely solely on the voltage drop of the LED to match up like this. I've seen it done but rarely.

 

I think his point is, when the sink is off, the LED forward drops should prevent significant current from flowing from 8.2V to 5V through an available current path (presumably a PN junction).
I would have to see the LED datasheet before I would ever attempt it, and even then I wouldn't be comfortable with it.
Perhaps something like a 1A Schottky diode from each sink output to +5V would protect the IC, if necessary.

 

The interesting point is that LEDs drop voltage no matter what. It is a little less at very low currents.

There will never be even just 5 volts at the end of the LED chain.

It is very rare that LEDs latch in a way they short circuit. I have seen it a few times for single 20mA LEDs. So far never for multichip LEDs.

Normally you can almost count LEDs latch open. But as I say, I have seen 20mA LEDs which latched short circuit.

Think of mass products, after some years statistically the first failures occur for various reasons. So if you have a big expensive PCB, use a polyfuse or even a melting fuse, save the expensive PCB, prevent fires and all that.

If the LEDs never heat up to the margin and experience overcurrent, there is very little chance for them to latch short circuit (in my opinion).

 

Indeed, this is my assumption too - an LED will voltage drop and not conduct until greater than something like 1.5v (for a red LED). The actual LED I plan to use is here. Sadly no info on how it behaves below 1.9v/14mA (approx).

I tried hooking up chains of blue LEDs with a resistor and a 4.66V (measured) supply to test the theory. These have forward voltages of 3.2V typical at 20mA. They're cheap LEDs, so the datasheet doesn't have a voltage/current graph.

With 2 LEDs in series, they light faintly and have 2.3v drop each at 0.007 mA.
With 3 LEDs in series, they don't appear to light and have a varied voltage drop across each (0.82, 1.82 and 2.02v) and <0.001 mA (my multimeter doesn't go lower).

The circuit itself is low cost and the LiPo contains short protection.

Would considering how the IC clamps the current help decide if it is safe or not? (I couldn't glean much from the data sheet.)

 

They probably don't clamp, I expect they are an open drain pin (ie a FET drain pin).

The LED may get some small leakage current under 1.9v, that might expose the pin to overvoltage condition. Adding a high value resistor from the pin to ground should stop that.

Alternatively you could expose a pin to 6v or 7v through a high value resistor, and see if the pin has an internal clamp diode that will limit th epin voltage to about 5.7v or so.

 

The boards arrived yesterday. I've been running the circuit as described in the first post, but with a 9v source. The IC seems fine and operates as I would expect. I've been pulsing full current and using PWM without issue.

 

Just a quick update. I've now been running this config without any issues on 5 different boards for many weeks. They run different voltages and LEDs. No issues at all.

 

Having just posted my last, I had my first failure!

The board was running at 8.4V with two different LEDs strings:
- 2x 3.8v drop white LEDs - chip sinking 0.8v
- 3x 2.4v drop red LEDs - chip sinking 1.2v

After a late night soldering, I managed to short one of the whites and one of the reds! The net result was normal function for a few minutes, then a puff of smoke and two channels failing.

Upon inspection:
- channel 4 was running 1x 3.8v drop white LEDs - chip sinking 4.6v
- channel 1 was running 2x 2.4v drop red LEDs - chip sinking 3.6v

When fully on, this is within the permitted voltage range for sinking (though in my opinion way too much energy to dissipate in the chip). The absolute maximum is 6v. So, obviously as the LEDs started up and the current / voltage drop increased, sufficient current at >6v hit the chip and fry it.

It's quite interesting in the way it failed too - it continued to limit the current, but the control logic for the channel failed open. For my application, this is actually a 'good' failure.