Driving an LED with 1.8V

[leonhart88]

Hey guys,

Quick question...I'm designing a lower power BLE device running off a CR2032 coin cell battery.

All my peripherals can operate at 1.8V so I've chosen a 1.8V regulator (STLQ015M18R).  The only thing I'm concerned about is driving of indicator LEDs (APDA3020LSECK/J3-PF).  This LED has a forward voltage of 1.8V, but I'm planning to drive it at a much lower voltage (0.2mA, instead of 2mA).  Reading the graph off the datasheet, the forward voltage would be somewhere around 1.7V.

This is cutting it pretty close to the supply voltage so I'm worried it won't be reliable (please correct me if I'm wrong).  From what I know, I could try the following:

• Use a LED driver like BCR 402R E6327...but these have a minimum output current of 10mA and I'm not sure if you can just tack on extra series resistance.  One thing I could do is to provide power from a microcontroller pin and PWM it to achieve a lower average current
• Use a boost converter to up the voltage from the 1.8V for the LEDs
• Increase my main regulator voltage
• Use a constant current source/driver for the LED

Just wondering what people suggest.  I'm trying to minimize current consumption so I'd like to run the LEDs below 1mA, but I'm also trying to keep parts count down if possible.

Thanks.

[Audioguru]

The graph shows a "typical" LED that you cannot buy unless you buy thousands at different times, measure them all then select the voltage you want.
The text in the datasheet says that at 20mA some are 2.2V, others are as high as 2.8V and others are less than 1.8V. The voltage of an LED changes when its temperature changes.

Therefore an LED is driven with current, not with voltage.

[mikewhy]

Maybe consider driving it with something like a Joule Thief to pump the voltage. Basically, an inductor driven by a relaxation or other oscillator.

Sent from my SM-T810 using Tapatalk

[sleemanj]

You are powering the device from a CR2032 right.  So you have 3v nominal available. 

3v+resistor+led+mosfet

[leonhart88]

The graph shows a "typical" LED that you cannot buy unless you buy thousands at different times, measure them all then select the voltage you want.
The text in the datasheet says that at 20mA some are 2.2V, others are as high as 2.8V and others are less than 1.8V. The voltage of an LED changes when its temperature changes.

Therefore an LED is driven with current, not with voltage.

Yes, this is why what I've spec'ed is too close for comfort.

Maybe consider driving it with something like a Joule Thief to pump the voltage. Basically, an inductor driven by a relaxation or other oscillator.

Sent from my SM-T810 using Tapatalk

I'll look into that.

You are powering the device from a CR2032 right.  So you have 3v nominal available. 

3v+resistor+led+mosfet

The only issue with driving directly from the battery voltage is that my brightness will change with the voltage level which is undesirable.

[wraper]

The only issue with driving directly from the battery voltage is that my brightness will change with the voltage level which is undesirable.

As a workaround, if you have MCU with ADC, you could measure battery voltage and PWM LED from MCU accordingly to keep the same brightness.

[leonhart88]

The only issue with driving directly from the battery voltage is that my brightness will change with the voltage level which is undesirable.

As a workaround, if you have MCU with ADC, you could measure battery voltage and PWM LED from MCU accordingly to keep the same brightness.

That's a good idea.  That might allow me to get rid of a regulator entirely and just feed the CR2032 directly into my peripherals as I basically just have a BLE module and a digital accelerometer.  The regulator was mainly there to keep constant brightness, but also, I wasn't sure how other peripherals would behave if the voltage supply fluctuated.  I know inside my BLE module (nrf52 based), there's an internal regulator, but I'm not sure how the accelerometer or other peripherals like FLASH would behave.

[tooki]

Thanks to the internal resistance of a 2032 cell, you can put an LED right across it with no resistor at all and the LED will be happy. If you add a resistor to further current limit it, I suspect you’ll be just fine. Who cares if the brightness changes, as long as it’s visible? I would expect that, as the battery voltage drops, the BLE module would stop working before the LED.

[leonhart88]

Thanks to the internal resistance of a 2032 cell, you can put an LED right across it with no resistor at all and the LED will be happy. If you add a resistor to further current limit it, I suspect you’ll be just fine. Who cares if the brightness changes, as long as it’s visible? I would expect that, as the battery voltage drops, the BLE module would stop working before the LED.

I suppose that's a fair question...I guess it doesn't really matter if the brightness changes over time since it will be such a gradual effect.  The only thing is I want to keep current consumption at a minimum.  So if I spec out my current limiting resistor such that at 2V the LED is still visible, at 3V it will be drawing more current than desired.

[leonhart88]

Thanks for everyone's suggestions.

I think at this point I'm going to just power everything directly from the battery and eliminate regulators entirely from my board.  I'll power the LED directly from the battery voltage and potentially PWM it in order to keep a constant brightness as well as a constant average current draw.

Cheers.

[T3sl4co1l]

You don't even need a level shifter / switch.

Just connect +3V to LED anode, LED cathode to resistor, resistor to port pin.

When the port pin is in the '1' or 'Z' state, the LED sees 1.2V or less: not enough to light up (well, a few nanoamperes worth).

Note this won't work very well for IR LEDs, will be marginal for red, will work excellent for GaP green (but they're very inefficient and you'll need more than 0.2mA anyway), and won't work very well for GaInN/GaN green, blue and white (because they are 3-3.6V parts and really need a bit of boost to be sure you've got enough headroom).

Tim

[leonhart88]

You don't even need a level shifter / switch.

Just connect +3V to LED anode, LED cathode to resistor, resistor to port pin.

When the port pin is in the '1' or 'Z' state, the LED sees 1.2V or less: not enough to light up (well, a few nanoamperes worth).

Note this won't work very well for IR LEDs, will be marginal for red, will work excellent for GaP green (but they're very inefficient and you'll need more than 0.2mA anyway), and won't work very well for GaInN/GaN green, blue and white (because they are 3-3.6V parts and really need a bit of boost to be sure you've got enough headroom).

Tim

Yup that's what I was thinking of doing, no seperate FET or switch.

I am considering adding a step-up regulator so that I could use an RGB LED actually.  But will have to see how much of a cost that will add to the system.

[leonhart88]

For those that might be interested, I decided on running my peripherals straight off the battery, and then using a boost converter to get 5V for an RGB LED using the following parts:

TPS61222DCKR
CLX6D-FKB-CMPQSGKBB7A363

Thanks everyone for your inputs.  Cheers.

[ebclr]

Basic circuit for double the tension