Help with small LCD LED backlight driver

[globoy]

I am designing a device using a 3.5" LCD module (the ER-TFT035-6 from buydisplay.com) that has a 6 LED backlight, wired as two sets of three LEDs in parallel (as shown in the attached image).  The spec says Vf is typically 3.2V at ~18mA/LED.  I am powering the device off of a single LiPo cell and currently have both a regulated 3.3V rail and the raw battery voltage.  I will probably limit the lower end of the battery voltage to around 3.4-3.5V (low-battery shutdown).  The high end is the typical 4.2V but could reach up to around 5V if the battery is removed but the device powered from USB (power-path charger).

I am looking for a very low-dropout linear current regulator since my on-board voltages can be nearly as low as Vf.  I'd like to limit current to around 16 mA per LED (~50 mA per set of three).  I would be happy with one regulator (both parallel strings in parallel) or two regulators (each driving one 3 LED set).

I've done some searching around and found one part that looks promising, the Diodes BCR430UW6.  Unfortunately in the current parts environment it is unobtainium.  I've also seen some simple transistor based current regulators and current mirrors.  Getting the very low drop-out seems to be difficult however.

Just curious if anyone has an ideas up their sleeves.  Thanks!

[Zero999]

Why not just use an op-amp, such as the good old LM358 and a transistor such as the BC337?


V_SET can come from the 3.3V power supply and a potential divider.

[ajb]

An op-amp constant current source like Zero999 mentioned may be a good way to go.  With only 200-300mV of headroom you'll want to use a FET instead of a BJT as the pass transistor and that means compensating the loop for the gate capacitance.  200mV at 50mA means max 4R for the sense resistor, transistor, etc, so the FET will need to be a fairly low resistance relative to the actual power handling ability you need and gate voltage you'll be able to apply, so a bit of care will be needed in selecting it.  In reality you might have more headroom than that, but it partly depends on how the module is built.  You'll need a a reference for the drive, of course, which can be divided from a regulated rail, or if your MCU has a built-in bandgap reference that you can bring out that could work.  Even better if you have a DAC you aren't using, then you can even make the backlight current programmable. 

[tooki]

I am designing a device using a 3.5" LCD module (the ER-TFT035-6 from buydisplay.com) that has a 6 LED backlight, wired as two sets of three LEDs in parallel (as shown in the attached image).  The spec says Vf is typically 3.2V at ~18mA/LED.  I am powering the device off of a single LiPo cell and currently have both a regulated 3.3V rail and the raw battery voltage.  I will probably limit the lower end of the battery voltage to around 3.4-3.5V (low-battery shutdown).  The high end is the typical 4.2V but could reach up to around 5V if the battery is removed but the device powered from USB (power-path charger).

I am looking for a very low-dropout linear current regulator since my on-board voltages can be nearly as low as Vf.  I'd like to limit current to around 16 mA per LED (~50 mA per set of three).  I would be happy with one regulator (both parallel strings in parallel) or two regulators (each driving one 3 LED set).

I've done some searching around and found one part that looks promising, the Diodes BCR430UW6.  Unfortunately in the current parts environment it is unobtainium.  I've also seen some simple transistor based current regulators and current mirrors.  Getting the very low drop-out seems to be difficult however.

Just curious if anyone has an ideas up their sleeves.  Thanks!

In my current project, I'm using the TPS63021 buck-boost converter to power the 3.3V system logic (including LCD backlight), since it'll allow input voltages both above and below the output. It's a delightful little part that consumes very little board space. What could work for you is the adjustable-output version, the TPS63020, since you could adjust the output voltage to achieve the current you want.

But if you already have regulated 3.3V available, why not just use a small current limiting resistor?

[SiliconWizard]

I've never used a constant current source/sink for LED backlights as far as I can remember, but I can understand your need here if you're going to power BL directly from the battery (something I don't think I've done but that makes sense for efficiency reasons.)

Problem here is your requirement of low drop-out (we can assume you'll want as close to 0V as possible.). Simple transistor+opamp-based current sinks will have a few hundred mV drop-out depending on transistors as mentioned above. Would that be OK? 200-300 mV is not negligible when you deal with a 3.4 V to 4.2 V range...

Just another thought. I'm assuming from your figures that you'll be using a LiPo battery? 3.4 V as a cut-off voltage looks a big high to me. You might "waste" something like 10% to 20% of capacity depending on the battery and your current draw. Maybe that's OK for you. I'm further guessing that it might directly power some LDO generating a 3.3 V rail, which would explain the 3.4-3.5 V threshold? Usually a better approach for optimizing capacity is to use a buck-boost instead, but admittedly it comes at a cost. And, if you chose the 3.4-3.5 V threshold just because of the backlight and not for the reason above, that would be even "worse" IMHO. Using a good buck-boost (some have high efficiency > 90%) to power the backlight would solve the need to use a current sink (a single resistor would do), and would allow you to pick a lower cut-off voltage. Just a thought here. I might be missing some other requirements you have.

[globoy]

Thanks for the suggestions and comments everybody.

@Zero999 and @ajb - This approach seems promising and inexpensive.  I did a little more searching based on your suggestion and found an example circuit from Linear and a general-purpose load that also might work designed by Nick Johnson (of ReLoad and ReLoad Pro fame).  I will scour my parts bin and see if I can prototype something in the next day or so.  Here is an article from TI about a circuit with compensation as ajb suggests.  The Linear circuit and Nick's schematic are attached.

https://www.ti.com/lit/an/slaa868/slaa868.pdf?ts=1627317245231

@tooki -  A simple resistor seems possible (and easiest/cheapest!).  2 ohms would provide 50 mA @ Vf=3.2 with very little power dissipation.  Maybe I too quickly discounted this approach because I worried about variations in Vf across a population of displays resulting in brightness variation from unit to unit.  I should check into that more.  Thanks for the pointers to the TI parts.

@SiliconWizard - Your insightful comments are exactly the things I've been trying to balance.  I have used a boost-buck converter in a previous system but worry about parts availability steered me to a low-dropout buck converter that seems to be available now.  The 3.4V minimum comes from the buck converter and the 3.4-3.5V range accounts for total system error in measuring the battery voltage across varying loads.  It does seem to also provide a little room for some kind of simple current regulator too which in my mind was an acceptable trade-off for some lost battery capacity (total parts cost is also a consideration).

[Zero999]

A MOSFET will work, but it's a little more tricky to stabilise. You need a device with a low threshold and a rail-to-rail output op-amp, for it to run down to 3.5V

A BJT won't be that bad. The BC337 is specified to saturate at 50mV, with a collector current of 50mA. Another advantage is a cheap LM358 can be used, as it needs to output <1V. I know a MOSFET will have a  much lower voltage drop, but I doubt it's worth the trouble.
https://www.onsemi.com/pdf/datasheet/bc337-d.pdf

I was thinking something like this will do.

Note that, it doesn't have to be perfect. Don't worry if the current falls a little, when the low battery threshold is reached. It will be hardly noticeable.