RGB LCD backlight currents & resistors

General project help for Adafruit customers

Moderators: adafruit_support_bill, adafruit

Please be positive and constructive with your questions and comments.
Locked
User avatar
roo123
 
Posts: 26
Joined: Sat Apr 04, 2015 11:42 pm

RGB LCD backlight currents & resistors

Post by roo123 »

I have the 20*4 and 16*2 LCD character displays, both fitted with RGB backlights.

Noticing the bigger display was much brighter I measured the voltage across the resistors on the back.

The bigger display seems to have 40mA per colour channel; the smaller display around 9mA (one channel is a bit higher).

I'm guessing that the bigger display has two LED per colour, each LED consuming 20mA; that the smaller display has one LED per colour and that it's running at half power (around 10mA).

Would it be a good idea to desolder the 220 ohm resistors on the smaller display and replace them with 100 ohm values that will allow the full 20mA per LED?

thanks

User avatar
adafruit_support_mike
 
Posts: 67454
Joined: Thu Feb 11, 2010 2:51 pm

Re: RGB LCD backlight currents & resistors

Post by adafruit_support_mike »

There will only be one LED (or one string of series LEDs) per resistor.

You never want to put LEDs in parallel. The forward voltages don't match exactly, and an offset of 60mV produces a 10:1 variation in current. It sounds like the large display is running a bit over spec for the LEDs. That isn't fatal, but will reduce the working lifespan of the LEDs.

Changing the smaller display's resistors to get 20mA per channel is safe, but you might find the blue LEDs overpowering the others. You'll get a better balance by keeping the total power around 35mW per LED.. 20mA @ 1.7v for red LEDs, 11mA @ 3.3v for blue ones.

User avatar
roo123
 
Posts: 26
Joined: Sat Apr 04, 2015 11:42 pm

Re: RGB LCD backlight currents & resistors

Post by roo123 »

Great advice Mike.

Thanks.

It's odd that on the 20*4 the resistors seem chosen very carefully (R: 75ohm, G:47:ohm, B: 47ohm) yet give 78mW, 124mW, 124mW per LED respectively: uneven and a huge overload.

Whereas on the 16*2 there's simply a standard-sounding 220ohm on each channel and the results are perfectly 27mW per LED!

I found the following doc on the Winstar site which recommends on page 1 to supply 16-20mA per LED. Yet the lifetime figures quoted on page 3 are for less than 15mA ! And on the 20*4 I'm measuring 40mA. All very odd!

http://www.winstar.com.tw/download.php?DID=4

User avatar
adafruit_support_mike
 
Posts: 67454
Joined: Thu Feb 11, 2010 2:51 pm

Re: RGB LCD backlight currents & resistors

Post by adafruit_support_mike »

roo123 wrote:It's odd that on the 20*4 the resistors seem chosen very carefully (R: 75ohm, G:47:ohm, B: 47ohm) yet give 78mW, 124mW, 124mW per LED respectively: uneven and a huge overload.
That sounds a bit off.. a blue LED's forward voltage should be about a volt higher than a green one's. If both LED-resistor pairs are connected to the same supply voltage, you shouldn't get the same power value for the same resistance.

What's your supply voltage, and what voltage are you seeing across each resistor?
roo123 wrote:I found the following doc on the Winstar site which recommends on page 1 to supply 16-20mA per LED. Yet the lifetime figures quoted on page 3 are for less than 15mA ! And on the 20*4 I'm measuring 40mA. All very odd!
That's part of what makes reading datasheets a challenge.

Datasheets are a negotiated compromise between the engineering team, the marketing department, and the legal department. The engineers want to give users useful information, and the legal department insists that everything in the document be true, but marketers are masters of "not lying, just telling the truth the right way." You have to pay attention to the footnotes, test condition specifications, and names/units on the graphs. After you read enough of them, you learn to sniff out the places where they're working hard to avoid saying something.

In this case, the important information is combined in the almost-unreadable graph of LED lifespan at the bottom of page 2 and the first footnote at the bottom of page 3.

The graph is a masterpiece: the title is "Life Span" but the units on the horizontal axis are degrees C and the units on the vertical axis are "percentage of maximum current". The graph is actually saying that as the temperature rises, you have to lower the current to get the lifespan specified at the bottom of page 3. It's wonderfully silent on how fast the LEDs will wear out if you run them at higher current levels.

The footnote at the bottom of page 3 says "we lowered the current to get this lifespan value."

So when page 1 says. "best operated at 16mA to 20mA", that means "best for us because you'll be buying lots of replacement units soon."

User avatar
roo123
 
Posts: 26
Joined: Sat Apr 04, 2015 11:42 pm

Re: RGB LCD backlight currents & resistors

Post by roo123 »

OMG

Supply to 20*4 character display: 5v.
Red 75ohm 2.98v (accross resistor)
Green 47ohm 1.9v
Blue 47ohm 1.9v

Regarding the temperature graph it's not clear what "maximum current" is, but if it's 20mA (from their 16-20mA range) then it looks as though above 25C one can easily get into danger. Since ambient temperature is often 25C and given the LED are embedded they might well be hotter so an appropriate driving current even lower?

Further down page 3 they give even more lifetime expectations, colour by colour for RGB displays. This time they are using 15mW (below the 16-20 recommendation!).

In the same table in the section on luminosity they feed the R LED less current? That, at least, is coherent with the choice of resistors / mW observed in our 20*4 display.

Just to round off the confusion, I checked back to the product page https://www.adafruit.com/products/499 and the copy writer states that there are 2 LED per colour hence 40mA: so has somebody been naughty and put them in parallel after all?

Encouraged by your remarks I've been reading around how people drive LED and getting more and more perplexed. For example it seems that using an active "constant current" driver is commonly regarded as a good idea (as opposed to a limiting resistor). Yet if the Forward Voltage of a LED increases as the LED ages doesn't constant current mean that the power dissipation by the LED is going to increase too, hence the brightness and the temperature-degradation effects? So the problem is not solved.

User avatar
adafruit_support_mike
 
Posts: 67454
Joined: Thu Feb 11, 2010 2:51 pm

Re: RGB LCD backlight currents & resistors

Post by adafruit_support_mike »

roo123 wrote:Supply to 20*4 character display: 5v.
Red 75ohm 2.98v (accross resistor)
Green 47ohm 1.9v
Blue 47ohm 1.9v
Yeah, those numbers check.

I'm not sure what they're doing with the green LEDs, but one decent measurement outweighs a ton of theory.
roo123 wrote:Just to round off the confusion, I checked back to the product page https://www.adafruit.com/products/499 and the copy writer states that there are 2 LED per colour hence 40mA: so has somebody been naughty and put them in parallel after all?
I suppose it's possible.

LED manufacturers have options the rest of us don't. They have to test the LEDs as they come out of the fab anyway, and often bin them by wavelength. It's fairly easy to adapt those tests and binning procedures to get batches of LEDs whose forward voltages match well enough that they can run in parallel.
roo123 wrote:For example it seems that using an active "constant current" driver is commonly regarded as a good idea (as opposed to a limiting resistor). Yet if the Forward Voltage of a LED increases as the LED ages doesn't constant current mean that the power dissipation by the LED is going to increase too, hence the brightness and the temperature-degradation effects? So the problem is not solved.
Constant current doesn't solve any problems with LED lifespan.. those are inherent to the nature of the device.

LEDs fail because current flowing through the junction generates heat and applies a certain amount of physical force to the silicon atoms (it's called the 'electron wind'). The combination of heat and force pushes some atoms out of position, causing lattice defects. Those defects increase the resistance through the junction, producing more heat. That increases the rate of defect production for the same amount of current, etc.

The process is exponential, so it starts with literally one or two atoms, then builds up to become a problem from there. It takes a long time to get started, but snowballs quickly once it hits a certain level.

The only way to control that process is to limit the amount of current flowing through the junction. The mechanics are complicated, but the relationship between current and lifespan is very roughly N^4.. doubling the current speeds the process up by about a factor of 16, cutting the current in half slows the process down by about a factor of 16. That's why you'll see so many small indicator LEDs running at 2mA to 5mA. You can still see them, but their lifespan is extended to decades of continuous use.

Constant-current drivers solve the problem of having to care about the supply voltage or LED forward voltage.

Whack together a circuit like this:
mirror.jpg
mirror.jpg (19.84 KiB) Viewed 180 times
and all the LEDs get close to 20mA regardless of their color or the supply voltage. Better yet, the only component value you have to calculate is the single 36 ohm resistor.

The NPN to the right of the 36 ohm resistor controls the base of the NPN above the resistor, and together they form a feedback loop which holds the voltage across the resistor near 0.7v. That holds the current through the resistor near 20mA.

The PNP above the resistor has its base tied to its collector, holding its base voltage at whatever will allow 20mA to flow through it. The PNP to the right has the same base voltage, so 20mA will flow through that one as well. The combination is called a 'current mirror'.

The 20mA flowing out of the PNP on the right goes through the NPN to the right of the resistor, so the two NPNs adjust the exact voltage across the 36 ohm resistor to "whatever allows 20mA to flow through the NPN on the right". All the other NPNs have the same base voltage, so they're all set to allow 20mA through as well.

The supply voltage and LED forward voltages are almost irrelevant. Changing the supply voltage from 3v to 10v changes the current through the 36 ohm resistor by a couple of milliamps. At any given voltage, the current through red, green, and blue LEDs will differ by less than 1mA.

And that isn't a very good circuit.. you can make the matching and stability about 40 times better by adding resistors between the emitters of the other NPN transistors and GND. It makes the calculations a bit more complicated though, and for driving LEDs it really isn't worth the extra parts and effort.

User avatar
roo123
 
Posts: 26
Joined: Sat Apr 04, 2015 11:42 pm

Re: RGB LCD backlight currents & resistors

Post by roo123 »

Amazing circuit Mike.

I just looked up "current mirrors" and yours is by far the most sophisticated!

Got to try it. And for sure will be reading about "electron winds" and trying to understand the aging process a little better.

Many thanks for all your insights.

User avatar
adafruit_support_mike
 
Posts: 67454
Joined: Thu Feb 11, 2010 2:51 pm

Re: RGB LCD backlight currents & resistors

Post by adafruit_support_mike »

Shoulders of giants..

Current mirrors are largely a solved problem, with Robert Widlar being the guy who got things started. He invented several circuits that are basic furniture these days. George Wilson built on top of Widlar's ideas, and the circuit above is a simple variation on a Wilson current mirror.

WRT electron wind, you'll probably find 'electromigration' to be a stronger search term. That's usually associated with the movement of copper in integrated circuits, but the forces are pretty much the same for silicon.

Locked
Please be positive and constructive with your questions and comments.

Return to “General Project help”