Need help making LCD display backlight dimmer

[valley001]

Hello all,

I am using one of these displays and the dimmest setting is still too bright. 

https://www.adafruit.com/product/912

Unfortunately the unit I have, what I believe is the backlight driver is unmarked.  The other chips are video processor (center) and dc-dc converter (left).  I believe the backlight driver is on the upper right. 

I have attached a higher resolution image of the board.  EDIT: Good grief there is a improperly soldered resistor in there. The one pictured is not in use it is a spare but ill have to attend to that resistor. 

Is there any way to dim this with hardware?  I did look up some backlight driver datasheets and I think some of them set brightness with a voltage reference range.   I was hoping it would be as simple as a resistor change but I don't think ill be that lucky.  I guess a filter in front of the panel is a last resort.

[inse]

Not regarding your question, but is the capacitor right next to the quartz misplaced on purpose, or is this a case of tombstone effect?

[DavidAlfa]

What do you expect providing a blurry image? Most of the parts are unreadable.

[ajb]

The adafruit page you linked to has a link to the datasheet for the TFT, which shows that the controller has a built-in boost controller for the backlight.  The shitty application schematic shows that it only needs an external transistor, diode, inductor, and current sense resistor.  Should be pretty straightforward to trace that out and swap the current sense resistor for a higher value, which will give a lower current. 

The unmarked IC looks to be connected to the serial interface on the panel, so I would guess based on the datasheet and adafruit description that it's a little MCU to handle the one-button setting of backlight value via serial commands to the internal boost controller.

[coromonadalix]

the dtasheet show you the needed voltage for driving the led backlight ...  you can build an current limitted driver,  like lm317 circuits  etc ...

Or you have to reverse engenier the pcb circuit to see  wich kind of control they have used, in the datashhet you see at least 2 ways of doing so,  probe the vled pins to see if its a pwm or dc control voltage ??

Adafruit could send you the schematics if you ask them ??

[valley001]

Thank you all for your helpful input. 

That LCD in the supplied datasheet has a 50 pin connector where this device has a 40 pin connector, so I am not sure what's going on there.  The datasheet for the 1.5 inch version does have a 40 pin connector.  I think they linked the wrong datasheet.  It looks like they basically use the same driver board for the 1.5, 2, and 2.5 inch displays. 

Before I noticed the pin discrepancy I did my best to trace out some of what I think may be the LED driver circuit, see attached image.  Since I dont have a proper pinout for this LCD I can not confirm the connections to the LCD are related to the backlight.

D1 is marked KL4 (Schottky diode)
Q1 is marked Y1 (NPN transistor)

I cant find C8

Not sure if I am entirely off base here.  As a last resort I will join the Adafruit boards and see if they will supply info.  If I am correct in my tracing out the circuit, R1 is the sense resistor correct?  Its a 1K resistor. 

[ajb]

Hmmm, bummer adafruit has the wrong datasheet.  But the circuitry you've traced out does match with the application schematic in that datasheet.  R2 is the current sense resistor, the one below the other transistor labeled "330" would be the prime candidate except you wouldn't expect it to be connected to another transistor like that.  Unless they're meddling with the sense resistor to adjust the brightness instead of using the serial interface to the panel?

[valley001]

Ill try to trace some more of it out and post it.  Interestingly, the 2 inch panel specs (datasheet is behind some paywall) indicates it does not have an internal backlight driver.  The 1.5 inch is directly driven without a driver I think.   I really wish I could get the label on that one chip, exhaustive image search did not yield any clear enough pic.  I know its a Microchip part. 

[valley001]

Yet more tracing out, see pic, getting a bit jumbled sorry.   The 330 resistor is indeed connected to ground and a pin of the display, and also to that transistor.  I traced it through the resistor next to R1 and to the mystery chip.   Thoughts?

[ajb]

I'm pretty sure the mystery chip is an MCU, it's the only thing I can think of that would be connected to the LCD that way, presumably for initial configuration of the panel.

Are you sure you have the connections for that second transistor correct? It would make more sense if it were in parallel with the backlight LEDs. A parallel transistor driven via PWM is a common method of dimming LEDs driven by switch-mode constant current drivers like this. When the transistor is off the constant current driver operates normally, when it's on it shorts out the LED string, but since the current is kept constant and the voltage across the transistor is very low very little power is dissipated.

[valley001]

OK, I spent some more time on it and found the second transistor (now marked Q2) that is connected to R2 and the LCD is tied to ground (pic updated in the prior post).  So both sides of R2 are tied to ground.  There may be more that I am missing of course.

[valley001]

Any ideas now that I discovered that ground connection?  Could I swap R2 for a higher value?

[ajb]

I don't think that's right.  There's no point in having both sides of that resistor connected to ground, and if that's truly how it's connected then it can't be an effective current sense resistor. 

How are you tracing these connections out?  If you're using a multimeter on continuity mode, it may be that the meter's threshold for what counts as a short circuit is too high (that current sense resistor is only 30/33 ohms), and/or it's confounded by parallel connections in the circuit, including diode junctions in some of the active components.  Manually confirming connections by tracing them out visually is a good way to confirm.  If you can't see the connections you could try using the meter on resistance mode, so you can tell when a connection is actually direct as opposed to just lowish resistance, or try taking two measurements, swapping the leads of the meter in between to rule out parasitic diode connections.  It may also help to disconnect the LCD from the board, to eliminate any connections that are made through the FPC/display controller. 

[valley001]

Thank you for your insight and patience.  You are right I have been using continuity check.  I went back and verified my connections visually and with the resistance mode and confirmed all either confirm visually or read .1 ohms.  I removed the connections made in error, as you surmised I was reading a 33 ohm connection as continuity.  I hope this revised version makes more sense.

[valley001]

Just bumping this up to see if my latest version of tracing the circuit out in the post above might yield some clues. 

[ajb]

Okay, so based on that it looks like they're still using that second transistor to PWM the backlight, but it's in series with the LEDs, which is the wrong way to PWM the output of a constant current driver. When the transistor is off, the driver will see no voltage at the top of the sense resistor (because the LEDs are disconnected, so no current is flowing!) and will increase its duty cycle in an attempt to drive more current into the LEDs to try to get the feedback voltage up to what it expects.  Of course this won't have any effect until the PWM transistor turns on again, at which point the LEDs get whacked by the driver's best effort to drive current through an open circuit. The feedback loop should recover pretty quickly, but then the PWM transistor is going to get turned off again and the whole cycle repeats.  This will subject to backlight LEDs to unnecessary stress and may shorten their lifespan, and the effect will generally be worse at lower PWM duty cycles. 

In any case, that still leaves the resistor marked "330" as the sense resistor. If you don't care about adjusting the backlight you could bypass that second transistor to disable the PWM dimming, but either way increasing the value of the sense resistor will decrease the brightness. 

"330" as a conventional resistor marking would mean 33ohms (3, 3, and then zero 0s), but not all chip resistors use that marking style. In this case it's pretty close to the expected 30 Ohms of the sense resistor from the application schematic so that's probably correct but it never hurts to verify. The relationship between resistor value and current is linear, but the relationship between LED current and brightness is not, so you might need to experiment to find the best value.

In general with current sense resistors you want to make sure that you select a part with an appropriate power rating. Here that's not likely to be a big deal, but if you measure the voltage across the resistor in operation (and not PWMed!) you can use that to determine the power dissipation: P= (V^2)/R.

[valley001]

Thank you very much for this, I owe you a beer or two for muddling through my amateurish circuit tracing lol.  I think I will experiment with higher values and leave the rest of the circuit intact so I can still increase/decrease brightness with the button.   That resistor does indeed read 33 ohms. 

[valley001]

Well... After dusting off my SMD soldering skills I experimented with several different values, and did not notice much of an improvement until I went up above 100 ohms, at 180 ohms I have something that is becoming useable.  I gather through, that I may have a problem with dissipation correct? 

I measured the voltage across the resistor, and at the lowest brightness I get 1.1 volts.  At the highest brightness I get some 10mv.  Bear in mind its still PWM so maybe none of those measurements are valid? or maybe I am going about it wrong all together?  I would like to lower the brightness further but don't want to ruin my display either lol.  At least the lowest brightness setting is now more or less useable. 

[ajb]

at 180 ohms I have something that is becoming useable.  I gather through, that I may have a problem with dissipation correct? 

Probably not, the power dissipated by the resistor is inversely proportional to the resistance in this case (because the backlight driver holds the voltage across the resistor constant), so at 180Ω you have about 18% of the power dissipation that you had with the 33Ω. 

I measured the voltage across the resistor, and at the lowest brightness I get 1.1 volts.  At the highest brightness I get some 10mv.  Bear in mind its still PWM so maybe none of those measurements are valid? or maybe I am going about it wrong all together?   

 

You're not going to be able to measure this with a multimeter while that PWM is happening.  You would need to use an oscilloscope and look at what the voltage is during the on time.  If the highest brightness setting is full on (no PWM) then you could measure under that condition, but you'd again need a 'scope to tell if there's still PWM at that setting or not.  You're very unlikely to hurt the display with a higher value current resistor, though, so I wouldn't worry about it.

[valley001]

Thank you.  I do have a scope, maybe ill solder some leads and look at it.

If I want to remove the PWM, would I remove the second transistor (labeled Q2 in the attached) and connect the resistor to the lead that goes to the LED- connection. Is this correct?  Could I install a pot there for adjustment?

Attached is a cleaned up version of the trace out.  I did find the datasheet, so I was able to label the pins leading to the display. 

[valley001]

I want to bump this up to see about an undesirable phenomenon that remains when changing out the current sense resistor.

Before changing the resistor, along with the display being way too bright in general, at the mid-high brightness settings the screen is almost completely washed out with almost no contrast.  My hope was that reducing the brightness would also improve the quality of the display in the higher brightness settings.  What I noticed, however, is that while changing the current sense resistor to a higher value does indeed reduce the brightness of the backlight, the display quality remains the same IE, although dimmer the display is still washed out in the mid to higher brightness settings.

Is this a symptom of the PWM method of dimming the display?  If so, would eliminating Q2 (in my diagram from replay #19) and connecting LED- directly to R2 eliminate this effect?  Could I then install a pot in place of R2 for brightness control?

Thank you.