Dim LCD display - Sharp microwave

[Peabody]

The LCD display on my Sharp microwave is pretty dim.  All the segments work, but it's just not bright enough to see clearly.  For some reason I thought it was a VFD display, but it's LCD.  And what's more, I found the schematic.  Relevant pages attached.

So it appears the backlight consists of four green LEDs in series, and they are on all the time.  The display itself has 22 SEG lines and 3 COM lines.  Since all the segments are equally dim, I'm hoping the problem is the backlight, and not the display itself, the connections to it, or the processor.

Looking at the page 8 circuit, it seems something must be wrong with the LEDs, or the Q1 transistor, or with the power rails, which would probably be a capacitor issue (possibly the big cap shown on page 7).

I'd like some guidance on the possibility that I could actually fix this thing.  Does it sound like a backlight problem, or do LCDs go dim for other reasons.  I've seen videos on fixing individual dark segments, and that's usually a contact problem.  But I don't have any of those.

By the way, it appears the positive output of the bridge rectifier is connected to ground, and the negative output is what's called VR.  So it appears to a negative voltage power supply, at least with respect to the chassis.

Anyway, I bought this thing in 2000, so it's barely broken in, and I'd like to fix it if that's possible.

What do you think?  Any suggestions will be appreciated.

Sorry about the PDFs.  That's all I have.

[shakalnokturn]

A good start would be to estimate whether the backlight is bright enough by itself.
If it isn't check C1, C2, R4 (gone high?), LEDs they're in series so one defective can dim the string.

Edit: My bets are on C2...

[TheMG]

Most likely the LEDs themselves have gone dim. This is a normal failure mode for high-hour LEDs. 20 years turned on 24/7 is 175,000 hours. Especially if they are driven at or very close to their maximum rated current.

[Peabody]

I was able to remove C2 and check it, and it's fine.  So I'll have to rig up a separate AC supply so I can get some readings.  But at this point it really looks like it may be the backlight LEDs, or possibly the LED display itself.  Do displays go bad?  The dimness is uniform on everything, including segments that are almost never lit up.  So I keep coming back to the backlight.

Attached are pictures of the back of the display circuit board where the LEDs are, and a closeup of one of the LEDs.  If I need to replace the LEDs, how do I search for them?  I mean, they are surface mount LEDs, but beyond that how would you describe them?

[TheMG]

Most likely just standard 20mA SMD LEDs (you can measure in circuit the current of the existing ones if you want to be certain, it should be unaffected by the worn out LEDs).

For the sizing they follow typical SMD sizes similar to SMD resistors (ie 0603, 0804, etc). Just measure and compare dimensions.

Also nothing wrong at all with going with higher rated LEDs as long as they physically fit and the forward voltage is the same. Overrated LEDs will last a very very long time compared to lower rated ones operating at full current.

[james_s]

I'd bet money that the display is just fine. The problem is either tired LEDs, or a fault in the current source that is resulting in reduced current, in which case I'd agree with the person who said it's probably the capacitor. Those are just bog standard SMT LEDs, probably 0805 but I can't tell for sure from a photo. You can buy LEDs that will work in a rainbow of different colors, avoid phosphor based ones (mostly white) for applications where it will be on all the time as the phosphor tends to wear out.

[Peabody]

Yeah, it was a real disappointment that C2 turned out to be ok.  That would have been a very simple fix.  So I'll have to rig up a temporary mains supply to the board so I can measure voltage when it's running, and see what's going on.

I wonder if someone could explain the LED driver circuit to me (picture attached).  Is it a constant current circuit?  If so, what is the function of R11?  And why would constant current be needed?  Is it in case one or more LEDs fails short, or is there some other reason?

[Peabody]

I was able to do some voltage readings, and based on the drop across R1, the LED current is about 6.2mA.  The average Vf per diode is 1.95V.  They are just plain green LEDs as far as I can tell.  They measure 1.8x1.2 mm, which appears to be 0805.

I tried adding a 10K resistor in parallel with R3, then a 470R in parallel with R1, but neither had  any noticeable effect on brightness.  So maybe the LEDs are putting out all they can.  But 6.2mA doesn't seem like very much.

It would be nice to change the circuit to increase the current through the LEDs and thereby increase the brightness without having to replace the LEDs, but based on my experiments, that doesn't look like it will work.  Or maybe I just wasn't aggressive enough with my parallel resistors.

Looking around at Digikey, I found only one 0805 green LED that appears to have exceptional efficiency - a lot of light intensity for your milliamp - and it has about the right Vf.  However, it has a domed top whereas the orginals are flat.

https://www.digikey.com/en/products/detail/kingbright/APTD2012LCGCK/7043073?

So I will probably include four of them on my next Digikey order unless someone has a better idea.  I will say that everything in the circuit seems to be working fine, including the Nichicon electrolytics.  But the LEDs are just too dim.  My memory may be faulty, but I don't remember feeling the display was too dim when the microwave was new, so it seems the LEDs are just getting old, and can't perform anymore.

[james_s]

6mA should be quite bright with most modern green LEDs. Keep in mind the intensity is going to depend on the spread, so that one may have exceptionally high intensity *because* of the domed top focusing the light. In a pinch 0603 LEDs will usually fit ok on 0805 pads.

I'm not quite sure how that driver works, it might be easier to tell if more of the schematic around it was present.

[floobydust]

I would replace them with different LED's - that APTD2012L has a bubble lens and is narrow angle 40°, but your old one is flat lens wide angle 140°. You need that if the LCD is lit from the back, not the edges.
The transistor is a simple constant current source- but it looks lazy, you should be seeing around 18mA, not the 6.2mA you are measuring. It might have low hFE due to old age. I would replace it.

[Peabody]

I see the driver circuit as kinda constant current, which may be designed to provide a standard brightness with variations in Vf of the LEDs, but with the addition of R11 to carry most of the current, and the heat, and allow the transistor to manage the current at the margin without having all the current going through it.

But what's interesting about floobydust's comment is that if the transistor wasn't there, the voltage at the bottom of R1 would be 0.52V, which is very close to what I measured.  So the transistor may have a truly degraded hfe, and in fact could actually be completely dead.   Well, not completely dead because there is clearly some base current, but maybe dead in the sense of no collector current.  However, I don't understand where the 18mA number comes from.  What's the math for that?

With regard to the lens on the LED I picked, you guys are saying the high mcd spec results only from the dome focusing? I didn't know the rating worked that way.  Well, in case it really is brighter than the others, would it be possible to cut the dome off with a razor blade, or melt it off?  I'm not sure how much spread I need.  The LEDs are an inch behind the display, and there's a plastic diffusion panel behind the display that seems to spread things out.

I will breadboard the driver circuit with a 2N3906, using a 12V/5V supply with 3 LEDs, and see if anything interesting turns up.  Of course if the dimness is caused by the transistor being dead, I can fix that by adding another resistor in parallel with R11, and just let more current through.

[floobydust]

You can do the math or simulate it in LTSpice, which is easier. R2/R3 junction would be -2.65V if Q1 base presented no load to the voltage divider, and assuming hFE around 200 around 0.1mA normal base current, it's up to -2.44V but you measure -1.22V which is 0.6mA load from the transistor's base. That's too high to only get 4mA collector current. R11 adds 2mA to give the 6mA total. It's an odd resistor there, it just makes the current source vary with the raw DC rail, so mains voltages fluctuations would be seen a little bit in the brightness moving around.
So I figure the transistor has aged, at 8,760hrs/year the parts are running 24-7. I'd change the LED's out and the transistor.

[Peabody]

I don't understand 4mA through the collector and 2mA through R11.  I calculate the total current through R1 at 6.83mA, and the base current at .61mA.  That leaves 6.22mA for the LEDs.  But the voltage drop across R11 would produce current of 6.31mA all by itself.  So it's not clear to me that the collector is contributing anything at all.  (The numbers don't quite match, but the parts probably aren't 1%, and my meter may not be all that accurate.)  Could you explain your 4mA/2mA values?

I'm tempted  to just reduce the value of R11 to get more current through, and hopefully be done with it without a Digikey order.  But I guess I shouldn't be modifying the circuit design of a microwave oven.  The transistor is available in low-gain (hFE = 120 - 270) and high-gain (hFE = 180 - 390) versions.  No way to tell which was used originally.  Do you have a preference?

I'm still not clear what to do about the LEDs.  There aren't many green LEDs with Vf around 1.9V, and I don't know if the circuit will still work with a higher Vf LED.

[Peabody]

I breadboarded the LED driver circuit using three green LEDs with -12V and -5V rails, and a 2N3906 transistor.  With the collector disconnected, simulating a dead transistor, I get essentially what I got on the display board - about .5V at the emitter, 1.2V at the base, and a bit over 6V at the bottom of R11.  And dim LEDs.

Then I connected the collector to the circuit, and the emitter voltage increased to 1.56V, the base became 2.26V, and the bottom of R11 dropped to 5.58V.  And the LEDs became much, much brighter.  In terms of current, I have 19.02mA through R1, less .12mA base current, leaving 18.9mA for the LEDs (up from about 6mA before, so no wonder it's brighter).  It looks like R11 is passing 4mA, and the collector is providing 14.9mA.

It's pretty clear to me that the transistor has failed open.  The EB diode still works, but I don't think there's any collector current.  And by the way, whatever the function of R11 was supposed to be in terms of controlling current, one benefit it provides is leaving the user with at least some display, although dim, if the transistor should fail.

I think there's a chance the LEDs are still ok.  I found a brand new 2SA1593T in my junque box, which is total overkill for what's needed.  But I have no other use for it, and if I can physically remove the transistor from the board and solder wires to the pads, I should be able to wire this new one in.  And I can worry about the LEDs after that if the display is still too dim.

[james_s]

If the current is really that much lower than it should be then the LEDs are probably ok. Removing the transistor is easy, ideally you'd use a hot air pencil but it can be done with an ordinary soldering iron too, just heat the joint on one side and gently lift that lead out of the solder, then heat the other two pins at the same time using a wide enough tip. Or if you're confident that it's bad, you can snip the leads off and remove them individually.

I'm actually surprised they bothered with the current source, a simple resistor in series with a regulated power rail ought to have worked fine, but maybe they wanted to make sure it would look consistent with LEDs from different batches.

[floobydust]

Just to back up a bit, the original sim was hasty and I wrongly used pure green/white/blue LED's with high 3.3V Vf and the current-source was starved.
The (posted) sim is using yellow-green AlInGaP LED's with about 2.1V Vf. That shows 6mA LED total, the transistor is pretty much doing nothing, numbers match your measurements. It's hard to know the exact failure mode of the transistor, it seems to be taking base current without doing much for collector current, as if the hFE is below 10. Any general purpose transistor would work, I prefer higher gain parts around 150-200 up.

What I noticed is the current-source barely has enough compliance voltage to power four high efficiency (pure green/white/blue) LED's with their higher Vf.
Around 3.2V each consumes ~12.8V of the 14.65 you have and works giving ~20mA. That is 1.64V lost on R1 82R, and 0.15V only across the transistor C-E and R11.
But when I went up to 3.3V per LED (someones meh LED model), the current-source did not have enough output voltage giving very low current (R11) for the LEDs.
The problem is the 14.65V raw DC will sag when the oven is running, or if mains is low that day. You don't want the LED's cutting out (dim with R11 only).
It was tempting to put in white or pure green LEDs as they are much brighter but would not work well and who needs the display lighting up the entire kitchen 

[wraper]

LEDs do degrade. I've seen severely degraded LEDs in car dashboards which became so dim there was barely any light from them. And not like some individual LED but all of them altogether. 6mA vs 20mA will not make that much apparent difference in brightness. Also area around the LED die does not look right.

[Peabody]

I'm actually surprised they bothered with the current source, a simple resistor in series with a regulated power rail ought to have worked fine, but maybe they wanted to make sure it would look consistent with LEDs from different batches.

It's not a regulated power rail.  It's just a transformer, bridge rectifier and big cap.  And it has a good bit of ripple.  But then the 560R resistor and second capacitor after the LEDs smooth things out very nicely.  In my breadboard test I tried it with five LEDs as well as three, and the brightness didn't change noticeably.  So the driver circuit seems to handle even extreme Vf variations quite well.

Just to back up a bit, the original sim was hasty and I wrongly used pure green/white/blue LED's with high 3.3V Vf and the current-source was starved.
The (posted) sim is using yellow-green AlInGaP LED's with about 2.1V Vf. That shows 6mA LED total, the transistor is pretty much doing nothing, numbers match your measurements. It's hard to know the exact failure mode of the transistor, it seems to be taking base current without doing much for collector current, as if the hFE is below 10. Any general purpose transistor would work, I prefer higher gain parts around 150-200 up.

What I noticed is the current-source barely has enough compliance voltage to power four high efficiency (pure green/white/blue) LED's with their higher Vf.
Around 3.2V each consumes ~12.8V of the 14.65 you have and works giving ~20mA. That is 1.64V lost on R1 82R, and 0.15V only across the transistor C-E and R11.
But when I went up to 3.3V per LED (someones meh LED model), the current-source did not have enough output voltage giving very low current (R11) for the LEDs.
The problem is the 14.65V raw DC will sag when the oven is running, or if mains is low that day. You don't want the LED's cutting out (dim with R11 only).
It was tempting to put in white or pure green LEDs as they are much brighter but would not work well and who needs the display lighting up the entire kitchen 

In my breadboard test with three green LEDs, I got 1.56V at the emitter, 5.58V at the collector, and 2.27V at the base.  That's net LED current of 18.9mA, which is 4.02mA through R11 and 14.88mA through the collector.  Average Vf was 2.15V.

But then I added two red LEDs.  That gave me 1.43V at the emitter (so less total current), 1.9V at the collector, and 2.16V at the base.  That's net LED current of 17.27mA, which is .47mA through R11 and 16.8mA through the collector.  Average Vf was 2.03V with the reds in there.

So the curcuit adjusted pretty well for the additional LEDs, but with only .47mA now going through R11, adding a sixth LED might be close to not working at all.  But what surprised me was that I couldn't detect any difference in brightness in the green LEDs when I added the reds.  Either the difference in current wasn't enough to notice, or the green LEDs reach their maximum output at a lower current.

LEDs do degrade. I've seen severely degraded LEDs in car dashboards which became so dim there was barely any light from them. And not like some individual LED but all of them altogether. 6mA vs 20mA will not make that much apparent difference in brightness. Also area around the LED die does not look right.

In my breadboard tests, going from 6mA to 19mA made a big difference in brightness.  Of course those tests were with new, modern LEDs, and the ones in the microwave may not behave that way.  But they have been coddled at 6mA for quite a few years now, so maybe they still have some life in them.  In any case, surely the right approach is to replace the transistor and see what happens.  I can do that without having to order anything.  If I ultimately have to replace the LEDs, I can do that, but since I have to replace the transistor anyway, I'll do that first and see if it solves the problem.

[Peabody]

Replacement of the transistor worked, and the display was somewhat brighter, but not a lot.  However, I managed to melt part of the flexible circuit board connecting the keypad to the board, and there's just no way to repair that.  So I'll have to get a new microwave.  I'm usually pretty good at repairs, but just wasn't paying attention.  Anyway, I got 20 years out of this thing, which I guess is ok.

I did test the old transistor, and it had no collector current, as I suspected.

[james_s]

Well that's annoying.

I wouldn't be so quick to give up hope though. I came across a damaged flexible keypad strip in an TDS420 scope and I was able to fix it with some tape and then masked and painted a line of carbon paint over the damaged area. Put a piece of tape over that and called it good, works perfectly.