Bulbdial - Was working fine - now has odd LED groupings

myx_2000

Hey all,

I got my BD kit for Christmas. I built it Christmas night. Everything has been working great until this afternoon. I noticed the clock going orange occasionally. Once I finally got a chance to look at what was going on, I realized that the 12 position blue LED was not lighting. I went into the alignment mode and found that all sorts of stuff was now goofed up. 

Things like on Blue... the 10ish position both this and the 10:30 position LEDs light ( 10:30 being brighter). At 10:30 , 10 and  10:30 lights with 10:30 brighter, 12 straight up sort of flickers but is off mostly and the next LED lights and so does the 12 and 9 position on the red ring. There are other combinations of blue LEDs on at the same time. There are many points on the red ring where a red LED will light but just barely. 

On Green from 6-11:30 are fine, then from 12-6 there are 2 LEDs on at a time. They are spaced by apart by 6 LEDs. They move together each press with the proper LED being the brighter of the 2. Then as it gets farther around the dial the dominant LED gets brighter and the earlier LED gets dimmer. Not sure if this makes sense. 

On Red, 12-3 are goofed up. 12 and 3 are lit together and 1 & 2 are lit together. This is the same with the proper LED being the brightest of the 2. Also this is the dimmest doubling of the 3 rings where the dimmer of the 2 lit is nearly off. It is on by just a hint. Normally I would have suspected a soldering error (which hI have not ruled out), but I was able to fully align everything and have it all work for the first 2 days.
I have not played with the code at all, so it is the source that shipped with the clock and I seriously doubt that this is a chronodot issue. 

I have no idea where to start on this one. Any ideas?

myx_2000

After submitting this, my wife was asking what was wrong. As I was showing her, I noticed that it was different combinations that were goofed up. LEDs I just stated as being out (blue 12 straight up) now light and others are mis grouped. It is like a charlieplex mismatch or something. 

Windell Oskay

If just one of the ten LED lines is not wired up correctly-- for example, if a solder joint has failed --you will start to see these kinds of funny lit-up patterns.  And yes, it's an effect of charlieplexing. 

When you're describing the LED positions, I'm not sure if you're describing the locations of the LEDs that are lit up, or if you mean the opposite-- the LEDs that light to indicate those times.  But in any case, you should check each of the vertical LED line jumper wires, and especially the connections at the microcontroller and the resistors-- those are the most likely places for the issue to be located.

myx_2000

Good point. I meant the actual physical bulb, not the projection bulb. After I posted I realized that being that there are 30 LEDs, that my 10:30 example didn't really work either. But the approximate position.I have done a visual inspection of the boards and everything looks good, so I will re flow the boards a ring at a time starting with the blue and hitting the resistors first and testing in between steps. Do to a mis following of instructions, I had not read not to snip the vertical resistor leads on the green board. Having been in bass snipping mode due to all the LEDs, I snipped them. After realizing my mistake, when I got the red ring in place I used several snipped LED leads to bridge between the green and red boards. Those were the first things I went after to test as I suspected a bad connection.

Actually I am curious why you chose to use 0 Ohm resistors instead of a wire jumper both on the board and between layers. It is not at all a criticism, just curious from a design standpoint. I am in the middle of a personal board design and I am always interested what brought someone to a decision point.

Thanks Windell

Windell Oskay

If you have a multimeter handy, you might check to see if there's a short circuit between lines LED1 and LED2, or between either of those lines and LED10.    There could be a solder blob or stray wire somewhere causing the problem.   I'm having a little trouble following most of your specific LED locations, but that would be my guess from your description on the red ring. 

The Zerohm jumpers really are a specific type of wire jumper.  They are not actually resistors; we did not choose them "instead" of wire jumpers.    They are inexpensive, easy to count out, convenient, easy to identify, and their shape makes step 33 of the instructions pretty easy.   Without that "bead" in the middle, we'd need to figure out some other way to hold them in place, and make sure that there was enough length on both sides.

myx_2000

Hey Windell,

Ok, I spent the last hour reflowing all the solder points. While it solved the red lights coming on with the blue.

But it seems that it made matters worse for the blues. There are many more blues on at the same time and they also space out like the greens are. Now D21 is nothing but very dim. No flickering. I thought it might be that it was reversed or something, but the LED innards are all identical all the way around the clock. I checked this for all three rings. Blue and Green and all the same. Red runs the opposite direction even though the polarity stays the same. I checked my cell pix that I took during the build (I was posting them to Facebook). I checked to verify that the leads were in the proper way. Anyone who says that high megapixel count is not important on cellphones can suck eggs. It is these details that keep me from pealing the paint from the walls in when troubleshooting this kind stuff. 

Sorry getting off topic. Since this is not clearing up easily, I will start to document which specific components are giving me headaches. Also, if I were to record a small video and post it up on YouTube, would that be helpful at all?

myx_2000

BTW I have been metering places where I can figure out where the traces go, but there is a lot of flipping and flopping here. I did meter all the vertical 0 ohm lines and they all seem fine.
Also while reflowing everything, I checked for miscellaneous blobs, wire, chunks -o- matter and can not find anything out of the normal.

Windell Oskay

If it was working properly to begin with, you can be assured that all of the LEDs are facing in the correct directions; they don't spontaneously reverse direction.  And, if you really need to look, the right way to verify the orientation is by looking at the flat face on the collar of each LED, rather than at the internal wireframes; those vary between different types of LEDs.

Reflowing all the solder joints is usually *not* the way to find a problem, but it can easily create new ones if you overheat the LEDs, damage the traces, or create dry joints out of good ones.   Instead, try to look and see where the problem actually is, and only resolder where absolutely necessary. 

I'm not sure that I understand what you mean by "flipping and flopping." You should measure the resistances with the power off; nothing should be changing while you do that.

If LED D21 is very dim, that could also point to a possible connection between lines LED1 and LED2.  Please check the resistance between these lines; they should not be connected.

A youtube video would probably not be helpful.  Checking the resistance between the LED lines would be.

myx_2000

Sorry, work kicked in and I haven't had a chance to work on this more until tonight. 

The flipping and flopping was referring to trying to follow the traces above and below to figure our what goes to what. 

The reflowing was done quickly. Sorry. Won't happen again. ;0) My thinking was that in running through all the points, if there was a short, blob, stray wire, unsoldered joint connection... I would see it. 

LED1 to LED 2 shows 690ish OHMs.

Just out of curiosity I checked between all the LED lines (not D14, D24, D34 etc, but LED1-LED10) and there is quite a variety here. There are some that show OL, several that read between 30k-50k and LED1-LED2 being the lowest at 600-700, but I am just shooting in the dark. Windell, I want to troubleshoot this thing and not waste your time, but charlieplexing is always a mess when it goes south. But, just being honest, I do not have a clue where to begin on this one.

I am also questioning myself here. I would have swore that this was working properly when I was done because there is an entirely different look when one LED per color is on vs. multiple LEDs on. But... Aside from ooohing and ahhhhing at it, it was just sitting there. So, I do not know what the heck happened. 

Once I get this finished and working again, is it possible to get a clear front plate cut without the circle hole in the center? I am noticing that dust is getting in there and I want to seal it up as much as possible. I am looking very seriously at the rear projection clock face so the absence of a hole will not detract. If this is possible, what would the cost be? 

Thanks

Mike

Windell Oskay

Yes, 690 ohms sounds very low.  It definitely sounds like there is something wrong there. However, it may be hard to diagnose, because it seems as though something has failed partway-- not all the way open or shorted --either of which would make this pretty straightforward.   In the absence of any better guidance, and if D12 seems to be working more-or-less correctly, I'd suggest replacing D21, as it seems like that is the likely point of failure.   

If you'd prefer, you might want to consider sending it to us for analysis; we'd be willing to replace that LED for you, and figure out what the problem is otherwise.   

As far as the faceplate without a hole goes, you may not need one.   A common procedure when installing the rear-projection plate is to use the original opaque clock face to cover the big hole in the other side.  Just use it instead of the washers on that side.

myx_2000

Good idea for the hole, but I am in love with this massive array of LEDs. Also if someone asks how it works, the clear is ideal for this. But I won't worry about this for now. I still have an extra LED from the build. I don't mind replacing it. If it is still goofed, then I could send it to you. Or are you interested in why it is goofed at this point? I won't mess with it until I hear either way. Thanks.

Windell Oskay

It's better for everyone if you can get it fixed on your own, but if you do get really stuck, we're here to help.

myx_2000

Hey Windell,

Sorry, I have been crazy busy with work again. I finally got a little time today so I replaced D21 like you recommended. I have the same result. D21 does not light and I still have several sets of lights that light simultaneously. So, I need to ask for further help. Thanks.

Windell Oskay

So, your two basic approaches at this point would be either (a) carefully document what the actual status is right now, or (b) send it back for us to take a look at it.  Seeing as we're getting on several months now, I'd tend to advise option (b) at this point.   If you'd like to do that, please contact our store directly.

myx_2000

Right. Sorry. Message sent.

Windell Oskay

Received clock today and found the problem:  D12 had an apparent connection between the leads.  Replaced it, everything seems to be OK now.

myx_2000

Wow. Thank you Windell. I received the clock today. Was this something I could see? I am perplexed how this would have happened after it was all put together. I promise, I really did look all over to find the issue. Hopefully this did not take you long to fix. 

So one LED was enough to goof all that up? So, the charlieplexing can be a blessing and a curse all at the same time. But none the less thank you. It is SO cool. 

Windell Oskay

Well, we knew from the set of LEDs that were apparently connected that there was a partial short between LED1 and LED2, or a defect at D21 or D12, the two LEDs between those lines.  When we received it, I checked those two LEDs.  D21 worked fine, but D12 did not, so I clipped one of its leads to see if that LED was the issue, or if it was external.  Turned out that even isolated, D12 would not light up, so it was apparently the culprit.  

I'm not sure what went wrong with it; it could have been stressed during lead-bending or soldering.  But, I'm glad that it's working now.

myx_2000

Got it (mostly ;D) Thanks again. 

myx_2000

Hi... It's me again...
Hey, I have had this on my right speaker in my living room since it returned. This morning I noticed that I had orange for a moment. Going into alignment mode I found that LEDs D46 and D64 are now out or barely lit (it seems to fluctuate). But this seems to be causing double lighting of other blue LEDs. It testing mode the problems only exist in the blue ring. Please advise. Thanks.

Windell Oskay

So, you may have a similar problem at one of those two LEDs.  Those two lines are only involved in the blue, as you can see here:  http://www.evilmadscientist.com/2010/on-the-design-of-the-bulbdial-clock/