Eleco reel LEDs fixed

[websherpa]

HE'S ALIVE! ALIVE I SAY!
http://www.websherpa.ca/images/hanab...i_reelfix2.jpg

I soldered in the second LED in a pair (as specified below) and now they both glow as bright as a full Don Chan Moon! But a lot toward the blue green side of white (and I think that's possibly because of the hyper orange "R" LED in the triplet), certainly they are much whiter than the "whitest" yellow working LEDs on the other reels.

Here's what it looks like with the door open and just the two middle on the right (third) reel fixed (above and below that the LEDs are now too dim to see). The top two leds of the first reel and the bottom two on the middle reel are working original LEDs.

http://www.websherpa.ca/images/hanab...i_reelfix4.jpgAt first I thought they might be a touch too bright, but actually they are a touch too blue and not red/yellow enough.

Here's a shot from the side so you can see the LEDs (working and partly working) a bit. Note that the working ones are yellow cast, the fading fast ones are bluish cast (but look bluer here than to the eye), and the replacement LED far right looks much intenser blue/green-white to the eye. On my reels the top leds on the middle reel are blued and go out (and sometimes on again), the bottom two yellow ones on the first reel blink as you play.

http://www.websherpa.ca/images/hanab...i_reelfix1.jpg


With the door closed, the new LEDs don't show as intense through Don Chan's red garment as do the original working "yellow" LEDs. And the blue contrasts a bit with the more yellow cast of the surrounding flourescents used. The Blue Fans glow nicely though, so it's not bad, just a little less intense and different and Don Chan's eyes look a little more intense and serious tinged in blue.

In this shot you can just tell the difference in the amount of red wavelength punching through Don's garment, but you can definitely see the blue cast from the one replacement pair (again, the LEDs above and below it on the third reel are out almost completely - and the photo shows a pretty close color rendition to how it looks to the eye).

http://www.websherpa.ca/images/hanab...i_reelfix5.jpg

and a closeup direct comparison of the two versions (old & new replacement). [ PS: If you read this far, it took me HOURS of play to get a Regular Bonus in order to take these shots for y'all! ]

http://www.websherpa.ca/images/hanab...i_reelfix3.jpg

After yesterday's disappointing start, this is elation, however. I think re-soldering the joints on the old LEDs (which I did try as well) did nothing good, yeserday or a re-try today except exacerbate the LED chips' existing problems. After playing with these I've got an educated guess that the problem is stress cracks in the LED housing caused by heat, vibration and age that cause intermittent failure, moisture infiltration, etc.

Now to go and replace the other four chips in the reel set and pray. My only problem, as this was an experiment and I did a miscount, didn't order enough chips to replace the entire three reels worth. Also, I'm also not sure if I like the blue cast and would prefer to correct the colour balance toward the yellow/red end of the spectrum, and that either means replacing some resistors (with help from those more knowledgeable), or trying a different LED (which I might do as well).

So the bottom line is that these are "close" matches for the original, but I'm doubting they are exact due to the spectrum output difference. It seems tha the red or yellowish cast was chosen and "tuned" in purposely with the resistor choices so as to highlight Don Chan's garment, and blend the main reel lighting with the surrounding flourescent illumination.

Kingbright makes another version with a more intense red AAF5060PBE-SURVG, but this wasn't a regular stock item from Mouser.com I've got a quotation order in to find out how much to bring them in on a special order.

Being able to use a regular one-colour White or Yellow high intensity LED would save a lot of money for this fix. Each of these LEDs cost me about US$4.49 retail (not including tax, admin and shipping), so three reel sets would require 18 chips or $80.82 in parts alone (less if I can get them in bulk locally). And if the spectrum doesn't match you'd have to replace them all, and since the old ones are precariously fragile, it's easy to lose working ones in the process.

Once I have ironed out the kinks and found a cheaper alternative, hopefully with fronzdan's priceless help I'll repost a more concise version of the fix instructions in a new thread.

Perhaps after the fix is done and tested for a while I may offer to solder new LEDs (or exchange boards if I get any spare LED boards) for those Pachilanders who need this fix but may be soldering iron challenged.... but for now I need to experiment more to get the right colour balance, and a cheaper supply of replacement LEDs.

[fronzdan]

Nice work! It looks great!

It's appears a little more bluish than the old ones but certainly acceptable. Right now there should be around 20mA on the blue LED branch. You can lower the intensity of the blue by reducing the current. If you swap the 820ohm for a 1k you should get around 15mA. The only thing is, the overall intensity will suffer a bit too. So it might look less blue, but it also may shine through Donchan less.

It might be worthwhile to breadboard two of these like the schematic and vary the resistances wtih a couple 1 watt 1K pots. Of course the price of the fix will go up.

I should have some white LEDs next week sometime. I'll try them out and post pics for comparison.

[slotter]

Thanks for those pictures, websherpa. Nice work!

[websherpa]

You can lower the intensity of the blue by reducing the current. If you swap the 820ohm for a 1k you should get around 15mA. The only thing is, the overall intensity will suffer a bit too.

Yes, I'd drag out a breadboard and try this though I don't have a convenient 24v source though I guess I could just tap off the existing power harness. Without doing the math (I hate it) I am wondering whether lowering the 750ohm resistors a little might also do the colour balance trick without losing intensity (at risk of over stimulating the LEDs?).

Upon closer examination I just discovered that my close-up examination picture is a little misleading, since my third reel lands a little off centre alignment (ok, ok, I admit it I set up the three Don Chans in a row by hand for the shot...) - if I carefully align LED, Don Chan and eye together I can see that although the original yellow is a better match all around and better red through the garment, the intensity of the old vs new LEDs is approximately the same and not as dim as the picture makes it seem, so that makes me even more curious to try out the OTHER Kingbright SURVG LEDs mentioned (but not currently available through mouser).

[noobie]

Does anyone have a way of measuring the color temperature of the LEDs? It might make it easier to find correct replacements.

[Darren]

http://i26.photobucket.com/albums/c1...ations_red.gif

NOW I'M GOING TO HAVE TO GET SOME FOR BEAST SAPP

Cheers
Darren

P.S. Great Pictures

[fronzdan]

You can lower the resistance of the R/G LED chain without overstressing them. They are being used well below spec. But I have no idea if you'd be stressing the device that is sinking all this current on the other side. Right now there's around 30-something mA total per pair. You could drop the current on the blue and increase the current on the R/G.

/What if you just made them all the same? Then there'd be nothing to show a contrast. It might even be nicer with the bluer tint all around....

[websherpa]

Well I set up a quick experiment on a breadboard, though I had some issues using trim pots to vary the resistances and may have blown one of my LEDs (the red diode, of course)... I think I fed too much current for too long. And I'm on an upward learning curve...

In any case, fronzdan you can see your schematic in action here - thank you very helpful.

http://www.websherpa.ca/images/leds_breadboarded.jpg

I set it up with a 110AC/24DCvolt wallwart that supposedly puts out 600 mA (I measured something like 18.9 volts and the original circuit pulls about 29.3mA - though I am using an old "CrappyTire" multimeter - maybe it's time for a new one - I have some old radios to refurbish and could use a capacitance meter - but that's another story.

I set the breadboard up with the original circuit and resistor values, and the new LEDs - it performed as the it does on the machine's pcb. Good start.

Second I added two mini 1k trim pots, bad idea, smell warming plastic. I think I slipped and one was rotated to no resistance, that's when I damaged one of my RGB LEDS (and now there are 5, one shy of a complete board darn it!). Rather than continue to monkey with two variables at the same time, I rememberd an axiom from electronics 101 and opted to fix one resistor and vary the other to start.

Next I added a 1k Ohm Trim pot in series with the existing 820 so I could measure how much resistance necessary trim the Blue diode back to the same perceived intensity as the others. I was able to acheive a visually white lite at close to 2K Ohm. A bit dimmer, of course because I'm turning down the current on the Blue LED. Now, do you think I would remember to measure the current feeding the LED pins so I could equalize that? No, I had it in my head that I was varying voltage by adding resistance (what a maroon) and took voltage measurements across the pins (in the original circuit, and now this). Doh, they're the same, of course. Diodes vary brightness with current and are biased by the resistors. Equal currents wont gurantee me equal brightness for 3 different colours, but the relationship between current and luminous intensity is linear, so halving the current should half the intensity.

For reference iin my test circuit the forward voltages were 1.6v across the Red, 2.8v across Green, and 3.2v across Blue, well within the specs of these Kingbrights.

Next I put back in the 820 Ohm resistor only on the B diode series and used a 1K trimpot on the R/G diode. I can increase the luminous intensity of the R/G diodes, but doesn't seem to do much until the resistance drops way down near zero and it seemed to stress the chip (I had some issues either with losing the contact points on the trim pots - which don't sit well in a breadboard, or perhaps the pot was just not very good. I tried some fixed lower value resistances, but I didn't have anything smaller handy (in the 1 watt range) than about 630 Ohms so it didn't make much difference. And next time I'll try measuring the currents across the LEDs and calculate the wattage so I can assure myself I'm not about to blow them up, or maybe use some theory.

The Absolute Max DC Forward Current for the Red branch of the LED is 50 mA (so am I correct in thinking that at 24volts I shouldn't lower the resistance below 480 Ohms?) - but it's tied in series to the green LEDs (AMDCFC = 30mA) - for 4 total or so I have no idea how that affects the lowest choice of resistor, possibly somewhere in between 480 and 800 Ohms? (help!)

Oh look what I just found: http://led.linear1.org/1led.wiz (also a multiple led calculator) and http://www.kpsec.freeuk.com/components/led.htm - what an amazing resource. Now that I have the actual formulaz I could do some better reverse calculations.

For 4 leds in series at 24volt source, 2.5 volt forward voltage (max for Red) with a 30ma forward current (max for Green) the "wizard" suggests a 470 Ohm 1 Watt resistor, so maybe 480 Ohms is as low as I should go.

Solution 0: 4 x 1 array uses 4 LEDs exactly
+----|>|----|>|----|>|----|>|---/\/\/----+ R = 470 ohms

The wizard says: In solution 0:
each 470 ohm resistor dissipates 423 mW
the wizard says the color code for 470 is yellow violet brown
the wizard thinks 1W resistors are needed for your application
together, all resistors dissipate 423 mW
together, the diodes dissipate 360 mW
total power dissipated by the array is 783 mW
the array draws current of 30 mA from the source.

The 820 Ohm 2 watt resistor is theoretically driving the Blue diodes near their max perhaps, so at least a 1k - 1.8k Ohm resistor would cut back the intensity a bit. It doesn't help me that the resistance varies logarithmically with the current.

OK, I didn't get enough time to finish the experiment properly, but I did try and learn a couple more useful things:

1) Other bright 3-4 volt LEDs work fine in the circuit. I tried both a 4000mcd and 7000mcd white LED (in my breadboard circuits) and either runs nicely off the 820Ohm resistor. Perhaps too bright, white light, not yellow (but also not blue), cheap, and I'll have to test, directly in the machine, but that leads me to observation #2;

2) It's likely that these type of RGB LEDs were chosen for 2 reasons, first the colour can be tuned as necessary (though they're not really being used to their full potential here - but I suspect that it has to do with matching the colour of the flourescents and because the LCD screen has a bit of a yellow filtering affect to it), and most importantly they have a very wide viewing angle of 100degrees, the average LED has a viewing angle of about 60degrees and on the ones I tested it's noticeably narrow, that may be an issue when it comes to backlighting pachislo reels.

I'm further convinced that the more red spectrum SURVG would be a better first choice replacement LED, though until I find a ready source to test, I will continue to tweak the SEEVG versions (well two banks anyway), and try out the cheaper single diodes if I can find a cheap white wide angle version?

Until next time. "Owe my eyes hurt." "Then why do you keep staring into those little lights?"

[websherpa]

/What if you just made them all the same? Then there'd be nothing to show a contrast. It might even be nicer with the bluer tint all around....

This brings up a good point, unless you're already used to the "mixed" coloration of dieing Eleco LEDs, it would be hard to justify fixing only one or two LEDs per reel since the colours won't match (at least until I figure out how to get them closer).

[Tulsa]

With all the holes, shouldn't it be called a swiss cheese board?

[noobie]

Would I be correct in assuming that if we could find the original parts supplier, none of this other tweaking would be necessary? I'd hate to see a fix for the machine that costs almost as much as the machine itself, not to mention having to add resistors and such to get it working.

[alstonj1]

at this point wouldnt it be easier to piggy back two white leds to get the viewing angle your looking for instead of using the tri color leds ?i mean since were going for a white light not colored lights as would be the real use of a tri colored led.and the cost of the white leds should be a much cheaper mod.

[vespacar]

Would I be correct in assuming that if we could find the original parts supplier, none of this other tweaking would be necessary? I'd hate to see a fix for the machine that costs almost as much as the machine itself, not to mention having to add resistors and such to get it working.

Even if you were able to get OEM parts, would that really fix the problem or just delay it. Anyway to ship the parts over and the cost of the parts would end up being higher that the fix that the members are comming up with anyway. Personally I think the blue looks better anyway!

Cheers,
James

[websherpa]

Would I be correct in assuming that if we could find the original parts supplier, none of this other tweaking would be necessary? I'd hate to see a fix for the machine that costs almost as much as the machine itself, not to mention having to add resistors and such to get it working.

That would be ideal (though replacing all the LEDS with the same 6 dip type will still be expensive - don't forget that the original machines may have cost more than $1000 to make in the first place, so the cost is relative to your need for speed!)

Also, a machine manufaturer like Eleco Aruze can order the LEDs to be made custom for their application (since they are ordering many thousands of units). Standard off-the-shelf values may not match. However, if we could contact a pachislo repair technician, they may have a source (if these machines are even repaired as opposed to swapping boards).

Replacing only those LEDs that are burned out would keep the cost down, but then the LED would have to be a very close colour/intensity match for aesthetics - and age may have taken its toll.

1) Having lived with the blue pair for a couple days now, it's not as objectionable, it starts to grow on you - but I still think it's not red/yellow enough for my taste. The LED 6 dip chip swap is fairly straightforward but at US$4.49 per chip (6 chips per reel board x 3 reels = 18 chips or $80 per machine) it's a relatively expensive fix.

2) It's possible that the more red balanced Kingbright (or another manufacturer's chip) is a better colour match and that would preclude the need to swap out any resistors, I'll test this once I can source them.

2) in a brief observation of the remaining "good" yellow LEDs in my machine, I think I've noticed that the blue diode isn't on or isn't being driven very brightly (compared to the "blue" dead LEDs) - I haven't pulled those reels yet for a direct observation - but if I am incorrect about what a "normal" colour temp state for the LEDs are, then possibly a more white or blue balanced colour is the norm. [Note: I am supposed to receive a better Eleco in a couple weeks, so hopefully I will be able to compare to that one], my instinct says a white to yellow light would work best as they are trying to pop through the "red" in Don Chan's garment, and simply balance the colour of the Reel backing with the surrounding flourescent lights.

3)After sitting down and doing the actual formulaic calculations, in the next phase of testing I am going to use these bluer Kingbright LEDs that I have, but swap the original R/G 750 Ohm 2watt resistor with a 500 Ohm 1 watt resistor (1 Watt should dissipate the heat with the 4 LEDs also in series along with the resistor should dissipate about 660mW), if my calculations are correct, this should raise the brightness of the R/G led's by 50%. (a value close to 433 Ohm would drive the green at it's maximum forward voltage and get a bit more yellow (red + green) into the mix). I'll leave the Blue diode brightness alone as I think I need to keep and increase brightness, not reduce it.

4) A cheaper fix could be acheived by replacing the existing RGB 6 dip LEDs with a single higher intensity (say 4 volt white 2000mcd with 60degree or better viewing angle, I haven't pinned down the needed intensity yet but I don't think more than 2000mcd is needed) white LED (again, I haven't put these in yet - so once I do, I'll report back on the perceived colour temp) connected to the original pins 1 & 6 in the PCB (the 820 Resistor should give the correct forward current (20mA) for a white LED. At about $1.00 a piece, this would make fixing all the boards only around $18. The disadvantage here is no way to tweak the colour temp and the narrower viewing angle may look unnatural.

[websherpa]

at this point wouldnt it be easier to piggy back two white leds to get the viewing angle your looking for instead of using the tri color leds ?

Yes, you could put two white 4volt LEDs in series (I seem to remember parrallelling leds on one limiting resistor is wrong - but I'd have to look it up), but each would be driven at half the current (intensity) as the one alone. But I don't think it would be as aesthetically effective increasing the viewing angle as simply sourcing a single LED that has a wider viewing angle (say 100 degrees).

[websherpa]

Personally I think the blue looks better anyway!

Cheers,
James

When I get more time, I'm going to try to set up three reels done differently to note the differences. The bluish light is cool in some respects, it does make the reel "glow" more behind the LCD display animations (which may not be what the manufacturer intended).

Given that you can use these (or the more red versions) and tweak down the 750 Ohm resistor for not much more cost to get them a bit whiter... there are several ready options depending on how much you want to spend, and how confident at soldering you are.

There's no guarantee that these "new" Kingbright LEDs won't eventually suffer the same fate as the originals (except maybe they will last longer with less use and abuse).

[noobie]

A cheaper fix could be acheived by replacing the existing RGB 6 dip LEDs with a single higher intensity (say 4 volt white 2000mcd with 60degree or better viewing angle, I haven't pinned down the needed intensity yet but I don't think more than 2000mcd is needed) white LED (again, I haven't put these in yet - so once I do, I'll report back on the perceived colour temp) connected to the original pins 1 & 6 in the PCB (the 820 Resistor should give the correct forward current (20mA) for a white LED. At about $1.00 a piece, this would make fixing all the boards only around $18. The disadvantage here is no way to tweak the colour temp and the narrower viewing angle may look unnatural.

This sounds like a much more reasonable solution. If you subtract shipping costs, I paid about $80 for my machine. (good old Gaku!). I'd be hard-pressed to spend another $80 to fix the LEDs as opposed to just buying another machine and using it for parts. No soldering involved that way, either. Although I suppose eventually I'd end up having to do it again. Ah well, you can never have too many pachislos, right?

Regardless of what the best solution ends up being, KUDOS to everyone here for all their hard work in making this happen.

[alstonj1]

Thats what the forum is for and why we come here and thanks for taking on the task websherpa great job so far im very impressed with your findings and sure you'll get it at this pace.plus exploring your machine this way you really get to know how it works instead of thinking of it as a magical box with light bulbs .by the way i didnt mean piggy backing the power through one resistor you could use the red and green resistors i ment piggy backing the light from 2 leds as apposed to the tri-colored package.giving a 120 deg viewing angle as opposed to 60 .

[websherpa]

Thanks Jeff,

This was/is certainly a collaborative effort, all of the clues and groundwork were laid by other Pachitalk members, I have simply collected them together, learned a little theory at others prompting, invested a little time and cash, and put the theory to some practical tests. I am motivated by my desire to learn about and "tweak up" these "great value" toys, and to share the knowledge (and reap the reward of others knowledge).

They talk about linking home computers together and "borrowing" cycles to create giant "thinking" computers, but the Internet's greatest acheivement has always been to link up and take advantage of the greatest computers of all, the human brain.

[websherpa]

Quick update.

I tested one of the old removed RGB LEDs against a new Kingbright (AAF5060PBESEEVG) in the breadboard circuit (hooked up as in the original circuit). What I can confirm is that indeed the failure is with the epoxy lens, putting point pressure on it will cause the contacts inside to mate and the diodes (if they're not burned out) will work again. When I got all three diodes lit, the resulting light was a nice pure white.

So assuming this isn't a coincidence because I had the old one in series with a new one, my original assumption about "yellow" lights is is wrong, the original LEDs are balanced "white" (and also not blue balanced), so"blue" is less of a crime, but white would be better.

Again I dialed up the pots but wasn't able to figure out a safe combination of reistances that would produce a white light mix. Any help here would be appreciated. But I suspect that unless the exact RGB LED replacement is found, then you'd have to use three resistors and unmarry the red and green diodes. There's still hope for the Kingbright SURVG package, but I'm not holding my breath. I have posted the problem and the circuit diagram to Kingbright's support department - we'll see if they can help.

I think the next thing to do is to try to source a low cost, wide angle (80-100 degree) 4v White LED (~1000mcd). Any ideas?