445nm diode MURDER thread

[ElektroFreak]

I am willing to donate two diodes as long as this project exists and as long as the tests are comprehensive.

I do feel that the data that will be gleaned from a dedicated murder test will be gleaned naturally in short order due to the sheer number of these diodes that are being bought, thereby rendering a dedicated murder thread moot, but as long as there is interest in the idea my offer stands.

 

[daguin]

So the project gets cancelled. EF is going to donate 2 diodes!

I'll gladly do the longevity test. We are just very spoiled by IgorT's torture chamber and the "anal" nature of his personality. I will NOT be producing the graphs and readouts he did. I will not be posting page long explanations of the data. I think that FlamingPyro has a tourture chamber as well, but I do not know what all it will record.

That being said, I am also with Mike on this as well. We are not "hired researchers" working to figure out what members here can do. As many peole as can, should build and test.

We have develped a false sense of confidence from IgorT's testing. The results from only a couple of diodes will NOT tell us what a "safe current" is. We need dozens of tests just to approximate confidence in the knowledge. Two is a good start, but it is not enough

Paece,
dave

 

[ElektroFreak]

I almost mentioned Flaminpyro, but I wasn't sure if he has an LPM..

Those are all very good points. The real testing is the complete compilation of data that will be posted over the next weeks by many members here.

 

[Deleted member 8382]

So, do you want to murder, not testing, murder some diodes? How many? By who?

 

[ElektroFreak]

Is "murdering" not a form of testing? There's no point in murdering the diode if the event is not going to be carefully observed and the parameters recorded. I have no plans on intentionally killing any of my personal diodes, so I won't be conducting any murder testing myself, and If we're killing diodes just for kicks then I might not be so willing to donate..

I had suggested Flaminpyro, but I don't want anyone to feel like this idea is an attempt to get others to do things that we all should be doing. This thing wasn't my idea, I'm merely offering diodes because Albert here approached me with the idea, and as long as it's for the community I'm game as long as there is some real point in doing the testing.

 

[daguin]

So, do you want to murder, not testing, murder some diodes? How many? By who?

I think that we are trying to "push" this decision prematurely. Let's see what FP says. He's not at home right now. I don't know what all he brought down with him. Would he even be willing to do it if he did bring his torture chamber down? If not, would my more simple proposal be acceptable? Is there somebody else that can step up to do it?

We have the generous donation of two diodes. We do not need to hype people up to donate. Let's slow down with the decision process. "Hurry up" rarely leads to good science.

A second consideration here compared to the 8X and 12X Murder tests is the relative cost of the diodes. We are not looking at $450 plus international shipping for two diodes with these diodes. There should be many more members that can afford to test one of these. The question that Mike puts (without actually saying it) is:

"How many of the members here are willing to actually contribute to the science and how many of them just want to be told what to do, without risk, so they can say the got a pretty light?"

Peace,
dave

 

[Deleted member 8382]

The point is exactly there. I believe that even if the diodes are very cheap, noone is going to push them enough to test if they die or not. I'm not talking about turning a pot until it pops, but what about 5 hours at a current, 50mA more, and like this until it pops? Since they are that cheap we can repeat it varying parameters almost as many times as we want! Sure people is going to provide their own graphs, but until what point? We thought they were 500mW, now they can produce 1W without poping. what if they are 5W? Until when would we notice?

It was just an exaggerated example, but it's obvious that some "offical testing" is always for good

I said Flaminpyro because he haves a torturer, if he agrees I think it's the best person to do it. Also, I would like this to be a real murder, not like the other ones in which after 200 hours of life we left it alive. I want it to die, to know the exact point at which we must stay away

Maybe we can even add 2 or 3 more by donations and make a test with 4 or 5 diodes?

 

[daguin]

The point is exactly there. I believe that even if the diodes are very cheap, noone is going to push them enough to test if they die or not. I'm not talking about turning a pot until it pops, but what about 5 hours at a current, 50mA more, and like this until it pops? Since they are that cheap we can repeat it varying parameters almost as many times as we want! Sure people is going to provide their own graphs, but until what point? We thought they were 500mW, now they can produce 1W without poping. what if they are 5W? Until when would we notice?

It was just an exaggerated example, but it's obvious that some "offical testing" is always for good

I said Flaminpyro because he haves a torturer, if he agrees I think it's the best person to do it. Also, I would like this to be a real murder, not like the other ones in which after 200 hours of life we left it alive. I want it to die, to know the exact point at which we must stay away

Maybe we can even add 2 or 3 more by donations and make a test with 4 or 5 diodes?

You missed my main point. You will NEVER get the kind of "confident" knowledge you are requiring, from only two or even four diodes. This is the "false" sense of confidence I talked about. Diodes are too inconsistent for that to work.

The following is going to be a MAJOR over-simplification

To discover where they will run a certain diode, manufacturers/researchers run MANY diodes at the same current. They look for the point where 50% of the diodes fail after the desired lifetime. When they reach that point (50% fail) they record it. Since the diodes in the projector were running at ~650mW and the projector is "rated" for 10,000 hours, it is probable that at 650mW about half of the diodes will be dead at 10,000 hours of use.

Understanding the way diodes are rated and the sometimes exgtreme variability in them, how can seeing that two diodes ran at any length of time at any current will assure you that any other diode will run for that time at that current. What if yours were ones that would run longer than "rated"? What if they were ones that would run shorter than "rated"? What if you got one of each?

Even IgorT's 12X experiment exposes the problems inherent in this kind of "research." Before IgorT got a 12X to run at the desired current, TWO OF THEM FAILED AT LOWER CURRENTS! However, instead of accepting this limit, we and he kept trying. We finally got one that acted the way we wanted and all breathed a sigh of relief that we had finally gotten a diode that was not defective. People now jumped on the higher "safe" current as proven by IgorT's testing. This despite the "fact" that 2/3 of the diodes tested failed well below the succesful one.

So how many diodes do we need to kill? At what current should we run them? How many hours of life at a given current will be acceptable? What good is "killing them" at all? We will know at what current those particular diodes failed after a given number of hours. However, what will that tell me about MY diode?

Peace,
dave

 

[Morgan]

Just to add Hallucynogenic... Some of the data gathered and posted from previous testing has proved to be questionable with later batches of some BRs like the GGW. Later batches have performed far below the original, "safe", limits. True, these may have been, as Dave put it to me once, 'Bottom of the Barrel', engineering examples but it showed that, with the best will in the world, initial testing didn't necessarily reflect a single diode bought by an individual.

With so many graphs and readings expected from all concerned, (and I suspect there will be literally hundreds considering other forums are also hungry for results), it really just needs an analyst to gather all the information and produce documentation from them.

I bought 1 initially from PL, (pre shipped to UK drove this decision), and then bought another in case I got curious to perform some detrimental tests. So cheap I didn't really have to think too hard about it. At $50, (with rumours of 1/2 that soon), more than enough data will be available. Just needs someone to collate it. IMHO.

There are already enough specs to get started and the, 'pretty light', is going to be released well below any dangerous limits too. Well, dangerous to the diode anyways! Start at just over 4V, 200mA threshold, (from figures I have seen but will confidently try), and climb from there...

M

 

[Krutz]

..yes, they may "enhance" the manufacturing process to have them closer to their requirements. in case this is cheaper, anyway.
the difference between the BR-diodes and the (whatwastheirname) 445nm ones is the relative power we drive them at. with the BRs, it could have been 4 times (wild guess) their rating, at which they were run in the sled. if they now use cheaper, less powerful diodes in the newer sleds, we will push them over their (new) limit.
here, with the 445nm ones, their original output in the projector seems to be 700mW (peak after lens). whatever happens, all diodes in any newer revision will surely survive those 700mW aka 100%.
engineering samples, which were never meant to be built into projectors, and maybe never meant to be run anywhere their (projector-) 100% are a different case.

and finally: not a single diode died, yet. with maybe 100 in the hands of excited experimenters. some of them have several diodes, and they are cheap enough to sacrifice some. didnt happen yet. maybe there will be one more surprise about those diodes soon..

manuel

 

[Deleted member 8382]

Krutz, I'll say it today. These are gonna go at 2W.

 

[Morgan]

Krutz, I'll say it today. These are gonna go at 2W.

Only with TEC cooling for sure.

@Krutz - Not to be misunderstood, I was talking about poor GGW performance due to engineering samples. I wasn't sure if I was clear on that.

I totally agree that because these new blues are of such high power to start with that safely driving them to shocking outputs, (for handhelds anyway), will not unduly stress them. I don't think we need to know absolute limits right away. We can all hopefully now get at least half a Watt for next to nothing with long, safe life! (Hmmm, wonder what the authorities will do about that one?!!!:thinking

M

 

[ReNNo]

Is there anyone testing this diode at 700mA? (only with heatsink)
Can this diode survive few minutes, few hours, or few days on this current?

 

[HumanSymphony]

Is there anyone testing this diode at 700mA? (only with heatsink)
Can this diode survive few minutes, few hours, or few days on this current?

People are already running these at 1A 700mA is probably the nominal current for these.

 

[Morgan]

With a small TEC one has been run up to 1W for 45 minutes inputting around 1A and was steady. Pushing it higher though started to see output drop away to 925mW.

Hope that helps. Not the answer to a simple heatsink output but gives clues.

M

 

[ReNNo]

Looks like HumanSymphony is right.
700mW might be nominal current