I was about to write a long post explaining what should and should not be done and why, but i have a whole lot of work to do and two weeks worth of emails to answer, so i'll be short.

I never intended to stop the first experiment for anything other than re-plotting, nor change it's conditions... In this hobby, the "useful life" of a diode ends when a diode pops... And we usually don't change currents after a laser is built.

The experiments are meant to give builders an approximate idea of what to expect. And that's why the experiments should relate to what we usually build.


Also, the 100h mark was just one of my suggestions for a standard of "reliability". In the meanwhile i realized even half as much is plenty for a hobby pointer.

But i don't want to stop the test because i want to know:
- How long the diode will survive this current
- How fast it will degrade with time
- How much it will degrade before dying

All these bits of info will be of EXTREME importance during the following tests, including that of the 12x!

And we can't get to them, unless i leave the first diode as it is, and test it often enough not to miss any big changes in it's behavior.

Since degradation is slowing down, i won't even re-plot it until the 60h mark, unless of course i notice a kink developing, which is why i still check it's beam profile twice a day.



I was mostly wondering what current to test the second diode at. And when i thought about starting the second experiment sooner, it did not imply stopping the first (the torture chamber can handle multiple diodes at once, if i add sensors and counters).


When it comes to the second diode, there are several things that interest me.

Obviously i want to know how long it can live at higher currents/powers.
But believe it or not, i am just as interested in how long it would survive at the same current (i'll explain why)!

Problem is, i can't test both. And testing higher currents/powers is of interest to many of us. That's why i was thinking of a middle ground. A current, which would give answers about higher powers, but not make the second experiment completelly unrelateable to the first one...


And the reason i can't test both - as Dave already explained - is, that if the diode is tortured at one current first, and then the current is raised, the data from the second or third current does not tell us how long the diode would have lived there! It only tells us how long it lived there AFTER having been tortured at lower currents first.

Unfortunatelly every diode will degrade at a different rate and to a different level before dying, making it impossible to extrapolate data from a single step. In fact, we've seen that degradation slows down with.... ....degradation, making things even more complicated!



Unfortunatelly, my experience with torturing and killing diodes is mostly limited to PHRs, a small number of 4x's and some GGWs...

PHRs would usually degrade by around 3% before dying, when overdriven hard.. The 4x i cycled seemed to degrade a lot, but it fell into a kink, making the numbers meaningless.. My first GGW degraded almost 5%, but then i set the current higher, after which it died rapidly, and now i don't know how far it would have degraded if i left everything the same...


With GGWs, i can rely on the efficiency being a good sign of health, there higher efficiency diodes will outlive lower efficiency ones, when set to same currents. With PHRs (and 4x's) kinks made things a little more complicated.

But since those were all reject sleds, and evidence exists some of them were even prototype sleds, i can't say for sure the same will hold true with 8x's from drives...


It's unlikelly, but in theory at least, a higher efficiency diode could kill itself faster, due to the higher optical flux it produces at the same current.
While i'm pretty sure efficiency is a sign of diode health, it's not the only factor involved...


I have killed many diodes and quite a few in stupid ways, but none of them cost almost $200 a pop...


We have a very limited number of diodes available for sacrifice to the lazor gawdz, and that's why we need to squeeze them for as much info as possible.

But for the same reason, we also have to squeeze very carefully.


P.S. And yes, this was the short version.

gz 4k posts

Latest Po(@time)/(Po initial) versus minutes curve.

hmm, do we even know if diodes degrade at their rated output?

manuel

Of course they do. That's why they also have a rated lifetime.

Peace,
dave

..if we knew how much diodes degrade at their rated output (datasheets), we maybe can find out what an unknown diode is rated for? i am sure manufacturers rate their diodes depending on their degrading, so the degradion-to-rating-formula should be somewhat common for all diodes?

manuel

Uhm, sorry, but i doubt about that ..... it may be too easy

Seriously, no, i don't think, the diode chips are produced from different manufacturers, and the degradation have too much variable and possible causes, for this ..... maybe if you can test 20 or 30 diodes for each serie, then at the end you can have a medium-shape curve valid for all the serie, but just from one or two diode, i doubt it works .....

That would, first, require us to be able to really identify the diode , and second, require us to be able to get the datasheets from the manufacturer.

We have been very ineffective at accomplishing either of those tasks

I wonder if PBD's university association could convince the drive manufacturer (Pioneer and/or LG) to give us that info . . . . . .

Peace,
dave

but but but!
but it should be possible to find out how diodes degrade at their rated output? from the manufacturer, datasheets, pressreleases? i expected its simply stated in datasheets?
we dont have to identify unknown diodes. we simply try & error until we know at what output they degrade like 1% every 100 hours? (if thats what the other datasheets say about rated output)

sure, if we cant find out what the "rated degradion" is like, we are out of luck again.

perhaps the datasheet wouldnt help too much anyway? we want to know at what output the diodes survive many hours, not at what rated output they survive several thousand hours..

manuel

All of that begins with the correct identification of the diode and getting the datasheet. "8X" is not a "rating." Without those two things all we have is what we do. We buy what we can afford and we experiment with them. This "fund" and the 12X fund are the first time I know of that the forum members have thrown together this way. Usually it is one or two of us taking the chance and blowing the first couple of diodes.

Peace,
dave

The diode just reached the 60h mark!

Just stopped it for a quick test, to record degradation patterns, but the diode still seems to have some life left in it!

Couldn't be happier with the results so far!

Igor;

60 Hours is very good news.

This diode is surprising in durability.

It should make 100 hours at the current rate of power loss (~7.5% @ 60hr.).

I am starting to think that other problems,
like electrostatic discharge are responsible for some premature "in the field" failures.

LarryDFW

I was thinking about the same thing, ESD damage. I'm actually considering connecting a Lasorb, or maybe IgorT has another idea on the matter.

bryce007;

Based on some precautions from the laser diode manufactuers,

I have been using Schottky diodes wired opposite the normal current flow thru the diode.

This limits static discharge to less than 1/2 of a volt in reverse polarity.

Diodes usually have a maximum reverse voltage of a few volts.

LarryDFW

That sounds like a good idea, thanks!

*poke*.. i'm still alive... somewhere.

Couldn't do this sooner, too busy..

Here are the 60h degradation plots:






Diode Stats:
- Ith = 37mA
- Po @ 300mA = 330mA (92.7% of initial power)
- Slope Efficiency = 1.254mW/mA
- Actual Efficiency = 18.81% Peak / 18.51% Avg.



Degradation has slowed down to a point, where it's hard to discern PI plots, even tho almost 20h passed since the last one.

The slope efficiency was actually calculated to be the same as before. The only reason the new plot ended somewhat lower is, that Ith was measured higher than before.
Basically, the slope is the same as before, only difference is, that the entire plot is shifted to the right by 1mA....

Interestingly, the efficiency plots show much more of a difference, but i should mention, that the previous one (yellow line) doesn't fit in.

This is most likelly due to temperature differences. It resulted in some errors, that are hard to explain.

The temperature was higher when i recorded the 42h plot - 26°C instead of 21°C. This should have resulted in a lower efficiency. But the Vf drops first as the temperature climbs, while the power drops much more slowly. Due to a lower Vf, the El. input power was lower, while the output power was almost the same, so the calculated efficiency actually came out higher!

This is somewhat confusing, i'd expect it to be the other way around...
I'll have to do some plots of a healthy diode at different temperatures, to explore this behavior in more detail.


Anyway, this is it for now, the diode is still alive and well. I might have to spread out the re-plotting intervals even more, to detect measurable differences next time.


Back to work for now, otherwise i'm getting parts to expand the torture chamber tomorrow...

It looks like each step down it power output takes twice as long as each previous step.

I think that there is definitely something else in the die that is degrading besides the front facet...

latest degradation plot

It looks like it is beginning to curve out again near the bottom. Probably that means it is beginning to enter the last phase of degradation before it pops.... just my opinion...

That assumption appears to be absolutely random. What causes you to think this?

Peace,
dave

The first 3 or 4 points of data in CHP's graph have a downward curve until the actual data line matches the projected line. Then the last three points start to curve outward again slightly, indicating that the degradation is slowing down less than it was before. Does that make sense? At least, that is the way I interpreted it....

That is a fair description of what is happening. My question is about why you think that this "flattening out" indicates the end of the diode? AFAIK we have never observed this before. The facts are not being questioned. The interpretation that this is "the last phase of degradation before it pops" is.

Peace,
dave

25mW down from 0-60... is that good, great, ...or bad?

Uhm ..... depend ..... 25mW of loss on that power, in all that time with the diode overloaded, looks good ..... but the real question may be made if it is good or bad respect what, exactly ? ..... i mean, this is the first diode cycled and plotted so long time (at least, the first one for which we have data), so we don't have any other data to compare with this one and try to guess ..... if the next one die in 10 hours, these results are good, if the next one last the double or loss half of this, the result is bad ..... just not enough data, for now .....