First Casualty: dead BDR-209 diode

[Jstr]

(Edit) Summary of thread: I set the current too high (765 mA), but the diode was fine until the battery voltage sagged and put the flexdrive out of it's specs range, and that is the believed cause of death. Even with initial input voltage of 4.2V, the flex was technically already out of it's range at that current, but it did not kill the diode until a good 20 min of battery drain. (it was a 3100 mAh battery, so at an estimated 4.0 V, maybe??). So, hopefully we will get some scope data on this driver and confirm this theory.

(Original post)

Hi guys,

The purpose of this thread is to isolate the cause of death and hopefully add to the limited information available on this diode. I welcome any thoughts and comments that might provide helpful insight.

*There is not much to see other than sinner's spectacular handiwork, so I will not post pics unless necessary.

Build:
- MS-SSW-III host (check it out here)
- Flexdrive set to 750 mA
- 1x18650 orbtronic cell + blank 18650
- G-2 lens
- DTR silicon wire


What Happened:

So, I recently got this diode from another member here (who got it from DTR) because I have been wanting a 405nm build for a while. Unfortunately, the diode died before I could even add it to my sig.

The sinner host seemed great for extended duty cycles, and (while it worked, it never even got warm).

The build part was pretty standard, just soldering some wires (slowly, I admit). I stuck the flexdrive in the pill, heatsinked with copper wire and protected by electrical tape- thankfully I just used tape to secure the driver in the pill. All, assembled, it worked great. Nice, visible 405nm beam. Really great divergence, too, I was surprised. I used it sparingly at first, keeping cycles to ~1 minute on/off (I had to, it burned everything it touched).

BUT, then I wanted to get it warm to see realistically what the duty cycle should be. I timed it, and 3 minutes 15 seconds into the test, the diode suddenly stopped lasing. No warning, just LED'd (I assume this is the correct term). I immediately clicked the clicky switch off.

When I turned it back on, I saw that the diode was still emitting a faint light, but the divergence was terrible- at 10 yards, the focused spot was about the size of printer paper and barely visible in complete darkness.

I opened up the pill and lowered the current by ~70 mA. Same result. I haven't done anything more to it since then, but soon I will disassemble. But first, I need to clear up some issues.


What I suspect could be the problem:

I have two theories, and I do not think the diode is to blame.

Either a). I soldered the driver poorly, which caused the output to gradually increase, which destroyed the diode from current overload.

Supporting Evidence: I noticed while setting the current that this was in fact occurring. The current would start out consistently (I tested several times) at 765 mA and VERY slowly it would climb. I barely noticed- I thought it was just slight fluctuation. But then I conducted a longer test. After 7-8 minutes at least of constant run, the current came up to 784 mA when I stopped, again, after 8 minutes of constant run because the test load was getting pretty warm.

Refuting evidence: The diode died very suddenly, without any signs of dimming or heat (when I felt the host immediately after, it was barely warm). If it was current overload, one would expect dimming at least. Although my soldering is not professional by any means, it's not exactly easy to mess up assuming you're careful and patient, and my job looked solid- I was very thorough. Also, the current ramp-up was extremely slow, and the power graph on DTR's site showed the peak (power) at around or above 850 mA.


Or b). the host has a weak contact point in the tailcap, which caused a current inrush every time the clicky was pressed, which eventually caused a spike that killed it immediately.

Supporting Evidence: While I was setting the current, every once in a while, I would click it and the multimeter would still read zero. I had to click it again (twice, to be "on"). Then, with the diode connected, the problem persisted. At points, it seems I had to click it several times to get the mechanism in the right spot. This was pretty annoying, but I did not think much of it. The clicky is metal, and the spring inside it *feels* (I have not opened the clicky) like it is not secure, as in when the host moves, the metal button rattles. The sudden manner of death seems to fit this(?). I have read multiple accounts of bad clicky switches and also that not having secure contact points can kill diodes.

Refuting evidence: It's a sinner host . And it worked for a little bit. I can't think of anything right now.


Anyway, thanks for reading. Please share any thoughts, ask questions, whatever. If nothing else, I want to know if I can still use the host, driver, or both as is, or if I need a new tailcap, maybe. Cuz, well, it sucks to lose a diode, and doubly so to lose a diode, driver, and host together. I also hope that when I get the money, I can buy a new sled, get the diode from that, buy a diode press, extract the old diode, and press in the new one (if I trust myself to do that much).

Hopefully the information in this thread will add to the info available for this diode, though I doubt that the diode itself was to blame, given its performance with others.

 

[IsaacT]

That sucks man :/ I just killed two red diodes tonight and have no idea why they wont work. My guess is ESD but who knows? The wiring was right and I didn't let the diodes get too hot from my iron.

On yours my guess would be that the internal temperature of the diode got too high after that 3 minute run cycle and caused damage to the die. That duty cycle just sounds way too long IMO.

 

[sinner]

405nm diodes barely gets warm even in a small host like the one I have here I did a similar test for a 5-minute LPM graph , but I was running it at a very conservative current, In my experience you dont want your 405 to get warm, not even slightly because unlike the 445nm's these are not so forgiving, at all, they dont dim, even if you over drive them they tend to run super stable, until they go POOF! and you wont even know it, no dimming no warnings nothing.

here's the graph on mine.

It is 810mW flat out, It will not go down even if the diode gets hot so you wont know for sure when to stop.

 

[rhd]

(A)

I can't think of a way to "poorly" solder a driver such that it would cause the output current to gradually increase.

In fact, I suspect that the current wasn't actually increasing, but rather that the resistor on your test load was simply getting hotter, which increased its resistance, and led to a larger voltage drop. You measured a higher voltage drop and interpreted it as resulting from higher current, rather than simply a higher resistance.

(B)

I have trouble imagining that a current in-rush would get past the driver. Wouldn't the Flexdrive filter that out?

(Instead)

I am inclined to think you should look first to the driver. FlexDrives may not be great options for boosting voltage to run 405s at 750mA. Remember, just because a FlexDrive can run a 445 at 1A, doesn't mean they can do a 405 at 750mA. The voltages of these diodes are very different. I suspect that 750mA at the Vf of a 405 is a much tougher task than 1A at the Vf of a 445.

I don't know much about the specs of a FlexDrive, but it may be that the output gets dirty when operating close to its limits. We might not have noticed this with 445s, because even with a FlexDrive pushed to its limits (1A with a 445?) the 445 itself would still have quite a margin to eat up additional current spikes. Maybe the combination of a 405 running near its limit + a FlexDrive running near its limit = spikes that the 405 can't handle.

I would suggest trying a better driver than the Flex. Something from Angelos would be a good route (preferably a buck, since your host has room for two 18650s anyway).

 

[Jstr]

Thanks for your interest.

Isaac,
I understand your concern for the 3 min cycle, but the host is massive, with a copper heatsink. I removed the lens and felt the module immediately after it burned out, and the heat was barely detectable. I was hoping for a five min cycle actually, and the heat transfer in the host should have handled it fine imo. Also, I know that IWIRE posted something about leaving his on for five min for beamshots.

Sinner,
That worries me. My current setting was ambitious. So, that might be the cause. But, I only put max 15 min total "on" time on the diode. Correct me if I am wrong, but this is not that much of an overdrive, it's only slightly above DTR's max current. Even overdriven like that, I would expect a shorter lifespan, but not a 15 min one!

rhd,
(A) That is a very good point. I didn't think of that. So, that should rule that out.
(B) Also, true. I did not really think it was inrush, but I remember reading about instances when bad connections caused death. And I killed a 532 module (driver attached) by only taping the connections. So I'm not sure how bad connections kill diodes, but this still is a possibility I think.
The Flexdrive is supposed to be capable of 1.5 Amps. The spec sheet says to go with the highest range possible for best efficiency, so I went with that (spec sheet range: 500mA~1.5A). Also, I believe IWIRE used this same driver for his build, albeit set at almost 100mA lower. I was recommended a boost due to the higher voltage drop of 405 diodes, but if I keep the host, I will take your advice and go for an x-drive.

Just making sure, but this flexdrive is still safe for another build, right (if it's not the best choice for this one)?

If the tailcap is a nonissue, can I still fix it so that the clicky works every time? It was pretty annoying. How would I take it apart?

Thanks!!

 

[DTR]

I would put my money on it just being too much current for the poor guy to handle. I suggest a max of 700mA and to me I think that is running it very enthusiastically. I am surprised we have not seen many issues at these currents. Really to be safe if you go 650mA-700mA you will most likely hit 1W with this diode and be able to maintain some type of life expectancy. Keep in mind we drive these diodes hard and don't expect 5000 hours out of them in portable units. Since we don't have a datasheet we cant be sure but we are probably near doubling the rated current. I have one running @ 750mA with a Flex for quite some time now with no issues or degradation but we know they can vary on the top end from the diode to diode.

The flexdrive is fine for 405nm diodes. It won't boost for 520nm 50mW and 120mW diodes but it work just fine with 405nm. The spec sheet on the flex does say 5.5V out but Drlava told us a while back that he had a buffer in there of about 1V and they are fine to use with 405nm diodes and in the past when I don't have microboosts I have always used the Flex with no issues.

Anyway good to get some feedback on this one but sorry to hear you lost it.:yabbem:

 

[rhd]

The Flexdrive is supposed to be capable of 1.5 Amps. The spec sheet says to go with the highest range possible for best efficiency, so I went with that (spec sheet range: 500mA~1.5A).

No. The FlexDrive is capable of 1.5 Amps when using a lithium ion to power a 2.0V load. In fact, the datasheet even notes that "not every combination of input voltage, output voltage and output current is useable." You don't have a 2.0V load, you have a 6.6V load (assuming a target current of 750mA).

To calculate the FlexDrive's safe operating range, the FlexDrive's own datasheet says to keep:

(Vout (Volts) * Iout (Amps) ) / Vin (Volts) < 1.2​

Your parameters are:

(6.6 * 0.75) / 3.6 = 1.375
which is not less than 1.2​

So you're running the FlexDrive out of its spec. This confirms my suspicion that the FlexDrive is not an appropriate driver to be using here. EDIT: As an FYI, I just glanced at the MicroBoost datasheet, and it employs the same formula, so the same concern arises with a MicroBoost at this current.

I was recommended a boost due to the higher voltage drop of 405 diodes, but if I keep the host, I will take your advice and go for an x-drive.

Unfortunately, you may have received some bad advice (or you misunderstood the advice given). That doesn't make sense to me. They have it backwards.

If your diode has a higher voltage drop, that will in general make a boost driver less appropriate. A boost driver powered by a single lithium ion will have more difficulty driving a diode with a 6.5V drop than one with, for example, a 5V drop, all else being equal.

Conversely, a buck driver powered by 2x lithium ions will produce better results with a diode that drops 6.5V, than one that drops 5V. This is because the duty cycle (of the switching driver itself, which is not the same "duty cycle" we talk about when referring to laser on/off times for cooling purposes) will be closer to 100% when the VIn and VOut are closer together. 6.5V is closer than 5V is to the input voltage of 2 series lithium ion cells, so the formula Vout / Vin will be closer to 100% when you are driving a 6.5V drop diode, than when driving a 5V drop diode. That will generally imply less output ripple. In general, all else being equal, a buck driver is favored when the Vf of the diode is higher.

Just making sure, but this flexdrive is still safe for another build, right (if it's not the best choice for this one)?

I think not. Others may disagree. But at the point where you're already running the FlexDrive outside of the parameters in its own datasheet, I think it's time to find another driver. I'm not up to date enough on Lazeerer's X-Drive revisions (he has a lot!), but I'm sure that if you talk to him, and give him your VIN / VOUT / and Current parameters, he'll be able to tell you what the safest X-series driver for this diode is. He may have boost drivers capable of safely running this diode. I know he has numerous bucks that will be more than adequate.

EDIT: I see now that you said "for another build". As long as you mean for a different diode, then sure, it may be just fine. People certainly use FlexDrives. So for a less demanding task (like perhaps a lower current 405, or a single mode diode), it should be re-usable. No need to trash it

 

[Jstr]

Thank you DTR and rhd

So rhd, your math makes sense and confirms that no, technically it is not within the driver's range, and I should have checked that.

But DTR and others (as seen by several builds here with both microboosts and flexdrives) seem to have no problem with it. DTR said above that there's a 1V buffer and he has a similar build with a flex @ 750mA.

It makes sense that boost drivers have a more difficulty with higher currents, but why then do most people (that I have seen) prefer microboosts/flexdrives/x-boosts for green and x-drives for 445 (I assume this because DTR's site has these Driver/Diode combo options available)?

I set my test load to six diodes when setting the current, which should have been adequate? I think this fact should further prove that the flexdrive is capable of driving a 405, but I'm not positive.

Thanks for your input rhd, your advice makes sense, but I am just confused because it contradicts what a lot of people are doing and have been doing.

DTR, would you recommend an x-drive? I am leaning toward a different, 1x18650 host now, so it may not matter, but still, out of curiosity, what is your take?

I might switch to a microboost then because it can boost up to 13V.

So, if nobody thinks it's the tailcap, I will dismiss that as a possible cause, which leaves the current. I guess the current was too much for this diode (maybe I got one with slightly lesser capabilities), and that killed the diode. (In this case, DTR wins )

Can anyone confirm that these sleds on ebay/amazon are indeed the same BDR-209 diode?

Thanks!

 

[rhd]

I set my test load to six diodes when setting the current, which should have been adequate? I think this fact should further prove that the flexdrive is capable of driving a 405, but I'm not positive.

I never said a flexdrive couldn't drive a 405. It obviously can.

Whether it can do so at a new (higher) current, is the question. This new higher current is out of the driver's own spec. Coupled with the reality that 405s are a lot more sensitive than 445s - you're asking for trouble.

There is a huge difference between a driver being capable of driving something, and being capable of doing so safely. For example, look at the early version of the Open Boost 2. It can drive a 445 at 2.4A from a single cell. But when we spectro'd it, the ripple was huge. That wasn't a problem for 445s. It would have killed a sensitive 405.

You're pushing the flex / microboost beyond its intended operating specs driving this diode at this current. That's a bad idea with a single mode diode that is sensitive, and to be clear, the flex / microboost has NOT been widely used in a scenario where it's driving a sensitive diode, while operating outside of its own specs. This is new territory.

 

[Jstr]

You misunderstand me. Sorry if its not clear but I agree with you completely! I plan on a.) Switching to microboost and b.) Lowering the current substantially.

I'm also interest in hearing more opinions on how substantial the difference between buck and boost is in this case because the microboost is able to boost to 13v which is plenty. But if the consensus is that buck is better, I will for sure do buck. Thanks!

 

[rhd]

You misunderstand me. Sorry if its not clear but I agree with you completely! I plan on a.) Switching to microboost and b.) Lowering the current substantially.

I'm also interest in hearing more opinions on how substantial the difference between buck and boost is in this case because the microboost is able to boost to 13v which is plenty. But if the consensus is that buck is better, I will for sure do buck. Thanks!

I don't think you read my post if you thought I was suggesting the MicroBoost.

It has exactly the same limitation formula in is datasheet as the flex. AGAIN, just because it can boost to 13V, doesn't mean it can do that at 750mA. It can't.

Read the datasheet.

 

[Jstr]

I did read your post, i realize that you say buck. please I am trying to determine how I should make my next build with this diode. As I said before I now wantto use a single 18650 so I NEED a boost. This is the ONLY reason i want to go with a boost driver. I also said I would substantially lower the current setting.

Edit
I do not think you read my posts either.
I also have read both data sheets thank you. Just because its on paper does not mean it's correct. The microboost is proven to work as I have said repeatedly

 

[DTR]

I did read your post, i realize that you say buck. please I am trying to determine how I should make my next build with this diode. As I said before I now wantto use a single 18650 so I NEED a boost. This is the ONLY reason i want to go with a boost driver. I also said I would substantially lower the current setting.

I have built around 50 of these @ 700mA with the microboost without issues. If you want to drive it more conservatively run it @ 600-650mA and you will be fine.

 

[rhd]

The microboost is proven to work as I have said repeatedly

Just because you said it repeatedly, doesn't make it true.

To my knowledge, it has not been proven reliable at 750mA and 6.6V or higher output. We haven't had diodes matching those specs for very long.

You're very certain that you're correct, that the datasheet is wrong, that I'm wrong, and that it's safe to use this driver in a configuration that doesn't have any substantial track record and exceeds Lava's formulaic parameters. So I'm not sure there's anything else for me to add. Sure, you're probably right....

You can put the issue to bed by scoping the driver, at this current, this v-drop, and with a lithium ion for input. I suspect you'll find ripple too intense for a sensitive diode.

But what do I know.

 

[DTR]

That is not a bad idea to have one scoped to see where they are with these diodes. I think I will shoot one over to Lazeerer and see what he comes up with.

 

[Jstr]

Just because you said it repeatedly, doesn't make it true.

To my knowledge, it has not been proven reliable at 750mA and 6.6V or higher output. We haven't had diodes matching those specs for very long.

You're very certain that you're correct, that the datasheet is wrong, that I'm wrong, and that it's safe to use this driver in a configuration that doesn't have any substantial track record and exceeds Lava's formulaic parameters. So I'm not sure there's anything else for me to add. Sure, you're probably right....

You can put the issue to bed by scoping the driver, at this current, this v-drop, and with a lithium ion for input. I suspect you'll find ripple too intense for a sensitive diode.

But what do I know.

Please READ what I am saying!! How many times have I said you are wrong? Zero. I agree that high currents should not be used with these boosts! This is the 3rd time I will say it: I would be fine running it with a microboost at a substantially lower current (I.e. 650 mA). At this current is what I mean when I say that these drivers are proven!

Edit:
Here is the math.

(6.6*.65)/3.7=1.159

I would like to say that I know you are far my superior when it comes to lasers. I started four months ago. The only thing I am trying to say is that people seem to be running these diodes with boosts despite the math, which I know you are right on, so how are they getting away with it?? I want to understand so I don't lose another diode. And I do appreciate your feedback.

On another note, is it plausible that my battery voltage was dropping and caused the above formula to exceed 1.2 and thus kill the diode? That would make sooo much sense.