Extracting Diodes From Nichia diode bank

Okay everybody here's what I learned. I chickened out on heating it. I didn't want to take the chance of killing every diode even though it probably wouldn't. So I just used a bench grinder and carefully removed the metal around the diode. It kept getting hot so I had a bag of ice to keep cooling it down. I ground it down enough to expose about 2/3 of the back of the diode and then I cracked it loose. The diode bank I had was NUBM06 445nm 8x bank. I wouldn't worry too much about the vibrations hurting it. At one point I got frustrated and took a hammer and chisel to it in attempt to Crack the block. It didn't work. Whatever alloy these things are cast from is very tough. It's hard and brittle. I used a pair of cutters to chop off chunks of the block. The main thing I think is be careful not to bend the frame of the diode. When I punched it out from the back it bent in and warped into a convex shape. It spread the posts and the bond wires must have been damaged.
So once you have about 2/3 of the diodes base exposed at that point it seems safe to punch it out without bending the frame.
As for this NUBM06 nichia it's a pretty good diode. I put about 4.5 amps to it and it is almost as good as the NUBM44. It is rated for 4 watts and the NUBM44 is rated for five. So it's definitely a pretty good buy. And if you manage to get all eight out without killing any it's a pretty good deal. Be aware though it's a open diode. I prefer the ones with cans, but this one is pretty good. If you are going to build a multi diode laser these would be a good choice. That's what I'm planning to do with these. 2 into 1 hand held. Should be pretty nice. I will be selling them on eBay or here once they let me. They said I didn't have enough posts yet to sell. But I have quite a few lasers for sale. All in Ehgemus hosts.
So in conclusion these banks are definitely worth buying however do expect to work for them. Unless someone out there wants to share techniques with the rest of us here. I guess we can just keep refining our methods and sharing our results.

I see what you are calling scratch marks, I've worked with solder my entire adult life and to me, they just appear to only be lower areas in the form of lines where the solder didn't fill in, appearing as scratches when they aren't.

How much do you pay for the NUBM06 block?

I paid $550.00. I ordered strait from the land of China. I paid an extra $20.00 for faster shipping. I got it in about 3 days.
I haven't seen anything about this diode anywhere. Does anyone have experience with this diode? I'm running it around 4.5 amps. It seems pretty efficient. It might be brighter than NUBM44 at the same amps and volts. I can only guess because I don't have a LPM. It's only slightly less powerful that a NUBM44. It would be a good choice for beam combining or knife edge.

Just like the NDB7875 was an upgrade to the M-140 I expect there will be an upgrade to the NDB7A75/NUBM05T, I hope this NUBM06 is that because the NUBM44 has 3 times the line width as the 7875 at least with a 3 element but not 3 times the power.

With a 3 element at 14 feet

M-140 @ 1.8 amps = 2.0 watts with a line width of 5/16

7875 @ 2.4 amps = 2.5 watts with a line width of 5/16

7A75@ 4.0 amps = 4.25 watts with a line width 0f 3/4

05T ----SAME AS 7A75-----

07E @ 3.5 amps = 3.25 watts with a line width of 3/4

44 @ 4.5 amps = 5.5 watts with a line width of 7/8

What you do see is an increase in efficiency in power output per power input but a decrease in divergence efficiency, that is wider divergence per watt and less energy density.

So as focused light is the goal second to energy efficency I suspect an improved G-ball NUMB05T could be in the works and the old canned diodes are going to go away. I don't expect a successor to the NUBM44 but who knows, it requires more downstream lens and alignment so I think the G-ball will take it's place, so I am hoping just maybe this NUBM06 is an upgrade.

Maybe the 06 is an upgrade to the 05T or just a horse by another name.

Can you test the 06 next to a 44 at 15 feet to see what the divergence and burn time looks like?

I will try it next to a 44 later today and let you know.

I slept on this question and woke up thinking the answer is what was suggested earlier, just super cool it and pop them out, the melted but completely flattened solder appears to be from when it was originally soldered in and the clean flat surface from being popped out when the bond flatly separated from the heat sink due to being so cold.

As evidence this is what he did, you can see the solder is melted, yet completely flat, except for some slivers which didn't pull off cleanly. Obvious to me now this is what he did, if he heated it, the solder wouldn't be completely flat on the back side. I bet he is thinking DAMN! Why did I put that photo up!

As further evidence, you can clearly see a flat piece of solder in the upper left hand corner of the back side of the diode which didn't melt, it flaked off from being so cold and popped out by sharp low impact blow to the back side by a special tool for an exact fit, I imagine.

Dissimilar materials expanding/contracting.

Put the housing in the freezer then power the diode on for 60-120 seconds, remove the power from the pins and tap with a punch between the pins to pop it loose, the cooling time and using the diode to heat it's self time are the key formulas.

Add a light tap with the power removed from the pins, that is a tap between the pins on the diode back as it's expanding from the heat of being run for 60- 120 seconds and the case super cooled, that tap could pop it loose and leave that smooth pattern in the bonding agent.

I edited my post while you were working on yours to add the likelihood he used a tool on the back side too. I think it might pop out without having to turn on the diode to heat it while the heat sink is super cool, but maybe that is needed.

I have a NUBM44 I just got and the solder on the back looks globby unlike the one in the picture. It was definitely heated. I think he heated it on some type of temp controlled hotplate and when it was hot enough punched it from behind to break break the friction press. I'll take a picture of the one I have.

Also if someone has a infrared temp gun. I don't know if you could get a true reading but if you could attempt to find out how hot a NUBM44 is at the chip maybe we could have an idea of what kind of temperatures the diode can withstand.

The one in the picture looks completely flat to me like it separated from something else flat without being heated. I can't imagine that uber thin flake of solder I pointed out on the upper left side of the diode being there like that if he had heated it. Although that doesn't mean he didn't change how he extracts them later, but which process is he using now? No way to know, he might still have stock from the older process.

Well if you could avoid heating the die/pn junction that would be good because storage temp is higher than operating temp.

I have a dead diode here, I know we all have dead ones, why not use a temp controlled plate and slowly turn it up to see how low that stuff melts on the back of a dead diode first.

If you want to PM me your email I will send the pics off the one I just got.

You could make a tool like a socket to fit the notches in the edge of the diode and with a rotation stop take up the slack in the tool so as to allow just a couple degrees of rotation to help break the diode free without breaking off the pins, there is a little room for 2-3 degrees of rotation. Here is the back of my Dead 44 I never understood why the diodes stick to a magnet like they are ferrous, steel is not a great heat conductor, some alloy?

By the way, does anyone know what projector the diode blocks for the NUBM07E were made for or come from? If so, please PM me.