(Updated July 16) Review/Test: Illumination Supply Direct-Bonded 16mm Copper Star with Cree XM-L2 U2 1C

Illumination Supply Direct-Bonded 16mm Copper Star with Cree XM-L2 U2 1C

Reviewer's Overall Rating: Review Incomplete


[Review subject photo goes here]

Summary:

Diameter 16mm
Thickness 1.5mm
Material Copper with upper dielectric, lower nickel plating
LED Type: Cree XM-L2, Flux Bin U2, Chromacity Bin 1C
From: Illumination Supply
Date Ordered: Review sample provided by Illumination Supply

Pros:

Cons:

Features / Value: TBD

Design / Build Quality: TBD

Overview

The Illumination Supply Direct Copper Bond (DCB) emitter stars from Illumination Supply offer improved thermal transfer characteristics over traditional aluminum stars with a dielectric layer. When attempting to push an emitter to extreme power levels, heat begins to negate any potential gains. By bonding the emitter thermal pad directly to the star base material, heat is transferred to the star and on to the heatsink at a much faster rate. The result is a cooler emitter die at the same current, therefore more photons out the front.


Setup

First, a little bit on the setup. The emitter is mounted to a 900g aluminum heatsink with large fins on the back. A fan is blowing on the back to maintain as close to an ideal heatsink as possible. This setup is close enough, but not perfect.

In addition to Luminous Flux (lumens), measurements of Emitter Current (IF), Emitter Voltage (Vf), Emitter Temperature, and Heatsink Temperature were taken. Measurements were taken at 0.2A increments from 0.2A to 6.2A.

In this first image, you can see where Vf and Emitter Temperature were measured.

I used duplicate wires to feed back to the voltmeter; no wire drop was measured. The goal was to measure the voltage drop across the star being tested.

I used a standard thermocouple glued to the edge of the emitter, as per recommendations found in a Cree app note (a reference link will be provided).

I will note that my initial naked-eye evaluation of the emitter placement did not show any misalignment, there does appear to be a slight misalignment in the emitter placement in this macro image. I will confirm this with an even closer macro before concluding anything.

This second shot shows the three meters used to capture Emitter Temperature (Vichy VC99), If (Metex M-3800), and Vf (Metex M-3800).

The Heatsink Temperature was measured right under the emitter placement, between the heatsink fins. This measurement was taken with an IR temperature probe (the heatsink is anodized). In other words, as close as I can get to the base of the star without needing to move the setup between measurements.

Here is the complete setup (power supply is out of frame to the left)

Closeup of heatsink in position.

And finally, the light meter:

Performance Tests

The first test is with the emitter/star in stock form, as shipped.

Test Results Data:

IS DCB 16mm Star with XM-L2 U2 1C Emitter (Stock)
If Vf Pled Lumens
0.2 2.77 0.55 98
0.4 2.85 1.14 187
0.6 2.91 1.75 278
0.8 2.96 2.37 360
1.0 3.01 3.01 440
1.2 3.05 3.66 516
1.4 3.10 4.34 593
1.6 3.13 5.01 663
1.8 3.17 5.71 730
2.0 3.20 6.40 796
2.2 3.23 7.11 857
2.4 3.26 7.82 918
2.6 3.28 8.53 973
2.8 3.31 9.27 1029
3.0 3.34 10.02 1080
3.2 3.36 10.75 1131
3.4 3.38 11.49 1176
3.6 3.41 12.28 1221
3.8 3.43 13.03 1264
4.0 3.45 13.80 1304
4.2 3.48 14.62 1342
4.4 3.49 15.36 1375
4.6 3.51 16.15 1407
4.8 3.54 16.99 1438
5.0 3.55 17.75 1464
5.2 3.57 18.56 1489
5.4 3.60 19.44 1537
5.6 3.61 20.22 1560
5.8 3.62 21.00 1567
6.0 3.66 21.96 1582
6.2 3.68 22.82 1590

It is still early, but this data looks promising, although slightly off the results found by Match on a SinkPAD. My setup is not calibrated to his, so there could be some variation. Once I get to the stage The star appears to be doing its job at getting the heat away from the emitter and keeping it as cool as possible.

Once the test was complete and things cooled down to room temperature again, I retested the emitter at 0.4A to look for signs of fatigue or wear. The results were identical to the first pass. I will note that the emitter was not run at 6.2A for more than a minute.

Next test is with the star lapped flat. This took a while, as the bottom is not already flat, it is slightly concave. This image was after lapping for a while with 2000 grit sandpaper.

I move to 400 grit, lap it flat, then proceed to lap down to 2000 grit for a smooth finish.

Test Results Data:

IS DCB 16mm Star with XM-L2 U2 1C Emitter (Lapped)
If Vf Pled Lumens
0.2 2.76 0.55 94
0.4 2.85 1.14 184
0.6 2.91 1.75 268
0.8 2.96 2.37 351
1.0 3.01 3.01 431
1.2 3.05 3.66 509
1.4 3.09 4.33 577
1.6 3.13 5.01 649
1.8 3.16 5.69 718
2.0 3.20 6.40 781
2.2 3.23 7.11 841
2.4 3.25 7.80 899
2.6 3.28 8.53 957
2.8 3.31 9.27 1010
3.0 3.33 9.99 1062
3.2 3.36 10.75 1113
3.4 3.39 11.53 1159
3.6 3.41 12.28 1203
3.8 3.43 13.03 1246
4.0 3.45 13.80 1287
4.2 3.47 14.57 1325
4.4 3.49 15.36 1360
4.6 3.52 16.19 1391
4.8 3.54 16.99 1424
5.0 3.56 17.80 1452
5.2 3.58 18.62 1475
5.4 3.60 19.44 1522
5.6 3.62 20.27 1545
5.8 3.64 21.11 1560
6.0 3.66 21.96 1575
6.2 3.68 22.82 1590

No significant change in output (virtually identical), and a marginal change in emitter temperature. The temperature probe remained in the same place between tests.

At this point I desolder the emitter to move it to a different star for comparative testing, and I notice the thermal pad was not completely soldered down:

The solder paste used was an insufficient quantity to fill the pad gap, and it did not bond well with the base. Instead of moving on, I decided to reflow the emitter again and retest.

Test Results Data:

IS DCB 16mm Star with XM-L2 U2 1C Emitter (Reflowed)
If Vf Pled Lumens
0.2 2.76 0.55 94
0.4 2.85 1.14 184
0.6 2.91 1.75 267
0.8 2.97 2.38 351
1.0 3.02 3.02 428
1.2 3.07 3.68 504
1.4 3.11 4.35 575
1.6 3.15 5.04 649
1.8 3.19 5.74 716
2.0 3.23 6.46 781
2.2 3.26 7.17 843
2.4 3.29 7.90 902
2.6 3.32 8.63 962
2.8 3.35 9.38 1020
3.0 3.38 10.14 1075
3.2 3.41 10.91 1128
3.4 3.44 11.70 1181
3.6 3.47 12.49 1228
3.8 3.49 13.26 1278
4.0 3.52 14.08 1322
4.2 3.54 14.87 1366
4.4 3.57 15.71 1410
4.6 3.59 16.51 1450
4.8 3.62 17.38 1490
5.0 3.64 18.20 1552
5.2 3.67 19.08 1590
5.4 3.69 19.93 1627
5.6 3.71 20.78 1664
5.8 3.74 21.69 1694
6.0 3.76 22.56 1724
6.2 3.79 23.50 1754

Much better. Over 150 lumens gained. Emitter temperature is lower but not dramatically so. I suspect that each time I apply the probe, I could get different thermal contact and this will skew that data.

Here is the solder bond from that test. Much better, but it appears the base is still resisting fresh solder paste. There is a very small gap between the emitter and the base, which I have seen on other DCB stars. This is just one possibility.

One curious change from the previous two tests is the change in emitter Vf. It jumped by about 100mV. I have no explanation for this change. The reflow was not done to the Cree recommended temperature curve (realistically, none of us have the equipment for that), but I would not have expected this. Another possibility is a less secure solder joint that I didn't notice. Just possibilities, this is as yet unexplained. I'll return the emitter to this star and recheck Vf again at some point.

Comparison to Other Emitter Stars

I have also tested Noctigon, SinkPAD, and aluminum 16mm stars with the same emitter. The results are very comparable, when comparing apples to apples (properly mounted emitters on lapped stars). Since I do not have pre-mounted emitter products available for the competitor stars, I am comparing the lapped reflowed data (what I consider apples to apples).

Note: This particular aluminum star did not perform well. I have another that I want to try at some point.

Closeup of 5.0A to 6.2A for the three leaders:

These results (3.7% variance) are probably well within my measurement tolerance. I would give a very slight edge to the SinkPAD and Noctigon, but it's so minor that it is probably unnoticeable in all but calibrated equipment. In practical use, these three DCB stars seem to give comparable results.

Dimensions

  • Thickness: 1.57mm
  • Diameter: 16.04mm
  • Widest diameter: 16.91mm, stamp edges may need minor trimming in small pills

Conclusions

I have no conclusions at this point. The IS-DCB star definitely does perform much better than an aluminum star. The emitter is definitely capable of serious light output. However, I do have some issues to resolve before I can recommend this product as a whole.

First, and most glaring, would be the possibility of improperly reflowed emitters. The poor solder bond could be a singular faulty unit, or a systemic problem with production. Calvin at Illumination Supply is looking into the stock to see if more units have similar problems.

The second, apparently minor, issue is the non-flat base. This didn't show a significant problem in the comparative tests, but the reflow issue may have masked it. This problem is resolved with a minute or two of lapping. Because of this, I'm not concerned about this one.

More to come as we get to the bottom of the reflowing issue.

Thanks for reading! searchID8934

Thanks for doing this test, its looking good so far!

Hi relic,

I’m really looking forward to the comparisons you listed!

Jim

Thanks relic38. Looking forward to the rest.

Very useful info. A Subscribed!

Thanks relic!

These measurements look very well done, thanks for doing the job! Apart from the performance of the IS copper board, it gives some nice information about the XM-L2 emitter as well. I am looking forward to the pictures of the light measurement set-up, and to the rest of the tests.

I would have thought it would have scaled better……

Thanks!

The lumens are a bit under, 1080lumens at 3A for the XM-L2 U2.

+1 - Nice Job! Who's missing Match? I'm also interested in seeing how you did the lumens measurements.

I noticed that too, and I believe part of it is due to the emitter no longer at 25C when at those drive currents. The rest may be tolerances in my measurement equipment. I’m sure Cree are slightly better set up than I. :slight_smile:

Great job relic! Regardless of any minor inaccuracies in your calibration, your direct comparisons should eek out any differences between test samples. I wonder how an aluminum sinkpad would fare since they are the same in construction as their copper counterparts (direct thermal path).

Do you plan on removing and reflowing the same emitter to each board you test?

Thanks FP. I am planning to reflow the same emitter in each test. I am aware that repeated reflow cycles are not advised my Cree, however I am confident that the stress this introduces will create less variation than trying to do the test with different emitters. I also have full control of max reflow temperature so I do not overheat the emitter any more than I have to.

I do not have an aluminum SinkPAD. Maybe I should fill out that survey…

Thanks very much Relic for the hard work.
I’m still missing Match though, I believe he’s part of members that give some inspiration for many others here. Will be glad if he back, that’s not mean he have to do some testing :bigsmile:

Great info and report. Thank you.

Just don’t plot lumens/watt, that might depress me.

My thoughts exactly about production variations between emitters. You’ll notice a sudden plunge in luminosity should your emitter start to fail after a reflow and turn-on readings should all be close.

The survey is how I got my AL sinkpads. Just a hunch, but I bet they will perform far better than most people would expect.

The Vf and lumens are consistently lower than Match's tests on a SinkPAD - I think he used a U2 1C as well. Now this difference isn't a bad thing at all - just interesting to note. More important is it's in the neighborhood and your lumens curve looks really, really good and match's up with Match's, accept for that hitch at 5.2-5.4A. So could be just the measurements/calibration, emitter differences, or the copper star is that much different from a SinkPAD - no telling yet, still early... Very interesting to watch this data fill out...

FP, I just filled out the survey at SinkPAD, one more product to compare to (eventually).

Tom E, that blip at 5.4A is when the light meter went from the 20,000 lux scale to the 200,000 lux scale. I believe the 2,000 and 20,000 lux scales read slightly low on my meter. Nothing drastic, but noticeable.
I’m avoiding interpreting any data at this point. Once I have more data from more tests and can compare to other direct-bonded stars.

One important note is that in statistics, a sample size of one is mostly useless. There could be some undetected defect in that sample, or it could be a stand-out performer. Either way, the test results from one sample are likely not going to represent the average performance of a mass-produced product. Product characterization requires several samples from several production batches to average out those anomalies.
But alas, we live in the budget world. I would not expect a supplier to sent multiple samples from two production runs, nor would I have the time to test that many. We work with what we have and as long as we are aware that the data is limited, all will be OK.
One thing I would not like to see is someone read through the review, see the positive test results, order a bunch and have them comparatively under-perform and then blame the review for misleading them. This potential issue exists with any product review of one sample. Most reviews out there are just that.
[\rant]

I will make one observation though; these stars perform much better than a standard aluminum star. That I am confident in saying.

I was curious too about the Vf data. Seems good cells (i.e. Sanyo 2600mAh) in single cell lights could sustain regulation at drive currents up to about 2.8A.

-Garry

relic - +1, agree on all points... Scale change explains the 5.4A hitch.