6x 20mm XML-ledboard comparison

73 posts / 0 new
Last post

Pages

djozz
djozz's picture
Offline
Last seen: 3 hours 25 min ago
Joined: 09/07/2012 - 17:04
Posts: 17954
Location: Amsterdam
6x 20mm XML-ledboard comparison

1) common aluminium board, in fact the board the XM-L2 3B led was bought on, from Fasttech.

2) copper board with traditional dielectric layer under the central heatsink-pad, from Fasttech

3) same type of board as 2), but with the dielectric layer removed under the central pad, the gap filled with solder

4) aluminium Sinkpad, version1 (with the shallow gap on the underside of the board where the central pad was punched up)

5) aluminium Sinkpad, version2 (flat on the bottom, the edges of the board a bit rounded, reveiling that it is a punched-out board )

6) copper Sinkpad, version1 (with the shallow gap on the underside of the board where the central pad was punched up)

EDIT: correction: rollinstone157 pointed out correctly that I swapped the two alu Sinkpads in the above picture.

In another thread BLF member rollinstone157 offered me some Sinkpad samples he had sitting around doing nothing. It later appeared that he could not find them anymore, so he was so kind to ask Sinkpad for some new samples especially for me to test. They agreed but asked for some quite extensive testing in return, especially to compare them to the various copper boards around that -unlike the Sinkpads (and Noctigons and IS boards)- have a dielectric layer under the heatsink-pad. I first wanted to do a limited test but was eventually happy to do the whole lot because I was curious about it, and so are some BLF-members I understand.

Relic38 did some very precise comparisons already (more professionally than me) between the different direct-bond copper boards around here and here. This test I did now is different types of Sinkpads versus non-direct bond boards, using the same led reflowed six times successively on the different boards and testing the performance.

I have recently bought a new lineair power supply. The display on the power supply shows a current reading that I really wanted to use for tests, because it just is easier, it saves me putting an extra amp-meter in the circuit. So a thing to do before testing these boards was checking the current reading of my power supply against my (mid-quality Amprobe) multimeter. For who wants to see the numbers:

The power supply and DMM agreed on the current within 0.1A, even closer at higher currents, I was happy enough with that. I also made some new test leads out of thick speaker wire, the bottom shows a voltage drop measurement of the black one (70cm). These long leads are suitable for use with the power supply, but I will make shorter ones for doing current measurents on flashlights with the DMM.

One of the -to be tested- boards was a cheap copper board with dielectric layer (I bought mine from Fasttech), with the layer removed, exposing the copper core of the board, and then the gap filled with solder. I used a scalpel to cut out the center pad, with a forceps removed the solder pad, then scraped the dielectric layer away with a small screwdriver, and on the reflow block filled the gap with solder. The work was tedious enough for me to happily next time pay a bit more and buy a proper direct-bond copper board, but in the true BLF-spirit this is the way to go of course Smile . Some pictures of that:

 

 

Everything is ready for the comparison. The 'reflow station' that I use for more than a year now is a 12V solder iron on a small power supply converted with an aluminium block, it has no temperature control other than that I know that at 9V reflow does not work because the solder will not melt, and at 10.7 V reflows go fast and smooth. I guess that the temperature of the led will not get excessively high in this this set-up.

The test set-up is my usual one, but with new power supply Smile , led-board screwed down on chunk of aluminium, tiny bit of Arctic Silver5 in between, board is directly connected to the power supply, current reading comes from power supply, voltage is measured at the end of short and thick wires soldered on the led-board. On top of the led is a C8-reflector facing the ceiling. Light bounced from the ceiling is measured with a (cheap but proven reliable) lux-meter.

The luxmeter was taped in position and between different measurements just the led-board was removed from the set-up so that the position of the board next to the luxmeter was exactly the same in all measurements. Normally I calculate the lumen-numbers from the lux-numbers using this-set-up-specific conversion-factor 1.22 , but the C8-reflector blocks a bit of the ceiling bounced light, so it is not accurate in this case. So I only give the lux-numbers in the graph, for comparison between the boards. (I do suspect from this test though that my led was a lower bin than T6...)

I started the test with the led on the original aluminium board. I first turned on the current to 5A for a moment to 'burn in' the led a bit, probably did not do a lot, but I thought that it was fair that also the first test was done with a 'used' led. After each measurement series I removed the led from the board and reflowed it onto the next board. All those reflows did not leave the led-dome undamaged, it got a bit 'bobbly' over time and at the end of the copper Sinkpad test, at 7A, something on the led-dome burned and smoked.

 

I had to cut a bit of the dome off with a scalpel, and measured the led again on the same board: I got the same output, apparently damaged domes do not really influence output (but will affect the beam a bit of course). I ended the test-series with reflowing the led on the original aluminium board to see how much the performance of the led had suffered from the test-series (lux2 and V2 in the picture below, and measurement2 in the graphs).  Here's the numbers that I gathered:

I made an output graph, and a voltage graph:

I always have doubts if my improvised set-up is good enough to do such measurements, so I was again surprised to see in the produced graphs that the measurements are not all over the place, and make very much sense Smile

I am also glad to see that after all the abuse the led reflowed on the original board shows a very comparable performance to the first measurements on that same board. It tells me that the measurements in between can be reasonably compared to each other, that I am not looking at a slowly dying led.

 

I think the results speak for themselves, but to mention some finds that I think are interesting:

-the cheapy alu board that the led came on performes quite well up to 4A, the output curve compares well to match's finds., exept that I find  lower voltages. I can not compare these voltages with my own measurements of the XM-L2 I did last year because I used the power supply voltage then, which is higher than the voltage that is actually over the led (even measuring over the led wires as opposed to the actual solder pads of the led gives slightly different values, as Relic38 found).

-the cheap copper board with dielectric layer performs much better than I thought!

-that same cheap board with the dielectric layer removed and a thick layer of solder under the thermal pad performs as well as a copper Sinkpad! And even with a bit lower Vf, making it a bit more efficient. This I find interesting: perhaps because of just a bit worse thermal path it warms up a bit more than a copper Sinkpad, this lowers the Vf but does not really reduce output: the thermal management of the Sinkpad may work a bit too well in this test (no worries: in an actual flashlight there is more than enough compensation for that Wink )

-the aluminium Sinkpads perform slightly worse than the copper one, but in real life it is the same, version 1 or 2 does not matter.

 

I hope you liked these tests. I keep to my belief that every measurement error under 5% is nothing to worry about Wink .  But I am always open to questions and to suggestions of how to do better testing within my humble possibilities (I am working on a integrating sphere right now, not for more accuracy but because it makes measuring a bit easier, and stops the neighbours from wondering what that lunatic is doing all the time pointing idiotic bright flashlights at the ceiling).

Edited by: djozz on 04/17/2014 - 14:56
MRsDNF
MRsDNF's picture
Offline
Last seen: 1 month 4 days ago
Joined: 12/22/2011 - 21:18
Posts: 13473
Location: A light beam away from the missus in the land of Aus.

Orsm effort djozz. Thank you for the effort. Lots of valuable info here. I'll PM my address to you to send me the second sphere your building for me.

 

djozz quotes, "it came with chinese lettering that is chinese to me".

                      "My man mousehole needs one too"

old4570 said "I'm not an expert , so don't suffer from any such technical restrictions".

Old-Lumens. Highly admired and cherished member of Budget Light Forum. 11.5.2011 - 20.12.16. RIP.

 

xmint
xmint's picture
Offline
Last seen: 6 years 11 months ago
Joined: 02/28/2013 - 21:57
Posts: 389
Location: Surrey, UK

Just the sort of experiments that are so useful :bigsmile: , puts numbers and results to some subjective thoughts I had. Very much appreciated Beer

The views in this post are mine, they are views and beliefs only, and not to be taken as facts.

MRsDNF
MRsDNF's picture
Offline
Last seen: 1 month 4 days ago
Joined: 12/22/2011 - 21:18
Posts: 13473
Location: A light beam away from the missus in the land of Aus.

I'm surprised at the results from the ally Sinkpad, n/o 5 or is n/o 6 an under achiever?

 

djozz quotes, "it came with chinese lettering that is chinese to me".

                      "My man mousehole needs one too"

old4570 said "I'm not an expert , so don't suffer from any such technical restrictions".

Old-Lumens. Highly admired and cherished member of Budget Light Forum. 11.5.2011 - 20.12.16. RIP.

 

ImA4Wheelr
Offline
Last seen: 5 days 3 hours ago
Joined: 02/03/2013 - 14:51
Posts: 7935
Location: SC

Awesome, very informative testing.  Thank you djozz.  You the man.

I use the cheap FT copper boards in under 5 amp builds.  They are in some of my documented builds here.  I don't cut out the whole dialectic layer.  I gouge slot through it with a straight screwdriver.  So I get some direct copper and some copper spreader over the dialectic layer.  Seems to work good (don't see signs of thermal sag).  Now, I see from your testing that it may actually work better than I hoped.

djozz
djozz's picture
Offline
Last seen: 3 hours 25 min ago
Joined: 09/07/2012 - 17:04
Posts: 17954
Location: Amsterdam

Thanks for the appreciation Smile

MRsDNF wrote:

I'm surprised at the results from the ally Sinkpad, n/o 5 or is n/o 6 an under achiever?

No, it shows that the bottleneck for the heat escaping the system is not so much the board being aluminium or copper (conducting very well, and very very well respectively), but the thermal conductance very close to the led, and that situation is identical in all Sinkpads.

FlashPilot
FlashPilot's picture
Offline
Last seen: 5 hours 1 min ago
Joined: 05/10/2010 - 16:07
Posts: 4798
Location: USA

Ive always wondered how well the AL sinkpads conduct heat (as compared to the copper version), and now we know. Thanks for your hard work and a great post. Im glad I didnt throw out my samples.

TexasToasted
TexasToasted's picture
Offline
Last seen: 20 min ago
Joined: 10/25/2012 - 08:48
Posts: 1326
Location: Texas Gulf Coast

This is great stuff! I have a few of the copper boards with the dialectric layer and wondered how well they would work. I’m happy to see they do work well and I got a good deal.

Nightbird95
Nightbird95's picture
Offline
Last seen: 1 year 7 months ago
Joined: 01/03/2013 - 01:16
Posts: 1153
Location: N 11º 08', E 124º 37'

Wow! Thanx for these tests djozz! Smile

I’ve been planning on drilling a 2.5mm hole through the dielectric layers of my cheap copper MCPCBs for XM-L and fill the hole with an AWG 10 copper wire. I tried it with my hand drill before but I just can’t bore trough. I’ve contacted a friend who has a drill press and we’ve scheduled the boring after the Holy Week. Based on your tests, these cheap copper MCPCBs should be just like (almost?) SinkPads or Noctigons. :bigsmile:

dct73
Offline
Last seen: 2 years 6 months ago
Joined: 06/29/2012 - 21:40
Posts: 384
Location: wisconsin

Thanks. Your efforts are really appreciated!

WarHawk-AVG
WarHawk-AVG's picture
Offline
Last seen: 1 year 3 months ago
Joined: 01/04/2014 - 06:47
Posts: 5071
Location: H-Town

Awesome findings…proof positive that the direct heat path to the star does actually increase the heat transfer resulting in less heat sag and more light…and it doesn’t matter how that direct heat path is made, whether it be by sinkpad pressing up the center or cutting out the dielectric from a cheap pad and filling with solder

Great post man…great findings!

How long did you let the light run before you took your readings? Did you do a first on, then 30 second later test once things got nice and warm…just to compare what happens when the heat builds up

This test along with a thermal test of the emitter to see how hot it is would be a good identifier of which is better at moving the heat away from the emitter

Either way…AWESOME test man!

And would you say spending the extra $.75 for an actual sinkpad over the cheap dielectric covered FT copper star is worth it rather than doing all that surgery?

M4D M4X
M4D M4X's picture
Offline
Last seen: 3 weeks 4 days ago
Joined: 03/19/2014 - 05:17
Posts: 8821
Location: Austria (GMT + 1)

very cool sry, hot tests you did!

PLEASE NOTE
i do not work in "reviews, deals and codes" for the time being
maybe M4D M4X will return one day, but until then:

THANK YOU FOR YEARS OF YOUR SUPPORT AND FRIENDLY CONTACTS!

djozz
djozz's picture
Offline
Last seen: 3 hours 25 min ago
Joined: 09/07/2012 - 17:04
Posts: 17954
Location: Amsterdam

WarHawk-AVG wrote:
How long did you let the light run before you took your readings? Did you do a first on, then 30 second later test once things got nice and warm...just to compare what happens when the heat builds up This test along with a thermal test of the emitter to see how hot it is would be a good identifier of which is better at moving the heat away from the emitter

Very relevant question of course. Each test started at 0.5A, 10 seconds wait, voltage and output reading was written down, current was turn up to 1A, 10seconds wait, etc. up to 7 amps or max output. So the power went up gradually during the test. I monitored the luxmeter of course, and found that nowhere during the tests the readings were fast declining, when waiting longer before taking the reading the number only very slowly went down.  This test with so many numbers  could not have been done with after each new current waiting everything to settle to the max, instead I waited for the same 10 seconds each time, and I think it gave a good enough number that did not alter very much over time.

WarHawk-AVG
WarHawk-AVG's picture
Offline
Last seen: 1 year 3 months ago
Joined: 01/04/2014 - 06:47
Posts: 5071
Location: H-Town

Yeah, appears that way, so by the time it got to 3A it had already been “cooking” for a bit, even longer at higher currents…so the heat sag wouldn’t be an issue with the readings…nice!

So do you think it’s cost effective to get sinkpad over the cheap FT copper stars and cut out the dielectric layer, or just too much hassle

leaftye
Offline
Last seen: 9 months 5 days ago
Joined: 07/25/2012 - 17:43
Posts: 4278
Location: San Diego, CA

Thanks for the test.  This is very good information.  Much more useful than I thought it would be thanks to those additional boards you tested.  Now I wonder why the copper board with the dielectric layer performed so well.

The low mode should be lower.

djozz
djozz's picture
Offline
Last seen: 3 hours 25 min ago
Joined: 09/07/2012 - 17:04
Posts: 17954
Location: Amsterdam

leaftye wrote:

Thanks for the test.  This is very good information.  Much more useful than I thought it would be thanks to those additional boards you tested.  Now I wonder why the copper board with the dielectric layer performed so well.

Exactly, that is what I wonder as well: the minor difference between the aluminium and copper Sinkpads suggest that the board material is not critical -the bottleneck is closer to the led-, but when a dielectric layer is present, copper and aluminium are miles apart, makes no sense Undecided . Perhaps the dielectric layer in the copper board has a lower thermal resistance as the dielectric layer in the aluminium board? I scraped that layer in the copper board away: it was not exactly a thin layer, much thicker than I expected actually.

ryansoh3
ryansoh3's picture
Offline
Last seen: 7 months 3 weeks ago
Joined: 08/04/2012 - 08:21
Posts: 3890
Location: US

Subbed. Thanks for the review!

BLF ≠ B-grade Flashlight Forum

 

PH_Oton
PH_Oton's picture
Offline
Last seen: 2 months 4 days ago
Joined: 01/21/2014 - 08:15
Posts: 158
Location: Croatia

Amazing job djozz and thanks for all the effort! You helped me a lot with your charts and tests and I’m sure lot of other people share my thankfulness.

Ledsmoke
Ledsmoke's picture
Offline
Last seen: 1 year 2 months ago
Joined: 08/08/2011 - 16:05
Posts: 1995
Location: Denmark

Super tester! Thank you for these tests. That is just so useful and revealing. Thank you thank you thank you Smile

~ Ledsmoke ~

Dutch humor:

[quote=djozz]

 I do not think that the BLF-community ben

Pulsar13
Pulsar13's picture
Offline
Last seen: 10 months 2 days ago
Joined: 02/05/2012 - 10:23
Posts: 1909
Location: Malaysia

Thank you so much djozz. This is a great effort. The only issue is that you misspelled “.” with “,” on your notes all the way.

Sorry can’t help myself. Just kidding. Smile

I am actually quite surprised to see copper+dielectric is handling its own up till higher amps. I have a bunch of them from Aliexpress and am still thinking of the best method to punch through to copper layer. And then…

ImA4Wheelr wrote:
I use the cheap FT copper boards in under 5 amp builds.  They are in some of my documented builds here.  I don’t cut out the whole dialectic layer.  I gouge slot through it with a straight screwdriver. 

Seems to fit my style. How deep did you need to dig/punch? Got any pics?

Nightbird95
Nightbird95's picture
Offline
Last seen: 1 year 7 months ago
Joined: 01/03/2013 - 01:16
Posts: 1153
Location: N 11º 08', E 124º 37'
djozz wrote:
… the aluminium Sinkpads perform slightly worse than the copper one, but in real life it is the same, version 1 or 2 does not matter…

The aluminum SinkPads turned out to perform well. Would ordinary aluminum MCPCBs perform the same when its dielectric will be removed just like the copper MCPCBs? :~

Jerommel
Jerommel's picture
Offline
Last seen: 1 year 4 months ago
Joined: 01/04/2014 - 13:18
Posts: 6380
Location: the Hague, Netherlands

Probably…

Copper hype… Smile

And compliments to Djozz for this interesting topic.

ImA4Wheelr
Offline
Last seen: 5 days 3 hours ago
Joined: 02/03/2013 - 14:51
Posts: 7935
Location: SC

Pulsar13 wrote:

Seems to fit my style. How deep did you need to dig/punch? Got any pics?

Deeper than you would think.  Those combined layers are thicker than I imagined.  It may also be that I am compressing the underlying copper, but I try not to do that.  Not sure if I have any lose ones around.  If I do, gouge it and post a pic in the next day or two.

I have no idea how it performs.  I'm guessing somewhere between the djoss' number 3 and 4 bases.

TexasToasted
TexasToasted's picture
Offline
Last seen: 20 min ago
Joined: 10/25/2012 - 08:48
Posts: 1326
Location: Texas Gulf Coast
Nightbird95 wrote:
djozz wrote:
… the aluminium Sinkpads perform slightly worse than the copper one, but in real life it is the same, version 1 or 2 does not matter…

The aluminum SinkPads turned out to perform well. Would ordinary aluminum MCPCBs perform the same when its dielectric will be removed just like the copper MCPCBs? :~

The problem is, how are you going to solder the LED to the bare aluminum?

rollinstone157
Offline
Last seen: 6 months 3 weeks ago
Joined: 12/31/2013 - 13:08
Posts: 300
Location: NJ-PA Area

Thanks DJozz! Though I think you might have your #‘s wrong in the first picture. IIRC, the one with the tc cutout in the dielectric layer is the SinkPad 2. Backed up by the fact that the XM-L2 is of course a newer LED than the XM-L, it would seem odd to me for the “newer” SinkPad 2 (that you have listed as 5) to ignore the XM-L2, while the “older” SinkPad 1 would have it listed on the layer. I’ll pull apart my SupFire M6 to check—I need to lap those puppies anyway.

EDIT: Not really necessary, if you zoom in on the photo, the one you have labeled as 4 has SinkPad II written on the bottom, but listed as SinkPad V1 in the key, and the one you have label as 5 has SinkPad written on the bottom, but is listed as SinkPad V2 in the key.

Not a huge deal though Smile

djozz
djozz's picture
Offline
Last seen: 3 hours 25 min ago
Joined: 09/07/2012 - 17:04
Posts: 17954
Location: Amsterdam

You are completely correct, I mixed the two alu Sinkpads up in the first picture, luckily just in that picture, not in the tests themselves Smile

I will make a note in the OP.

RaceR86
RaceR86's picture
Offline
Last seen: 5 years 6 months ago
Joined: 02/22/2012 - 08:32
Posts: 3777
Location: Norway

WOW. Djozz. Once again, FANTASTIC work!

Very useful information to have.

Thank you!

BLF LED database – collaboration spreadsheet and latest news about where to buy LEDs
http://budgetlightforum.com/node/19342

Nightbird95
Nightbird95's picture
Offline
Last seen: 1 year 7 months ago
Joined: 01/03/2013 - 01:16
Posts: 1153
Location: N 11º 08', E 124º 37'
TexasToasted wrote:
Nightbird95 wrote:
djozz wrote:
… the aluminium Sinkpads perform slightly worse than the copper one, but in real life it is the same, version 1 or 2 does not matter…

The aluminum SinkPads turned out to perform well. Would ordinary aluminum MCPCBs perform the same when its dielectric will be removed just like the copper MCPCBs? :~

The problem is, how are you going to solder the LED to the bare aluminum?

Yeah I’ve been asking myself that same question too. :~

So I think I’ll do the same thing to aluminum MCPCBs what I’ll be doing to my copper MCPCBs – drill a 2.5 mm hole at the dielectric area then fill the hole with copper from an AWG 10 wire which is about 2.58 mm thick. That should do the trick (I hope). Smile

leaftye
Offline
Last seen: 9 months 5 days ago
Joined: 07/25/2012 - 17:43
Posts: 4278
Location: San Diego, CA

Nightbird95 wrote:
TexasToasted wrote:
Nightbird95 wrote:
djozz wrote:
... the aluminium Sinkpads perform slightly worse than the copper one, but in real life it is the same, version 1 or 2 does not matter...
The aluminum SinkPads turned out to perform well. Would ordinary aluminum MCPCBs perform the same when its dielectric will be removed just like the copper MCPCBs? :~
The problem is, how are you going to solder the LED to the bare aluminum?
Yeah I've been asking myself that same question too. :~ So I think I'll do the same thing to aluminum MCPCBs what I'll be doing to my copper MCPCBs - drill a 2.5 mm hole at the dielectric area then fill the hole with copper from an AWG 10 wire which is about 2.58 mm thick. That should do the trick (I hope). :)

Electroplating

The low mode should be lower.

nikanon
Offline
Last seen: 5 years 10 months ago
Joined: 08/25/2012 - 08:50
Posts: 439

Thank you for this test and the well presented results even a non-tech person like me can understand easily.

I got one question, how good/bad will a 20mm noctigon board perform in comparison to the sinkpads ?

Pulsar13
Pulsar13's picture
Offline
Last seen: 10 months 2 days ago
Joined: 02/05/2012 - 10:23
Posts: 1909
Location: Malaysia

nikanon wrote:
Thank you for this test and the well presented results even a non-tech person like me can understand easily.

I got one question, how good/bad will a 20mm noctigon board perform in comparison to the sinkpads ?

About the same. From our members’ tests, pretty sure we can establish now that the limiting factor is almost always the dielectric layer. Once that factor is taken out, even aluminium base gives performance not too far from Noctigon/Sinkpad.

Pages