Luminus CFT-90 Testing - The Mother of all LEDs

239 posts / 0 new
Last post

Pages

The_Driver
The_Driver's picture
Offline
Last seen: 16 hours 38 min ago
Joined: 10/20/2016 - 05:51
Posts: 1423
Location: Germany
Luminus CFT-90 Testing - The Mother of all LEDs

sma in the German TLF forum has tested the Luminus CFT-90, the new high-power LED from Luminus. This LED has a 9mm^2 Die, no dome and is rated for up to 27A. It’s factory mounted on a large copper pcb. sma tested it up to 45A (I had guessed beforehand that it goes up to at least 40A). See here.

As I predicted it has a suprisingly high luminance for such a large die. It managed up to 230cd/mm^2 which is better than any XP-G2 and almost as high as the Osram Black Flat at maximum current, but with 9-times the die size! At 40A, 4.86V, 194W it produced 5700 Lumens. Of course a XHP-70.2 can be much brighter, but it also has a much larger die and thus a lower luminance. There is also the fact that this Luminus LED is actually designed for continuous high-power use. The XHP-70.2 and XHP-50.2 and only designed for a maximum of around 30W and 15W.

Using this LED in a “flashlight” is of course very costly and requires a lot of effort. The LED already costs 100$+. A 20-40A Buck driver is also hard to get (I only found one for 90$). It will also be rather large. Keeping the LED cool is of course the most difficult part though. Active cooling is probably a must if one wants to run it continously at maximum power.

Edited by: The_Driver on 06/20/2018 - 18:04
The Miller
The Miller's picture
Offline
Last seen: 1 year 7 months ago
Joined: 12/14/2015 - 12:08
Posts: 9630
Location: Charente France

Do you know if sma is on the list for a GT?
Edit at below, yes, ah well, would have been cool to have him mod the GT with this

The_Driver
The_Driver's picture
Offline
Last seen: 16 hours 38 min ago
Joined: 10/20/2016 - 05:51
Posts: 1423
Location: Germany

You mean sma? I don’t think so, he mostly prefers HIDs.

L4M4
L4M4's picture
Offline
Last seen: 1 hour 44 min ago
Joined: 12/08/2013 - 13:43
Posts: 1307
Location: Germany

“194W it produced 5700 Lumens”

30Lm/W?
Which year is it?

Plus it’s extremely expensive and big.
no competitor for a XHP70.2

djozz
djozz's picture
Offline
Last seen: 7 hours 24 min ago
Joined: 09/07/2012 - 17:04
Posts: 15955
Location: Amsterdam

The_Driver wrote:
You mean sma? I don’t think so, he mostly prefers HIDs.

..and helicopter surplus stuff… Big Smile
The_Driver
The_Driver's picture
Offline
Last seen: 16 hours 38 min ago
Joined: 10/20/2016 - 05:51
Posts: 1423
Location: Germany

L4M4 wrote:
“194W it produced 5700 Lumens”

30Lm/W?
Which year is it?

Plus it’s extremely expensive and big.
no competitor for a XHP70.2

Every LED is inefficient when driven with a high power density. The XHP70.2 is no exception there. The Luminus LED has much higher luminance and thus will get you more throw. It will produce an amazing beam, that most here are not used to.

djozz
djozz's picture
Offline
Last seen: 7 hours 24 min ago
Joined: 09/07/2012 - 17:04
Posts: 15955
Location: Amsterdam

L4M4 wrote:
“194W it produced 5700 Lumens”

30Lm/W?
Which year is it?

Plus it’s extremely expensive and big.
no competitor for a XHP70.2


It is the 3×3mm die that does the talking, a XPL has a ~4 mm2 die, a XHP70.2 4 times that: 16 mm2. So with 45% smaller die it produces just 32% less light than a XHP70.2 at maximum current (~20 A).

So it throws better than a XHP70.2, by 20%

(numbers scraped away from various places, correct me if I made a mistake)

The_Driver
The_Driver's picture
Offline
Last seen: 16 hours 38 min ago
Joined: 10/20/2016 - 05:51
Posts: 1423
Location: Germany

When comparing them one needs to take into account that the XHP70.2 will be dimmer after de-doming. So the difference in luminance is larger. The XHP70.2 has been tested up to 145W, so I will compare both LEDs at this power.

 

XHP70.2 P2 - 20A - 7.25V - 8300lm @ 85°C with dome - 20% less without dome would be 6640lm - ~149cd/mm^2 (XP-L2 de-domed is 3.55mm^2, so the XHP70.2 de-domed is 14.2mm^2) - 46lm/W

CFT-90 - 38A - 3.81V - 5600lm @ ??°C without dome - 228cd/mm^2 - 39lm/W

I don't know what temperature the LED had in sma's test though. The real values might be a bit different.

 

So the CFT-90 has a roughly 50% higher luminance compared to the XHP70.2 at 145W. Thats a noticeable difference! Both have a very low efficiency at such high power levels. The difference between the two is rather small in this regard. Luminus has definitely caught up! When considering power density the CFT-90 is actually much better compared to the Cree LED.

 

The temperature is also important. The XHP70.2 has a thermal resistance of 0.9°C/W without the PCB and the CFT-90 has 0.45°C/W with the PCB. So in reality the value of the XHP70.2 will be a bit higher. Both have a maximum allowed junction temperature of 150°C. Lets say the heatsink (flashlight) has a temperature of 50°C @ 145W (maximum temp that is pain free). Here's a rough calculation (ignoring the LED-heatsink interface etc.):

CFT-90: (145W * 0.45°C/W) + 50°C = 115°C

XHP70.2: (145W * 0.9°C/W) + 50°C = 180°C

So with this heatsink the XHP70.2 would have a high chance of failure and the Luminus LED would have no problem even though it has a much higher intensity and is less efficient. That's a large advantage if you ask me.

 

At this power level the heatsink of the XHP70.2 needs to stay cooler than 20°C (to stay under 150°C). You would need powerful active cooling to achieve this. The heatsink of the Luminus LED on the other hand only needs to stay under 85°C. That will be difficult but doable. 

mrheosuper
Offline
Last seen: 2 days 4 hours ago
Joined: 09/30/2016 - 12:44
Posts: 1488
Location: Ho Chi Minh City, Vietnam

last time i checked the price of sst-90 is very high, so i bet this led is no exception

Forgot my pen

The_Driver
The_Driver's picture
Offline
Last seen: 16 hours 38 min ago
Joined: 10/20/2016 - 05:51
Posts: 1423
Location: Germany
mrheosuper wrote:
last time i checked the price of sst-90 is very high, so i bet this led is no exception

It’s much more expensive compared to the SST-90. >100$, but I find this genereally reasonable. It is designed to replace Xenon short-arc lamps. In the long run it is much cheaper compared to them because they have a rather short lifetime and require more power. It is also by far the most powerful officially rated LED on the market.

DB Custom
DB Custom's picture
Offline
Last seen: 14 hours 19 min ago
Joined: 01/13/2013 - 22:28
Posts: 20324
Location: Heart of Texas

Interesting. Burn 27A worth of power to get what the 70.2 can do with 8A or so. I have a 70.2 in my Convoy L6 making 9200 lumens off of 2 26650 cells, commonly available driver and the emitter is less than $20. Not really seeing the appeal of the luminus.

I’ve liked the CBT-140 for a while, but it, like other Luminus emitters, is so inefficient it’s hardly worth the triple digit price point.

Edit: Not to mention that the extremely common and cheap XM-L2 is capable of a mile in throw (actually a measured 815Kcd) from a reflectored light. At only 4.6A. From a single cell. Simple FET driver.

The_Driver
The_Driver's picture
Offline
Last seen: 16 hours 38 min ago
Joined: 10/20/2016 - 05:51
Posts: 1423
Location: Germany

DB Custom wrote:
Interesting. Burn 27A worth of power to get what the 70.2 can do with 8A or so. I have a 70.2 in my Convoy L6 making 9200 lumens off of 2 26650 cells, commonly available driver and the emitter is less than $20. Not really seeing the appeal of the luminus.

I’ve liked the CBT-140 for a while, but it, like other Luminus emitters, is so inefficient it’s hardly worth the triple digit price point.

Edit: Not to mention that the extremely common and cheap XM-L2 is capable of a mile in throw (actually a measured 815Kcd) from a reflectored light. At only 4.6A. From a single cell. Simple FET driver.

The XHP70.2 P2 does not produce 5000 lumens at 8A when de-domed, maybe 4000. The Luminus LED will produce much more throw. More than the XHP70.2. More than any XM-L2 de-domed. It beats them all. In addition to this it produces a lot of lumens. The XM-L2 will do up to around 1700 lumens when de-domed. That’s quite a bit less. So compared to a de-domed XM-L2 you will get a hotspot that is 2.5x as large and at the same time probably around 30-45% more Candela.

The CBT-140 is basically the 5-years old predecessor. The CFT-90 is new, much more efficient and has a much higher luminance.

The appeal is this: it’s the best. The best always costs a premium. If you want over 100W of power in a single power-LED continously and not worry about the reliability, this LED is the only option (because of the lower temperature, see my post above).

Jerommel
Jerommel's picture
Offline
Last seen: 5 months 1 week ago
Joined: 01/04/2014 - 13:18
Posts: 5721
Location: the Hague, Netherlands

DB Custom wrote:
Interesting. Burn 27A worth of power to get what the 70.2 can do with 8A or so.
The xhp70.2 needs about double the Voltage, so you should look at U × I = P (Power in Watts).

2Q19

Agro
Agro's picture
Online
Last seen: 39 sec ago
Joined: 05/14/2017 - 11:16
Posts: 4737
Location: Ślōnsk

I wouldn’t compare the numbers at the peak output. The inefficiency is indeed terrible. But at around 30A it’s much better in this regard (which doesn’t make it good actually). And it will still out-throw nearly anything on the market. And certainly anything close to its output.
I don’t see the point comparing it to XHP70.2. It’s much more powerful and efficient, but much bigger. XHP50.2 is much closer, actually pretty much the same size. And max output is close too, 5000 lumens with the dome. 50.2 leaks light to the sides, so it’s much less throwy with the same output. Efficiency? ~80W@5000 lm for XHP50.2 (with dome) and ~110W for CFT-90. Dedome and you’ll arrive somewhere in the 4000-4500 lm range with the same 80W consumption. Scale CFT-90 down to this level, it takes ~70W@4500 lm and ~60W@4000 lm.

KawiBoy1428
KawiBoy1428's picture
Offline
Last seen: 7 hours 47 min ago
Joined: 04/11/2014 - 18:05
Posts: 3826
Location: The Motor City

The only thing I can see this LED being easily used in is a Big Aspherical Lens/ Fresnel build, can’t see a massive reflector with a massive aperture being cost effective.

$200.00 for LED/Driver with active cooling in an Aspherical Light with the cost of materials, machining and powering up to 40A….I don’t like aspheric lens lights, so it’s not attractive to me.

But I hope some one with some $$$$ bucks goes forward with it…. Wink Thumbs Up

KB1428 “Live Life WOT

DB Custom
DB Custom's picture
Offline
Last seen: 14 hours 19 min ago
Joined: 01/13/2013 - 22:28
Posts: 20324
Location: Heart of Texas

Oh, don’t get me wrong… I’m not saying I wouldn’t like to try it out and see, but at the price point, the offset COB design, and the extremely high amp requirements, I don’t see myself doing anything with this one for quite a while.

Enderman
Enderman's picture
Offline
Last seen: 9 hours 41 min ago
Joined: 11/03/2016 - 22:42
Posts: 3965
Location: Vancouver, Canada

When you want max throw, you can throw efficiency out the window.
The point is to get the highest possible numbers, and this is currently #3 on the list of highest intensity LEDs.
What’s so cool about it is that it can do this while also outputting a huge amount of lumens, which the first two don’t do.

Perfect LED for applications such as moving heads, which are starting to transition to LED:

And sky beams which are currently all short-arc:

.
Neither of these are “flashlights” and neither of them are “budget” so the majority of people on this forum won’t really get excited by this LED Silly

yclongthrow
Offline
Last seen: 1 week 2 days ago
Joined: 08/20/2011 - 14:21
Posts: 368
Location: USA

So hypothetically, what kind of numbers would be seen in a BLF GT sized reflector?

Enderman
Enderman's picture
Offline
Last seen: 9 hours 41 min ago
Joined: 11/03/2016 - 22:42
Posts: 3965
Location: Vancouver, Canada

yclongthrow wrote:
So hypothetically, what kind of numbers would be seen in a BLF GT sized reflector?

2.3Mcd, ~3km throw
~4k lumens in spot, ~1.2k in spill.
265m spot diameter at 1km
10 degree beam divergence.
djozz
djozz's picture
Offline
Last seen: 7 hours 24 min ago
Joined: 09/07/2012 - 17:04
Posts: 15955
Location: Amsterdam

I indeed did not take the domelessness (is this a word? Innocent ) of the CFT-90 into account in my calculation, that makes a big difference.

If any host is suitable for this led it must be the BLF-GT, in fact its bulkiness and that it takes 8× 18650 batteries makes it more tailormade for the CFT-90 than for the XHP35, as if the GT temporarily is fitted with a XHP35 but in reality was waiting for this led Smile .

Enderman
Enderman's picture
Offline
Last seen: 9 hours 41 min ago
Joined: 11/03/2016 - 22:42
Posts: 3965
Location: Vancouver, Canada

High thermal mass doesn’t necessarily mean high thermal dissipation, if this LED was driven anywhere near its max in a GT it would overheat instantly.
I wouldn’t recommend it for any ‘normal’ flashlight form factor, more for something with AT LEAST a fan cooled heatsink, or preferably liquid cooling.

djozz
djozz's picture
Offline
Last seen: 7 hours 24 min ago
Joined: 09/07/2012 - 17:04
Posts: 15955
Location: Amsterdam

Enderman wrote:
High thermal mass doesn’t necessarily mean high thermal dissipation, if this LED was driven anywhere near its max in a GT it would overheat instantly.
I wouldn’t recommend it for any ‘normal’ flashlight form factor, more for something with AT LEAST a fan cooled heatsink, or preferably liquid cooling.

Mind that a well-fixed Q8 runs at over 20A already, at a comparable voltage, so half the current of the CFT-90 at maximum output, and does that for 3 minutes without getting worryingly hot. With its larger mass and surface area the GT should be able to run the CFT-90 at over 40 A for three minutes too before stepping down.
Enderman
Enderman's picture
Offline
Last seen: 9 hours 41 min ago
Joined: 11/03/2016 - 22:42
Posts: 3965
Location: Vancouver, Canada

djozz wrote:

Mind that a well-fixed Q8 runs at over 20A already, at a comparable voltage, so half the current of the CFT-90 at maximum output, and does that for 3 minutes without getting worryingly hot. With its larger mass and surface area the GT should be able to run the CFT-90 at over 40 A for three minutes too before stepping down.

First of all, the heat from a Q8 is spread out over 4 points, the CFT90 is a single LED so the junction temperature will be 8x higher than a Q8 (since it will be producing 2x the heat output).
For this reason you would need to use a large copper base heatsink, or a TEC.

Second, the surface area of the GT is probably close to 10% that of the heatsink used in SMA’s test. Heatsinks have tens or hundreds of times more surface area than a flashlight like the GT with its few shallow fins.

Third, the mass doesn’t matter at all for heat dissipation, it’s just an energy reservoir.
If all you care about is a few seconds of high output then sure, the GT will work, but for practical purpouses and actually using the light for more than a few minutes on max it needs to have much higher thermal dissipation.

Personally I think any light that can’t maintain its max output for long periods of time is basically useless, it may as well not have that max setting.
Which is why I hate lights that are direct drive.

djozz
djozz's picture
Offline
Last seen: 7 hours 24 min ago
Joined: 09/07/2012 - 17:04
Posts: 15955
Location: Amsterdam

^ It is an ongoing discussion on BLF how desirable high output/fast stepdown flashlights are. For maintaining the high output of the CFT-90 for longer time the GT will not suffice, but I’m sure someone will do the mod Big Smile and I’m curious what runtime on max it will have Smile

The Miller
The Miller's picture
Offline
Last seen: 1 year 7 months ago
Joined: 12/14/2015 - 12:08
Posts: 9630
Location: Charente France

Ah come on if those 25K+ lumen lights offer enough fun for a few second to such extend people buy it, sure seeing this monster led in the GT should be awesome Big Smile

AlexGT
AlexGT's picture
Offline
Last seen: 7 hours 30 sec ago
Joined: 06/07/2012 - 17:39
Posts: 4372
Location: Texas

djozz wrote:
I indeed did not take the domelessness (is this a word? Innocent ) of the CFT-90 into account in my calculation, that makes a big difference.

If any host is suitable for this led it must be the BLF-GT, in fact its bulkiness and that it takes 8× 18650 batteries makes it more tailormade for the CFT-90 than for the XHP35, as if the GT temporarily is fitted with a XHP35 but in reality was waiting for this led Smile .

+1 hope the price comes down and efficiency go up…

Enderman
Enderman's picture
Offline
Last seen: 9 hours 41 min ago
Joined: 11/03/2016 - 22:42
Posts: 3965
Location: Vancouver, Canada

The Miller wrote:
Ah come on if those 25K+ lumen lights offer enough fun for a few second to such extend people buy it, sure seeing this monster led in the GT should be awesome Big Smile

Might as well just make a flashlight with a flash bulb so you get those 100k lumens for 1/10th of a second…
People buy dumb things just because of marketing, such as 25k lumens for 30 seconds.
Lexel
Lexel's picture
Offline
Last seen: 9 hours 30 min ago
Joined: 11/01/2016 - 08:00
Posts: 5473
Location: Germany

Enderman wrote:
djozz wrote:

Mind that a well-fixed Q8 runs at over 20A already, at a comparable voltage, so half the current of the CFT-90 at maximum output, and does that for 3 minutes without getting worryingly hot. With its larger mass and surface area the GT should be able to run the CFT-90 at over 40 A for three minutes too before stepping down.

First of all, the heat from a Q8 is spread out over 4 points, the CFT90 is a single LED so the junction temperature will be 8x higher than a Q8 (since it will be producing 2x the heat output).
For this reason you would need to use a large copper base heatsink, or a TEC.

Second, the surface area of the GT is probably close to 10% that of the heatsink used in SMA’s test. Heatsinks have tens or hundreds of times more surface area than a flashlight like the GT with its few shallow fins.

Third, the mass doesn’t matter at all for heat dissipation, it’s just an energy reservoir.
If all you care about is a few seconds of high output then sure, the GT will work, but for practical purpouses and actually using the light for more than a few minutes on max it needs to have much higher thermal dissipation.

Personally I think any light that can’t maintain its max output for long periods of time is basically useless, it may as well not have that max setting.
Which is why I hate lights that are direct drive.

this is not true, the CFT90 has 0.45K/W to the heat sink, while a XPL has 2.2K/W to solder pad
the MCPCB mount adds a bit to that so probably 2.5K/W

so at 20A the CFT90 die stays cooler

wolfdog1226
wolfdog1226's picture
Offline
Last seen: 5 months 1 week ago
Joined: 10/09/2013 - 20:47
Posts: 2120
Location: Acquasanta,Italia/ Valley Forge, Pa. USA
Enderman wrote:
The Miller wrote:
Ah come on if those 25K+ lumen lights offer enough fun for a few second to such extend people buy it, sure seeing this monster led in the GT should be awesome Big Smile
Might as well just make a flashlight with a flash bulb so you get those 100k lumens for 1/10th of a second… People buy dumb things just because of marketing, such as 25k lumens for 30 seconds.

+1

If a light can not sustain MAX output for at least a few minutes it is USELESS to me. I categorize it as a WOW light which is impractical for normal use at that level.

A lot of people buy these modded lights from V54 and others just for that[The WOW factor]. These people do not even use the lights that much! Within weeks or a few months a lot of them are up for sale!

To each his own.

Those lights are useless to CAPO and I on the trail. The longer a light can sustain MAX,,,,the better. That is my preference.

I know several people who probably have a 10K plus collection primarily for that WOW factor and not for every day use. They are entitled to their preference also although I do not understand it.

Solitude breeds contemplation which creates clarity. 

Environment molds a person. Perseverance changes them. 

                                                                   WOLFDOG 

Agro
Agro's picture
Online
Last seen: 39 sec ago
Joined: 05/14/2017 - 11:16
Posts: 4737
Location: Ślōnsk
wolfdog1226 wrote:
Enderman wrote:
The Miller wrote:
Ah come on if those 25K+ lumen lights offer enough fun for a few second to such extend people buy it, sure seeing this monster led in the GT should be awesome Big Smile
Might as well just make a flashlight with a flash bulb so you get those 100k lumens for 1/10th of a second… People buy dumb things just because of marketing, such as 25k lumens for 30 seconds.

+1

If a light can not sustain MAX output for at least a few minutes it is USELESS to me. I categorize it as a WOW light which is impractical for normal use at that level.

A lot of people buy these modded lights from V54 and others just for that[The WOW factor]. These people do not even use the lights that much! Within weeks or a few months a lot of them are up for sale!

To each his own.

Those lights are useless to CAPO and I on the trail. The longer a light can sustain MAX,,,,the better. That is my preference.

I know several people who probably have a 10K plus collection primarily for that WOW factor and not for every day use. They are entitled to their preference also although I do not understand it.
Enderman wrote:
The Miller wrote:
Ah come on if those 25K+ lumen lights offer enough fun for a few second to such extend people buy it, sure seeing this monster led in the GT should be awesome Big Smile
Might as well just make a flashlight with a flash bulb so you get those 100k lumens for 1/10th of a second… People buy dumb things just because of marketing, such as 25k lumens for 30 seconds.
Turbo mode on my D4 steps down after 10 seconds. Unless I turn it off before that, which happens half of the time. And I do because I use it to take a glimpse of a place that’s dark with the mode that I use at the time. So yes, this turbo is actually useful for some people.
The Miller
The Miller's picture
Offline
Last seen: 1 year 7 months ago
Joined: 12/14/2015 - 12:08
Posts: 9630
Location: Charente France

Despite not being able to cool this monster led on max, the GT still seems the best flashlight to try it in, just image what beam at not max amps will give Big Smile

Pages