Вставка
- Опубліковано 7 сер 2024
- PCBWay: $5 for 2 layer PCBs with 24hr turnaround at www.pcbway.com/
Support the Channel with Patreon: / sdgelectronics
A big thank you to all of my Patreon supporters!
-------------------------------------------------------------------------------------------------
Soldering Tools and Equipment:
Solder Paste: bit.ly/2E1P5jR
Solder Flux: bit.ly/2MQn7we
Quick 857DW+ Hot Air Station: bit.ly/2OWDvf4
T12 Soldering Station: www.banggood.com/custlink/3KK...
AC Version: www.banggood.com/custlink/3GK...
T12 Iron Tips: www.banggood.com/custlink/3DG...
Daniu Tip Cleaner: bit.ly/31VHwWM
KSGER T12-BCM3 Tip: bit.ly/2xgYXCQ
Test Equipment:
Mustool MDS8207 Multimeter: bit.ly/2kKHr7A
FY6900 Generator: bit.ly/2yZI40n
Differential Probe: bit.ly/31xNTOF
Microscope Equipment:
Eakins Barlow Lenses: bit.ly/3aBY5tR
Eakins Black Microscope: bit.ly/2VEcYaW
Eakins Camera Microscope: bit.ly/2x86hnT
Camera Adaptor for Microscopes: bit.ly/2KBZQMQ
Amscope Microscope Head (US): amzn.to/2YmZWNx
Amscope Trinocular Microscope (US): amzn.to/2YdBE8k
Amscope Trinocular Microscope (UK): amzn.to/2V8w9Gh
Swift SS41-WF20 Binocular Microscope US: amzn.to/2VLWEpb
Swift SS41-WF20 Binocular Microscope UK: amzn.to/2GYGYFV
Visit my website: www.sdgelectronics.co.uk/
#led #mcpcb #cree #ElectronicsCreators - Наука та технологія
When I was beginning my adventure with power leds (Cree CXA arrays), I was having leds die on me (silicone cracking ripped bondwires), but the thermocouple didn't show anything abnormal. I quickly realized that the only way to get a reasonable picture of what's going on was to buy a thermal camera. This is also true with smaller leds, because thermocouple can give substantially false readings because of absorption of light.
Thermal camera helps a lot. For one, it explained me that the silicone gets very hot, probably because of energy loss in the phosphor. It also clearly showed me that my led array was heating abnormally in some spots, where i probably had either bubbles in thermal grease, or dents in the heatsink.
One more interesting conclusion was that I shouldn't use thermal-conducting epoxy - it degrades in about a year of use, probably due to massive amount of light getting through the white ceramic base of the arrays. Ultimately, I abandoned these arrays - they proved too unreliable. I have bought massive amount of relatively low-power leds, and they proved incredibly reliable even at high temperatures (110 deg.C) on regular FR4. (still not a single one of 2048 has died after 5 years or so of everyday use)
Interesting, in almost all my house i am running cxa1304 and cxa1507 for +-3 years and for now all the LEDs don't have any problem. However i am running them without thermal compound(for the 04 ones) both on 5°C/W heatsinks so they remain cold-ish. I am using screws to mount the LEDs (the wrong way, but didn't want to use glue/epoxy on them)
I am a fan of COB LEDs due to ease of mounting to a heatsink, even thought they are more expensive(comparing to 2835 0.005€/ea everlight LEDs)
Looks like your FR design was spot on. Fantastic overview.
I used metal core PCB's in my applications and they are very useful if you need electrical isolation between the board and the heatsink as the Aluminium core in this case is not electrically connected to the LED pad. There are also metal core PCB's with quite different characteristics in terms of thermal resistance between the copper layer and the aluminium core.
That's true, I forgot about the isolation aspects. PCBWay offer some options on the MCPCBs - the ones selected had the best thermal conductivity that they offer (2 W/mK).
Interesting tests, thanks Steve 👍️ Before the test I would have guessed there'd have been more of a difference between the two materials.
Cheers
John
That shows that the heat radiations from a aluminium PCB is about the same as a FR4, but without a heatsink on the back it do not say anything about how well the heat is transferred to the back.
Lovely video, thanks dude 😁 I've been gobbling up these videos with skin and all every time I see a new one pop up in my feed! 🤭 It's great to finally see the scope ring light saga (almost?) come to an end with all those unforseen issues and such! I'm hoping the ring light case doesn't give you too much trouble because I'm really curious about all those awesome new projects the future holds 🤓🥺
Do the same test with a fan on the heat sinks and you will find the led junction temperature will be lower on mcpcb once the leds temperature has stabilized. This is a nice project look forward to the redesign of the 3D printed bezel.
That might be worthy of a test.
@@sdgelectronics also you performed this test with heatsinks facing down. Since passive cooling revolves around convection, I wonder if that condition and close proximity of test subjects could affect the difference.
In case of active cooling result may still be the same if insulation layer of MCPCB cripples its thermal conductivity. Or it may be that the heatsink used is the real bottle neck for a steady state temp result and OP would seem right.
Excellent! That will save a bit of money :) I was actually considering the thermal performance as I saw this video appear and was thinking about getting thinner boards in FR4 to lower the thermal resistance but I don't know if its significant. Certainly there's no structural requirement for the boards to be as thick as they are.
The only reason in this application to set the FR4 boards to 0.8mm was the little pegs on the lenses would interfere with the heatsink if the PCB material was any thinner. If not using those, you could go for the thinnest PCB you can get if you can ensure good contact with the heatsink.
Good video
This is a very interesting outcome. I've been working with high power LED flashlights for a long time and I always considered PCB heatsinks to be inferior to metal core PCBs. But as you mentioned, the PCB must be designed properly to achieve this level of performance.
Very high power flashlights often use copper core PCBs for an even better heat transfer. With a copper PCB they can use a "direct thermal path" design, where the top FR4 layer has a cutout in the place of the LED heatsink pad. Therefore the LED gets soldered directly to the copper plate, bypassing the thermal resistance of the FR4 PCB layer.
Thanks !!
I think the real question should be which pcb's thermal conductivity is better - when put on the same heatsink and at around 3A (3A for 6V xhp50, or 6A for 3V version), I doubt that fr4 would transfer heat quicker that xhp50 melts that bismuth ;D ... anyway, that stuff is important for flashlights, where you need max conductivity in a very limited space, not in these applications where you have one room space for cooling...
Very interesting video. Thanks for the test Steve. Could the small difference in temperature be due to the black vs white soldermask? Matt black being the best emitter of IR.
Could you try measuring how the Al PCB fares if used without a heatsink? (Both short term and stabilized over time.) In that test, orientation as well whether the lens is attached should make a difference as well.
Thank u..
Are the pcbway drawings you made for the FR4 and this board available somewhere? How would I get them and or the parts list?
Good job, How do you do V-Cut at Proteus? Can you share the gerber files?
Did you make the fiducial points? Or did the manufacturer do it?
I wonder how closesly matched the two LEDs are in respect to light output, current consumption and thus thermal activity?
They are supposed to be binned and from the same batch. I did mean to compare light output but I wasn't sure if the black PCB would skew the readings.
This result seems odd, in almost all applications notes I have seen, mcpcbs tend to outperform FR4 with unfilled thermal vias by a decent margin.
Did you check the power consumption of both devices? Are the copper layers the same thickness? Are both pcbs the same thickness? Did you use thermal interface material between the metal and the heatsink when you mounted it?
This actually makes sense. If the power dissapation is same, on long term the temps will always equalize. It's not possible that one system runs cooler than other, unless the coupling of the heat source and heat sink is so bad, that considerable amount of heat is dissapated via the device itself into air instead of the heat sink. Clearly in this case, the coupling on FR4 is adequate for the application.
How did you order LED chip from Cree LED? possible order from Asia?
What was the little capacitor for on the LED PCB's? I was thinking it was ment for a anti flicker?
when used with a DC-DC converter, it helps reduce the ripple current at the LED
I understood that the MCPCBs are more for LEDs without thermal vias, especially when those are packed up in quite a big amount to reach luminous requirements. Correct me if I am wrong.
Its also for when the thermal pads can be electrically conductive, these Cree LEDs hes using have 2 configurations, the 3-5 volt configuration makes the pad isolated while the 12 volt config makes the pad not electrically isolated
This is OK for LEDs that have an electrically isolated thermal slug. But certain LEDs have the thermal slug electrically connected to the anode, so having a bunch of series-connected LEDs is problematic with the FR4 board and much better with aluminum-core PCB.
It would have been interesting to cover the stack up of the aluminium boards, are they .8mm FR4 bonded to .8mm aluminium and is 2 layer possible with them ?
Yes, 2 layer is the limit with the metal core either in the middle or on the bottom side. Not sure how vias are handled on the PCBs with the metal core in the middle though.
@@sdgelectronics I thought it would be 2 layer .8 FR4 bonded to .8 aluminium ?
Can you share that Lumileds article?
I was trying to find it yesterday, but they redesigned their website. If it recall correctly, it was a paper about cooling on their "new" Luxeon Rebel LEDs which at the time were the highest power LEDs on the market
@@sdgelectronics
I think I have found article.
lumstatic.com/En/0i/WQb7sgG5kseBR4dwjA.pdf
CNC kitchen of electronics..lol
Too close together... I think they are reaching mutual equilibrium
I love it when people just assume things like that and then a simple test proves them mostly wrong. People are SO SURE about things that they never think to do a test. Why would they test? it's obvious! Except it's wrong. Aluminum PCB material DOESN'T conduct heat to a heat sink better in any meaningful or significant way than copper vias.
Better than what?
V-Sscoring is designed for a PCB guillotine/de-paneler to separate the boards - not for hand wiggling.