MCPCB: help to build MCPCB sheet

Matley

Newly Enlightened
Joined
Feb 4, 2012
Messages
3
Hi all.
I'm new and i don't know a lot about led.
I'd like to build a MCPCB sheet.
I work in a galvanic industry and in a recent study I discover how to metallize over hard anodizing.
The aluminium is a 6000 alloy. The hard anodizing have a thickness of 40 micrometers.
From literature the breakdown voltage of 40 micrometers hard anodizing is more than 1000V. How can I test the breakdown voltage??
I chemically deposited a layer of 10 micrometers of nickel.
The next step is to electroplating a layer of copper.
My doubt is about the etching of the copper+nickel. What do you use to etch the copper of standard MCPCB?
The iron Chloride don't dissolve the nickel.Ammonia too. I etched the nickel layer with success with nitric acid but it require at least 20 minutes. I have to try it also with copper. My doubt is that nitric acid dissolve copper to fast and run underlayer.
The last question is about prices. How much cost actual MCPCB? i think my sheet could cost about 1,5 euro dm2. is it cheap?
Thank you in advance.
 

pretmetled

Newly Enlightened
Joined
Jan 30, 2012
Messages
59
From literature the breakdown voltage of 40 micrometers hard anodizing is more than 1000V. How can I test the breakdown voltage??

Apply increasingly high voltage and see when it breaks down. ;-)


I chemically deposited a layer of 10 micrometers of nickel.
The next step is to electroplating a layer of copper.
My doubt is about the etching of the copper+nickel. What do you use to etch the copper of standard MCPCB?

Don't they use the same process for the copper layes of MCPCBs as for the regular stuff? Also, check out this thread: http://www.candlepowerforums.com/vb...Has-anyone-ever-designed-and-had-made-a-MCPCB

The last question is about prices. How much cost actual MCPCB? i think my sheet could cost about 1,5 euro dm2. is it cheap?
Thank you in advance.

If you can make MCPCBs for that price then that's pretty cheap methinks.

You may however want to consider just using boring old double sided PCBs, and via the crap out of it. I recently looked into both MCPCB and double sided regular PCB, and unless someone can tell me why I shouldn't then double sided cheapo board with via's makes perfect economic sense. ;-)
 

Steve K

Flashlight Enthusiast
Joined
Jun 10, 2002
Messages
2,786
Location
Peoria, IL
I work in a galvanic industry and in a recent study I discover how to metallize over hard anodizing.
The aluminium is a 6000 alloy. The hard anodizing have a thickness of 40 micrometers.
From literature the breakdown voltage of 40 micrometers hard anodizing is more than 1000V. How can I test the breakdown voltage??

Testing the breakdown of insulators is not uncommon, and is commonly referred to as "hipot" testing, where "hipot" is short for "high potential" voltage. Hipot testers are found in electronics production facilities. Maybe there is a place near you that would be willing to work with you?

Steve K.
 

Matley

Newly Enlightened
Joined
Feb 4, 2012
Messages
3
Thank you both for the fast reply.
What parameters do I must consider to measure the characterisitic of a PCB dielectric?
I have the following information about hard anodizing (alumina) dielectric:
- dielectric constant is in the range of 7 to 10
- thermal conductivity is about 21-63 W/m
- voltage breakdown up to 1500v for 40micrometers of dielectric.

What king of etcher is used in MCPCB??
After etching how to solder over copper??
 

CKOD

Enlightened
Joined
Aug 3, 2010
Messages
708
Don't they use the same process for the copper layes of MCPCBs as for the regular stuff?

Nope, normal ones are aluminum with an insulating layer applied(i.e. applied vs grown with anodizing which would be much thinner and more thermally conductive), and then copper foil laminated down.


To measure the breaktown voltage, you would measure with an insulation tester as mentinoed. They can be pretty expensive, but to really qualify it, are needed. They will read out the insulation resistance in mega/giga ohms as they apply 250/500/1000/5000v etc to the device under test. Be sure your test coupon has edge clearance so that way it wont arc over from one conductor to another.

http://www.springerlink.com/content/k7440411158n6w76/ suggests that Ferric chloride can be used for etching nickel. However if its a lot slower than copper, it may be difficult to get precise etches that are even and not undercut. Normal profesisonal PCB manufacturing actually uses a negative resist, that after exposing and washing away, they plate tin onto the copper to use as an etch resist rather than the polymer. The rest of the negative is washed away, the exposed copper etched, the tin stripped, and you're left with precise copper traces.

Since nickel is more active than copper, perhaps you could do similar. I.E. after you apply your nickel coating, apply a masking material, expose it, develop it, and plate copper into the uncovered areas of nickel, and then use an etchant that will eat the nickel but not the copper.

Solder is applied to the copper by means of "hot air solder leveling" where the board runs though a solder bath, and then a stream of hot air blows excess solder from the board/traces.
 

Kinnza

Enlightened
Joined
Nov 15, 2005
Messages
223
Location
Spain
The process of making circuits on MCPCB and alumina is different. While typical PCB process is substractive (copper etched), circuit on alumina are additive, traces and pads are printed on the alumina (and then fired).

Search for thick film conductive pastes.
 

pretmetled

Newly Enlightened
Joined
Jan 30, 2012
Messages
59
While typical PCB process is substractive (copper etched), circuit on alumina are additive, traces and pads are printed on the alumina (and then fired).

Search for thick film conductive pastes.

Thanks, that explanation made perfect sense.
 

pretmetled

Newly Enlightened
Joined
Jan 30, 2012
Messages
59
- voltage breakdown up to 1500v for 40micrometers of dielectric.

Related question: how does this 40 micrometer layer compare to regular heatsinks? As in, say you buy an anodized aluminum heatsink from Chinese company XYZ, how thick is that anozide layer typically going to be?
 

alpg88

Flashlight Enthusiast
Joined
Apr 19, 2005
Messages
5,282
unless you make pcb that has heatpad of the led sit directly on the star, without insulating film between the star and led's heatpad, i don't see much sense making one at all, i mean what can you offer that is better than 1000's of stars already on the market??
after i tried copper star with led sitting directly on the copper, i'll never buy any different star, too bad it is only made for xm-l at this time.
insulation is tested by meter called megger.
 

alpg88

Flashlight Enthusiast
Joined
Apr 19, 2005
Messages
5,282
Related question: how does this 40 micrometer layer compare to regular heatsinks? As in, say you buy an anodized aluminum heatsink from Chinese company XYZ, how thick is that anozide layer typically going to be?
you only need layer that is enough to work with voltage that is in the circuit, it is counterproductive to make thicker that you need layer of anodizing (we talking about heat sink, not outside of the light), since insulation not only insulates voltage, but heat as well, it decreases heat transfer. something you don't want in a led star, it is not big number, but all adds up, star insulation, anodized heatsink, junctions... all thing together do make a difference. that is why i generally don't like stars, IMO one should avoid them as much as possible
 

pretmetled

Newly Enlightened
Joined
Jan 30, 2012
Messages
59
Thanks! As for stars, I presume they're popular because they do the job "well enough" and are easy to work with. Wouldn't know, never used one, but that is the impression I got. Personally I'd ratther use a 0.8 mm PCB with thermal via's, since this is way cheaper to get fabricated. The main worry there is the electrical insulation between the bottom side of the pcb and the heatsink. Haven't really found the easy solution I was looking for...
 

Matley

Newly Enlightened
Joined
Feb 4, 2012
Messages
3
It is becoming too difficult to me, cuz i'm a noob, so I reverse the question.
Could Metallization over hard anodizing (alumina) be used for some application?
I think it could have great characteristic at a low price. Hard anodizing is a cheap, well known process. You can have a better dielectric compared with epoxy one used in MCPCB, with high adhesion to the substrate. Metalization with nickel is done at low temperature (<100°C). Over nickel you can do easily a copper layer.
Thank you all guys.
icon10.gif
 

Kinnza

Enlightened
Joined
Nov 15, 2005
Messages
223
Location
Spain
It is becoming too difficult to me, cuz i'm a noob, so I reverse the question.
Could Metallization over hard anodizing (alumina) be used for some application?
I think it could have great characteristic at a low price. Hard anodizing is a cheap, well known process. You can have a better dielectric compared with epoxy one used in MCPCB, with high adhesion to the substrate. Metalization with nickel is done at low temperature (<100°C). Over nickel you can do easily a copper layer.
Thank you all guys.
icon10.gif

Yes, it could be used, as alumina acts as a dielectric barrier with very good thermal transfer. Actually, there are many manufacturers offering it (just search at any chinese manufacturers portal, as alibaba).

The question is if there are better alternatives, and the answer is yes. Aluminiun Nitride (AlN) has 3x the thermal conductivity of alumina (AlO3), with similar dielectric strenght. A little more pricey, but once you choose a substrate for the best thermal transfer, probably you wont get a worse solution for saving very little.

On the other hand, lately most LEDs comes with the thermal slug electrically isolated, allowing to build circuits lacking any dielectric barrier just below it.
 

pretmetled

Newly Enlightened
Joined
Jan 30, 2012
Messages
59
The question is if there are better alternatives, and the answer is yes. Aluminiun Nitride (AlN) has 3x the thermal conductivity of alumina (AlO3), with similar dielectric strenght. A little more pricey, but once you choose a substrate for the best thermal transfer, probably you wont get a worse solution for saving very little.

How much more is a little more pricey? Because roughly 3 times thermal conductivity (compared to boring old anodized) would be interesting as long as the price doesn't enter Beryllium Oxide price ranges. ;)

And shamelessly lazy question: how does the breakdown voltage of AlN compare to that of AlO3?

You mention thermal slugs ... Unfortunately these are only used on high power leds, and probably with good (economical) reason. For those of us wanting to use lower power leds, what's the better pin to do thermal via's on? Anode? Cathode? Doesn't matter? I can imagine there being a preference for one or the other due to leds not being perfectly symmetrical. But I'll be damned if I can think of a reasonable argument to pick either one.
 
Last edited:

Kinnza

Enlightened
Joined
Nov 15, 2005
Messages
223
Location
Spain
Difference in price between AlO3 and AlN is small, no more than 20% plus, but likely you can get similar price for both. BeO on the other hand, can go about 5x more pricey.

Breakdown voltage for all is between 15 and 20 KV/mm.

BeO is very rough and needs to be thick (relative to lenght), which somewhat offset is superior thermal conductivity (>200W/mK, vs >170W/mK for AlN and >65W/mK for AlO3).

You can get plates of any of them already metallized with Ni/Ti or Cu. Check for example this chinese manufacturer, which offer all three. 0.1-1$ each plate (dimensions not specified), 0.2-5$ metallized.

Other manufacturers offers the AlN and AlO3 as a thin dielectric layer on top of a Al heatsink.
 

SemiMan

Banned
Joined
Jan 13, 2005
Messages
3,899
What you are discussing is already available on the market:

http://www.irctt.com/anotherm/index.aspx

Has been for some time. Not quite the commercial success they were hoping for I believe.

1) Cost of having it made
2) The reality that the heatsink usually forms part of the light structure and hence this may not buy you much
3) Cost/difficulty of assembling LEDs to it

Semiman
 

DIWdiver

Flashlight Enthusiast
Joined
Jan 27, 2010
Messages
2,725
Location
Connecticut, USA
You mention thermal slugs ... Unfortunately these are only used on high power leds, and probably with good (economical) reason. For those of us wanting to use lower power leds, what's the better pin to do thermal via's on? Anode? Cathode? Doesn't matter? I can imagine there being a preference for one or the other due to leds not being perfectly symmetrical. But I'll be damned if I can think of a reasonable argument to pick either one.

You want to put the heatsink on whichever terminal has the better thermal path. I've seen it go both ways.

The LED die will typically sit on metal or ceramic substrate. The bottom of the die is either anode or cathode, and has large thermal path to the associated terminal or thermal pad. Then there is a tiny gold or aluminum wire (sometimes you can see it inside the encapsulation) connecting the top side of the die to the other terminal. In high power LEDs there are multiple wires at this point.

The back side of the die always has better thermal path to the outside, but unfortunately (at least in low-power LEDs) there's no reliable rule as to whether this is anode or cathode.
 

pretmetled

Newly Enlightened
Joined
Jan 30, 2012
Messages
59
You want to put the heatsink on whichever terminal has the better thermal path. I've seen it go both ways.

The LED die will typically sit on metal or ceramic substrate. The bottom of the die is either anode or cathode, and has large thermal path to the associated terminal or thermal pad. Then there is a tiny gold or aluminum wire (sometimes you can see it inside the encapsulation) connecting the top side of the die to the other terminal. In high power LEDs there are multiple wires at this point.

The back side of the die always has better thermal path to the outside, but unfortunately (at least in low-power LEDs) there's no reliable rule as to whether this is anode or cathode.

Thanks for your reply! That all makes sense. Yesterday I came upon a datasheet from I think it was an UFO style led that essentially said "you better use the anode for thermal management". But your point about it being whatever happens to be the one on the back/bottom side makes sense.

My mild frustration is that there are posts and posts about Thou shallt use MCPCBs! in the context of high power leds, but not so much about thermal management in general. Plenty of "folk wisdom", not enough analysis for my tastes. Could very well be I'm blind though and just not finding the good threads. :p What "folk wisdom" you ask? Well, stuff like the rule of thumb I see here and there that you should use 10 square inches for each Watt of emitter power. Ey? Is that single sided FR4, no via's, 1 oz copper, solder mask all over? Is that for 3 Watt plus emitters, or also for 100 mW emitters? Because easy rules generally are based upon lots of assumptions, which sometimes get lost when Timmy learns this rule from Tommy. Case in point, I was checking out the datasheet of a led array with 144 mid-power leds on it. Total power of the array was 24.5 Watt. And amusingly enough the board size came to under 13.5 square inches, or about 5.5 square inch for every Watt of emitter power. Now I already was suspicious of that "10 square inch per Watt" rule, so no surprise there. And since this was from an array with plenty of warranty and from a Way-Too-Expensive company I would guess that they didn't cut any particular engineering corners. At a guess it looked like 0.8 mm FR4, and not even riddled with vias. Kindof hard to guess from the pictures. And the leds were all smd, so no throughhole for some free extra thermal conduction to the backside of the pcb.

Sorry for my thermal management rant. *cough* And thanks again for your answer, one less doubt to deal with while having fun with leds. :)
 

Kinnza

Enlightened
Joined
Nov 15, 2005
Messages
223
Location
Spain
If you want to go deeper understanding thermal management, you should go to the webs of the main manufacturers. Phillips Lumileds, Cree and Osram has excellent application notes about it.
 

pretmetled

Newly Enlightened
Joined
Jan 30, 2012
Messages
59
If you want to go deeper understanding thermal management, you should go to the webs of the main manufacturers. Phillips Lumileds, Cree and Osram has excellent application notes about it.

Yep, thanks for the reminder. No single app note seems to have all the info I was looking for, but by reading through <too_many> pdfs you can get a reasonable understanding. Esp Cree and Osram were good.cd Do
 
Top