LM3886T - alternative mounting - diyAudio
Go Back   Home > Forums > Amplifiers > Chip Amps

Chip Amps Amplifiers based on integrated circuits

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 8th January 2005, 02:12 PM   #1
wxn is offline wxn  Lithuania
diyAudio Member
 
wxn's Avatar
 
Join Date: Sep 2003
Location: Vilnius
Default LM3886T - alternative mounting

Hi.
I've came up with something new or so I suppose. After some searching and browsing I haven't found anything similar.

LM3886T has its case connected to V-, not ground as would be quite natural. So we need to isolate the chip from heatsink. The best solution discovered so far seems to be a thin mica layer (0.02-0.05mm), a bit of thermal grease and mounting like this:
http://sound.westhost.com/hs_fig5.gif

IMHO this is a bit clumsy to implement and even more so to take apart. And then after some thinking, epoxy glue came to mind. It is *quite* stiff, can withstand high temperatures and is one of the best insulators ever.

So why not just glue the chip to an aluminum (or preferably copper) bar under (heavy) pressure with mica or another insulator of <0.05mm thickness inside? This would ensure low thermal resistance chip>bar, (possibly) better heat spreading, easier mounting on the heatsink etc.

I'm attaching a picture of first prototype. The experiment was quite a failure (my fault) because of somewhat thick insulator - a peace of paper (0.15mm thickness as measured afterwards). Another bad thing was that the surfaces were not as flat as they should be, very far from shiny polished surface.

The result was quite satisfactory for average power levels up to 30W (tested with white noise and 100Hz sine wave), above that the chip>bar junction could not cope with the heat and eventualy SPiKe kicked in.

I suppose that correctly made, such approach would be perfectly suitable for all the power LM3886 can deliver and in the same time, it could solve some inconveniences. Any comments and thoughts are highly appreciated.
Attached Images
File Type: jpg pic(1).jpg (9.9 KB, 961 views)
__________________
Nothing is as simple as it seems
  Reply With Quote
Old 8th January 2005, 02:13 PM   #2
wxn is offline wxn  Lithuania
diyAudio Member
 
wxn's Avatar
 
Join Date: Sep 2003
Location: Vilnius
Here's another pic. Sorry for the quality
Attached Images
File Type: jpg pic(2).jpg (11.7 KB, 846 views)
__________________
Nothing is as simple as it seems
  Reply With Quote
Old 8th January 2005, 03:43 PM   #3
boholm is offline boholm  Denmark
diyAudio Member
 
boholm's Avatar
 
Join Date: Apr 2004
Location: Copenhagen
To make the right conclusions you must also mount a LM3886 as shown in http://sound.westhost.com/hs_fig5.gif. You must do this so you will have a reference to compare your experiments with. And you must make the same tests on both of them - and then compare.

Otherwise we cannot know if it really works.

But . . .

Mica and thermal pasta results in very low thermal resistance, which is important to lead the heat away from the component. Silicone pads can also be used and with those there is no need for grease. They come with a thermal resistace down to 0,2 șC/W which is pretty low.

I do not know how low a thermal resistance glue has. So maybe your experiments can tell us?
__________________
Best regards
Bo
  Reply With Quote
Old 8th January 2005, 04:16 PM   #4
diyAudio Member
 
demogorgon's Avatar
 
Join Date: Jul 2004
Location: Behind you.
Send a message via MSN to demogorgon
The best way to implement the lm3886 is actually to allow the chip direct metal contact and a very thin layer of artic silver between. this offcource does not allow for grounding of the heatsink, but it's a compromise i'd take any day.

what you have effectivly made here is a heatspreader, found on microprocessors as the amd 64 or P4. this is ok for something that has under a square cm of heat conducting area and up to 150w of power dispation (not unusal for overlocked cpu's), but for semiconductors it's limiting the thermal resistance quite a lot as it allready has heatspreader in the form of an pacaging. (to-3, to220 ect)

the more you put in between the heat sorce and the cooling, thee higher temps you aturally get.

what i would suggest is applying thermal epoxy glue like artic silver ilumina glue wich is ceramic based, and sand teh chip until only a very thin layer remains if you must have the heatsink grounded.
  Reply With Quote
Old 8th January 2005, 04:48 PM   #5
Sheldon is offline Sheldon  United States
diyAudio Member
 
Join Date: Dec 2003
Location: San Diego
Default Re: LM3886T - alternative mounting

Quote:
Originally posted by wxn
Hi.
I've came up with something new or so I suppose. After some searching and browsing I haven't found anything similar.

LM3886T has its case connected to V-, not ground as would be quite natural. So we need to isolate the chip from heatsink. The best solution discovered so far seems to be a thin mica layer (0.02-0.05mm), a bit of thermal grease and mounting like this:
http://sound.westhost.com/hs_fig5.gif

IMHO this is a bit clumsy to implement and even more so to take apart. And then after some thinking, epoxy glue came to mind. It is *quite* stiff, can withstand high temperatures and is one of the best insulators ever.

So why not just glue the chip to an aluminum (or preferably copper) bar under (heavy) pressure with mica or another insulator of <0.05mm thickness inside? This would ensure low thermal resistance chip>bar, (possibly) better heat spreading, easier mounting on the heatsink etc.

I'm attaching a picture of first prototype. The experiment was quite a failure (my fault) because of somewhat thick insulator - a peace of paper (0.15mm thickness as measured afterwards). Another bad thing was that the surfaces were not as flat as they should be, very far from shiny polished surface.

The result was quite satisfactory for average power levels up to 30W (tested with white noise and 100Hz sine wave), above that the chip>bar junction could not cope with the heat and eventualy SPiKe kicked in.

I suppose that correctly made, such approach would be perfectly suitable for all the power LM3886 can deliver and in the same time, it could solve some inconveniences. Any comments and thoughts are highly appreciated.
The approach could work. For reliabilities sake, I'd suggest a copper heat sink, or something with a very similar coefficient of expansion. The base for the chip is copper or a copper alloy. If you use something that expands at a different rate, you may eventually see the chip detach from the sink. That's one reason a thermal grease is used, it can go with the flow - so to speak.

Sheldon
  Reply With Quote
Old 8th January 2005, 04:54 PM   #6
wxn is offline wxn  Lithuania
diyAudio Member
 
wxn's Avatar
 
Join Date: Sep 2003
Location: Vilnius
A friend of mine recently tested the same heatsink with a similar setup and the usual mica+grease. He drew about 80-90W average power into 4 ohms and he could still hold his finger on the chip. That's more than anyone could wish and absolutely more than I will ever need because this is not a 200W sub amplifier, most likely it will run in a 2x paralleled or 2ch biamp setup. Of course, this heatspreader setup will probably perform a bit worse than that, but imho if done rightly, it will look much better.

Afterall, I'm tempted to try this thing out simply because nobody else seems to have done so And having in mind that I got these chips for less than 3$ each, not much of a sacrifice in case something goes not so well.

One more thing, what would you guys suggest as the insulator between chip and the heatspreader (great word, btw)? Mica seems to be the thinnest thing around but it may split (as the glue itself won't provide any pressure after it stiffens) and degrade the thermal conductivity. I have a few peaces of very thin (almost transparent) paper(!) at hand, it's about 0.04mm thick. That's less than the "standard" mica washer thickness (0.05mm). Also, the epoxy will soak through it (possibly) improving it's thermal conductivity. One might think that mica should be a better heat conductor by default as it is used everywhere, but keep in mind that all dielectrics have low thermal conductivity by nature.

All this should be tested I think. Any suggestions are welcome though.
__________________
Nothing is as simple as it seems
  Reply With Quote
Old 8th January 2005, 05:01 PM   #7
wxn is offline wxn  Lithuania
diyAudio Member
 
wxn's Avatar
 
Join Date: Sep 2003
Location: Vilnius
Default Re: Re: LM3886T - alternative mounting

Quote:
Originally posted by Sheldon


The approach could work. For reliabilities sake, I'd suggest a copper heat sink, or something with a very similar coefficient of expansion. The base for the chip is copper or a copper alloy. If you use something that expands at a different rate, you may eventually see the chip detach from the sink. That's one reason a thermal grease is used, it can go with the flow - so to speak.

Sheldon
I absolutely agree with You. Not only could epoxy crackle because of different metals, but copper is also almost twice as good at transferring heat (the whole point of this thing). Unfortunately, copper is very hard to come by here. I have recently contacted a (probably the one and only in this area) supplier and will probably make the next heatspreader of copper not aluminium if everything goes well.
__________________
Nothing is as simple as it seems
  Reply With Quote
Old 8th January 2005, 05:09 PM   #8
sam9 is offline sam9  United States
diyAudio Member
 
sam9's Avatar
 
Join Date: Jun 2002
Location: Left Coast
Quote:
IMHO this is a bit clumsy to implement and even more so to take apart. And then after some thinking, epoxy glue came to mind. It is *quite* stiff, can withstand high temperatures and is one of the best insulators ever.
IHO, all methods of attaching transistors or IC's to heatsinks are clumbsy. In the case of the LM3886 I have a pair bolted to the same heatsink using just a bolt and wide washer with insulation provided by a strip of Kapton tape and a smear of RS thermal grease. The heatsink is pretty skimpy but they have been in nearly daily operation for three years with no problems. I actually expected the internal thermal thermal protection to trip at least once but it has not. However, to be candid I chose a PS that makes this 50W/ch system as I'm highly suspicious of trying to push max power out of any of these IC-amps. Nonetheless, I think one of the virtues of the LM3886 is that you don't need to enage in heroic heatsinking methods.
  Reply With Quote
Old 8th January 2005, 11:03 PM   #9
diyAudio Member
 
thomas997's Avatar
 
Join Date: Feb 2004
Location: Vancouver
Have you tried thermal tape?

You can get it fairly thin, but then its probably not reliable in the long run as insulation..
  Reply With Quote
Old 8th January 2005, 11:18 PM   #10
wxn is offline wxn  Lithuania
diyAudio Member
 
wxn's Avatar
 
Join Date: Sep 2003
Location: Vilnius
Quote:
Originally posted by thomas997
Have you tried thermal tape?

You can get it fairly thin, but then its probably not reliable in the long run as insulation..
You mean smth like silicon pads? No, that's exactly what I am trying to avoid.
__________________
Nothing is as simple as it seems
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
10 pcs LM3886T daedalus Swap Meet 3 28th March 2009 04:39 AM
Wtb :- LM3886T PCB poynton Swap Meet 7 6th September 2007 05:18 AM
LM3886T for sale mat02ah Swap Meet 18 1st June 2006 08:20 AM
Lm3886t Nobodysfool Chip Amps 10 22nd May 2006 07:26 AM
what is the mounting depth and mounting diameter of this? lemans23 Multi-Way 2 26th October 2005 11:52 PM


New To Site? Need Help?

All times are GMT. The time now is 09:31 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright ©1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2