Heatsink approval needed for lm3886

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
help please

I can locally source this heatsink

7.8in (L) x 3in (W) x 1.9 (H) heatsinks. there are 20 fins in total.

Would it be suitable for 2 LM3886? If not, how about one?

I was looking at the PDF mentioned below (AN-1192) and the heatsink is longer in length (about about 2.4in) and taller as well (by 0.9in) compared to the heatsink recommendation for the BPA100 and BR100.

I am aiming for a +/- 35VDC supply which should be about 60W (according the graph in the LM3886 datasheet)

Thank you very much :)

ps. if it matters, I have non-insulated LM3886. does non-insulated has better thermal transfer? thanks again
 
Yes, the non-insulated package has better thermal dissipation. You should insulate the package from the heat sink with a very thin thermally conductive dielectric. I use very thin kapton tape for this.

What is the thickness of the baseplate? It would need to be at least .6cm (1/4") thick, the thicker the better. If it is much smaller than this, I would not feel comfortable running full power into even one LM3886 on it.

Running the LM3886 at +/-35VDC is pushing the envelope of the thermal dissipation ability of the package, especially if your load is not at least 8 ohm (most speakers dip below this for a significant portion of the spectrum). I'm sure you can find alot of experience with supply voltages on this forum. Maybe I'm too cautious. Just know your load well, exercise due diligence, and make a decision based on your needs.

The heatsink recommended for the bpa100 or br100 is intended for two chips in bridge or parallel operation. If your heatsink is larger than the recommended one in AN1192, then you should be very comfortable with 2 LM3886 on each heatsink given a reasonable operating environment and proper orientation of the fins. Just try to give each chip a bit of space between it and the other chip (say, 10 times the thickness of the baseplate as a general distance apart, about 5 or 6 cm).


Best regards,
David
 
Hi Jarthel,
can you find the heatsink in a catalogue?

I have some 0.6C/W sinks, they are 7.9"long by 3.94"high by 1.57" deep (200mmx100mmx40mm) the backplate is 10mm thick and the 22 fins are 30mm long.

Your sink is of similar proportions so 0.6C/W to 0.8C/W could be a close guess. Less tall but deeper fins.
You must de-rate it for low delta T. Probably 1.2 for DeltaT =40Cdegrees resulting in 0.7 to 0.96C/W.
Bolting two chips on this, spaced apart to ensure good heat spread, will be an effective 1.4C/W to 1.9C/W when running at 60 to 70degC. The charts in the datasheet seem to suggest your sink is big enough for two chipamps.

The cooler the chipamp runs the better it should perform, in particular the less often the Spike protection should trigger on transients.
 
this is the heatsink: https://www.soanarplus.com/getProductWebDetails.do?prodId=HH8546

it's not mentioned in the webpage but the thermal resistance is .0552C/W.

looking at the photo, the based is definitely around 6mm or more.

Would it be better if I use 2 heatsinks? I can get another model. thermal resistance = .0784C/W. it's almost the same dimensions except that this one is shorter by 50mm and there's a flange.

==================

just another question: The 2nd heatsink I mentioned comes in 2 types. flange (it's called bracket in the catalog) and flat base.

the flange type has lower thermal resistance.

maybe if I fashion a flange (screwing an L-shaped piece of aluminum to the base)for the 1st heatsink, the thermal resistance will increase?

Thank you very much
 
That heat sink should be very good for a 2 chip application.

The heat sink in AN1192 was conservatively chosen and was designed to dissipate a moderately strenuous application.

Looking at the thermal dissipation equations, with an 8 ohm load and +/- 35V supplies you will actually be dissipating 31W of heat at full power output of 50W into the load. That is actually less than my application which will dissipate 40W of heat at full power with a 4 ohm load and +/- 28V supplies. I guess I wasn't being as cautious as I thought!

I do not think you need to be too concerned about the mounting flange, unless it makes mounting to the heatsink more convenient for you. Some have mounted the non-isolated chips directly to a bracket like the ones you described, which was then isolated from and bolted to the heatsink. This would most certainly be overkill for your application. I recommend directly mounting the heatsink to the chip, with a thin dielectric to isolate them electrically.

Neither do you need to use two heatsinks, unless you prefer to make the sides of your amp out of them and want to do so for aesthetic purposes. It certainly would not hurt to use two of the smaller heatsinks you described, one for each chip on each side of the enclosure. Of course this is assuming that you just have one stereo pair to mount.

My advice from having been in your position and spent too much time analyzing the application before starting work on it, is to go ahead and get your feet wet. It is much more fun to do than to plan. You have done due diligence in my opinion and have a solid heatsink selected. Choose either the one or two heatsink configuration based on your personal aesthetic preferences and proceed!

Good luck and have fun,
David
 
jarthel said:
this is the heatsink: https://www.soanarplus.com/getProductWebDetails.do?prodId=HH8546

it's not mentioned in the webpage but the thermal resistance is .0552C/W.

looking at the photo, the based is definitely around 6mm or more.

There's a misprint in there.
0.05C/W is impossible from that size of sink.

That sink is metric 200x75x48 (show conversions for those that can't)

If the base is only 6mm thick then the heat dispersion will extend efficiently about 60mm radius around the chip. Beyond that 10:1 radius the thermal resistance makes the back plate and fins run significantly cooler and the whole heatsink dissipates less heat than specified. The manufacturers use a constant temperature over the whole back plate surface for testing Rth s-a. This sink will dissipate more heat if two chips are spread out across the width. Even better on a discrete output stage with 4 devices spread out for a shorter radius and four contact surfaces, but we don't need to go there.

If the corrected Rth s-a is 0.784C/W then the equivalent per chipamp is 1.568C/W. Well within the National table values. Go ahead.
 
It is 0.557 - and it is a chunky heatsink - probably sufficient for 4 chips.
not recommended.
have a look at the table and identify the operating conditions where maximum Rth s-a >=2.23C/W.
Low voltage PSU into high load impedance and while ambient temperatures must all be towards the cooler end of the heatsink requirement.

The table shows National's minimum requirement for the heatsink.
I recommend that you run the chipamp much cooler than the maximum, when Spike Protection will trigger much earlier to prevent temperature adjusted SOAR from being exceeded.
 
professional Kapton isolator
 

Attachments

  • kapton.jpg
    kapton.jpg
    57.3 KB · Views: 172
AndrewT,

What do you think of using the adhesive style Kapton tape? I have always used this out of convenience, but if there is a better method, I would love to know. Would it be better to use non adhesive Kapton with a gap filler on both sides? I would suspect the answer is yes, but it sounds like you have experience in this field, and I would like to use the best practical method!

Regards,
David
 
Hi,
I have no experience with Kapton.
The only thickness I found in the UK was far too thick (0.006" if I recall correctly).
The adhesive is often 0.002" thick (and worse). I don't know how thermally conductive the adhesive is that is used. It would be nice if the suppliers/manufacturers would supply thermal data for both parts of the isolator and what the combined performance is with actual "To" sizes. But the few that specify don't tell us.

What is the big secret? Why all the competing measurement units? Consistency and comparability would help.

I have a feeling that 0.001" Kapton with a normal thermal compound on both sides will perform as well as the best out there.
But that solution will not appeal to the assemblers. It is as unclean and slow as mica.
The next best would be 0.001" Kapton tape + 0.001" to 0.0015" thermal adhesive (total thickness ~0.05mm). I suspect this may be outperformed by the best of the deformable but a lot cheaper.
 
A general rule for when to use thermal grease is: Hard surfaces -- Yes, soft surfaces -- No. But many apparently hard surfaces (like many of the polyimide aka "kapton" foils) have a polyphase coating that gets soft and fills the gaps when reaching a certain temperature and gets hard again (but keeping its shape) when coming below that phase change temp (on the order of 50°C). With such polyphase coating, use of thermal grease is not recommended. Of course one needs it with mica and Al2O3 washers or with direct mounting as those are permanet hard surfaces. Compressible thermal pads like those made from silicone may only be used with components fixed to the sink at a single point (single bolt mounting). Components with multiple fixing points can easily be damaged when using compressible pads -- bendings stresses can cause the substrate to crack.

Some details on thermal accessories, catalog excerpts in a PDF:
http://www.fischerelektronik.de/fischer/uploadfischerfcool/Fischer/E.1.E.pdf

Similar stuff from another manufacturer:
http://www.heatmanagement.com/kunze/00english/products/kataloge/Products.pdf

Regards, Klaus
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.