| lowlevel |
I know in theory that heat sinks work better if they are black but what is the current thinking.
I note that almost all small heatsinks on the market are pianted as stock but many of the large items available as parts and almost every amp I've had the lid off in recent time the big heatsinks are plain aluminium.
Whats the current cutting edge thinking?
I have also heard many arguments in the dim past about heat transfer using diferent types of washers under you power devices.
Berilium was suposed to be real funky in the late seventies early eighties (if you didn't mind a slow lingeriing death)
I was braught up on mica washers and silicon grease.
It was considered better to omit the washer if design allowed it.
How are the grey silicon washers stacking up for transfer and long term reliability?
I have seen a few of the grey fellas blown thru in TV flyback circuits.
cheers chaps |
|
|
| jaycee |
Well, it isnt paint, its anodising which is a chemical process involving nasty acids and dyes. Paint would produce a relatively thick layer that would be too poor a thermal conductor etc...
The choice du jour for mounting transistors where heat transfer is important now seems to be Kapton. The sil-pads (grey is one variety) are usually used in mass produced stuff, and there are better quality ones than the grey ones but they are normally quite expensive and it seems to work out cheaper to buy a roll of Kapton tape and cut it up into needed chunks. I haven't tried this myself yet - I use the grey sil-pads. |
|
|
| kilowattski |
| Actually they are not painted. They are either Anodized or Hard Coated (Hard Anodized). From what I understand Mica insulating washers and thermal transfer goo seem to work better. If you Hard Coat the heatsink, the coating is about 2 to 3 mils thick (about the thickness of the mica insulator), non conductive and has excellent thermal transfer characteristics. Simply mounting your transistor directly on a Hard Coated Heatsink with thermal transfer goo seems to be the best senario. I have done this successfully in the recent past. |
|
|
| uli |
| quote: | Originally posted by jaycee
Well, it isnt paint, its anodising which is a chemical process involving nasty acids and dyes. Paint would produce a relatively thick layer that would be too poor a thermal conductor etc...
|
The dissipation factor of painted alu is better than anodized alu.
You need to use special paint (ovens).
Uli
:nod: :nod: :nod: |
|
|
| Eva |
Feel free to correct me if I'm wrong...
There are two different heat transfer phenomena between heatsinks and air : Radiation and contact. Radiation is improved when the radiating surface is black, but contact transfer is reduced when using paint
Also, the thermal resistance from the transistors to the heatsink appears to be bigger if there is a layer of paint in between, so painting the entire heatsink is not a good idea
Anodizing is said to be a good compromise since it's only a fine 0,1mm layer of oxidized aluminum created by electrolytic processes, it protects aluminum from corroding and it also provides 100V or more of electrical isolation
It appears that contact heat transfer dominates in forced air cooling applications and thus anodizing or painting is not worth the extra cost. Forced cooling heatsinks are usually raw aluminium
But in natural air flow applications, the increase in Rth from anodized to raw aluminium is 15% to 20% according to some heatsink manufacturer. This makes a difference for small heatsinks, I think |
|
|
| anatech |
Hi lowlevel,
Mounting transistors using just grease and relying on an anodized surface is risky. A scratch and you have a short. I still use mica and grease. It is messy and takes longer, but it works. When the sil-pads are used I still use a little grease. Just make sure the mounting surfaces are flat and smooth. Sand them with 600 grit wet / dry paper if you aren't sure. A sanding block helps (mine are titanium).
Make sure there are no burrs as well.
-Chris |
|
|
| Mr Evil |
Just to clarify what has already been said: The three methods of heat transfer are conduction, convection and radiation (and evaporation is sometimes classed as the fourth, but not important here). All three are involved in a heatsink. First the heat is conducted from the device to the heatsink and through the heatsink to the fins. Then it is both radiated and convected away to the atmosphere.
Convection is the dominant transfer method. With forced air cooling radiation can be ignored. With natural convection it makes a small difference which may be worth bothering about if you want the best performance.
Two factors make radiation work better: Surface area and colour. More surface area and darker colour are better. Matt finishes have more surface area. Increased surface area also improves convection.
The surface coating, be it paint or anodizing, decreases the effectiveness of conduction through the heatsink, but as long as it is a thin layer it doesn't have to have a big effect.
The only way to be sure if a surface coating will be beneficial or not is to measure it. I just happen to have an article on thermal resistance measurement on my website.
As for washers: I prefer to go without insulation at all where possible and use just thermal paste (high quality silver-based stuff like Arctic Silver is well worth it). This may mean separate heatsinks for each transistor and each heatsink insulated from the case and each other, but it can make a significant saving in thermal resistance, allowing for smaller heatsinks overall.
For non-critical applications I like to use those sil-pad things, for simplicity and cleanliness, but their performance is relatively poor.
Hard anodized heatsinks can provide good electrical insulation without washers, but normal anodizing should not be relied on in case of scratches.
In addition to surface smoothness as anatech mentioned, mounting pressure also has a big effect on thermal resistance. You want the highest pressure possible applied as evenly as possible. Be wary of using the mounting holes on plastic devices as too much pressure there tends to lift the opposite end of the device off the heatsink. |
|
|
| johnnyx |
| The mica insulation on a friends power amp broke down, and was arcing between the heatsink and TO3 power transistor. The heatsink was bare aluminium. How it happened is a mystery, (there was only about 40v - unlike the high voltages in a TV circuit) but since then I have been careful to use good insulation. :) |
|
|
| Mr Evil |
| quote: | Originally posted by johnnyx
The mica insulation on a friends power amp broke down, and was arcing between the heatsink and TO3 power transistor. The heatsink was bare aluminium. How it happened is a mystery, (there was only about 40v - unlike the high voltages in a TV circuit) but since then I have been careful to use good insulation. :) |
Could it have been dust? It can be a problem when using thermal paste that crud sticks to it. If the insulating washer doesn't extend far enough past the edge of the device, or if the thermal paste is splurged everywhere, then it is possible for enough stuff to accumulate to bypass the insulation. |
|
|
| Eva |
From time to time I've seen very small solder balls and small particles of aluminum from heatsink machining trapped between the insulating silicone washers and the heatsink or the transistor. They were causing otherwise unexplainable intermittent shorts. Sometimes manufacturing quality is not as good as expected...
Once I even found a small piece of TO-220 transistor leg trapped between a TO-3 device and his insulating mica washer in a 1980s Spanish manufactured PA amplifier. That ill-fated TO-3 device was a driver instead of an output device and was not blown, but it had almost unity current gain due to years of thermal cycling to extreme temperatures. The amplifier showed hard to explain distortion and humm in one of its channels but the mica insulator resisted without causing a short |
|
|
| djQUAN |
| quote: | | . When the sil-pads are used I still use a little grease. |
using grease with silpads increases their thermal resistance. |
|
|
| sam9 |
Lowlevel,
If you are looking for really, realy good heatsinks in your own part of the world check www.conradheatsinks.com. The prices are impressively reasonable, too. |
|
|
| kilowattski |
Anatech,
Never ever sand a surface if you want to have a good transfer of heat to it, especially if you plan on using a Silpad. |
|
|
| Mr Evil |
| quote: | Originally posted by kilowattski
Anatech,
Never ever sand a surface if you want to have a good transfer of heat to it. |
What makes you say that? I find that most heatsinks absolutely need lapping to smooth out irregularities. Most transistors could be smoother too. Smooth surfaces are vital to achieving the lowest thermal resistance: It's possible to achieve a significant measurable improvement by spending a few minutes with the wet & dry. Smoother surfaces require less thermal paste too, which saves a bit of money. |
|
|
| anatech |
Hi kilowattski,
It's all about surface flatness and roughness. If you look at many heatsinks, you will find the mounting surface can be anything but smooth. Look at 600 grit sandpaper, sand some aluminum. You will find an immediate improvement in smoothness. Now, I use a titanium block to sand with. That darn thing is flat (machined).
With my thermometer (Omega 450AET, E type thermocouple & calibrated to well winthin tolerance with probe), I have measured improvements. Never has the interface become worse. Now, if this isn't enough, and we are better now that Motorola specs, we can begin lapping the surface (as in tape head). For a mirror finish you can use the back of the paper. Use the wet / dry kind.
Our goal is to transfer heat to the heatsink. Anything that improves this is fair game. As Mr Evil stated, the best is to grease the part directly to the heatsink. You should consider the heatsink live. I don't like to go this far due to the dangers involved. Fingers and tools come in contact easily.
Concerning insulator puncture, I've seen all manner of things under transistors. The original Carver Lightstars had a real problem in assembly with cleanliness. Solderballs were the most common objects piercing the insulators. It's always fun to see what some repairers can get stuck in there. The best of these people use the original grease. Nice eh?
-Chris |
|
|
| kilowattski |
Anatech,
I stand corrected. I was always taught that the surface should be milled flat but after some research I see now that lapping with some 400 to 600 grit wet/dry is perfectly acceptable. |
|
|
| anatech |
Hi djQUAN,
Thermal resistance of silpads is improved (lowered) with a small amount of heatsink grease. I have measured this (see above) and there are studies at either IR or ON Semi (can't remember which ) to support this. Of course, we are talking about a small amount.
Generally, in the read world, mica and grease are the best except for exotics. Sil pads are used to speed the manufacturing process. Period, end 'o story. I accept that their performance is improving.
-Chris |
|
|
| anatech |
Hi kilowattski,
Sometimes the parts can use it too. Anyhow, no problem. I guess you had visions of some guy with 220 grit wadded up scratching back & forth. :drunk: Got shivers up your spine yet?
I have seen this.
-Chris |
|
|
| Illusus |
to get more performanse out of mica insulators you can split them with a razor blade. It can be a bit tricky but a little practice goes a long way. I've split a single mica washer into four before...after breaking a half dozen first. In my new project I'm using seperate sinks for each transistor isolated from chassis by using nylon bolts and washers. To keep hands and tools from causing a short I'm surrounding the assy. with a Faraday cage, other than that I'm just going to have to be careful... So, no dialectric between transistor, just a bit of grease.
As to painting heatsinks...I've used ceramic exhaust manifold paint before with excellent results. It goes on thin and sticks like poo to a wool blanket. It does need to be baked in an oven for a few hours though. You should be able to find it at any hot rod supply shop, even a lot of standard auto parts shops carry it. |
|
|
| lowlevel |
Thats what I was interested a spirited discussion.:D
I've always been very fussy when assembling heatsinking. In my opinion a large portion of electronic failures can be put down to poor hygeen.
I have filed many heatsinks with a sharp clean mill file and been interested in how flat they weren't prior.
I have also seen items where the devices didn't even contact the sink properly.
I am certainly aware that there are a number of ways to make something black.
I always stock about 5 types of black paint (i work in the entertainment industry)
I have had good proven results on theatre lanterns with black stove paint, they run much cooler with a good consistant coat of black.
Stag used to make a product called "heat dispersant" for the automotive trade for painting brake drums & the like, never used any.
Falt black nitrocelulose laquer is a favorite of mine & it goes on nice & thin if you spary it right.
I would usualy mask the contact areas.
Lets put the mill finish thing down to commercial reality.
so who's done tests silpads V's mica & grease??
cheers chaps |
|
|
| uli |
| quote: | Originally posted by anatech
Generally, in the read world, mica and grease are the best except for exotics. Sil pads are used to speed the manufacturing process. |
Today this is not quite true anymore as even silpads without (!)
grease have a better thermal transfer capability than mica &
grease. In fact is the resistance of mica (&grease) with
0.65K/W/inch² not top of the line anymore.
Aluminium Oxide (exotical) and especially Kapton foil
(without grease!!!) are far superiour.
Kapton foil has a thermal resistance of 0.07K/W/inch² !
Data source:
Fischer electronics
Uli
:nod: :nod: :nod: |
|
|
| Mr Evil |
| quote: | Originally posted by uli
Today this is not quite true anymore as even silpads without (!)
grease have a better thermal transfer capability than mica &
grease... |
Stop using that old silicone paste and start using newer stuff, like the Arctic Silver I mentioned earlier! It can give a thermal resistance only one tenth that of standard paste, although that advantage would be proportionally less if using a mica washer. |
|
|
| uli |
Hi Miles,
the advantage of Arctic Silver is already well known to me :D
but its gone using mica which has just mediocre thermal resistance.
Uli
:nod: :nod: :nod: |
|
|
| Stabist |
Hi!
Interesting topic! One question - I have heatsink with aprox. factor 0,5K/W - the "base plate" is only 5mm thick - but accordingly from catalogue same heatsink but with 10mm thick "base" - has thermal factor at same height arround 0,35K/W ... So I wonder - If i add some copper plate and mount transistors on it - will I lower the thermal factor of the heatsink??
Thanks |
|
|
| Tensop |
conrad(makers of some very, very good heatsinks) have this to say about theirs:
Coating Material
Textured black polyester powder coating has been chosen as the standard finish on all coated Conrad heatsinks and provides:
* a quality, durable and attractive finish capable of withstanding elevated temperatures,
* increased thermal dissipation in the order of 5% to 8% (depending on the heatsink) under natural convection. |
|
|
| wrenchone |
The very low figure for thermal resistance for the kapton insulator is only for a 0.001" film. This is pretty thin, even for something as tough as kapton, and you want to be careful of burrs and such on the heat sink to avoid punch-through. Most vendors supply kapton insulators using around 0.002" thick film to help avoid punch-through.
Mica has horrible thermal resistance. The only reason it works well as a heat sink insulator is that one can use a reasonably small thickness. Alumina is great, but it is pretty brittle, so you need to be very careful about heat sink flatness and controlled torque for tightening down the screws in order to avoid breakage. I've busted a lot of alumina insulators, usually right near the screw hole. Obviously, I need to invest in a small torque wrench. Any production facility that really knows what they are doing uses torque wrenches to tighten down devices on to heat sinks. |
|
|
| AndrewT |
Hi,
Wrenchone; I can only find .006" kapton, which is much inferior. Can you quote some sources of .001" or .002" kapton film?
regards Andrew T. |
|
|
| Mr Evil |
| quote: | Originally posted by Stabist
Hi!
Interesting topic! One question - I have heatsink with aprox. factor 0,5K/W - the "base plate" is only 5mm thick - but accordingly from catalogue same heatsink but with 10mm thick "base" - has thermal factor at same height arround 0,35K/W ... So I wonder - If i add some copper plate and mount transistors on it - will I lower the thermal factor of the heatsink??
Thanks |
It may help, depending on a number of factors. You must be able to bond the copper to the heatsink very well, as the interface between two materials adds a lot of thermal resistance. Also it will be most likely to help if the transistor(s) you are mounting on it cover only a small proportion of the heatsink's base area, then the benefit of the extra copper plate spreading out the heat is more likely to outweight the increased thermal resistance from having extra material in the way and another material interface. |
|
|
| wrenchone |
| Andrew T - Aavid/Thermalloy makes kapton insulators for standard semiconductor packages using 0.002" thick material. A previous post in this thread gives a link to Fischer, who makes the 0.001" thick insulators. |
|
|
| bear |
Ymmv, but...
Imho, copper + aluminum = trouble.
First off they have different thermal expansion factors.
Fastened closely together, two flat surfaces (required for any
decent heat transfer in the first place) - and we mean FLAT - you
get a bi-metalic element = bend.
After that you get the lovely electrolytic corrosion factor.
So not a great idea to use *iron* or ferrous *steel* hardware with aluminum anyway! :- )
...better to use more aluminum and a more effective combination of mounting + transfer area concept. Or all copper heatsinking.
As far as painting heatsinks with anything, I am rather baffled as to how the paint could possibly improve the heatsink to air transfer, as most all paint has a binder that is not particularly good at transferring heat at all compared to anodized aluminum. Powdercoat is usually thick and plastic (polyesters, urethanes, nylons, etc... not good thermal stuff at all). Baffled on that one.
_-_-bear :Pawprint: |
|
|
| Mr Evil |
| quote: | Originally posted by bear
...Imho, copper + aluminum = trouble... |
You highlight some real problems, but the fact that there are good commercial aluminium heatsinks with copper bases available shows that they are not insurmountable.
| quote: | Originally posted by bear
...As far as painting heatsinks with anything, I am rather baffled as to how the paint could possibly improve the heatsink to air transfer... |
Any surface coating will add some thermal resistance by impeding conduction, but it will at the same time reduce thermal resistance by improving radiation. As long as the improvement to radiation is greater than the effect on conduction then there will be a net benefit. |
|
|
| anatech |
Actually, I feel we are splitting hairs here. You should not be designing so close to the edge. As long as you are not dry mounting the transistors with a mica spacer your heat transfer should be okay.
Most of the high performance flexible insulators are one time use only. This may present a service issue if you are a manufacturer. This is coupled with the average tech "cranking" the part down way to tight. This applies to mounting hole fastening. With mounting clamps, overtightening still occurs. Striped threads and bent clamps are the result. Never mind cut through on all examples.
I appreciate that we want to run as cool as possible. Just remember real life situations and give extra cooling capacity where possible. And remember ... the laws of physics rule, no matter what the graph says.
-Chris |
|
|
| Stabist |
| quote: | Originally posted by Mr Evil
It may help, depending on a number of factors. You must be able to bond the copper to the heatsink very well, as the interface between two materials adds a lot of thermal resistance. Also it will be most likely to help if the transistor(s) you are mounting on it cover only a small proportion of the heatsink's base area, then the benefit of the extra copper plate spreading out the heat is more likely to outweight the increased thermal resistance from having extra material in the way and another material interface. |
The Al heatsink base is 350 x 150mm2 and the cooper plate I used is 250x50 mm2 - and I used 12 m3 nuts to bolt it to heatsink - so I guess it will be good enough ...
| quote: | Originally posted by bear
Ymmv, but...
Imho, copper + aluminum = trouble.
First off they have different thermal expansion factors.
Fastened closely together, two flat surfaces (required for any
decent heat transfer in the first place) - and we mean FLAT - you
get a bi-metalic element = bend.
After that you get the lovely electrolytic corrosion factor.
|
Hmm - I completely forgot about "bimetal" effect - that could be more serious problem, you're right ... :( I must calculate how much does copper and aluminium "stretch" apart/together with +/-40°C temp rise/fall ...
But the other problem - electrolytic - I always thought that:
- you need also some kind of acid as a meddium
- aluminium is eloxated so I guess that also avoids Al/Cu corrosion??? |
|
|
| anatech |
Hi Primoz,
All you need is water for corrosion to start, moisture over the longer term. Do you have any heatsink grease between the sheets. That will help on all counts.
-Chris |
|
|
| Stabist |
Yeah, I intend to use termal grease between Cu and Al ...
The thing is - this was originaly supposed to be a MiniA - with around 60W dissipation per channel - so 60W*0,5K/W//channel gives around 30°C overtemperature ...
But eventually MiniA project converted into A3 - so I need to dissipate around 100W now :( - and those Cu bars seemed like a good option to me ... I've allready made everything necessary to mount them - I just need to put the thermal grease between and to try out how it works ... And at that point I've found this topic - so I decided to get 2nd opinion about my idea ... A little bit late tough ... Now I'm a little scared because of that "bimetal" effect ... |
|
|
| Stabist |
So I've found the thermal coeficient (a) for Cu and Al - for Cu it is 16,5*10^-6 and for Al 23,8*10^-6 - and if you put that in formula:
d L = L1*a*(t2-t1) - at temp difference of 40°C I get around 0,1mm difference between Cu and Al ... Hmm - at one hand - it doesn't seem a lot - but on the other - we're always talking about microscopic rises etc that decreases thermal flow efficiency ... And from "microscopic" pointz of view - 0,1mm is enormously big ...
So?? What do you guys think? |
|
|
| moving_electron |
| quote: | Originally posted by anatech
Actually, I feel we are splitting hairs here. You should not be designing so close to the edge. |
I don't know. If one can get twice the transfer through the insulator isn't that the same as if the heatsink was twice as big? Seems like a big effect. |
|
|
| anatech |
Hi moving_electron,
No, it's not the same thing. The heatsink can only dissipate so much given the same area. Excess energy will raise the temperature (and give greater transfer) at the expense of higher junction temperatures.
The difference is more pronounced when greater power is being dissipated and there is a small performance gain but you still only have X amount of radiating surface.
My point is that if you are that close, get a bigger sink.
-Chris |
|
|
| lowlevel |
Discussint these matters with a mate who droped in.
Aparantlt a little grease will improve the transfer with silpads but in the long term the grease atacks the pad material leading to failure. He asures me the result is most unpleasant to clean up.
He has done a lot of high power service work & is a firm mica washer supporter. |
|
|
| Mr Evil |
| quote: | Originally posted by anatech
...My point is that if you are that close, get a bigger sink.
-Chris |
A few points here:
1) It's not always possible to get a bigger heatsink, because of size, weight, cost etc.
2) Even an infinite heatsink may not be big enough if too much power is being pushed through an inadequate interface. At high power the interface becomes very important.
3) Even if the heatsink is already adequate, a little improvement can't hurt!:) |
|
|
| anatech |
Hi Mr Evil,
I don't think we are disagreeing here at all. Your first point lists exceptions to the general rule. Your second point is something I fully agree with as stated in an earlier post and your last point is something else I've mentioned before in this thread. I really only wanted to have people use common sense and look at the whole picture.
Hi lowlevel,
Whether the grease attacks the sil pad or not depends on the composition of each. I use the Wakefield stuff (120-8) and it hasn't done anything nasty that I have seen. Besides, some sil-pad is a mess to remove all by their little lonesomes. We are talking razer blade time followed by solvent.
I too am a firm mica / grease supporter, having worked on lot's of high power amplifiers. Messy but reliable.
-Chris |
|
|
| Stabist |
Tommorow I'll get 12psc of Kapton insulators - I intend to use 4 for MiniA and 8 for A3//A5mix project ...
But what about using thermal grease - some say you shouldn't use it at all with Kapton, some say it's a must, some say it's whatever ... ?!?!?! :confused: |
|
|
| anatech |
Hi Stabist,
The safest bet would be to follow the manuafacturers recommendations. With regard to mounting the output transistors, follow the recommendations of that manufacturer. You can't go wrong.
Work on the mounting surfaces to make them as smooth and flat as possible.
-Chris |
|
|
| Keld |
| quote: | Originally posted by Stabist
SNIP What about using thermal grease - some say you shouldn't use it at all with Kapton, some say it's a must, some say it's whatever ... ?!?!?! :confused: |
I dont know any reason why not to use thermal grease. |
|
|
| sam9 |
| quote: | But what about using thermal grease - some say you shouldn't use it at all with Kapton, some say it's a must, some say it's whatever ... ?!?!?!
|
I use it with Kapton tape. I do not use it with the Berquist K10 or KA-10 type Sil-Pads which incorporate Kapton. The tape does little to conform to surface irregularity so I think a thin layer of grease is advisable, |
|
|
| PetarLD |
| Hi, I build my P3A amps a long time ago and I didn't have enough mica washers for all 4 devices, so I just cut out a rectangular washer out of a yogurt plastic cap, or the cup itself. Used some silicon paste from Radio Shack between the output transistor and the heatsink I have driven those things to 90 degrees C ~200 F, I couldn't touch the heat sink, well I did and burned my skin, anyhow I drive them that hard every time and for a year and few months they are still sounding the same as the day I build them. and surprisingly the plastic washers haven't melted yet, that was the only thing that I was scared of. Do you guys have any ideas for a very low budget supplements for the mica washers? Since I don't know where to buy them from, had to scavage the ones I had from an old PC PSU. |
|
|
| kilowattski |
| Sorry to be so blunt, but what you did was very, very foolish. IMHO it was not very smart to use a regular piece of plastic as an insulator for your power transistors and it is also very poor idea to run your transistors that hot. It sounds like a fire just waiting to happen. You can get Mica insulators at most any major electronic parts supplier. In the US they can be purchased from Digikey at $0.08 or 8 cents each. Maybe it's just me, but I just don't think it is worth risking a fire in lieu of purchasing an 8 cent mica washer. |
|
|
| anatech |
Hi PetarLD,
I think Radio Shack has insulator kits as well. I personally can't believe your amp didn't blow.
Anyhow, kilowattski is right, and now you know better. Sorry to be blunt, but you are very, very lucky. Fix it before your luck changes.
-Chris |
|
|
| jacco vermeulen |
Does anyone of you guys know a resource for mica insulators for MT200 devices.
I have loads of those in Toshiba and Sanken versions.
My dealer around the corner, bmm-electronics.nl ,sells them for nearly $2.50 the piece.
Silly price i think, considering what a mica TO247 costs.
Maybe even an idea if mica is sold in sheets and where to get them?
In case some of you dont know yet: Mica is a mineral called Muscovite.
Long time ago i studied geology at the university, spent hours watching stones with muscovite through a magnafying glass.
I never gathered why the name mica is used.
Can someone enlighten me? |
|
|
| Magura |
| quote: | Originally posted by WorkingAtHome
Just to add some new stuff to talk about, anybody used either of these products?
Aavid’s Ultrastick:
http://dkc3.digikey.com/PDF/T051/0638.pdf (top left corner of page) |
I've been using this for a couple of years. I rarely use anything else. The type I'm using is the Crayotherm from Bergquist. It's cheap and AFAIK the best there is. There is a version in sheets as well, it is somewhat easier to deal with. For insulation there is a version with Kapton sandwiched between 2 sheets of Crayotherm.
Magura :) |
|
|
| moving_electron |
I have not used it but I assume it is electically conductive? Pretty interesting where they say they have been applied. |
|
|
| sam9 |
| quote: | In case some of you dont know yet: Mica is a mineral called Muscovite.
Long time ago i studied geology at the university, spent hours watching stones with muscovite through a magnafying glass.
I never gathered why the name mica is used. |
Imagine we are back in the Cold War. What do you think happens when a defense contractor tells the government he's using "Moscovite"? |
|
|
| WorkingAtHome |
Here's a link to the Panasonic DataSheet for Pyrolytic Graphite:
http://www.panasonic.com/industrial.../AYA0000CE2.pdf
seems to have a resistance of 10k/cm.
Also, here's a link to a marketing doc:
http://www.panasonic.com/industrial...s_info_0105.pdf
They show it being used as a transistor insulator. There are different types (adhesives etc). None point out specifically that they are electrical insulators, however.
The heatpipe example they show is interesting, however. They use a piece of the material to conduct heat away from a device to a remote radiator, like a wire for heat. That could be very very interesting. |
|
|
| johnnyx |
| quote: | Originally posted by WorkingAtHome
[B
seems to have a resistance of 10k/cm.
[/B] |
This is conductivity, 10000S/cm ie Siemens/cm. Conductance is 1/resistance.
:) |
|
|
| PetarLD |
| Ok, I might get some mica washers for my P68 that I am planning on building, but I don't see any need for the P3A since the one that has mica washers used to get just as hot as the one with the plastic. Now I have them powering 8 ohm speakers so they barely get warm even after a few hours of intensive driving. The output devices are pretty durable from what I have experienced, and I only got the devices that hot just to see if the plastic would melt which it didn't so I can say that I have more than a 2 safety factor incorporated into the output. But as soon as I get my hands on some mica washers I would replace the plastic ones. Thanks for the concerns guys, no fire hazard yet. :hot: |
|
|
| petewilson |
Folks
I'm using On semiconductor 3281/1302 output transistors. I can't see any insulating washers for these devices or their package at Digikey.
Am I bline? please eductae...
Thanks
-- Pete |
|
|
| EWorkshop1708 |
Is there any amp stability issues when a transistor is electrically connected to the heatsink? Should any decoupling caps be used from the heatsinks to Ground?
The subwoofer amp I'm building now uses 5 pairs of output transistors.
5 NPN transistors are on one heatsink with a PNP driver, the other heatsink has 5 PNP with an NPN driver.
The power devices are just using grease and bolted direct to heatsink. The drivers are mounted using silpads because of my CFP design. I have no output resistors, so the heatsinks are electrically connected to each other as well.
IS this a problem, or should this be fine. |
|
|
| AndrewT |
Hi,
I have seen it said that the heatsink should be capacitor decoupled back to ground. |
|
|
| johnnyx |
| quote: | Originally posted by EWorkshop1708
I have no output resistors, so the heatsinks are electrically connected to each other as well.
IS this a problem, or should this be fine. |
So does this mean that the heatsinks are directly connected to the output? In this case, you cannot use a cap to ground from the heatsink, because it will load the output. Such a large area attached to the output will increase the chance of capacitively coupling back to the input stages. Maybe a grounded shield would help, and care over the layout. It should be OK if you are aware of the potential problems and take precautions, but this kind of thing is largely unpredictable, so you will have to build it and see. You can always add insulation if it proves too troublesome.:) |
|
|
| EWorkshop1708 |
| quote: | Originally posted by johnnyx
So does this mean that the heatsinks are directly connected to the output? In this case, you cannot use a cap to ground from the heatsink, because it will load the output. Such a large area attached to the output will increase the chance of capacitively coupling back to the input stages. Maybe a grounded shield would help, and care over the layout. It should be OK if you are aware of the potential problems and take precautions, but this kind of thing is largely unpredictable, so you will have to build it and see. You can always add insulation if it proves too troublesome.:) |
I've built it already, being tested so far.
Yes, the heatsinks are connected directly to the output.
Reason I ask, is the amp works fine in class B but when adding some bias the thing has no variation past 10mA before pulling max current. I was thinking it's maybe oscillation and not really a bias issue. I was wondering if the heatsinks are picking up stray frequencies.
My amp has a zobel with a 0.1uf cap and 4.7ohm resistor, but maybe I could add some extra zobel networks attached direct to heatsink or something. |
|
|
| anatech |
Hi EWorkshop1708,
That's what is happening. Use RF bypass caps to ground from each heatsink. 0.1 uF should do it.
-Chris |
|
|
| EWorkshop1708 |
| quote: | Originally posted by anatech
Hi EWorkshop1708,
That's what is happening. Use RF bypass caps to ground from each heatsink. 0.1 uF should do it.
-Chris |
Thanks
Would smaller values than 0.1 be ok? |
|
|
| johnnyx |
I am sure that adding capacitors to ground at the heatsink, i.e across the output, will make stabilisation more difficult, not easier.
Try a different way.:) |
|
|
| anatech |
Hi EWorkshop1708,
I just want to check what's going on with your amp. Collectors are hot directly to the heatsink? Collectors are at the supply voltage (common emitter output)? If this is the case, capacitors to ground are the correct proceedure. 0.1uF is in the ballpark but you can try anything from 1uF electrolytics (film are better) to 0.001uF. A smaller value may reduce the oscillation a bit but not eliminate it. In that case, increase the value. An oscilloscope really helps here if you can get to one.
-Chris |
|
|
| AndrewT |
Hi,
you are CFP? meaning collectors are to output and heatsink swings with output voltage? |
|
|
| EWorkshop1708 |
| quote: | Originally posted by AndrewT
Hi,
you are CFP? meaning collectors are to output and heatsink swings with output voltage? |
Correct. I was hoping CFP would save me from using pads between transistors and heatsink. I mean I could still do EF with them directly to heatsink, but I imagine you'd get a hell of a shock if you touched the sinks from the sum of both rails being at the potential of the heatsinks with an EF design. |
|
|
| EWorkshop1708 |
| quote: | Originally posted by anatech
An oscilloscope really helps here if you can get to one.
-Chris |
I agree. I wish I had an O-Scope, but I'm looking for one as my next toy. I've used them before but never have actually owned one......yet. |
|
|
| anatech |
Hi EWorkshop1708,
I looked at what I think is your schematic. You should try to decouiple the supplies. Not at the heatsinks as they are your output. You may need to install some small caps from collector to base. Maybe around 100pF - experiment.
-Chris |
|
|
| EWorkshop1708 |
| quote: | Originally posted by anatech
Hi EWorkshop1708,
I looked at what I think is your schematic. You should try to decouiple the supplies. Not at the heatsinks as they are your output. You may need to install some small caps from collector to base. Maybe around 100pF - experiment.
-Chris |
You mean collector and base of the drivers right?
Heck, if the miller effect works for VAS stages, then I see why it could be used for the output stage as well.
Speaking of the subject of this thread, my heatsinks are black anodized but I get a good connection between transistors and heatsink, and I do have silicone grease on all transistors. I don't think anodizing does much for the heatsink as far as conductivity is concerned. |
|
|
| anatech |
Hi EWorkshop1708,
Try both and stick with whatever works for you.
Black anodizing is for thermal transfer to the atmosphere and it works. You may get some electrical insulating properties from hard anodizing .. but don't rely on it. The case screws probably make the best electrical contact for you.
Heat transfer to the heatsink is mainly dependant on how flat the mating surfaces are. The heatsink compound is supposed to displace the air which insulates thermally. Use a very thin coat of compound. Observe proper mounting pressure to avoid warping the transistor package (ie do not overtighten).
That help?
-Chris |
|
|
| AndrewT |
Hi,
I agree with Anatech, your electrical connection is probably through the screw.
The anodise is an excellent insulator but a slight scratch will create a short once under voltage stress. |
|
|
| BrianDonegan |
| agree. Also, keep in mind that something that shows insulating qualities with a meter, will not necessarily be an insulator at higher voltages, so be careful... |
|
|
| EWorkshop1708 |
Actually I checked, the heatsink surface does make some contact, even with grease. I think it's the imperfect surface, all the small scratches in the anoziding, and these heatsinks are about 10 years old from some old amps.
The screws, however do not need washers and don't contact the heatsink because I'm using MJL 4302 and 4281 and those type transistors, the case insulates the screw, all they require is pads if you decide to use them. |
|
|
| amplifierguru |
Interesting thread, Lowlevel.
And good to see some very local enthusiasm.
Sam 9, I have concerns about powder coated heatsinks. But I like the idea of exhaust lacquer to do the black body job.
I use pressure clamps (e.g. foldback clips) for heatsink association - it's kool and a defined pressure thing. Simple and effective. |
|
|
|
