Question about silicone dielectric grease

So what caused the short to ground? Was it the mica pad or 40 year old dried up white insulating grease? I'm new and I'm not sure if I'm sending the pictures correctly,,anyway the marantz 2226b that had 150 ohms to ground from r776 to transistor q750.
The mica insulator on the left is the one I replaced and the one on the right is one I scavenged out of a cable box.. The gray one was twice that size originally and I cut it in half and this is what is left of it. The piece I put on didn't have a hole in it. Is the failure I found a common thing? Has anyone ever heard of a short on a transistor to heatsink being caused by either an old mica pad or old white insulating grease?
Thanks in advance for everyone's time and knowledge.
 
According to the test the one post refers to where he tested the thermal transfer ability between silicone dielectric grease and heat sink paste,the paste was better at transferring heat but only marginally better than the silicone dielectric grease.Now I know this was only one test and in the real world anything is possible,

Whatever purported "test" was done (citation?) is, of necessity, utter nonsense and the experimenter was completely incompetent. The physics of thermal conductivity explains how conductivity varies between materials, and we'd need to throw out a few hundred years of repeatable science to make silicone grease as good as a metal . This simply can't be. Silica is basically an insulator! I posted specific numbers showing differences in thermal conductivity for a variety of materials. The thermally conductive particles, either metal oxides and encapsulated metals, conduct the heat by mating the device and heatsink surfaces, and the grease is just a carrier to hold that thermally conductive material in place.

The metal particles do the work, NOT the silicone grease.

If silicone grease with no additives was only marginally better than thermal compound, then said compound would be totally unnecessary and engineers would have been specifying it for absolutely no reason. Or merely bribed by the compound manufacturers? It is not the case that "real world anything is possible"; it is the case that whatever experiment was done was totally bogus and the experimenter incompetent. Science has rational rules and is repeatable. If it isn't repeatable, it is witchcraft, not science.

The fact, and it is a fact, is that silicone grease is a poor heat conductor, and ZnO is about 7x better and silver is about 140x better.

I suggest you telephone any of the thermal compound manufacturers and ask for tech support. Explain your thesis and listen to the debunking.

I don't know why you want silicone grease to be as good as thermal compound, but you proceed down that path at your peril. I hear the ka-ching! sound of the cash register ringing up your purchase of new transistors when yours burn out from thermal overload.
 
Last edited by a moderator:
The use of heat sink grease is to fill in all very small holes in the heat sink and transistor as well as the mica insulation so that there is complete area available for heat transfer. I understand you will be bound and determined to use dielectric grease but they make heat sink grease for a reason and the data sheets talk about its ability to transfer heat. Dielectric grease doesn't have the heat transfer components in it so it is not the stuff anyone here would use. but it is your gear, do what you want and, well, get good at repairing them, you might be doing that occasionally. High temperatures at the p-n junctions can kill a transistor and the metal case properly mounted to a heat sink is designed to pull that heat away from that junction. Using improper materials can hasten the death of that junction.

I gotta say this is all spot on.

The worse the thermal conductivity, the higher the junction temperature and the shorter the lifespan.

This is all just correct facts and good science. The junction doesn't care about your theory, it just cares that it gets hot enough to be damaged.
 
It's worth pointing out that paste should only ever a VERY thin layer. Its purpose is to improve thermal contact between the contacting surfaces, by filling microvoids. You want most of the contact to be between the contacting surfaces, not the paste.

Therefore, the thermal resistance of the paste in bulk should be of little significance, as it is not used in bulk, but only in a very thin layer, which should give a low resistance. If there is a reduction in overall thermal resistance, due to improved contact, vs the resistance of the thermal paste, then we're winning.

Check the thermal resistance of that mica washer, for comparison...

I like silpads, because they are clean, simple, low resistance, and, being conformal, don't need paste.
 
While easy to use, the silpad thermal conductivity is terrible!

Datasheet lists thermal conductivity as 3.0 W/m-K, same as silicone grease with no particulate loading.
 
It's all terrible compared to metals, but the thermal conductivity number doesn't take thickness into account. Mica is about 0.7 W/m-K, four times worse than the Sil-Pad, but then you look at thickness. You can shave a mica insulator down to a few thou, whereas the Sil-Pads tend to be thicker. Hopefully the thermal grease layer will be down around 0.0002", so it's not much of a factor. If the parts aren't flat, all bets are off.

It would be nice if all manufacturers of insulators and grease were forced to give numbers based on identical tests, but it's up to us to sort out the truth.
 
It's all terrible compared to metals, but the thermal conductivity number doesn't take thickness into account. Mica is about 0.7 W/m-K, four times worse than the Sil-Pad, but then you look at thickness. You can shave a mica insulator down to a few thou, whereas the Sil-Pads tend to be thicker. Hopefully the thermal grease layer will be down around 0.0002", so it's not much of a factor. If the parts aren't flat, all bets are off.

While the polymer layer is thin and conformal, at the same time if it truly worked as well (or better) than the metal pastes we'd be using it on CPUs. I remember from back in the day that the heatsink pads for CPUs did not work. as well as ZnO which is what we used. I remember having a silpad on a Pentium and it was really bad for heat conduction. ZnO, all we had as an alternative, worked far better and dropped the temperature.

Transistors are not lapped and neither are the heatsinks. Hard to lap the base, but it can be done on either side of the pins. Heatsink is easier to lap.

It would be nice if all manufacturers of insulators and grease were forced to give numbers based on identical tests, but it's up to us to sort out the truth.

Sure, but at the same time we do know that the metal paste is what does the work and that grease by itself is not suitable.
 
OK, we get it: you're a die-hard Arctic Silver user and you won't rest until everyone is using Arctic Silver. Now please stop crusading and let OP get what he needs out of this thread.

No, that's your strawman. My point was that silicone grease and silpads are not as good as conventional thermal compound, even the ZnO which was widely used for decades.

The silpad was created for simplicity in manufacturing, since it removes steps and a messy procedure. But it doesn't work as well, as any electrical engineer can tell you. The datasheet sets forth the numbers. The OP asked about substituting for ZnO thermal compound with thermal grease. What's been posted is a great deal of misleading and deceptive information about silicone grease (which the OP calls dielectric grease) and silpads. The numbers don't lie. I've used all of those materials, and have researched the differences; have you?
 
I use Arctic Silver. Yeah, it's loaded with real silver. It's no more electrically conductive that any other good thermal grease. You shouldn't depend on grease to electrically insulate. If the sink is grounded you should be using mica insulators. So far nothing I've tried has beat Arctic Silver and .0007" mica.

The problem with silicone pads is you need to have uniform clamping force and crush the snot out of them to get their best performance. Tried a few. Mica and grease destroy silicone pads.
 
Last edited:
Hello all,I am the original poster and I wanted to share my thoughts. I have spent several hours researching this subject and here is my thoughts. As everyone knows the purpose for it mainly is to help transfer heat from the electronic component to the heat sink. Another reason is to prevent corrosion between the two metal surfaces. According to the test the one post refers to where he tested the thermal transfer ability between silicone dielectric grease and heat sink paste,the paste was better at transferring heat but only marginally better than the silicone dielectric grease.Now I know this was only one test and in the real world anything is possible,A few characteristics of the silicone dielectric is won't burn,won't melt,won't run,temp range up to 500 degrees f,non hardening,safe for plastics and rubber(if it came in contact)
Does anyone have any accounts of it not working for mounting components to heatsinks???
Wouldn't it be alot cheaper to buy and use???,,if it indeed posed no issues??

I haven't repaired that many items yet and like I said I'm learning and appreciate all the insight and information I get especially from this site.I have a question about a stereo I just repaired!
Marantz 2226b stereo that I was given and running it with the cover off I noticed alot of heat coming from the p800 board, specifically it was coming from the big 5 watt ceramic resistor (r776), touching it and it would burn you,,inspected and noticed the solder to the leads was melting,,so first thing I did was resolder the leads,,it didn't help,,a quick voltage test and I learned it was dissipating 12 watts,35.5 volt drop thru a 100 ohm resistor,,I tried to set the power supply adjustment to 35 volts and it would not adjust any higher than 29 volts with vr turned all the way,,much learning and reading later from info on this sight I figured I would replace the 2sd330 transistor,,long story short I measured the resistance to ground on the 5 watt ceramic resistor (r776),it showed 150 ohms on the side that goes to the transistor c terminal..the new transistor showed that the collector internally shorts to it's heatsink,,so just for the heck of it I unmounted the transistor from touching the heat sink but left it soldered to board.rechecked resistance to ground and the short was gone,,so I found a gray rubber looking really thick pad I saved from somewhere,removed the old thin mics pad,,cleaned up the old white grease,I then mounted it back to the heatsink with some silicone dielectric grease and reinstalled the metal clamp bar that's present on a 2226b and checked resistance to ground and it now showed 3.2m ohms,,so I fired it up and tested it,,no heat at all on resistor,,adjusted p/s from 35.8 volts to 35. It's been playing now for 8 hours with no issues and nothing replaced but an insulator pad
My question is what caused the 150 ohm short to ground?, Was it the old mica pad or was it that the old white grease had melted off over the past 40 years and left whatever particles it holds in suspension to cause a short???

Hello, I'll be sharing with you the datasheet for one pair of transistors my amp uses (Sony STR-DE310 from the late 96).

So in the file you see a lot of specifications for example about mica thickness used, silicone grease, as well as torque used for bolting the transistors which are BJT to the heatsink.

I've been thinking about the thermal conductivity out of the three types it suggests to use, which all of them are lower than 1 W/mK... My options where I live are PC thermal grease, MX-4 Arctic (8.5 W/mK) or the typical white grease which I don't have the specs, but I can share what the datasheet recommends to apply in the attached files.

If someone can help me, I would be using the stock mica pads that's on the amp and looking forward to replace the grease.
 

Attachments

  • mp1620.pdf
    212.7 KB · Views: 4
  • YG6260.png
    YG6260.png
    93.7 KB · Views: 4
  • G746.png
    G746.png
    215.9 KB · Views: 3
  • sc102.PNG
    sc102.PNG
    30 KB · Views: 3
  • Arctic MX-4.png
    Arctic MX-4.png
    133.9 KB · Views: 4
So in the file you see a lot of specifications for example about mica thickness used, silicone grease, as well as torque used for bolting the transistors which are BJT to the heatsink.
I've been waiting for someone with experience to weigh in on this, but my understanding from rebuilding a 1974 Scott and building a Heathkit around 1978 or so, is that silicone grease used to be standard operating procedure for mounting TO-3's and mica. The contemporary particulate loaded thermal compounds are better. But for the first two decades of the transistors era, it seems silicone was a common way to go. And designs calculated safe operating zones based on that thermal conductivity.

It's good to have that table of thermal conductivities, but I wonder if it is a measure of the material itself, or in use. I doubt there are order of magnitude gains to be had from a vanishingly small layer of compound. But I could be wrong.
 
I've been waiting for someone with experience to weigh in on this, but my understanding from rebuilding a 1974 Scott and building a Heathkit around 1978 or so, is that silicone grease used to be standard operating procedure for mounting TO-3's and mica. The contemporary particulate loaded thermal compounds are better. But for the first two decades of the transistors era, it seems silicone was a common way to go. And designs calculated safe operating zones based on that thermal conductivity.

You could, if you wanted, use google to trivially verify any of the facts I previously set forth.

The early days of transistors used a variety of metal-oxide (non-conductive) fillers, most commonly ZnO, as it is inexpensive and non-toxic. (ZnO is white and opaque, and was commonly used as a sunscreen; this, in fact, is the white layer on lifeguard's noses as seen in all the photos and films from the 1950s and 1960s.) Just ask anyone working on transistorized gear from the 1960s and 1970s who cleaned ZnO from heatsinks and smeared it on the new device while replacing transistors. Very messy.

Even beryllium oxide (BeO) was also widely used because the ordinary ZnO was insufficiently conductive. Heathkit, in fact, used the highly toxic BeO in some of its transmitters as an unspecified "thermal compound". Probably until the late 1970s, when the hazards became more widely known. Motorola similarly used BeO in its transmitters which were sold to hams and governments. Poisonous thermal compound is a well-known problem for hams retrofitting equipment. Special heatsink compound was used for these applications to improve dissipation via the heatsink. This is widely known and not disputed. If no advantage accrued, toxic fillers would never have been used. You may verify all of this with a simple search.

You may, furthermore, trivially verify, as I above set forth, "The fact, and it is a fact, is that silicone grease is a poor heat conductor, and ZnO is about 7x better and silver is about 140x better."

That is actual knowledge and experience which, again, may be trivially verified with a simple google search. I even quoted material from a respectable source! In No. 4 quoted from an industry publication:
https://www.audiokarma.org/forums/i...icone-dielectric-grease.832809/#post-11851271
Emerging Generation Of Thermal. Greases Offers Advantages
Greg Becker, Chris Lee, And Zuchen Lin
Advanced Packaging Magazine
June 2005​

It's good to have that table of thermal conductivities, but I wonder if it is a measure of the material itself, or in use. I doubt there are order of magnitude gains to be had from a vanishingly small layer of compound. But I could be wrong.

The table in Number 4 is from an industry publication describing the properties of various thermal greases. Equivalent sources may be trivially found.

But consider the argument from the perspective of simple economics. If the gains were negligible then no manufacturer would purchase the more-expensive ZnO thermal compound or, indeed, any of the more expensive variants. Ample data exists as to the superiority of filled grease versus grease alone. This is why all of the older electronic equipment has ZnO residue around the transistors. It was commonly applied!

Again, the purpose of the compound is to thermally mate irregular surfaces. I suggest reading about lapping for superior thermal mating. All of this was previously described in gory detail, and Number 4 quotes various conductivities from an industry publication. You may trivially verify the truth of the data and statements in that article.

All of this was clearly set forth in gory detail and need not be endlessly rehashed and reargued.
 
I used to build assemblies with thermoelectric coolers, which have very flat ceramic surfaces, and copper cold fingers and aluminum heatsinks. Every surface was lapped. There's no question that flat lapped and polished surfaces that mate almost perfectly, combined with a loaded thermal compound, are astoundingly good. At the other end of the spectrum we have low powered hifi equipment. I've disassembled low power receivers, < 30 Wrms, that I'm pretty sure had never been disassembled before, and they used a clear unloaded grease with mica under the TO3s. Fortunately for us, not every amp is designed at the thermal limit of the devices and heat sinks. Personally, I stick with the well known loaded thermal compounds & mica, or if the power levels are low, the higher end ($$) Bergquist pads.
 
options where I live are PC thermal grease, MX-4 Arctic (8.5 W/mK) or the typical white grease which I don't have the specs
Your Sanken document refers to the use of "silicone grease" but the recommended materials listed there are all heat sink/thermal compounds (which, however, likely have silicone grease as a base component). So, there may be just some issue with terminology here. I don't believe that anyone in this thread has recommended use of pure silicone grease over thermal compound.

The compounds recommended by Sanken are probably all more or less equivalent to the "PC thermal grease" you mention---by which I assume you mean "generic thermal compound"--- but if you don't have a specification for what that actually is it might be worthwhile to buy and use a branded thermal compound with specified properties.

It's not clear to me what you mean by "typical white grease," but if you mean white Lithium grease then using that would not be recommended.

I'm not an expert on MX-4 but as long as you are operating within normal parameters for the amp, compound with the thermal conductivity specified by Sanken should be completely adequate.
 
I used to build assemblies with thermoelectric coolers, which have very flat ceramic surfaces, and copper cold fingers and aluminum heatsinks. Every surface was lapped. There's no question that flat lapped and polished surfaces that mate almost perfectly, combined with a loaded thermal compound, are astoundingly good.

Yeah, the peltier coolers were certainly amazing. But the expense of the cooler — as Conrad knows, back in the day this was a specialty product sourced only at great expense — was not nearly as impressive as the staggering power consumption. I remember these for the Pentiums. Consumed as much, possibly more, wattage as the processor, which that was impressive on its own. Plus all that heat had to be itself dissipated, lest the junction fail.

At the other end of the spectrum we have low powered hifi equipment. I've disassembled low power receivers, < 30 Wrms, that I'm pretty sure had never been disassembled before, and they used a clear unloaded grease with mica under the TO3s.

Sure, as a cost-savings measure because the devices are not pushed and if it fails from thermal degradation after a few years, well, that's the customer's problem. I remember a Heathkit TV — circa 1979, so maybe the GR-2000 — including a small container of thermal compound, injection-molded plastic pot with a pull-off top, used for power transistors. Not the best option. Better than nothing, true, but still, far from the best option available at that time.

Unfilled silicone grease, as can be ascertained in the data I previously posted, is a very poor thermal grease and is not suggested for new builds or rebuilds. Fine as a base, but not so good as the only conductor of heat.

Simply because poor materials were once selected — either because nothing better was available or as a cost-savings measure — is no reason to today replicate that choice. By that standard if the original was BeO the replacement must also be BeO.
 
Your Sanken document refers to the use of "silicone grease" but the recommended materials listed there are all heat sink/thermal compounds (which, however, likely have silicone grease as a base component). So, there may be just some issue with terminology here. I don't believe that anyone in this thread has recommended use of pure silicone grease over thermal compound.

Yes, I'm sorry made a mistake about the terminology. I meant thermal compound. I'm looking for a heat sink/thermal compound that matches or has similar values that suggests in the Sanken datasheet, which are less than 1 W/mK, but modern thermal compound like the Arctic MX-4 exceeds the that value (8.5 W/mK) and I'm not sure if it's suitable for the job. Viscosity is another variable that's in my thoughts too.

I meant by typical white grease, cheap white thermal compund used in CPU's, sorry.
 
OK, I misread you as asking about 3 possibilities rather than 2 which are both "PC thermal grease".

The concern which has been raised (and which has also been disputed) with the "modern" CPU-type compounds is lower resistivity and lower dielectric strength (even if they are described as "non-conducting"). Arctic does not seem to give any spec for dielectric strength. Still, it may be better or worse than what you might get with "cheap CPU compound". Presumably if it was as high as SC102 then Arctic would publicize that.

MX-4's viscosity is also 3 times higher than DOW SC102, which I think translates, at least, to some difference in application procedure.
 
Back
Top Bottom