Philips k9 was a very old generation, no plastic trz...looking on some pictures I think I mixed up the model...maybe it was was an old Telefunken But I'm certain I saw such a ceramic coupling in one old CRT TV
Last edited:
Recalling several not so uplifting feedback from the past regarding aluminum oxide insulators sold on Aliexpress is that many of them are not flat, so I checked a few recent feedback and found a few again:
https://www.aliexpress.com/item/4000063252698.html
https://www.aliexpress.com/item/4000063252698.html
I didn't see any bad feedback for the few aluminium nitride I checked, maybe a safer bet.
What do you think of kapton tape as a horizontal spreader?Not an expert in it as they say there are many varieties of kapton from 5...56 conductivity, but i know people using it a lot with high power amplifiers as its 7.5 medium conductivity looks quite ok for its thickness while offering a high degree of electric isolation.
I asked about it after seeing something looking like a refference to this aspect although not sure about it.
For a material to act well as a spreader, it should have the thermal conductivity akin to aluminum.
Alumina can be used, but it is roughly 20% of aluminum. BeO is twice aluminum, aluminum nitride is about the same as aluminum.
We use kapton that is alumina filled for better conductivity, but is still isn't that good.
John
Alumina can be used, but it is roughly 20% of aluminum. BeO is twice aluminum, aluminum nitride is about the same as aluminum.
We use kapton that is alumina filled for better conductivity, but is still isn't that good.
John
Do you mean something along the lines of this study?
https://www.sciencedirect.com/science/article/abs/pii/0011227596887881
Unfortunately i can't download it...
Thermally the only thing I believe is better would be a graphene interface to a silver heatsink, or perhaps a high velocity turbulent flow cooling of the package bottom (with reynolds numbers in the 4K range and up), something like what IBM did with their computer modules and high flow helium gas.
The only caveat I see with soldering to the package is, normal TO packages with plastic encapsulation are not really rated to withstand the soldering temperatures well. The overmolding plastic is held by copper features stamped into the formed base, and high heat takes the plastic well over it's heat distortion temperature. Above that temperature (called glass transition temperature), the plastic goes soft and also enters a higher thermal expansion regime. Many transistors rely on the overmolding-to-base integrity to assist with protection of the die and bond wires from environment. The simple act of soldering can introduce halides into cracks opened during the high heat of soldering, cracks which close up when the temperature drops. Any protection of the die and bonds by glob top or other protection materials may not be enough, and I do not recall large power semi's being protected with nitride passivation.
Surface mount packages are designed to deal with those soldering issues, so may be a better option.
John
I did think of typing that, but how many carats would be involved with a diamond 1mm thick by 10 mm square?
I keep telling my boss that I now work to support buying the tools I want for my hobbies.. Multi-carat diamond heat spreaders are a tad outside my price range, especially for DIY. (unless I win powerball).
Just picked up a vacuum pump for epoxy degassing and vacuum impregnation, new toy to play with after the glass kiln and resin printer are fully integrated in my workshop, a dual purpose fume hood setup in the works.
jn
ps.. I could not believe how sensitive platinum catalyst silicones are when in contact with UV cure resin printed parts. The UV resins inhibit cure, despite my building a UV curing chamber and running it for 6 hours on clear resin prints. Tin catalyst silicones, no problem. This old dog learned a new trick..
I keep telling my boss that I now work to support buying the tools I want for my hobbies.. Multi-carat diamond heat spreaders are a tad outside my price range, especially for DIY. (unless I win powerball).
Just picked up a vacuum pump for epoxy degassing and vacuum impregnation, new toy to play with after the glass kiln and resin printer are fully integrated in my workshop, a dual purpose fume hood setup in the works.
jn
ps.. I could not believe how sensitive platinum catalyst silicones are when in contact with UV cure resin printed parts. The UV resins inhibit cure, despite my building a UV curing chamber and running it for 6 hours on clear resin prints. Tin catalyst silicones, no problem. This old dog learned a new trick..
Last edited:
Another aspect of power transistors that is rarely considered is the power cycling of the die attach.
While it is great to go massive cooling stuff, the power density can wreak havoc on the die attach solder. When the structures reach operating temperatures, each component will have thermally expanded at their different rates. Silicon is typically 3.3 ppm/degree C, while copper is 16.6 ppm/degree C. The thermal profile during high power dissipation is also significantly different from higher operational temperature. The die attach shear forces and cyclic fatigue are different.
I try to go very easy on the die when I need reliability. But that's just me, I'm still using amps, speakers, and ttables from '78..
John
While it is great to go massive cooling stuff, the power density can wreak havoc on the die attach solder. When the structures reach operating temperatures, each component will have thermally expanded at their different rates. Silicon is typically 3.3 ppm/degree C, while copper is 16.6 ppm/degree C. The thermal profile during high power dissipation is also significantly different from higher operational temperature. The die attach shear forces and cyclic fatigue are different.
I try to go very easy on the die when I need reliability. But that's just me, I'm still using amps, speakers, and ttables from '78..
John
IXYS have been producing power semi's since many years also using packaging technology incorporating a ceramic substrate, they seem to be the only one.
https://www.ixys.com/Documents/AppNotes/IXAN0022.pdf
https://www.ixys.com/Documents/AppNotes/IXAN0022.pdf
Excellent link.
I started using DBC back in '84 IIRC. Really neat stuff. Copper oxide melts 2 degrees C below the melt temperature of copper metal, the furnace profile to make the DBC has to be really really well controlled.
Because it's so close to copper melt, the final grain structure of the copper is really large, like 1/4 inch large, and dead soft.
And as stated, the TCE of the substrate is roughly 6 ppm, so much closer to the die expansion.
I suspect the grain structure is also an advantage for void control under the die. I remember Moto (I think) patented a stamped texture to attach the die to, the outgassing would pass through the texture gaps and not pool under the die. DBC by process has the grain structure "valleys" that do the same.
The DBC I used was nickel plated, and required a specific flux to cut the oxide, not sure how IXYS does it, but assume a belt or dap process is involved using no flux.
The alumina isolation is a major advantage...they mention aluminum nitride as well, which was just starting to arrive in the 80's as a material of interest. Not sure about the direct bond process for aluminum nitride, but it's thermal conductivity is almost that of aluminum and it's TCE is 4ppm, even closer to silicon.
jn
I started using DBC back in '84 IIRC. Really neat stuff. Copper oxide melts 2 degrees C below the melt temperature of copper metal, the furnace profile to make the DBC has to be really really well controlled.
Because it's so close to copper melt, the final grain structure of the copper is really large, like 1/4 inch large, and dead soft.
And as stated, the TCE of the substrate is roughly 6 ppm, so much closer to the die expansion.
I suspect the grain structure is also an advantage for void control under the die. I remember Moto (I think) patented a stamped texture to attach the die to, the outgassing would pass through the texture gaps and not pool under the die. DBC by process has the grain structure "valleys" that do the same.
The DBC I used was nickel plated, and required a specific flux to cut the oxide, not sure how IXYS does it, but assume a belt or dap process is involved using no flux.
The alumina isolation is a major advantage...they mention aluminum nitride as well, which was just starting to arrive in the 80's as a material of interest. Not sure about the direct bond process for aluminum nitride, but it's thermal conductivity is almost that of aluminum and it's TCE is 4ppm, even closer to silicon.
jn