TO-3's actually vary more in their construction and thermal impedance that TO264.
From strictly the viewpoint external to the package, it's the area of contact to the heatsink that counts, and the thermal interface material chosen. Area of the TO-264 is quite good, so it's really the chosen material (what type of film, or mica, etc) that matters.
Internal to the package, the chip size and package construction are the main matters. Look at specified Rth(j-c) (junction to case); TO-3 are made of many materials, stell, aluminum, or copper (mostly only TI used a copper alloy when they made power transistors); the Steel ones use some type of thermal spreaker, typically molybdenum, sometimes with copper, because soldering the silicon die to steel is not practical. This spreads the heat but somewhat increases Rth. Aluminum packages (such as Mot used to use) have fairly low Rth but terrible long term reliability due to CTE mismatch, and have been dropped.
I'd tend to vote for TO264, as the metal header is a copper alloy on which the chip is directly soldered; still, the chip size and construction is a factor also. The TO264 is much easier to mount, also. In all cases, bellview cup washers should be used with a torque wrench.
~Jon
PS- I'm a power semi apps enginer.
From strictly the viewpoint external to the package, it's the area of contact to the heatsink that counts, and the thermal interface material chosen. Area of the TO-264 is quite good, so it's really the chosen material (what type of film, or mica, etc) that matters.
Internal to the package, the chip size and package construction are the main matters. Look at specified Rth(j-c) (junction to case); TO-3 are made of many materials, stell, aluminum, or copper (mostly only TI used a copper alloy when they made power transistors); the Steel ones use some type of thermal spreaker, typically molybdenum, sometimes with copper, because soldering the silicon die to steel is not practical. This spreads the heat but somewhat increases Rth. Aluminum packages (such as Mot used to use) have fairly low Rth but terrible long term reliability due to CTE mismatch, and have been dropped.
I'd tend to vote for TO264, as the metal header is a copper alloy on which the chip is directly soldered; still, the chip size and construction is a factor also. The TO264 is much easier to mount, also. In all cases, bellview cup washers should be used with a torque wrench.
~Jon
PS- I'm a power semi apps enginer.
aavid thermalloy makes a series of semiconductor mounting clips, among which are the max-07 for t0-220 and the max-08 for to-3p. the to-264 is a close cousin of the to-3p, a bit larger in area but close to the same thickness. these clips give you single hole mounting and measured force over the heat transfer area. they would seem to be a good transistor mounting solution.
Mr JonMarsh s opinion seems to carry weight-----but------While brousing through
the forum I came across ----In the thread Leach Superamp vs Leach 4.5 , Mr
Samuel Jayaraj & djk discuss the following ,
samuel asked -----How many MJL21193/21194 (Plastic package) devices
should I use in the Superamp to operate reliably into 2 ohm loads at continous full
power? This is for serious PA applications.
djk replied--------Generalizations that can be made: Plastic is a waste of time and
money for big amplifiers. Assuming realistic transistor case operating
temperatures you will need twice as many MJL21193/94 as you would
MJ21193/94 because of the maximum junction temperatures. Then there are
thermal cycling issues, at the same junction temperatures the metal case style has
10X the life span as the plastic
samuel------Does the same rule apply to Mosfets also, ie., Metal can (TO3) being
able to handle more power than Plastic (TO3P, TOP264 etc.,)??
djk------
In the case of the Motorola parts the TO3 case parts are rated at 25% more
power at room temperature, plus there is a 50*C difference in the maximum
allowable junction temperature.
The Xicon/Magnetec/Semelab parts are rated at the same power in either case
style, plus the TO3 case has the same maximum temperature as the plastic. I still
like the TO3 package over plastic though. I have not seen any thermal cycling data
on these parts either. The data from International Rectifier on their FETs actually
showed the TO3 as being worse than their plastic for reliability (the number of
samples may have been too small to be statistically correct). Data from Harris
(RCA) and ON (Motorola) show the stainless steel cased TO3 to be way out front
of plastic
samuel--------Thanks again djk. I wonder why the IR parts are the only
exclusion/contradiction to an otherwise accepted fact that the TO3 package is
consistently better than Plastic cases (TO3P, TOP264 and ofcourse, TO220) in
terms of max.Junction Temperature and Dissipation characteristics; perhaps, it
may have to do with the internal structure.
P.S. One of the moderators of this forum could probably link this thread in such a
way that a Forum Search for "TO3 vs Plastic case" would show up djk's very
concise, informative posts.
I thought reposting the above in this thread will get more answers and will be more
informative to all.
Rajeev Luthra
the forum I came across ----In the thread Leach Superamp vs Leach 4.5 , Mr
Samuel Jayaraj & djk discuss the following ,
samuel asked -----How many MJL21193/21194 (Plastic package) devices
should I use in the Superamp to operate reliably into 2 ohm loads at continous full
power? This is for serious PA applications.
djk replied--------Generalizations that can be made: Plastic is a waste of time and
money for big amplifiers. Assuming realistic transistor case operating
temperatures you will need twice as many MJL21193/94 as you would
MJ21193/94 because of the maximum junction temperatures. Then there are
thermal cycling issues, at the same junction temperatures the metal case style has
10X the life span as the plastic
samuel------Does the same rule apply to Mosfets also, ie., Metal can (TO3) being
able to handle more power than Plastic (TO3P, TOP264 etc.,)??
djk------
In the case of the Motorola parts the TO3 case parts are rated at 25% more
power at room temperature, plus there is a 50*C difference in the maximum
allowable junction temperature.
The Xicon/Magnetec/Semelab parts are rated at the same power in either case
style, plus the TO3 case has the same maximum temperature as the plastic. I still
like the TO3 package over plastic though. I have not seen any thermal cycling data
on these parts either. The data from International Rectifier on their FETs actually
showed the TO3 as being worse than their plastic for reliability (the number of
samples may have been too small to be statistically correct). Data from Harris
(RCA) and ON (Motorola) show the stainless steel cased TO3 to be way out front
of plastic
samuel--------Thanks again djk. I wonder why the IR parts are the only
exclusion/contradiction to an otherwise accepted fact that the TO3 package is
consistently better than Plastic cases (TO3P, TOP264 and ofcourse, TO220) in
terms of max.Junction Temperature and Dissipation characteristics; perhaps, it
may have to do with the internal structure.
P.S. One of the moderators of this forum could probably link this thread in such a
way that a Forum Search for "TO3 vs Plastic case" would show up djk's very
concise, informative posts.
I thought reposting the above in this thread will get more answers and will be more
informative to all.
Rajeev Luthra
Please allow some comments from a longterm experienced guy in semiconductor device- packaging and reliability.
1. Be very carefull by comparing reliability data from different manufacturer. Very often the test methodes are not the same and thus the results not comparable.
Hitachi. Sanken, Toshiba, IRF and other leading power device manufacturer have longterm compared reliability data of metal cases with new designs like TO3P etc and have decided to use only plastic packages. In the most of the cases TO3 packages are replaced by plastic packages. Thus for the majority of devices no choice exist anymore.
2. The todays maximum junction temperature for modern power devices is generally 150°C and does not depend on device package type only on applied chip manufacturing technology. Some new devices for automotive application even can work at 180 - 200°C for more than 2000 hours.
3. RthU depend on the package heatsink surface and the used metal. TO3 as already mentioned is made from different metals like Ni/Fe alloys and others. These alloys show pour thermal conductivity, compared with the Ni-plated copper heatsinks used with plastic packages. In addition the heatsink planarity specification of plastic packages are much more tight than those of TO3, resulting in better heat transfer from package heatsink to power heatsink. In the case of plastic packages the chip has direct contact to the plastic encapsulating material, which in the most of power device cases is a high thermal conductive material ~2.6W/mK. Thus heat can also be direct radiated to the environmental and can be improved by using mounting clips which act as additional heatsink.
4. Anyway, if you build class A amps, don't forget that the silicon chip age if used at the thermal high end specification and alter his properties. Thus you should have enough spare parts if you like that your amps are still well sounding after several years. You should be prepared to replace your power transistors if used with high bias current at least after 2000 hours. Most of the Japanese power transistors use for PNP-devices a green or blue package material. If the green/blue color of these devices turn to brown at the place of the chip, it is time to replace the device with a new one because the plastic material has significantly changed his properties and negatively influence the device properties, mostly resulting in a leakage current increase.
These are only few arguments based on real facts of R&D and application experience.
Gerd
1. Be very carefull by comparing reliability data from different manufacturer. Very often the test methodes are not the same and thus the results not comparable.
Hitachi. Sanken, Toshiba, IRF and other leading power device manufacturer have longterm compared reliability data of metal cases with new designs like TO3P etc and have decided to use only plastic packages. In the most of the cases TO3 packages are replaced by plastic packages. Thus for the majority of devices no choice exist anymore.
2. The todays maximum junction temperature for modern power devices is generally 150°C and does not depend on device package type only on applied chip manufacturing technology. Some new devices for automotive application even can work at 180 - 200°C for more than 2000 hours.
3. RthU depend on the package heatsink surface and the used metal. TO3 as already mentioned is made from different metals like Ni/Fe alloys and others. These alloys show pour thermal conductivity, compared with the Ni-plated copper heatsinks used with plastic packages. In addition the heatsink planarity specification of plastic packages are much more tight than those of TO3, resulting in better heat transfer from package heatsink to power heatsink. In the case of plastic packages the chip has direct contact to the plastic encapsulating material, which in the most of power device cases is a high thermal conductive material ~2.6W/mK. Thus heat can also be direct radiated to the environmental and can be improved by using mounting clips which act as additional heatsink.
4. Anyway, if you build class A amps, don't forget that the silicon chip age if used at the thermal high end specification and alter his properties. Thus you should have enough spare parts if you like that your amps are still well sounding after several years. You should be prepared to replace your power transistors if used with high bias current at least after 2000 hours. Most of the Japanese power transistors use for PNP-devices a green or blue package material. If the green/blue color of these devices turn to brown at the place of the chip, it is time to replace the device with a new one because the plastic material has significantly changed his properties and negatively influence the device properties, mostly resulting in a leakage current increase.
These are only few arguments based on real facts of R&D and application experience.
Gerd
After reading the text of JonMarsh & GKU I am convinced that the comming time is of the plastic devices and the TO-3 will slowly phase out , I am sure djk will also change his opinion as the new devices are better in all respects . Other than the technical aspect they are also more convienent to use as they can be soldered on the pcb and all can be clamped on a heatsink in a row thus saving time which is spent in the mounting and wiring of the TO-3 devices.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.