Metal footprint area of an MT200 is also larger than the one of a TO-3 (a TO-3 has 4 holes, an MT200 only two)
The copper heat spreader of an MT200 :
- larger area
- thicker (2mm)
- rectangular shaped, and L/R symmetrically positioned
TO-3 :
- L/R heat distribution is asymmetrical (off-center die + copper, the copper has a bottleneck inbetween the holes for base/emitter leads)
The package of an MT200 is 6mm thick, much stiffer (Area Moment of Inertia of a rectangular beam of homogeneous material : twice as thick, 8 times stiffer)
I can bend the mounting plate of a TO-3 with my fingers, not so for an MT200.
Thermal conduction of an epoxy package is better than for air, inside of a TO-3 top is air.
Radiation for a dT of 25C-30C will not amount to much, reason for selling top-hat heatsinks for TO-3 devices.
EUVL measured less than 0.3C/W for a TO-264 package with a mica insulator (21mm x 26mm, metal part of the footprint area is 55 percent), see his Linear Audio article.
Footprint area of an MT200 is over 40 percent larger.
(mounting plate of an MT200 has a 2mm thickness, makes 4mm with the copper added)
Last edited:
TO3 > MT200 > TO-264
IMO TO-3 is old school and not recommended for new designs, they already are getting scarcer.
Inside air space, thickness of this, and top hats don't matter with real heat sinks, look at the Theta JC specs! TO-3 has 4 holes? if you can bend a TO-3 easily then we can still call you Superman 😱 besides all that not all TO-3 are created equal.
Top hats are only seen for devices without a HS and the delta Ts will be very high indeed.
Last edited:
Sorry, I don't buy these rabbits. If you want to sell, try painting them pink.
Seriously, 0.3C/W for TO264 with mica is against any data sheet or application note you may find. I don't know who EUVL is (and I really don't care) and I don't subscribe to any Linear Audio magazine (whatever that is).
Check this one (is Fairchild good enough?) https://www.fairchildsemi.com/application-notes/AN/AN-4166.pdf
Fig 7f:
TO264, no mica, greased, at maximum torque: 0.4C/W
TO264, no mica, dry, at maximum torque: 0.7C/W
Fig9:
TO264, mica, greased on both sides, at maximum torque: 1.0C/W
TO264, silicone pad, optimum torque: 2.0C/W
I'll hold the comments on the other rabbits. The one with bending a TO3 (like the MJE15022) with fingers only is to juicy to miss though, you must be an Avengers team member.
TO3's are (were) made with both steel and copper bases. The copper could well be malleable, not the steel. However they are essentially obsolete like many audio parts.
If you are going into the Wayback machine for this stuff use beryllium oxide washers Beryllium Oxide Ceramic - heat sink interface material . They also have the virtue of being more expensive than the transistor they are used with.
If you are going into the Wayback machine for this stuff use beryllium oxide washers Beryllium Oxide Ceramic - heat sink interface material . They also have the virtue of being more expensive than the transistor they are used with.
True, mounting plate/base of e.g. an (alloy) Hitachi TO-3 lateral MOSFET has a 3mm thickness, very stiff. Regular TO-3's are 2mm, the ends are flexible.
2 mounting holes, plus 2 for the base and emitter lead, which are to be subtracted from the total mounting area.
Total mounting area of a TO-3 is easy enough to measure for yourself, by drawing the outline on square millimeter paper, and count the squares. Back when, I used a planimeter at university. (used to measure the area of a ship hull cross-section)
(cheap TO-3 series only had a millimeter thickness, zero problem to rip the top of those. I weigh 250lb at the moment, btw, lost 15 thanks to a nose that's gone scent-crazy)
Last edited:
That fairchild paper has graphs of ThetaJC yet they are measuring effect of washers and gloop which is (according to their own diagram) ThetaCS. Which is being argued about?
Thank you for the link.
You really should get the EUVL article from Linear Audio vol 3, it is quite good.
I go into the thermal stack details in my article in LA #9.
That Fairchild app note is quite poorly written. They mix up thermal resistance, thermal conductivity, they do not explain the methodology to any degree..
Fairchild also fails to make the distinction of the die size and it's impact on TJC, nor the thickness of the copper on the bottom of the package; it has a tremendous effect on the overall thermal spread to the sink. Rather, they are rating the package type as the dominant, which is good enough if one uses terrible insulator conductivity like mica and grease. It would be far better to use the graphene sheet and either alumina or BeO as the insulating material.
I assume their thermal resistance measurement is junction to sink, but again, they really don't write up the methodology very well.
John
No surprise, the JEDEC standards for specifying this stuff are essentially useless.
Great, now go ahead and read it. Rthj-heatsink for TO264 isolated is about 1.0C/W, even in the most favourable case, in agreement with the Fairchild app note. Not even close to what you claimed (0.3C/W in #78061).
Last edited:
Agreed, too expensive.
Using alumina doesn't give too much of a hit on performance though.
A 1/2 by 1/2 inch piece 25 mils thick is only a .09 Deg C/watt increase over BeO, or 9 degrees C higher junction temp at 100 watts dissipation.
BTW, I seem to recall some moto TO-3 parts that seemed to be aluminum cases. I think my tigers use them.
John
Live long enough and everything you know will be found to be not true.
Like the speed of light being the fastest.......
CERN scientists 'break the speed of light' - Telegraph
THx-RNMarsh
Like the speed of light being the fastest.......
CERN scientists 'break the speed of light' - Telegraph
THx-RNMarsh
This a Joke? That article is over 4 years old. Since then a fault was found with the equipment, the speed of light remains unbroken and someone lost his job.
Last edited:
I have nothing against Italians or their priorities, but when the Gran Sasso stuff became public, Brad Plunkett and I had, independently, the same reactions, and concocted a very similar anecdote about the discrepancy in arrival times, having to do with when they had gone to lunch.
That sets a new benchmark for an audio amplifier speed. Anything not matching that should be considered mid-fi at best.
EUVL measured an Rth case to heatsink of 0.29C/W for a TO-264 package with a mica insulator (Table 1)
Rjs = Rjc + Rcs = Rth junction to case + Rth case to heatsink.
Rjc for a 200W Sanken in an MT200 package is 0.625 C/W (125C/200W)
Olympus F1
What actually happened was a fault in the Infra red sensor used to detect the earth and the satellite went into emergency sun pointing mode. The Italian ground control wanted to get away to watch the football so decided they could get it back on line quicker by ignoring the manual and left a thruster running for 20 minutes, with legendary effects.
A lot of grumpy people in the office I was working in when that happened. Used up 8 years of station keeping fuel getting it back. The fact that they recovered it was astounding.
- Status
- Not open for further replies.