High current transistor connections

[IconicPCB]

Looking for comments on how to connect to high current transistors supplied in TO-3 packages.

What to avoid and how to obtain reliable connections at say 50 A continuous.

[T3sl4co1l]

Like, new bought TO-3s?  Who does that?

Just regular old soldered connections is all I've seen.  No, you can't practically make a joint with wire that's rated for 50A in a conduit.  Fortunately, 14AWG will handle that just fine, for a little while at least.  The same can be said for TO-3s, so it seems fine to me. :-)

Tim

[nsrmagazin]

The transistor goes to a connector with very short PCB traces and very wide. The connector goes to where ever you want.

[IconicPCB]

The transistor is rated to dissipate 300W.
To be used in a linear power supply.

25Amps continuous, 50Amp peak at 28V output.

Problem is to make reliable connections.
The regulator driver will ensure the transistor SOA is maintained.

Fan cooled heatsink will be employed but never the less the connection needs to be reliable and since the unit may under some circumstances get sufficiently hot the soldered joint is potentially the weak spot.

[MagicSmoker]

Probably best to use a socket rather than attempt to solder directly to the pins. Also, you are almost certainly going to need to use multiple devices, anyway, since total R_theta from junction to ambient is unlikely to be under the 0.25C/W necessary to allow 300W Pd at 25C ambient.

[strawberry]

300W rated is optimistic value with perfect cooling (liquid cooling or peltier)
50A peak is rated for some milisecounds (to charge inductor or capacitor as example)

 300W/28V(collector emiter voltage drop) =10.7A continuous

[Doctorandus_P]

300W and 50A seems very optimistic indeed for a single TO3 transistor.
It's much more common (sens) to put several transistros parallel to share the load current.

What is the type number of your Transistor, and why are you using the obsolete TO3
Is this to be made in some kind of production environment, or is it a one-off with parts from your junk box?

[David Hess]

50 amp TO-3 transistors use a variation of the TO-3 case with 60 mil pins versus 40 mil or sometimes thinner pins.  High voltage TO-3 transistors like the 2N6545 use a variation where the glass feedthrough insulators are wider.

I would not use a socket at such high currents.  Wrap a suitable wire around each lead and solder it.  For extra mechanical strength, form a hook in the end of the lead first and maybe use low temperature silver solder.  Or use a printed circuit board but be careful about rigidly mounting the transistor and heat sink to the printed circuit board because thermal expansion can cause the soldered connections to crack; some mechanical compliance is required for long term reliability.

[Hydron]

300W is probably with Tc = 25C, which is tricky unless it's running outdoors in Antarctica

I'm currently using some FGY75N60SMD IGBTs which have large dies and 2 power ratings - 700W @ 25C, 375W @ 100C. Even the latter would be hard to deal with in practice - you're talking levels of power used by a serious GPU, and those have heatpipes/vapour chamber or sometimes even liquid cooling to move that heat away from what is a much larger die (albeit with a much lower Tj(max)).

Incidentally, it's not hard to find parts where the package/pins can't handle the specified max current that the semiconductor is rated to - I've got a SKM800GA126D sitting on my desk which is a lovely example - the datasheet contains this gem: "IDC ≤ 500A limited by terminals"

[David Hess]

When using high-power amplifiers, there are certain problems you just never have if you use a big-enough heat sink.  This heat sink's thermal resistance is lower than 0.5 C/W. - Figure 7.4, page 84, Troubleshooting Analog Circuits, Robert Pease

The photograph shows a TO-3 style part mounted to a cylinder head from VW engine.

[T3sl4co1l]

Yeah, this screams other design issues.  An SMPS with extra filtering is a strong contender here as well, even if you have a very tight noise requirement.

Tim

[MagicSmoker]

...
I would not use a socket at such high currents.
...

Normally I would advise the same, but not in this case... because the transistor is guaranteed to fail and a socket will make cooking its replacement easier.

[IconicPCB]

Thank You for Your opinions.

Allvalidaddressing the issues of TO-3 packaging.


A yes to all of the above comments.

I have seen some devices packaged in srew downterminal power packages. ON semi has only one such part number. Central semiconductor do not list anything.
Any other part numbers recomendations are welcome..

[David Hess]

...
I would not use a socket at such high currents.
...

Normally I would advise the same, but not in this case... because the transistor is guaranteed to fail and a socket will make cooking its replacement easier.

There are some big collet sockets made for 60 mil pins that might handle 50 amps.  I have seen them used on older Intel VRM designs among other things.  I cannot say that I have seen them used with power transistors but it would not be apparent unless they were removed.  I would not rely on the standard TO-3 transistor socket which relies on two points of contact with each pin.

[Doctorandus_P]

When using high-power amplifiers, there are certain problems you just never have if you use a big-enough heat sink.  This heat sink's thermal resistance is lower than 0.5 C/W. - Figure 7.4, page 84, Troubleshooting Analog Circuits, Robert Pease

The photograph shows a TO-3 style part mounted to a cylinder head from VW engine.

Ok, and 0.5C/Watt will will give a 150C temperature rise of the aluminimum @ 300W dissipation.
And we're not even talking about junction temperature here, nor about what's left on SOA for the transistor at such temperatures.

More glass and thinner pins also do not help with more current.

It somehow reminds me of MOSfets which can handle redicilous currents, even in TO220.
One of the datasheets I saw stated a current of 350A for the chip, but with an asterisk and a note that the pins start to melt at 150A or so.

[dmills]

I just love the marketing fiction on the first page of Mosfet datasheets, 350A from a D2PAK, yea, right!

At least the bipolar stuff sometimes has useful things like SOA curves (But usually only at 25 degrees C case temp).
Personally I favour TO3P, but at no more then about 80W and 5A or so per device, makes cooling actually possible.

The elephant in the room with TO3 is thermal expansion issues, you generally do not want to solder them straight to a PCB (In fact IIRC one of the IPC standards has quite a lot to say about this).

Some sort of sleeve from the likes of mill-max might be suitable, at those currents but I would be taking a long hard look at your design assumptions.

Regards, Dan.

[T3sl4co1l]

I have seen some devices packaged in srew downterminal power packages. ON semi has only one such part number. Central semiconductor do not list anything.
Any other part numbers recomendations are welcome..

ON Semi has one?  They make a dozen comparable parts for audio amplifier applications!

Shopping on mfg sites is an exercise in futility.  Shop at a supplier with parametric search.  You will find hundreds of variously suitable BJTs and MOSFETs, in all manner of manufacturing-amenable packages!

Tim