Can you mount TO-3 transistors on flat heatsinks?

[heatbreak]

Hi there,
    It seems that it's easier and cheaper to get flat heatsinks (you know those used to mount TO-220s).  Is there anyway to mount TO-3 parts onto those flat heatsinks?  I thought about mounting a bank of them onto some kind of L-bracket and then mount the L-bracket onto the heatsink.  Would that work?  Would that give the same heat transfer as if you mount them directly onto the heatsinks?

Thanks

[Ian.M]

Errr. . .
TO-3 are designed for external mounting to a flat heatsink surface.  You need a 27mm flat between the fins and simply drill clearance holes for the leads.

[retrolefty]

Is there anyway to mount TO-3 parts onto those flat heatsinks?

 There sure is, and it works much better then on curved heatsinks.   

[lwatts666]

I assume you mean one side fully covered with fins, the other side completely flat.

The angle bracket solution can work, but it introduces additional thermal resistance between the TO-3 and the heatsink, and hence gives greater die temperature for a given power dissipation. Depending on the situation this may become a problem.

A suitably thick custom drilled bracket, additional mounting screws, extra heat transfer compound, assembly, etc will probably cost more than the savings made in the heat sink.

I've seen it done, and done it myself, when mechanical constraints overrode the additional costs involved.

Preference should be with a heat sink designed for TO-3's if possible

[station240]

It seems that it's easier and cheaper to get flat heatsinks (you know those used to mount TO-220s).  Is there anyway to mount TO-3 parts onto those flat heatsinks?  I thought about mounting a bank of them onto some kind of L-bracket and then mount the L-bracket onto the heatsink.

It's been done before, a common 100W* amplifier kit in Australia had 4 TO-3's mounted to an L bracket, then the bracket mounted to the heatsink. Kit came without the heatsink, so you could select/supply one to suit.

* there was also a 50W version with only half the output transistors.

[Siwastaja]

L brackets are used quite often. You can even sandwich the bracket to PCB for direct PCB mounting of the TO-3 packages. You need to screw them down and tighten before soldering the leads. This construction limits the L material thickness so much that using copper instead of aluminium is highly recommended. In any case, you need to do the thermal analysis and an extra L bracket complicates it a bit.

[Ian.M]

You'd do better simply to get TO-247 package parts in the first place.  However if you *MUST* mount a TO-3 to a flat base, fully finned heatsink, and you cant afford a heavy top hat heat spreader bracket or milled interface plate, consider drilling two extra holes outside the outline so you can thread the B and E wires back through from the fin side.  You will need to be really careful to align the B and E leg holes with a gap between fins, and wont be able to fit any ferrite beads or small caps right at the pins that may be needed for stability.

[Pjotr]

L brackets are used quite often. You can even sandwich the bracket to PCB for direct PCB mounting of the TO-3 packages. You need to screw them down and tighten before soldering the leads. This construction limits the L material thickness so much that using copper instead of aluminium is highly recommended. In any case, you need to do the thermal analysis and an extra L bracket complicates it a bit.

That is not a good idea. Many audio amps are constructed that way. I have repaired many due to loose solder joints over time. The solder joints cannot stand the stress due to thermal expansion of the TO3 pins. If you do so, always solder the TO3 pins with a separate wire to the PCB to catch thermal expansion.

[rdl]

It probably doesn't matter as long as the heatsink has enough surface area and the transistor is bolted down with sufficiently good contact.

The HP 3610 power supply uses heatsinks where the fins are vertical on the back of the case and the TO-3 transistors are mounted horizontally (pins down) on a "shelf" that projects into the case (see EEVblog #166, fast forward to about the 12 minute mark).

[pmbrunelle]

L brackets are used quite often. You can even sandwich the bracket to PCB for direct PCB mounting of the TO-3 packages. You need to screw them down and tighten before soldering the leads. This construction limits the L material thickness so much that using copper instead of aluminium is highly recommended. In any case, you need to do the thermal analysis and an extra L bracket complicates it a bit.

That is not a good idea. Many audio amps are constructed that way. I have repaired many due to loose solder joints over time. The solder joints cannot stand the stress due to thermal expansion of the TO3 pins. If you do so, always solder the TO3 pins with a separate wire to the PCB to catch thermal expansion.

This is good to know!

I wonder if you can bend the leads in a Z-shape first, and then soldering to a PCB might be reliable.

[heatbreak]

The HP 3610 power supply uses heatsinks where the fins are vertical on the back of the case and the TO-3 transistors are mounted horizontally (pins down) on a "shelf" that projects into the case (see EEVblog #166, fast forward to about the 12 minute mark).

Looks the "L-Bracket" and the heatsink are one single piece on the HP-3610.  That situation is similar to the earlier Hafler audio amps.  We were all discussing the situation where you have two separate heatsinks: L-Bracket and a flat, finned, heatsink and the two is joined together with screws and heatsink compound.

[SL4P]

How about mounting the TO-3 packages 'properly' to a suitable heatsink for the required application - and running simple flying lead pigtails back to the PCB ?

This has been done countless times in the past and doesn't seem to have any downside.

[dom0]

Lead inductance.

Why not just use TO-247 AKA TO-3P (because it achieves very similar thermal performance as TO-3)?

[SL4P]

Lead inductance
Sorry, I'm didn't pick up the RF/HF requirement...  possibly routing the leads, or individual shielding as needed.
I'm not experienced in that area.

[janoc]

Depends on the application but most places where TO-3s are used the lead inductance is not really an issue - audio amps, power supplies, etc.

All low frequency stuff. The parasitics can be dealt with by keeping the wires as short as possible - i.e. don't tread the wires to your PA transistors all over the case!

[Kleinstein]

Even though frequencies are low in DC supplies and Audio amplifiers, the output impedance is low and loop bandwidth is the critical limit, not the actual signal frequency. So parasitic inductance in the 100 nH range can matter even in a DC supply. However modern supplies prefer TO247 or similar over the old TO3 style.

[Pjotr]

GBP of a PSU is usally very small due to the output caps. Otherwise sense wires do not work. It is different with audio amps indeed. But even there, GBP is no more than 10MHz at best.

[dom0]

Of the whole amplifier... not of the output stage.

--

The classic HP topology is also somewhat prone to RF oscillations, due to the high voltage and current gain in the pass element. With high speed MOSFET devices those things can oscillate at over 100 MHz. Loop speed mainly depends on choice of output capacitance => low output capacitance => need for dynamic regulation => fast loop.