TO-3 linear regulator replacement

[Alex Eisenhut]

Given that there are many TO-220 style switching replacements for linear regulators, is there one on a TO-3 type PCB?

Just so you can drop it in old power supplies let's say, like for vintage computers. Noise not so important, but running a lot cooler would be better.

I thought I saw someone made some but I really can't find it.

[radioFlash]

Murata makes some, the 78xxSR series. These have output up to .5A and "up to" 95% efficiency.

http://power.murata.com/data/power/dms-78xxsr.pdf

The 78SR 2 Amp series goes up to 2 amps:

http://power.murata.com/data/power/mpm_78sr-2a_a00.pdf

[paulie]

78xxSR=to-220 !=to-3

[SeanB]

I think he wants ones that will go into a 2 pin with steel ground package exactly like a TO3 package does. Literally a drop in and screw down LM340T 5.0 replacement unit.

[NiHaoMike]

If you have some blown TO3 parts, break off the cap with large pliers and solder a switching module to the pins.

[Alex Eisenhut]

I think he wants ones that will go into a 2 pin with steel ground package exactly like a TO3 package does. Literally a drop in and screw down LM340T 5.0 replacement unit.

Yah, coulda sworn one of the vintage computing folks did it. Perhaps 'twas but a Glengoolie-fueled dream.

Or maybe a sign I should do it myself?

[SeanB]

PCB with a TO3 pattern on it, and simply have a TO220 pinout on top for the other style of regulator. Niche product for sure, but usable. Perhaps somebody could put a SMPS into a form with a TO3 layout, though the capacitors would make it higher than the TO3 can normally is.

[Alex Eisenhut]

PCB with a TO3 pattern on it, and simply have a TO220 pinout on top for the other style of regulator. Niche product for sure, but usable. Perhaps somebody could put a SMPS into a form with a TO3 layout, though the capacitors would make it higher than the TO3 can normally is.

For sure it's a niche, but some old equipment sure could use a thermal break. As you note, it should be simple, the biggest problem I predict is how to get leads on the PCB that won't melt when they're soldered into the circuit. Either HMP solder or some kind of inserted contact with retention force.

[radioFlash]

78xxSR=to-220 !=to-3

Yep, I missed that.

[Alex Eisenhut]

Hmm, to a first approximation, it's not impossible.

[c4757p]

I bet you can do even better with a nice high-frequency switcher and a smaller inductor

[Richard Crowley]

Some people simply remove the center pin from a TO-220 and fold over the remaining pins to fit into the original TO-3 pattern:

[Alex Eisenhut]

I bet you can do even better with a nice high-frequency switcher and a smaller inductor

That's gonna be the second approximation. 

Some people simply remove the center pin from a TO-220 and fold over the remaining pins to fit into the original TO-3 pattern:

Yeah but then it's still linear. The TO-220-style switchers I have wouldn't fit drop-in into my intended application.

[Vgkid]

If you are not running a ton of power, could you trim down the metal tab to fit.

[Alex Eisenhut]

I bet you can do even better with a nice high-frequency switcher and a smaller inductor

TPS54232D and a 3.3uH... Gotta look into that.

[mariush]

If money is not an issue, LTC3604 looks like a nice chip : http://uk.farnell.com/linear-technology/ltc3604eud-pbf/buck-synch-15v-2-5a-16qfn/dp/2295094

Up to 15v input, up to 95% efficiency, up to 4 Mhz frequency so you use 1uH or smaller inductors and a bunch of ceramic caps, up to 2.5A output (which was typically normal with to-3 packages).. it would definitely fit inside a to-3 package with room to spare.
Might even be able to fit a secondary filter at the output to make it look more linear.

ps. For up to 1A, LMR12010Y looks good, and is much cheaper as well and easier to solder :  http://uk.farnell.com/texas-instruments/lmr12010ymke/nano-module-1a/dp/2064697

[Alex Eisenhut]

If money is not an issue, LTC3604 looks like a nice chip : http://uk.farnell.com/linear-technology/ltc3604eud-pbf/buck-synch-15v-2-5a-16qfn/dp/2295094

Up to 15v input, up to 95% efficiency, up to 4 Mhz frequency so you use 1uH or smaller inductors and a bunch of ceramic caps, up to 2.5A output (which was typically normal with to-3 packages).. it would definitely fit inside a to-3 package with room to spare.
Might even be able to fit a secondary filter at the output to make it look more linear.

ps. For up to 1A, LMR12010Y looks good, and is much cheaper as well and easier to solder :  http://uk.farnell.com/texas-instruments/lmr12010ymke/nano-module-1a/dp/2064697

Thanks, I will take a look at all these chips. It's going to be quite amazing the power density achieved here. The choice of inductor will be interesting, I'll have to be sure it still has rated inductance at the rated current.

[SeanB]

You probaqbly will want to look at aluminium core boards for that chip if you want 2A. Electrodacus has done some work with using MCB in his designs, you probably could do it using a MCB with the TO3 footprint and the power devices, and a mezzanine board with the rest of the components on it, joined probably using either a solder joint or with a SM connector.

[Alex Eisenhut]

You probaqbly will want to look at aluminium core boards for that chip if you want 2A. Electrodacus has done some work with using MCB in his designs, you probably could do it using a MCB with the TO3 footprint and the power devices, and a mezzanine board with the rest of the components on it, joined probably using either a solder joint or with a SM connector.

I was wondering about that, with the integrated transistor in SOIC packages. Some chips have 4 ground pins, some only have one. One step at a time, I'll see later where this leads me.