Why this Voltage regulator is so expensive ??

[elnaldo]

Hello. I need to replace a LM338K regulator (TO-3), and I see it sells for $60 each (or more). The TO-220 version is just $2 (I see they share the datasheet, specs seem to be the same, I'll use this cheaper part and bend the leads to use the TO-3 heat-sink)

Do you know why this difference in price? Is it some very special part? Is it due to availability? Is the metal TO-3 case too expensive to manufacture? Is the TO-3 version much better than the plastic TO-220?

Mouser link (No Affiliation)
https://ar.mouser.com/Semiconductors/Power-Management-ICs/_/N-wnwhZ1yzvvqx?Keyword=lm338&FS=True

DigiKey link (No Affiliation)
https://www.digikey.com/products/en...cs/pmic-voltage-regulators-linear/699?k=lm338

 

[Watthour]

My guess is the supply/demand. It's likely that not many are made in T-03 cases. It appears that the T-03 can dissipate 50W while the TO-220 is rated only to 25W, but both at 5A continuous. I'd also be getting the TO-220 and heat sink it as necessary.

 

[ConradH]

Wow, that's crazy. I'm guessing demand must be near zero.

 

[elnaldo]

The power supply is rated +- 17V 1.5A each rail, input voltage to the regulator is near 30V, output 17V.

I think it should dissipate 13V * 1.5A = near 20 W at max output. Is my math right?

I don't think the mixing console it feeds will draw 1.5A from each rail, but I need to check the specs.

 

[sKiZo]

I'd suggest sinking it directly to the chassis with an insulator and lots of thermal paste. You can always stretch the leads to fit the board. Maximize air space around it as well.

 

[petehall347]

obsoleted ... no longer manufactured .. http://uk.farnell.com/stmicroelectronics/lm338k/ic-v-reg-adj-1-2-32v-to-3-2-338/dp/1094254?st=LM338K

 

[petehall347]

contact first and ask questions if part is genuine . and if they ever had complaints .. http://nikkoe.com/lm338k-30181.html

 

[ivandezande]

Mixing console huh...I redid a Soundcraft Delta 200 power supply that had like 5 of those TO-3 guys in it. One was an NTE and I wanted to replace it but saw the prices you did and thought otherwise. Strange stuff.

 

[petehall347]

seems all T03 reg.s are expensive ... https://uk.rs-online.com/web/c/semi...gulators/?sra=p&applied-dimensions=4294509596

 

[audiodummy]

I've started collecting these obsolete TO-3 ICs as I run across them... LM317K, LM340K, LM350K...

You're probably best off retrofitting with a TO-220 however. Main problem of doing so is that the thermal resistance of the TO-3 is better than TO-220.

Oh... and *minimize* heatsink paste. Remember that heatsink paste only complements the contact - fill the air voids.. You want metal-metal contact ideally, not through heatsink paste. Torque down properly - don't overtorque and cause the flange to bend or too little, leaving air voids. Of course if you have to deal with insulation, you're screwed.

Other possibilities include adding a series pass transistor or diode drops so that the LM338T doesn't take the brunt of the drop. The former will require significant redesign however.

 

[petehall347]

or maybe if possible running 2 in parallel ...

 

[Lavane]

I'm running a dual regulator right at its max rating to regulate a pair of 300B tube filaments. Adding a much bigger heat sink mounted to the chassis top plate really helped. It's been a year so far and not one time has it shut down from over heating. If your close to the max extra heat sinking should work. You could mount the regulator somewhere else to make it easier to add one if needed as sKiZo suggested. That's what I did it worked fine.

 

[cpt_paranoia]

Of course if you have to deal with insulation, you're screwed.

That's where a silpad comes in handy.

 

[elnaldo]

It's a Soundcraft CPS-150 power supply, this is the heat-sink, devices are insulated. I don't have the plastic covers. I think I'm going the TO-220 route (I can get it locally) . I'll check the mixer specs to see if I need a second device in parallel, but I think with 25W it will be fine.

I can add some smaller heat-sinks inside the cabinet, using one of the TO-3 screws unused with the TO-220 device.

Pict from the internet, not mine:

 

[elnaldo]

contact first and ask questions if part is genuine . and if they ever had complaints .. http://nikkoe.com/lm338k-30181.html

Shipping is an issue, I need to get it from Mouser, or locally (locally the IC is cheaper, but expensive compared to the TO-220, and there is a risk of fakes ones even if they look OK)

 

[fernarias]

NTE935 is the replacement and cost less. Funny, I have a new old stock 935 that I purchased in 1988 for a power supply project and ended up using a to-220 instead.

 

[elnaldo]

Finally got one "fake" TO-3 device locally for $6. The regulator works. The PS is rated 1.25A, not 1.5. I'll try this one and see how long it lasts. If it blows this next month I'll modify the circuit to install To-220 devices , perhaps with a resistor in parallel.

 

[triode17]

Funny, I just needed an LM7812K in a TO-3 pkg. They're long gone out of production. That explains the price. But luckily, I had an LM7812K and 7912K in my old parts stock. I know its real because its from 1979. I'm pushing it hard and although a TO-220 pkg. will work, the hot spot is too intense.

 

[cpt_paranoia]

Check the 78xx regulator application notes for how to use a high power parallel bypass transistor to make a high current regulator.

 

[audiodummy]

Careful, bare metal contact is always better than adding anything in between - especially air voids. The soft silicone insulation at least removes some of the air voids, but still not better than metal-metal contact.

If you go the TO-220 route, you could indeed hook up two TO-220's in parallel with a low value resistor (say 0.1Ω) on each IC's output. You may still need to check whether the circuit will support the extra current from the ADJ pin.

You may be better off drilling new holes for the TO-220s, using the existing TO-3 holes may not work well.