Are modern 78xx/79xx better than old ones?

[stefon]

I have 40 year old equimpment with linear power supply. Now is working fine after I've replaced few blown tantalum caps, but I wonder, if 40 year old 78xx /79xx is as good as modern one? Did they also have thermal and short protection? Or this depends on manufacturer?

[james_s]

I don't think they've ever really changed. Certainly the protection has been there since day one.

[T3sl4co1l]

Some differences may include, die shrink (lower Pdiss, shorter thermal time constant), improved specs (ADJ/GND bias and dropout probably about the same, but Vout tolerance may be tighter in the average case?), and case variations (thinner tab?).  Maybe compensation as well (more or less tolerant of certain combinations of ceramic or tantalum capacitors on the output or input?).

You'd have to test them, or crack them open, to see who's is doing what.  (Some photos are already on zeptobars.ru and others, so it's not entirely that awful. )

Not that they were ever good for very much current or power in the first place, and if you need an ampere or more you really should be investing in either a bigger or boosted reg, or a switcher (which can be a valid replacement option).  Note that hacked upgrades can be a serviceable option for equipment that was designed on the margins of its components.

Incidentally, did you know "TO-220" allows, not only a tab as thin as the leads, but wire leads as well?  Yeah, it's a truly ancient mess!  The first time seeing those, you might think it's a joke, or a hideously cheapened part, but nah, it is, in fact, within spec.  Shop for TO-220AB (the improved version with flat leads, guaranteed thicker tab, and usually with scalloped sides for whatever reason) or others if you need better thermal performance.

I don't think they've ever really changed. Certainly the protection has been there since day one.

I recall a story in, a Pease column I think -- they had a use case that was blowing up their "unblowuppable" regulators.  Some dick engineer at Delco ran one so close to the limits that it passed tests, but later failed from thermal cycling.  I think the conclusion was some tweaks to operating conditions and/or process or materials improvements.  (One would hope the engineer got a slap as well, but that's regrettably rare when you're on the supply side.)

Tim

[digsys]

Having run extensive destructive testing on dozens of variations of the ol' 78xx, 79xx and LM317 + variations, Tescoll(sp) is perfectly correct. There is a damn WIDE variation now on these parts from ALL suppliers, even sometimes vary between batches of supposedly the SAME part #. It's a mess.
The thermal transfer behavior is the worst offender. We test 1 out of EVERY batch before production.

[Shock]

I did a repair a while back that had a TO-220 packaged 7805 that had become input voltage sensitive due to a faulty tantalum on the output. I felt a lot more confidant reusing a Motorola (which had reasonably low hours) off a scrap pcb than cheap Chinese plastic one that the local electronics store sells. Additionally I added a heat sink as it was in free air, I think the chance of it failing in the next 40 years is fairly small.

So not that I have empirical evidence it was better, but the quality of cheap random branded plastic components these days doesn't inspire confidence. Using quality components is always the way to go especially on good equipment that didn't cheap out on components when it was originally built.

As for comparing quality components we have datasheets. If there is no datasheet, no brand or purchased through some unknown third party rather than a supplier dealing with the manufacturer, then I consider it a junk component. In addition to what Digsys said, if it falls outside their own spec then you know it's going to be rubbish.

Do you have to go replacing these old components retroactively? No I don't think so unless you notice it's out of spec. The exception to that is known point of failures on common components like capacitors. Proactive repairs like replacing components that are loading down others, resoldering joints and cleaning then reapplying thermal paste etc all help extend the life of the device, otherwise ride it till it dies.

[floobydust]

Between manufacturers, I find popcorn noise is all over the place. Some makes are quiet, others very noisy. Some gear does not care.
Also I find the cheap "single gauge TO-220" package absolute junk if the vregs are on a heatsink.

[james_s]

When it comes to the cheap off brand stuff all bets are off. I've never had a problem with 78xx regulators bought from reputable suppliers though. I've never needed to run one near the limits or aim for the ultimate in low noise though.

[Hexley]

I wonder, if 40 year old 78xx /79xx is as good as modern one?

Last year I had cause** to open up a 1970s vintage Fluke 8600A DMM and measure the output voltage of its internal LM340T5 regulator. It had crept up to 5.48 volts, well above spec. The date code on the chip was from 1974. So stuff happened over the intervening 40+ years, it seems.

BTW, I replaced it with a modern and accurate ST “A-series” part. Measured 4.986 volts after the change.

---
** The cause was to search for the reason that the DMM was reading 7.5% low on some ranges. Turned out not to be the power supply, but a bad analog switch (CD4016, U17), that was injecting charge into the reference voltage divider. Replaced that 40-year-old chip with a (somewhat) newer chip, and the problem was solved.

[mcovington]

Did LM340T-5 have lower specs than 7805 in the first place?  I seem to recall it was earlier and 7805 replaced it.

[wraper]

Did LM340T-5 have lower specs than 7805 in the first place?  I seem to recall it was earlier and 7805 replaced it.

It's the same thing AFAIK specific to National semi (now TI), with the same datasheet and specs (LM340A has tighter spec) and still produced.
There really are parts where manufacturer sells exactly the same part under several names. One name company specific, other basically industry standard.

[floobydust]

LM340 was popular for a few years in the late 1970's, datasheet claims twice as good regulation as LM78xx parts yet I'm not sure why it fizzled out. Both IC's have very similar schematics pretty much identical so I'm not sure how it got more gain.

[David Hess]

Some differences may include, die shrink (lower Pdiss, shorter thermal time constant), improved specs (ADJ/GND bias and dropout probably about the same, but Vout tolerance may be tighter in the average case?), and case variations (thinner tab?).  Maybe compensation as well (more or less tolerant of certain combinations of ceramic or tantalum capacitors on the output or input?).

I have personally seen smaller dies and thinner lead frames but not in all models.

I recall a story in, a Pease column I think -- they had a use case that was blowing up their "unblowuppable" regulators.  Some dick engineer at Delco ran one so close to the limits that it passed tests, but later failed from thermal cycling.  I think the conclusion was some tweaks to operating conditions and/or process or materials improvements.  (One would hope the engineer got a slap as well, but that's regrettably rare when you're on the supply side.)

Early ones lacked hysteresis in the thermal protection so they could sit at the maximum power point during overload with a load which they could start if they were not already in thermal shutdown.  Later parts fixed this so that the regulator would cool down enough to do a hard start into a difficult load.