79L05 stability, output capacitance

It's not an LDO in the usual sense, _but_ it is a hokey PNP equivalent pass device made up of a lateral PNP plus an NPN, so it's potentially squirrelly. Disclaimer: I haven't used one in probably 25+ years ;-) ...Jim Thompson

I wonder what it'll say to a big fat ceramic cap on its output.

I raise you another 25 years :-)

Can't remember ever having used them on any of my designs. Back in my school days such regulators were unaffordable anyhow. I did almost everything with uA723 or home-brew switchers where the controller was a cheap logic chip.

Can they still not make a monolithic PNP power tranny?

Wasn't there a time when it was PNP discrete transistors that were cheap and readily available, but ICs could only have NPNs?

Must have been a pain in the ass era to design power supplies.

X-Fab and IBM have processes with real, high-fT PNP's, but they're expensive multi-layer processes. ...Jim Thompson

I had lateral PNP's available to me (personally) at Motorola in 1962. ...Jim Thompson

Some of the decisions I see in power supply design from Japanese equipment I've repaired from the 1980s don't make a lot of sense to me, at least from a modern perspective. Lots of discrete parts, zeners, pass transistors, etc.

It's like the were looking at a circuit book from the late 1960s, but this was Japan circa 1986. I'm like, couldn't you guys have just slapped a 7815 in there and called it a day? They couldn't have cost that much even then...

But you graduated college in 1962. You mean there was a time when people actually got jobs doing what they were trained to do with their degree, right out of school?

What halcyon days...

Actually Motorola offered the job in February 1962, but I didn't graduate from MIT until June. ...Jim Thompson

Why not use an LM337 instead? ...Jim Thompson

That was a really long time ago. Once monolithic regulators came around, you used LM104/105s with external pass devices, which were pretty nice for the era. You can do it with a 723 as well, of course, if you stand on one leg a bit.

I never use 78L/79L devices either, because their ripple rejection is the pits compared with the straight 78/79.

Cheers

Phil Hobbs

Not really. I remember it like it was yesterday ;-)

I never cared for the 723. I usually just used an OpAmp... certain vendors 741's had common-mode range including Vcc which allowed for self-starting configurations.

...Jim Thompson

The OnSemi data sheet shows an NPN darlington as the output device, with V- out from the collector. Ditto TI/National, but the stuff upstream is a little different.

Why not a tantalum? Derated 2 or 3:1 on voltage, they are reliable. A

Quote "1 ?F solid tantalum or 10 ?F aluminum electrolytic required for stability". That gives me the goose bumps.

It's also single-sourced in the small SOT223 package. I need small, prefereably less than SOT223.

I could, but then I might as well place a resistor in series with a ceramic and not have tantalum issues. The other concern is that all the bypass caps down the line add up and those must be ceramic. So I'd have to make the output soft by adding series resistance which is kind of yucky.

The 723 was the best linear regulator ever, mainly because of its low noise.

I only have to drop down from an already well-regulated low ripple 12V and -12V. But space is at a premium and regular 78-series stuff is too big.

One tant and a lot of ceramic bypass caps seems to be stable. Tantalum ESR seems to be sort of magic. You'd have to do the whole bode plot thing, or hack a breadboard, to be sure it's stable. I'm thinking that the cap that has the ESR should be a lot more microfarads than the other ones, but that isn't very scientific.

Pulse loading is a good way to evaluate a breadboard, see if it rings.

We use 79L05 and LM337 in SO8, maybe a little smaller. And lots of SOT23 regs, but they can'd dump much heat.

Use the 79L in SOT89 and a small surfmount tantalum. 10u, 16 volts, 6 cents.

I guarantee it!

I'll put in a 1206 footprint ceramic and a series resistor. Just in case.

Yes, pulse loading is my standard test.