little homebrew LDO regulator

It's only dropping a few tenths of a volt. The 1.5 is current limited to about 2 amps in the LTM8023 switcher upstream.

Do you always include overvoltage (crowbar?) protection on all of your regulators?

John

. . .

It's a good idea actually, and something you don't see on commercial LDOs. Nobody wants to sell an LDO that needs a separate high voltage input! But in fact in many applications there is a higher voltage available.

It's a standard ;-) ...Jim Thompson

And, one day, they come up in the wrong sequence ;-) ...Jim Thompson

-- [On the Road, in New York]

| James E.Thompson, CTO | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona 85048 Skype: Contacts Only | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at

Look at the referenced LTC datasheet.

I'm designing a critical controller for a machine that sells for $120 million per copy. It has analog i/o, picosecond timing, fiberoptic i/o, PCI Express interface, arbitrary waveform generators, 16 fast digital interfaces to other boxes, which we are designing too. There's an Altera 750-ball FPGA and a 270 or something ball ARM processor.

What are you up to lately? A 555 doesn't need a lot of different power supplies.

John

There are a few around, LTC and I think TI. It makes sense, when you need amps at around 1 volt out.

John

Could be a problem with excessive localized power dissipation due to nonuniform current densities from operating that FET at less than fully enhanced channel bias...

Actually, a '555 can be quite a challenge to power. If you want to make an RL oscillator instead of RC, for example, you will need split supplies, and the 'Threshold' pin is basically an internal reference supply, which is recommended to get its own bypass capacitor. The combination of analog and digital functions makes it both sensitive to small fluctuations, and capable of high slew rates (which create lots of transient power current).

Give a student a light bulb and 24VAC: connecting them is a simple exercise. Then, give him a NE555 and 24VAC. Not so simple.

The Zetex dual SO8 is rated for 20 volts Vgs-max. I've tested a range of mosfets for gate blowout voltage and most of them died around 70 or so.

John

The LTC part does need a separate bias supply, described by one silly old fart as "cheating". I am in fact getting the 1.5 from a buck converter from +12, using an LTM8023 micromodule thing. Maybe two in parallel, actually.

I may need an amp or so, so the LTC3025 wouldn't work. There are LDOs that would work, but it was just as easy to make my own dual LDO out of parts we stock. After all, I am a circuit designer.

John

And he going to prove it again, right here.

John

At 0.4 volts drop across the fet, it's practically in its ohmic region. The second-breakdown sort of localized heating seems to happen at higher drain voltages. This fet, half of a Zetex ZXMN6A25DN8, is good for 3 amps at 0.4 volts drop, according to the SOAR curve. But it's a good point that most modern fets lose power dissipation capability as drain voltage goes up.

John

I need to build 1.1 and 1.2 volt supplies, using a dual opamp and a dual mosfet. The case I posted is the 1.1 volt one, with a little step on the reference voltage to evaluate loop dynamics, tune the RC values around the opamp. 1.2 is about the same.

My first guess, 1 nF and 1K, was a little ringy. The 10K has prettier step response, but looks like worse transient load regulation for a big load step. Maybe I should try a faster opamp. Or just swamp the output with a lot of capacitance.

John