Hoorah for the patent system

Consider your brain fooly compensated for lack of such occurrences then. That can add up to quite some resistance... :-]

You won't be laughing so hard when you get sued for infringing that patent.

In 2007 I lost at least 4 jobs as patent expert when I commented that the patent in question wasn't worth the powder to blow it to hell ;-)

...Jim Thompson

It's not quite as simple as you say--they've patented 1. oscillator+buffer, 2. running off one r-c filtered supply, 3. where the resistor's supply is varied to compensate for the voltage drop across the resistor.

I'm too lazy to read the patent, but if #3 entailed varying the supply in a clever way to cancel noise--that _could_ be novel, but I'm skeptical too.

Charles Wenzel's website shows a cool way to reduce power supply noise:

Dunno if anyone bothers compensating for the resistor's d.c. drop though--I don't recall doing it personally. S'pose that drop might add a little noise at the low end, e.g., with changes in the oscillator's supply current.

Cheers, James Arthur

--- Indeed.

Have you noticed that inventions which should be in the public domain because they were developed with public funds are claimed to be proprietary to the government by the government?

Not a big surprise, eh?

NASA's "Tech Briefs" is a prime example, where they'll be willing to let you use stuff that's in the public domain for a price.

-- JF

Why not just bypass the 317's lower voltage-set resistor? That will have about the same effect.

John

If you want to add more parts, you can get it even cleaner. Basically it is like this:

----- ---------! 317 !--------+--------------------------------+---- ----- ! ! ! \\ -----/\\/\\--- ! ! / ! ! ! ! \\ ! /-!-------+--/\\/\\----+ +-----------+----!!--+-< ! ! ! C1 \\+/-----+-------+---- \\ ! ! / --- \\ \\ --- / ! ! \\ GND GND ! GND

Some care is needed to make it stable. C1 ends up with a small resistor in series. Over the band where it works, the output noise ends up less than the reference of the LM317 would otherwise give.

A friend of mine once suggested that he should patent the business technique of patenting existing technologies, then go after all of the big tech companies...

Really? I don't see it. (caveat: I was up all night working on some cool ideas. (That happens occasionally.))

a) ISTM bypassing the LM317's lower voltage-set resistor *does* cut the '317's noise gain but up to the divider ratio, but can't reduce output noise to less than the noise of the '317 itself.

b) Wenzel's shunt method is limited, 1rst-order, to the noise of a single bipolar transistor. I assume without even checking that a

(...Wenzel says later on that page he measured a 3-terminal regulator's noise as 330nV/sqr(Hz) @ 100 Hz.)

c) Wouldn't that bypassing destabilize the '317?

Foggy brains want to know...

Cheers, James Arthur

Hi, sleepyhead!

Just hang an aluminum cap across the lower resistor. That zaps the noise multiplication effect of the resistor network. At frequencies where the cap is effective, the noise is the same as if the 317 adj pin was grounded, namely the same as a 317-based 1.25 volt regulator. That's 1/8 as much noise as an LM317 10-volt regulator. Maybe less. That gives you a bit of soft-start for free, too.

Then hang another big cap across the output.

Or for really low noise, design a proper regulator circuit.

John

Or use a uA723 ... :-)

That means that particular claim and all dependent claims are

Moral of that is to never contradict (or seem to contradict) an employer. Let them waste their money on garbage patents. At worst, say "in my opinion, there might be a small problem in claim N; if you wish details, talk to Mr. ". Avoid telling them something they do not want to know, unless you can collect UP FRONT a very hefty fee.

Hmmm...reminds me of the fact that the patent(s) on RADAR were held SECRET for many years...

??? What is that thingis that the cap, etc are connected to?

If one has enough money, one could get away with that - at least for a while... BTW, toss in a patent for the paper clip...

Yeah, and then there was the diamond growth technique developed at NRL which GE claimed (I guess successfully, never having followed up that bit of bogosity) to have a patent on (but it should be just about expired - and DeBeers won't be happy - toss some methane and hydrogen through a welding torch at a temperature controlled target and grow a flawless diamond the size of your thumb - or even longer, if you like - single crystal).

Has all kinds of interesting engineering material applications when you get over the artificial prices the jewelry folks like to charge for charcoal that happens to have a nice crystal structure.

I don't recall exactly when I went to a talk about it, but it would have had to have been no later than 1994, so that patent should be run out in

Charcoal and coal are two entirely different things.

Diamonds do not come from charcoal.

That does look interesting. I wonder if it can make the LDO unstable.

For on-chip power supply isolation. I like to use a shunt regulator. The shunt regulator is paralleled (doh!) with the circuit you want to isolate. A current source from the top rail feeds the parallel combination. This of course is wasteful of power, but it does work well. I've used the scheme for internal voltage references, i.e. run the voltage output reference from a cell that is shunt regulated. It is as good as bootstrapped reference designs.

Makes perfect sense after a nap, & realizing/remembering the '317 senses across the upper resistor, not lower. A cap across that lower resistor gives an easy gain for almost no trouble. Agreed.

Wenzel's gizmo could still best that noise level, cleaning up most any commodity regulator's noise, or...

Indeed!

Cheers, James Arthur