LM317L 20% out of spec

On a sunny day (Fri, 05 Oct 2012 14:48:09 -0400) it happened Spehro Pefhany wrote in :

ceramic caps are often very temperature dependent, and also have sometimes high losses compared to poly or whatever else. I avoid ceramic caps everywhere except decouling if I can. AND they are sometimes microphonic to, make sound as well as are sensitive to sound. And tolerances tolerances... Did I mention size? No I did not, so size.

Leaded film caps are good medicine for a lot of low-level things. A bit of board vibration can produce a surprising amount of current from a 10 uF X5R.

Cheers

Phil Hobbs

At my first job we sometimes had to do piezo tests. Every time one of those heavy electric muni buses rumbled along we had to wait until it was far enough down Beethoven Street. After a while we almost knew their schedule by heart.

At my first job we sometimes had to do piezo tests. Every time one of those heavy electric muni buses rumbled along we had to wait until it was far enough down Beethoven Street. After a while we almost knew their schedule by heart.

My tube-style Keithley 405 electrometer has a 300-fA FS range. It takes a couple of hours' warm-up before that scale is usable, but once it's there, it makes a very entertaining demo system for the triboelectric effect in coax. Even with a BNC cap over the end of the cable, bumping it even a little bit pegs the meter.

And that's with the meter properly zeroed, so that shorting the input doesn't move the needle, so it's really triboelectricity and not just V*Delta_C.

Cheers

Phil Hobbs

We had a customer not too long ago make an inquire about a RG-62U product that make once in a while. It seems they had a customer that used it for some sort of microphonic application and the recent stock we shipped them wasn't so microphonic any more... Oh well.

Jamie

pump.)

Mine is Solid State!

I think it uses some old RCA small-signal mosfets in the front end.

It goes down to 1e-14 amps full-scale. Wonderful gadget.

I once built a wideband amp (14dB from a two transistor "double feedback" circuit, the kind apparently popular for this sort of task) with a 10uF

Tim

do

pump.)

It's actually sort of depressing that you can do better than that with an LMC660 and some Russian 10 gigohm glass encapsulated resistors (two for $8 on eBay). None of that sapphire standoff stuff that Keithley used to do.

Cheers

Phil Hobbs

do

precision

pump.)

I wonder if you could run a small-signal depletion mosfet at Vg=Vs=Vd to get zero leakage. Maybe Vd:Vs could be a very small AC voltage and Vg would modulate the channel conductance, with a closed-loop bootstrap thing.

was

wouldn't do

precision

pump.)

well.

bit

Sort of a resistive version of a varactor sampling bridge, as in the Philbrick P2. I suspect that the symmetry of the FET would make the effect quadratic near zero bias, though.

Cheers

Phil Hobbs

pump.)

it's too bad I didn't know you had an interest for G-ohm r's, we tossed a few new, sealed packages not to long ago. These were HV types used in a pre-existing design dynotron supply.

Company likes to do these house cleanings now and then, lots of good nice stuff gets tossed.

Jamie

do

pump.)

No worries, they're $4 each on eBay. If you toss any more of those nice Boonton 72s, though, I'd love a couple more.

Cheers

Phil Hobbs

was

wouldn't do

precision

pump.)

they

caps

well.

bit

their

Or just run the d-s voltage at, say, 0.1 volts. The fet runs in its ohmic region.

do

precision

pump.)

I have some samples of 0805 resistors up to 1T ohms.

was

wouldn't do

precision

pump.)

well.

bit

their

It would be interesting to know what the boundaries on cleanliness and relative humidity are, if one wants to keep, say, 0.5% accuracy through the life of an instrument. It wouldn't take much damp salty flux residue to blow the error budget.

Cheers

Phil Hobbs

was

wouldn't do

precision

pump.)

they

caps

well.

bit

their

takes

I posted an experiment here not too long ago. I made a maximally messy flux-gunked pcb and hung it on my Keithley. It pinned the neeedle on the 1e14 range, for a week or so until I lost interest.

bypass was

wouldn't do

high-Z

precision

charge pump.)

putting a

sense

they

filters

caps

well.

A bit

from a 10

of

it

their

takes

bumping

Plus another $8 to ship them if I located the ones you are indicating:

There are also these at about $3 each:

But you have to want 25. 5% tolerance.

Jon

was

wouldn't do

precision

pump.)

they

caps

well.

bit

their

takes

I recall. On the other hand, it might not have done as well in New Orleans or New York in the summer, especially if some assembly technician liked eating potato chips, or cleaned the board with drugstore rubbing alcohol instead of electronic-grade ethanol or IPA. (*) (Rubbing alcohol is enough to destroy the performance of a low-level board.)

Also flux isn't very stable with time. That's why I was wondering about how it does over the lifetime of the instrument.

Cheers

Phil Hobbs

(*) Isopropyl alcohol, not India Pale Ale. Even if it is Saturday, the sun's not even up yet, so it can't possibly be over the yardarm.