Drift mechanisms

On 08/11/2014 08:11, Tim Williams wrote:> What causes this? Is it cosmic rays? Can that induce enough charge > (~10pC?) to account for the observation? Is it a material property of the > capacitor or transistors?

Just guessing, could it be extreme 1/f transistor noise?

piglet

On second thoughts, I now suspect trapped charge in the 2n7002 gate oxide?

piglet

You can't just arbitrarily do this by "...cutting the op-amp bias..."

An example of how to do it properly is shown in my Patent 3,643,110 Sample and Hold Circuit... on the home page of my website.

This buffer not only has bias "cut", but reverse biases and holds-in-place any junction that might go wandering off. ...Jim Thompson

Could be air ionization. Pease has a column on that someplace.

Cheers

Phil Hobbs

Might be popcorn noise in the 2N7002. Or temperature changes ditto.

The storage cap is referenced to a large fraction of the 6 volt power supply, so power supply changes are coupled almost 100% into the output.

Why not use a cmos opamp?

Why not use just the pushbutton and a big cap? They make pretty good switches.

Is everything clean? Wash with 99% IPA and oven dry. It's amazing how conductive fingerprints are.

On Sat, 08 Nov 2014 08:23:47 -0800, John Larkin Gave us:

I like this supercap fire-and-recharge and wait for the next pulse idea. I guess cycle rate would be a factor, unless one could keep the cap pumped at all times. Then, its a power supply all over again.

They likely only need to strobe anyway.

If they are to be on in a 100% duty cycle, then their lifespan will likely suffer, and this element of the system will become a servicability/longevity issue.

We used to count pills similarly, but a parallax error could occur if two fell through the channel together. On a 16 channel feed cone, it would typically not express a single error in 1000 counts. The pill size mattered in that too. There were no spectral considerations in that however. Simple optical chop counts as the pills obscured the light from the sense elements.

Ooops... I thought this was the egg thread...

On a sunny day (Sat, 8 Nov 2014 13:07:32 -0500) it happened "Tom Miller" wrote in :

Some TV manufacturer used MOSFETs and I think it was poly capacitors to hold the volume and picture controls. There were up/down switches for each control that charged or discharged the cap from a high voltage via very high resistors with a neon is series, The set would hold the values for weeks if not longer. The MOSFET, cap, and neon was in a plastic module, clean, sealed + . . . . . . . . .|. . . . | .

-HV --R--0 down . neon | |--- . . === |--------> control . | C [ ] R . . | | . . -------| . . . . . . . . . .|. . . module | ///

It is probably patented, IIRC the cap was that MKH type? Not sure,

Not mine.

That's cool. Someone in this ng, some while ago, measured film cap leakages over a period of years. It was approximately zero, as I recall.

On Sat, 8 Nov 2014 13:07:32 -0500, "Tom Miller" Gave us:

In HV supply mfg, we had a saying...

"You can kill that supply with mere "coffee breath", be careful during assembly..." It was true too, since the potting would trap the failure mechanism in place. Assembler introduced finished product level infant mortality.

Like ESD events... after the damage occurs, the actual failure event will be down the road a bit in many cases. In some, it is immediate.

Brominated solvents rinse very clean as well, but cost more than IPA.

or you will soften your polymer based parts, and other problems can be caused with certain discreet components.

It's off scale?

Yup, on the 1e14 ohm range.

That little pcb has gloppy, uncleaned rosin flux soldering and all the fingerprints I could reasonably inflict.

"Jim Thompson" wrote in message news: snipped-for-privacy@4ax.com...

Uh... T1 base current?

T1 Vbe and D Vf make T2 Vbe=0, which isn't any different from what my circuit does; mine just does it in a more crappy way (pull-up/down resistors, which will lead to some offset, which is fine in this situation). It can hardly be said to "reverse bias" the junction(s).

So, yes, you're just "cutting op-amp bias", contradicting your first sentence.

I was specifically talking about long time constants. While this will do fine for the A-to-D it was purposed for, it's not suitable here.

Tim

I remember reading that one, but we're talking extremely small bits of charge, no? How could that affect a 0.01uF cap?

Tim

~45 years ago. I can do that in CMOS (or BiCMOS) now.

...Jim Thompson

That could be. CMOS gates are notoriously bad at this level, at least the ones not controlled for it? There are probably much better choices than

The apparent surge behavior is much too fast for thermal, but there are air currents around which definitely contribute to overall 1/f noise.

The 6V in turn comes from a switching power supply, which is regulated with a TL431, another somewhat noisy device. Though I don't recall that TL431 is exactly noisy to the level of popcorn, just that it has normal bipolar 1/f behavior.

Do you know of one with a bias input? ...A CMOS OTA? Also needs 30V supply range.

Or, I'd use LM13700, except it's almost not worth it given the darlington output stage; it's probably relatively leaky too, as there are no internal B-E resistors to keep it cut off.

On this (hobby) project, I'd just as well keep it discrete... so I did. (It's up to around 50 transistors total, not including what's inside the aforementioned TL431.)

That's quite true...

Tim

There, just washed it down with acetone. Umm, carefully to keep that away from the polystyrene cap, that is.

Construction is dead bug, using chits of PCB for soldered terminals. The cut FR-4 edges should be fine insulators, though with rosin caked around most things, maybe it will leak a bit.

Would rosin account for the apparent surges in behavior?

I don't think I've ever noticed that my breath or fingerprints are at all conductive (or corrosive), but that is definitely a problem that varies from person to person. Some do, some don't.

Tim

Tim - capacitors suffer all sorts of ailments, or features perhaps. The following is offered as possibly useful information. Capacitors can resume a charge that was previously on them. Working on a sample & hold device using polyester caps, a device holding zero volts wouls slowly drift back up toward the previous voltage. A switch to polypropelene units held that within limits. Another example, with a cap used on a power supply of several hundred volts was connected to capacitence measuring circuity. I was careful to short the device befoe hand. Half an hour latter, for some reason, I checked it again, but neglected shorting the cap. Pop! dead multimeter.

Hul

Tim Williams wrote: