Drift mechanisms

How about this?

When the pushbutton is open, there's no popcorn noise!

The PLL itself.

Just checked, the varactor is much too leaky for that. It has a time constant under a second, and that's with some RC lowpasses already attached.

Actually, I think I didn't even use a varactor proper -- it doesn't need much pull or linearity in this application, and that saves having to go out and buy some oddball component.

Tim

I'm aware of dielectric absorption, but it's not a big deal in this application: the sampling is the same every time, so after a few shots, it should "learn" to go away, more or less. It's a good catch in other applications, though!

Tim

-- Seven Transistor Labs Electrical Engineering Consultation Website:

Yes, two oscillators (one under control), one mixer. The mixer is DC coupled, so I can do handy things like PLL from it. Which has the curious property that, as a control loop, it doesn't care which polarity the error amplifier (this circuit in question) is wired in.

More background here:

Tim

• Talk to Einstein about that..

• Yes, cosmic rays might be the source. Take the instrument to another place, 30 or more miles away and look for similar long-term "average" and similar "pops".

• Do not know about the theory,but ages ago Intel had this PMOS RAM that used unusual voltages like -5,+12,+7 or some messy combination of supplies. In any event, one could write a pattern in some of them, unplug the RAM, and come back in a day or two and plug it back in and reliably read that pattern. Stored charge in the oxide (glass). I expect that air, a better insulator, may be worse because it can more easily be ionized by cosmics. A FET with a non-protected gate can store a charge for days. Fairchild data sheet makes no specific mention either way about the gate. Expect ACTUAL max leakage to be picoamps, not microamps. Do a test. Make a linear amp, meter on drain, charge on gate for some nominal current, and watch the meter - paint drying is more exciting.

Yes, an excellent article on S&H, leakages, how-to-do.

Not any off-the shelf switch; a DIY in-the-air switch. Dig up that Pease article..

A 2N7000 with a floating gate can hang partially on for days. Gate leakage must be not very many electrons per second. Here's a 2N7000 driving an LED partly on, floating gate:

You can take a small insulated screwdriver and tap gate/V+/gnd and transfer charge up/down onto the gate in tiny steps. Short drain to gate briefly to establish the partly-on state

Win Hill did this, for drain leakage:

I wonder what negative gate voltage might do.