A very silly oscillator

Sounds like ionic contamination. Did you use a water-soluble flux? If so, it will be hard to get the crud out from under the parts. Try a water-pic pressure toothbrush and deionized water, and really blast the chips from the sides, to flush underneath. That has saved us a couple of times.

Rosin flux, solvent clean, bake, conformal coat... seems to work pretty well.

John

That's more or less what I would assume. Have never built an electrometer amplifier, but remember the glass encapsulated resistors, ptfe standoffs and clear instructions not to touch the glass in some old ph meters. If you read some of the Fluke calibrator and kelvin varley divider manuals and look inside the box, there are the same sort of warnings. These are much lower resistance and they are on glass pcb's with pads > 1" apart.

Just an area I would take precautions if I were designing such kit. They don't do this sort of thing in commercial kit unless it's needed. It adds too much to the cost otherwise...

Regards,

Chris

I have an axial, epoxy-coated 1T resistor screwed into a dual Pomona banana plug. If I plug it into my Keithley electrometer, it reads just about right, after being handled like any other gadget.

John

cb

he

pA

I've used the water soluble flux, but only on through hole 10G resistors. They seem to work just fine with no special care.

I like rosin flux. Seems to cause the least amount of leakage, especially on those proto pcb's that get a lot of re-work.

What's a conformal coat?

George H.

..and stay away from the so-called "no clean" fluxes..

Raincoat for a populated board. A layer of poly (or other plastic) over the whole thing.

-- Les Cargill

Right, but we both know that would be unnacceptable for anything that was to be built into a product :-). Does the resistance rise if you clean the resistor and pomona with alcohol ?.

Back to the Keithley, I bet that thing is within a few percent of cal, even after 30 years and it's probably valved as well. There was something about they way some test gear was built then. It was built to be accurate and stay that way.

I'm looking for one of the Fluke hv 2Kv supplies, the one with the decade switches on the front panel. They were hybrid afaik, with a valve in the series regulator and solid state elsewhere. There are still making good money though, even on Ebay...

Regards,

Chris

Don't know... I'll check.

This Keithley

is all solid-state. The manual says to replace the high-val resistors every six months (!) but it seems fine, within a few per cent of a recently purchased 1T resistor. Cost $180 on ebay.

Yeah, there was an 8xxx series tube that all the Fluke and Kepco HV supplies seemed to use.

John

A layer of plastic stuff. Polyurethane varnish works well, or you can buy professional stuff. You can dip or spray or brush the whole board, or just brush the sensitive parts.

John

pcb

the

nd

I will second that. "no clean" fluxes are not suited for boards doing any kind of analog, low power, high speed, DC power, digital, RF or MEMs work. Other than that they are just fine.

Nice looking bit of kit. What tolerance is the resistor ?. My guess would be that the meter will have the last word, but may be wrong.

That sounds familiar and I may still have a couple of those. I'm pretty sure that was the type, well, two in parallel in fact, that was used on one of the first hp spectrum analysers as a voltage sweep output stage for the bwo, Had two of those over the years. The first one had a bwo that dropped out at points in the range, but turning up the heater regulator by just under 100mV fixed it. The second one was beyond repair and got broken up for spares. I remember feeling guilty at the time :-)...

Regards,

Chris