Bad resistors

Read the Agilent and Keithley specs and then you can start weaseling about the dictionary definition of "accurately."

I bet NIST measures gigohms to parts-per-million. I can do 0.1% with my home-brew rig.

The old tube GR megohm meter used a mechanical meter movement, and reference resistors weren't very good back in the 1950s, so they were probably in the 10% sort of range. Look up some old manuals if you're seriously interested.

You're probably not, for fear of actually learning something.

John

If they have to do a lot run simply because one order triggered it, they lose money if it isn't a pretty big order, and SMD has expiration dates due to termination oxidation issues so they cannot simply store the over-production stock either. Though I will take old parts because they are usable for proto builds.

One day in the not-so-far future you'll be able to buy thermal imagers at Home Depot for under $250, eventually under $100.

John

It was not an insult. Your question above, and your inability to find the answer yourself, proves to everyone just where your head is at, boy.

Sorry to be the bearer of the facts, Jack.

It usually is 'one over f' noise.

Right. I don't know if it was 1/f or wideband shot noise, or maybe both, but they were a lot noisier than the metal films, which seemed to be at the Johnson noise level, independent of bias.

Capacitance limited our home-made test rig, so the noise spectrum was hard to decode. The bottom like is that the thickfilms were noisy and the Dales weren't.

I was just thinking, in the shower, about developing a cool high-ohms resistance and resistor noise measurement box, USB interfaced. That would be fun. I probably won't do it, because it would be a fair amount of work with an unknown market.

John

You can't just solder it in, and not clean the board completely and start relying on test data. It being clean is the whole key. You can't do that at the bench without a brominated solvent or hot alcohol and a heat gun... hell... not at all. It needs to be clean.

Working with HV and with feedback loops, etc. we established a method of building the SMD circuit with the least leakage paths. It MUST be CLEAN.

'Bump soldered' It allows your precious degreaser to get all the way under the parts, and then encapsulation seals the path, which was how we finished off our HV sections. But the gap is the point being made.

part> ___ bump>___n_n______

No, I just wash my hands more often than you do.

I deliberately fingerprinted a Pomona dual banana plug that had a 1G resistor attached. I couldn't see the measured resistance change.

Try it.

John

As I said, they are axial RC07 types, not SMD.

They do stock 22M, available from distributors at a more sensible price, and 33M, also distributor stock maybe. The 50M ones are apparently not much in demand.

All *are* 1% parts. And we can measure them accurately.

I need a small constant current, and it's a battle between Johnson noise current and shot/excess noise. It's not easy to generate a nA level quiet constant current. The alternative was ten 10M surface-mount thinfilms in series, which we don't have room for.

John

Stop being silly. There is nothing special about 100 megohms.

John

Just checking :-)

Did you contact the guy from Voti? I don't know him but from what I've heard he is not into selling junk.

"crasic" "Phil Allison"

The bands can change color, especially with age or heat.

Or it could have been a batch of mismarked resistors that the equipment maker got for cheap

The maker musta saved about 5 cents ( or a few UK P) on a $1000 product.

... Phil

You do not understand. I am not talking about the equipment.

I would bet you the value of your entire firm that what they put their name on was an OEM device with their skin thrown on. Guaranteed. Wanna bet?

The fingerprints go on the resistor, not the banana plug.

And a 1G device as the dropping resistor for the feedback for an HV circuit will read fine on a meter, but in use, it will fail if it has those prints, or 'coffee breath' or other VOC across it.

What the meter reads may be fine for an LV application, but that 'dirt' is where part of your noise is coming from, and in HV, it makes for inaccuracy, and complete failure modes. Failure, especially once in the hands of the customer in a potted product, is not an option.

You're an idiot.

You left out one over f.

Bwuahahahahahaaha!

Not on any of the brands real men buy!