signal leads that pick up less ambient noise?

Wait, I'd need a differential driver too, yes? This signal generator I'm using doesn't have differential output, but the two output channels appear to be phase locked, so I guess I could pull an inverted signal from the second channel.

WWV used to be located in Greenbelt MD. But their neighbor, NASA-GSFC got tired of the 5, 10, 15MHZ, etc. being on every scope. WWV got moved to the wilds outside of Ft. Collins CO.

That shows you haven't got a parallel D.C. earthing path, which rules out one possible noise-injection pathway.

Those tests suggest that there is some degree of capacitive coupling between the sig-gen, the DUT and the 'scope, so that could be how the noise is getting in. As others have mentioned, plugging all three ito adjacent sockets on a distribution strip should reduce this noise path and putting a filter between the strip and the mains supply should reduce it still further.

Some cheap filters, especially those built into socket strips, only have capacitors, with no inductors. These will decrease the noise voltage at the expense of increasing the noise current; depending on how the noise is getting in, that may make things worse, rather than better.

At the low frequencies you are using you don't need BNC or coax. Twisted signal pair inside a grounded at signal source end screen might actually be better for what you want here.

At the low frequencies you are talking about some of the cables intended for broadcast quality microphones have the right sort of screening.

Those boards don't look at all sophisticated. When you say building it on breadboards do you mean Vero board or some other soldered prototyping medium or do you mean literally on a plug in patchboard?

Layout and signal screening will be important if you want to build an analogue computer that will actually behave itself. As a beginner you might be better off buying the kit of bits with the working PCBs.

It is incredibly easy to make small but important mistakes on prototype breadboards and the push fit ones can quickly become a rats nest. You can learn to etch and drill your own PCBs (no solder mask). That is how I got started as a teenager. The chemicals are a bit harder to buy now. OTOH laser printed masks are way quicker than by hand felt pen resist.

Have you checked that your bipolar 15v supply isn't the noise source?

The noise you are describing at around 40kHz looks awfully like an old school switched mode PSU sort of frequency.

But WWV were transmitting at quite a high RF power. Labs can be Faraday screened if you are working on sensitive RF stuff.

VLA is next door to White Sands Missile Base but they are quite good about avoiding the frequencies used for radio astronomy.

The biggest source of interference on site were reflections of distant cities off the underside of passing aircraft. Interference bursts went down after 9/11 and during Covid when there were hardly any flights.

That seems very likely to be from some power supply, either wall wart or built into some equipment.

A great EMI sniffer to sleuth out is an old school $5 pocket AM broadcast radio (the kind slightly larger than a cigarette pack and powered by 2 AA or 9v block battery).

Ah yes thanks for reminding me. I've used that method many times. Waving a problem detector around and then demonstrating that it goes silent when you put it in a closed metal box can be an excellent way of demonstrating that a metal box may be needed in some cases. You don't even need to read any text books by anyone at all.

I'm use the common white breadboards where you push the leads and components in. I'm not sure what the technical term is for those.

I checked just now, but the noise is not affected at all by turn on or off the bipolar 15 volt supply. As mentioned earlier, turning off computers at the desk, especially one older computer, eliminates most of the noise.

I also used those as a teen for very simple things including a cat flap alarm which frightened the neighbours' cats away. For the circuits you're making I'd use veroboard. Sometimes called stripboard.

I wanted to mention that this morning I moved the old PC computer box about ten feet away from my work station, and that reduced the noise down immediately to about 100 mV p-p, from the original 500 mV p-p.

After doing that, I tried running the signal from the signal generator through a simple lowpass RC filter, with R=200Ω and C=1µF. This drops the noise down some more to about 40 mV p-p.

I also moved the signal generator closer and am using some very short leads with alligator clips. This doesn't seem to have much effect on the noise, as far as I tell, but it seemed like a good idea in principle.

I like the idea of putting my analog computer inside a metal box, but I don't currently have a spare metal box, and also that would hide all the switches, trim pots, and terminals. I have an idea though of moving this direction later after building some kind of patch panel. I also wonder if it might be helpful at least to have the bottoms and sides of the analog computer surrounded by metal, i.e., set it inside an open metal box.

I found it essential to have a big die cast metal box with coax connectors and low-pass feed-throughs mounted in the walls when testing sensitive circuits. A desktop faraday cage. Plus lots of coax cables between test equipment and the circuit within the box.

.

Joe

Have a look at "Design Considerations" on this page:

Those plastic breadboard things are awful in many respects, noise being just one.

I make test circuits on double-side FR4, with a solid ground plane on one side, and soldered-on coax connectors. If it's super noise sensitive, it goes in a metal can, Altoids-size or Danish Butter Cookie size.

Could you clarify/expand on what exactly you are using for the low-pass feed-throughs and what that looks like? Are you talking about two separate things on the coax and feed-throughs, or like a coax connector attached to an RC filter?

Google for coaxial feed through capacitors.

.

.

With die cast aluminum boxes, it's best to use a nut to hold the feed through in place, rather than a threaded hole.

One can also mount probe pin jacks and the like in the box walls. There's lots of space.

Joe

The Danish Butter Cookie cans are solderable. The downside is that you have to empty them before you can put electronics inside.

Neither they nor the die-cast boxes shield 60 Hz mag fields very well.

Thank you, I've stored these links and started looking through them.

Question: With a simple C feed through capacitor, with no resistive element, do you have trouble with ringing? I had thought this was a reason to use RC filters rather than just C bypass filter, to reduce/eliminate ringing.

Capacitors and resistors don't ring. Adding inductance can introduce ringing. but enough resistance can make the resonant circuit critically damped and the voltages and current will decay monotonically.

Inductor are added to give you more attentuation but they can ring if the tank circuit created isn't adequately damped by enough resistance around the tank circuit.

One cm of straight wire has an inductance of about 10nH, so bad layout can introduce unintended inductance

So, when you use a coaxial feed through capacitors on your faraday cage, do you add a resistor right after the capacitor, to reduce/eliminate ringing? Or are you just trying that all your inputs on the board have resistors before whatever op amps or other components that they feed into?