signal leads that pick up less ambient noise?

I assume that your "leads" are a coax.

How are you measuring that noise? It's more likely to be ground loop noise than coax shield leakage.

Common-mode chokes, ferrites or toroids, can help. Just plugging the generator and the scope into the same outlet may help.

Frequency??

Without more information it's hard to say. At one extreme you might need different cables or different grounding. At the other you might need an electromagnetic screen room (Faraday cage).

That would be the direction I would look first. Remember the signal path is a loop, so if the earth return is not through the co-ax screen, you may pick up noise on the earth wire if it follows a different path. Even if you have the earth solidly joined along the co-ax screen, interference can be introduced if it is paralleled by another return path that picks up noise.

In the 'old days' , lifting the safety earth lead off the mains plug of one of the instruments was a way of cutting the parallel path, but that is frowned upon (and probably illegal) these days. Doing that meant that when the signal earth was disconnected, a dangerous voltage could exist between the two instrumnets because one of them wasn't properly earthed. With most modern instruments, which use double-insulated construction, there is no need for a safety earth and consequently no earth lead to lift.

What sort of noise and at what sorts of frequency?

Coax should be a lot better than that unless you have large fluctuating magnetic fields about and a closed loop of some sort.

How are you measuring this noise?

Yes, the loads something my co-worker threw together from old cable laying around. One one side is BNC that plugs into the signal generator. After than is a thin coax about 3mm thick. At the other end, he split the inner conductor and the shield into two leads, one with a pin at the end, and one with an alligator lead.

I have a project I'm doing building my own analog computer — which currently does not have any filtering installed on it, other than some

If I turn the computers off around my workbench, the noise becomes less, proportional to the percentage of computers I turn off. If I take the signal generator and the computer into another room with no computers, the noise almost vanishes.

I am planning to go that route, with the chokes and such, once some parts come in. But I was also wondering about the single leads themselves, which feed from the signal generator into analog computer inputs.

I'm not sure exactly, but when I zoom in to it on the scope, the noise appeared to have a regular ebb and flow of about 40 Khz.

Looking at in on the scope, the noise is thick, like a solid bar of spikes around the real signal. When I zoom in on it, it seems like it has a beat of around 40 Khz.

The way my coworker constructed it, the signal lead gets exposed only about an inch — ending at a pin — while the gnd lead gets exposed about

Sigh. What is the frequency of the *desired* signal?

Take your probe and pass it around your computers, see which item is causing the interference. It's best to find the source than to try to shield a circuit. For giggles you can swipe the probe around you scope LCD display and see the back light inverter noise.

Cheers

That sounds like the usual setup and doesn't normally cause any trouble, (although the lead to the croc-clip, described in another post as 8", would be a bit long if you were working at R.F.). Have you checked the continuity of the earth connection from sig-gen earth to croc-clip with an ohm-meter? (The earth leads sometimes fracture under the sleeving where they enter the back of the croc-clip)

The next test is to unclip the croc-clip and measure the resistance from the chassis of the device-on-test (or the 'scope) to the sig-gen earth. If it is low, that means there is an alternative D,C. signal path in parallel with the screen of the co-ax, which could be picking up interference.

If the D.C. test appears satisfactory, there may be an A.C. path through mains suppression capacitors, which passes the interference but doesn't showwith the D.C. of a continuity test. Does the signal drop and the interference get worse when the croc-clip is unclipped? If the signal doesn't drop, you have an A.C. path in parallel with the earth.

Do similar checks on the 'scope earth lead. .

That could be magnetic coupled from a nearby switched mode PSU.

Does it look like a PWM square wave or a sine wave?

I only ever see ripple if I stick my finger on an probe and my environment is really rather electrically noisy.

You haven't got one of those ultrasonic humidifiers nearby?

What happens if you put a 0.1uF capacitor in series with the coax shield coax so that there can be no DC current flow on the earth connection?

You can see the ground lead contribution by shortin the probe tip to the grounding lead: common mode is visible with shortest gnd lead and probe tip connected to same zero-voltage point of DUT, when ON/OFF, with sig gen connected/not connected, With sig gen ON/OFF.

This can be repeated iteratively, as the larger contributors are identified.

Commercial scope probes are typically supplied with a probe tip clip that allows probe grounding near the measurement point with minimal physical loop.

Don't forget other sources of noise - ccfl light bulbs and ballasts - printers, modems, smart phones, AC motors and fans and simple mains wiring(re-radiatioon).

Don't forget possible scope contribution - it is also following your measurement setup. There are some self-noisy scopes out there. Battery-powered scopes offer some interesting noise test setup options, in extreme situations.

Sig Gen output in coax as well. Same or similar precautions. Try different sig gen at some time.

Simplest interconnection and most familiar test environment is best.

I also recommend Morrison's book. Earlier editions prefered.

RL

Are you sure that it isn't a part of your analogue computer squegging?

That isn't a good sign. Scope leads don't usually let a lot of interference in. Try shortening the length on the earthing clip.

If you make your signals balanced differential drive then you can withstand a lot of ambient common mode noise without problems. The way that instrumentation amplifiers do it.

I think there is something funny going on with your setup.

See my other post in reply to Bill Sloman. If the noise is not being generated by the analog computer but is causing issues for it then a metal box and some feedthrough capacitors might be the solution. This assumes that your analog computer will fit in a sensibly sized metal box, doesn't have so many connections to make it impractical, and is a one-time design rather than volume production.

Checking just now, continuity from sig-gen earth to croc-clip appears to be good.

I'm a little fuzzy on how to do this test, as the DUT does not have a chassis (all on breadboards) and the scope is fully encased in plastic, except for the probe ports and such.

But if I connect just the signal pin to the DUT, leave the alligator clip disconnected, and then measure from DUT gnd bar to sig-gen earth, it is infinite resistance.

For this test, I put the signal generator leads in as input to DUT and scope as output from DUT. When I remove the crop clip from DUT gnd, the sinusoidal signal is still visible on the scope at the same amplitude, but the noise surrounding it increases in amplitude significantly.

Once again, the sine signal is still visible on the scope at the same amplitude, but the noise surrounding it increases — even more dramatically.

The analog computer I'm building is usually showing me simulations with oscillations usually not more than a few hertz. Input from the sig gen, when involved, is usually like 0.5 hertz.

Sine.

No. I know though that a lot of the noise is coming from one of the nearby older computers, because if I shut that computer off, the noise drops dramatically.

This does not appear to have any effect on the noise.