grounding lugs advice

[R.G.]

[...] Don't the signal grounds, e.g. 'ground' side of the cathode resistor/cap for the preamp tubes, need to be tied to the chassis someplace? I got as far as sort of understanding that the cathode resistor/cap should be tied to the return side of the power supply cap that feeds that tube. But what happens from there? Isn't there supposed to be a wire that goes from that junction to the chassis?

Grounding is a very slippery subject. There is only one One True Ground, and yes, it is the planet itself; in fact, not every place on the planet is a good ground. For fun, you might want to look up the grounding arrangements for large radio transmitting antennas. We're not going to be dealing with the OTG in amps, so we have to look elsewhere. It gets really tricky to say what "ground" means in a world where we don't use OTG, but instead "the point we have decided is the reference point for voltage measurements".

Yes, the cathode resistor/caps of preamp tubes, and eventually the output tubes as well, must connect to the chassis someplace to get quiet operation. Exactly how that happens is an exercise in Ohm's Law. We want the signal ground to be at the same voltage everywhere. If it's not, then "ground" here amounts to being a signal at someplace else, different from the remote circuit's "ground". That difference gets amplified.

The trick to making some electrically remote place have a zero voltage difference between it and your local "ground" is to make the current between them be zero. The simplest way to do this is to make sure that one and only one wire connects them. That makes it impossible for current to flow, so the voltage difference must be zero, no matter what the resistance of the connections is.

So we want the chassis connected to signal ground. We want the circuit's cathodes, cathode resistors, cathode caps, bias resistors, etc to be connected to signal ground. We want the power stage to be connected to signal ground, and often we want the output voltage to be connected to signal ground. Same for power supply ground. ACK!!

This can all actually be done. It's called star grounding. You pick one place that you define as the One True Ground, and connect all other grounds to that one by one and only one wire each. The wires may each have a current running down them, making each remote point be different from the One True Ground, but the fact that these are offset from one another doesn't make any difference to the circuits as the offsets from one ground to another can never be supplied to another stage for amplification.

In this sense, you want the chassis connected to your signal ground (and by extension to your One True Ground) by one and only one wire. Since no return path exists for current, no current can flow between chassis and signal ground, so the chassis IS at signal ground. If the chassis forms a Faraday Cage shield, incoming radiated EMI can cause circulating currents on the chassis itself, causing local spots on the chassis to be different from signal ground, but those differences never get transmitted to the circuits to be amplified.

So yes, you want signal ground connected to chassis. But only once.

[trobbins]

To alleviate long term corrosion in adverse locations (eg. near sea-side or in a chemical plant), it is common practice to apply a light smear of vaseline, or a commercial interface compound, to slow down long term air and moisture ingress to the metal interfaces in a bolted connection. The tight clamping pressure and toothed washers provide a stable low resistance connection, where the vaseline only acts to fill any minute voids that can show up as rust spots over years.

If you were somewhat paranoid, you could use brass hardware, or a conformal coating over the top, but in reality such a joint does not have self-heating or micro-arcing from high currents, and if safety was really a concern then you would be doing a PAT (portable appliance) type test periodically.

You can set up your own test (ie. pass 1-10A DC through the joint and try and measure the uV/mV drop), and yes most compliance checks require an earth continuity test with resistance no more than 0.1 ohm, but using a commercial tester with 10A operation can be a pain if it uses kelvin terminals.