While hum can be delightfully entertaining, I did note on a Leem amp 100W RMS that if you were stupid enough to run wires across the back of it, you got reminded the its PSU was switch mode.

The only way of getting rid of heater cathode leakage noise cheaply is run the heaters on DC and some tube amps do that. That also allows caps to be used to discourage nasties from riding on the rail. The twisted pair is not dead.

I have noted with some of the quieter radios that they also use a choke in both Negative & Positive of "B". On a couple of occasions, I have noted that a small transformer with the other winding shorted makes a choke, cost is often the killer with using wound chokes albeit they are superior. The electrodynamic speaker hum buck coil, works along that line.

Parallel wires carrying any form of AC can induct in to the other one, and that's normally the cause of CFL's switched of blinking. A lot of them are noisy and I sent a couple of LED spotlights back to point of sale for being like a symphony orchestra. On the odd occasion I have seen wires run in metal tubing.

Do note also that in some HMV sets and old and new. certain bits, or stages were in a Faraday cage. Some disk caps are very close to Microphones.

Marc

Marc, you have to know how to design and use switch mode power supplies to avoid these pitfalls.
When you do, they have many advantages.

As I indicated in my post, the heaters ARE running on DC, unlike the design on which this preamp was based. What caught me out was just how sensitive a vacuum toob is to a relatively benign magnetic field.

An ordinary copper and iron transformer feeding a bridge and a capacitor input filter can actually generate more rubbish down the mains than you would believe possible. Harmonics of 50Hz right up into the MHz range.

Don't believe me? Try this test:

Get a transformer capable of supplying 5 amps or so at 24 or 32 volts or thereabouts.
Connect it up to a bridge. One of those 25 amp chassis cooled types will do. (It does need to be silicon, not selenium)
Connect the output of the bridge to a capacitor of, say, 4700μF. Suitable voltage of course.
You will have a DC voltage roughly equal to the RMS value you put in, multiplied by 1.414.
Apply a few amps of load - a couple of old-school car headlamp bulbs that add up to the voltage will do.

Now, switch on an AM radio in the vicinity. See if you can pick up anything but buzzing noises!

Disconnect the load, the noise goes away.

Bad design can take many forms. See if you can tell me how you might quieten the power supply I just described.

The input bridge and capacitor is one of the main reasons a badly-designed SMPS is noisy, and why they have a reputation as such. The other big cause is a badly-designed or non-existent snubber on the active switching device.

For anyone interested in such things, LTSpice is a great free tool you can use to experiment with analogue circuits and power supplies without actually having to physically build your circuit. You can draw circuits and measure all the waveforms, voltages and currents using quite accurate mathematical models of a wide range of parts. Even including valves! It's great for optimising SMPS snubbers.

Well worth spending some time to learn it. There is an active user forum too. Some of the recent engineering graduates where I work tell me they used it at uni. as an aid for understanding things like FETs and filters.

I must be lucky; I have only had issues with modern switch mode power supplies and rarely the old conventional ones. My biggest issue is the RF riding on the mains from the long EHT runs. Built a few mains based supplies for battery sets over the decades without hum issues, other than heater cathode leakage.

One Fax machine with switch mode was doing a good job of getting into a 175kHz IF and locally Telstra traced an internet issue to on in a fridge.

I put the little SMPS module in. Hum is now gone.

I now have about 3mV p-p of random noise with a noticeable low frequency 10 - 20 Hz component. The two channels don't correspond at all so I don't think it's power supply related. That's a pretty fair result given the inputs are not connected to anything apart from their internal 47k.

There is a little random chirp of 3 or 4 cycles of 132kHz that peaks at about 5mV that happens between 10 and 50mS apart. It's probably coming from my simple little boost converter. I'm not sure I'll worry about it, it's very unlikely to be audible.

Funny thing, though, there is now 31mV p-p of 100Hz ripple on the 24 volts heater supply. I had replaced the two 4700μF caps in the passive filter with 100μF on the assumption that the filtering wouldn't be needed. It doesn't show up in the outputs though. In my experience it's not unusual for "plugpack" style converters like this one.

Now that we are working down in the realm of inherent valve noise, it has been interesting to see the difference between 12AX7s and 12AU7s, The ones we have to test are mainly used Aussie Philips and Mullard from the 1960s. They are quite consistent within type. The 12AX7s were far more sensitive to magnetic field hum pickup. The gain of the preamp is very close with either valve type due no doubt to the negative feedback. But the 12AU7s are quite microphonic, particularly an AWV one, which picks up my voice quite effectively!

Ah, analogue and valves! What fun!

I left it to soak overnight and measured the temp rise of the power supply components with a laser probe. I usually do this with a new design as a product life prediction exercise. The hottest part was the transformer core which went to 15 degrees above ambient. In this application it's loafing along.

That "little random chirp"?

Found it!

It's not in my box at all, it's coming from the LED light fitting over the table!

I turned the light off, the spiky bits vanished.

The light fitting stays though. Remember the light ring above the round table in the War Room in Dr. Strangelove? It looks like that!

So my vintage replica preamp is ready for production.

Interesting with the RFI from the rectifier. I may actually be killing it as there has always been RF riding on the miles of powerlines here and the number of radio stations contributing. That has for years, required steps to be taken to be rid of as much of it as possible.

The other issue on the Mains side is lightning & surges. It is now interesting as there are a number of LED lights replacing the inferior fluorescent types with electronic ballast's that fail constantly. House is 12 years old & 2 little used ones were the only survivors. The LED ones will work on 110V and continue working when the supply drops a phase.

The filtering has evolved over a number of years. First there were caps & some commercial inductive filters, then as the transistor era came, MOV's for spikes then duplicate by others now; Two MOV's for the benefit of spikes and RCD's (Earth leakage).

So Its now the triangle of caps for RF and they're also fitted to the inductive type on the TV and one MOV across the mains and the other Active to ground, to cause the RCD to trip: Totally killing the circuit. This does work.

A solar control box faulted & was throwing out surges & severe spikes (arcing). The unit on the TV killed the circuit and nothing on that circuit got spiked. Another also tripped the computer & peripheral circuit on a different sub.

There is a poorly advertised phenomena called "Silicon Ringing" that is why you see caps on some rectifier diodes.

Marc

Perhaps of interest I have, as mentioned, a 1935 Philco / Chrysler car radio in the line. It has some axial caps through the chassis in metal tubes and grounded topside. That is fascinating.