Best I have ever touched is a 1930's Autodyne factory special (Golden handshake) based on an AWA R301. First IF was a band pass filter.

Where I work we design and manufacture, amongst other things, LCD and LED signage for transport - trains, stations, trams and buses etc.

These things need to run from a range of DC voltages from 24 to 110 volts. The input power needs to be galvanically isolated from the case and they need to pass all the EMC, EMI, ESD and safety regulations for a world-wide market and exhibit mission-critical reliability.

Ergo they are not cheap.

Keeping the noise from radiating out of the box and down the power and ethernet cables requires great care in the design, given that in addition to the primary power converter, there is the boost converter for the LED backlights and several other smaller buck converters, all cackling away. I guess we have the time and money to spend to get it right, it can indeed be done with care and attention to detail.

Switched mode supplies are interesting, A couple of observations:

1. There is a very commonly used fixed-frequency power converter controller that runs at 630kHz. Explains the interference to ABC Newsradio! It gets totally wiped out in the Mona Vale Road - Pacific Highway underpass!

2. Most power converters run in "continuous mode" when supplying a full load, and use what is called "discontinuous mode" when the demand drops below a certain threshold. In this mode, the converter stops and starts in a pseudo-random manner. Explains the frying noise you are hearing.

3. The 1960's Sunbeam Mixmaster in my kitchen has a universal brush type AC motor. Its speed is governed by a centrifugal weight system that moves a carbon block in series with the AC to the motor, switching it rapidly on and off. Works very well too. A early form of switched mode controller!

On a side matter, switched mode supplies cause another not-often-discussed problem. I'll explain with an actual example:

Imaging you have a power converter supplying 100 watts at a fixed voltage to a load (a LED matrix sign under test with all LEDs on at maximum brightness). It is being supplied from a nominal 24 volts DC.

To keep the maths simple we will ignore efficiency and say that it is drawing 4.167 amps. (It will be around 10% up on this)

Say the supply cable and connectors have a series resistance of 1 ohm. Our supply at the unit's input now drops to 19.83 volts.

But wait! It still has to supply 100 watts! Our current has jumped up to 5.04 amps! The supply is now 18.96 volts!

The converter adjusts itself and pulls 5.27 amps. Volts are down to 18.73!

You get the idea.

Now, consider what happens if the nominal 24 volt supply takes a momentary dive to 15 volts. (Panto bounces off the wire, the bus starter kicks in or a marginal cable or connector join vibrates).

The unit immediately pulls 6.6 amps (without taking the wiring into account).

The voltage at the input to the supply drops to 8.4 volts and the unit now pulls 12 amps (or tries to!). There is a risk the internal "anti-meltdown" 6 amp delay fuse will fail.

Instead, an internal voltage monitor kicks in and shuts the thing down until the voltage returns to about 22 volts. If the cable were twice the length, you can see that the system would hiccup, between the upper and lower monitor points. And quite likely blow its internal fuse, even though there was nothing wrong with the device itself.

The moral of the story? Always take into account the ESR of your power supply and cable when powering devices with switched mode supplies.