Pretty much whole spectrum. Very high frequency can get in and produce audible frequencies by mixing on any non-linear element or by "rectification" in components, such like op-amps, that have different slew rate for positive and negative transitions. |
I'll second Kijanki's response, and add some experimental evidence:
I have a portable battery powered transistor radio which can tune AM signals from 150 KHz (long wave) to 18 MHz (short wave), plus the FM band from 88 to 108 MHz. If I tune it in between stations, so that its AGC goes to max gain, and bring it within about two feet of an i7-950 based computer I built, at moderate settings of the volume control loud noise is induced in the radio throughout that entire 150 KHz to 18 MHz range. There is perceivable but lesser noise on FM, although some or most of that, rather than being at 88 to 108 MHz, MIGHT be lower frequencies that are picked up in the radio at circuit points at or downstream of where the signal is downconverted to FM's 10.7 MHz intermediate frequency.
I should add that the two foot distance applies to the unshielded left side of the computer, nearly all of which is an acrylic window. But the noise can also be picked up on the shielded sides, although I have to bring the radio significantly closer on those sides.
Regards,
-- Al |
Thanks Kijanki. Just how high is high frequency? Sound Application is claiming usefulness bandwidth to over 2 gigahertz (about 2.5) and I would like to relate that to noise generated by digital products, power supplies and transformers. |
Pss, personal computer clocks might be in gigahertz range but computers also contain switching power supplies that operate in hundreds of kilohertz range and buses operating in hundreds of megahertz. In addition when signal is not continuous, but changes pattern (like data on computer bus) additional frequencies, called sidebands, are produced (since it is modulation). All these frequencies interact with each other creating even more frequencies - ending up in one big noise that Al, picked on his radio. I'm not sure what is Sound Application, but as I said before high frequency digital noise can affect audio band many different ways. |
Thanks Kijanki. Just how high is high frequency? Sound Application is claiming usefulness bandwidth to over 2 gigahertz (about 2.5) and I would like to relate that to noise generated by digital products, power supplies and transformers. |
Emitted is the key word here. Emissions are the same as radiated emissions. Most products are limited by Class B to less than 40bDuV/m, at least those that pass Class B testing. This is low enough so you will not have interference between components. Interference of clocks and SMPS supplies inside a component is certainly possible, but usually affecting only jitter in digital systems, not causing errors. Jitter is affected because signal noise levels are increased. Generally in good designs, the problem is more power supply electrical noise, magnetic fields and ground return paths that are more problematic, not emissions.
The thing to understand is that usually the only components that are even sensitive to radiated emissions are tuners, and these can pick up uV of EM radiation.
In typical audio systems, the component-component problem is ground-loops picking up EMI. Eliminate the ground loops and the problem is gone. If the ground-loop is retuned by making it larger diameter or smaller and maybe by twisting the power cords together to minimize the loop area, this can be a fix. Its usually when the ground loop is a size that acts like an antenna that serious problems occur. The area tunes a specific frequency and its harmonics.
Steve N.
Empirical Audio |
Any cable can pick-up electrical noise including speaker cable. Shielding is based on skin effect and does not work for lower frequencies. The key is to keep cables as short as possible, since they become effective receiving antenna at about 1/10 of the wavelength. Twisting wires help at lower frequencies but becomes ineffective for higher frequencies when pitch of twist is comparable to wavelength. Keeping computer away, plugged possibly into different supply phase, using short cables and good shielding is the key. Power supply conditioners can help since overall system noise increases jitter - hence analog noise. |
Thanks Steve. My understanding is that ac powerlines also act as
antennas for radiation being emitted. Perhaps I could ask what
frequency ranges designers commonly try to block going into or out
of a digital player being designed as high quality today? I'm not
looking for trade secrets, just the basics thanks. |
Ptss - Designers usually don't target specific frequencies and the good ones avoid using too many filters. Filters usually impact SQ. Sometimes designers must add ferrite beads or similar to pass FCC class B emissions.
Good design practices are the best way to avoid emissions and susceptibility. These include good ground-plane design, trace topology and impedance control, proper transmission-line termination and internal cabling impedance/termination. Using the correct connectors does not hurt either.
Filters can be useful and will not impact SQ if used on the ground-wires of cables that are differential or the ground-wires of power cords. Filters on power supplies usually impact SQ, even AC lines.
AC lines can radiate emissions if there is a generator present, such as SMPS that puts HF noise back into the power line. Some wallwarts are notorious for this. By themselves the frequencies on the AC service are too low to have much EMI. Magnetic fields are more prevalent with AC lines. |
Thanks Steve, I should have known that; since only a few companies have stressed power supply quality. FWIW, Alpha, Esoteric and Wadia come to mind. Also it may be why I have consistently found that isolation and conditioning have benefited every digital player I have tried in my Spectral system. |
I am not a fan of conditioners, with one exception, the Plasmatron from VHaudio.com. This really improves SQ for digital. I use one and many of my colleagues use one. The Plasmatron is an AC voltage regulator, not a filter.
Filters on AC generally reduce dynamics, so I steer clear.
Isolation is the best solution, both for digital and analog. I use both, so I have no ground-loops.
Steve N.
Empirical Audio |
Steve. I was just reading about the improvements of the Wadia 781 over the 581. Significant p/s improvements. Have you no idea what frequencies of noise are emitted from digital processors? |
