There are linearity issues in the cores, and some of the issues aren't monotonic, but the issue is so complicated that few folks even agree. I think cores can be made where the zero-crossing tilt of permeability covers the whole B-H curve working range, or close enough to be nominally monotonic.
All good fortune,
Chris
A simple light dimmer uses no or little filtering so the rise time has few limits. As an instantaneous rise time would have infinite harmonics this is limited just a bit by real circuit limits.
In professional dimmers some filtering is used. Not to limit EMI but to keep the lamp filaments from making noise known as singing.
The problem in pro use is that sound and lights are often part of the same production and lighting cables can get too close to the audio cables.
In semi pro productions cheaper dimmers are often used and cables may actually be run close to each other.
Star quad cables were intended for long microphone level runs in places like theaters where long cable runs of low level signals and light dimmers are used in a low noise location.
Last edited:
Back about the proverbial 40 years ago, the first home light dimmers I'd seen (here in the sticks) could make an audible buzz in Klipsch LaScala speakers (very sensitive, but still) with no external wiring connected. Yikes.
Much thanks, as always,
Chris
As a fun test, I have a modified GSM cellular device with debug firmware on it which can force transmit continuously at 2W in the 850/900 MHz bands and 1W in the 1800/1900MHz bands. The 217Hz TDMA buzz couples into everything.
The only audio devices I have seen deal well with it are cell phones because they need to follow strict guidelines to prevent failures in compliance testing. At a minimum they run no traces on the outer layers of the PCBs and put a shield can over almost every single active device.
The only audio devices I have seen deal well with it are cell phones because they need to follow strict guidelines to prevent failures in compliance testing. At a minimum they run no traces on the outer layers of the PCBs and put a shield can over almost every single active device.
I've tried to think about this and don't see an obvious answer. Maybe it's a question of being caught between conflicting polarized viewpoints: old/new, conventional measurements/ more elaborate measurements, all the stuff that appears on this thread. Maybe it's a Jared Diamond case with an excluded middle.
But it does seem to me to be a worthy topic for a cutting-edge thread, if only to dispense with it.
Much thanks, as always,
Chris
Seems to me one contribution is widespread ignorance/denial about common-mode noise. Hence to these denialists and duffers, there's no possible upside of including a trafo, only a downside (more distortion).
This could have been expressed more obsurely, but it would have taken a more concerted effort. To try again: if the low perm range about zero crossing is extended far enough to include all of design B-H excursion, then the core has no monotonicity issues. That's why there can be microphone level transformers, etc. And they work just fine.
Amplifier to speaker level transformers get into a more difficult and ambiguous region, but filtering the outside world of demons and dragons is easier here.
Chris
The contributions of The Neutron Guy are much missed. I (and I'll bet a lot of other geezers) learned a lot from his clarity in modeling. And, it's *all* modeling.
Much thanks, as always,
Chris
Disagreed. It's really the same op amp, but the 27 is unity gain stable, for which you pay with a slew rate of some 2.7 V/uS, while the 37 is decompensated, is stable at a gain of 5:1 but has a slew rate of around 12-15 V/uS.
While hardly a capital improvement, in my experience there is just enough benefit ro make the 37 more desirable. Especially in phono EQ amps, where we need lots of gain. Which is where I prefer to use it anyway due to low noise. I refere to the Burr-Brown engineers' passive phono RIAA eq amp, so far one of the nest such devices I have heard (published in a German magazine in 1990, I posted here a few times).
Unfortunately, I have never tried Scott's AD797, I understand its price is currently around $3. Yes, that is more expensive than your garden variety modern op amp, but in the grand scheme of things, the price difference will make the whole negligibly more expensive even assuming one uses quite a number of them inside.
But John's enthusiasm for it is contageous enough to make want to try it out. In between any and all test results, it's the word of the designer which carries the most weight, in my view.
But John's enthusiasm for it is contageous enough to make want to try it out. In between any and all test results, it's the word of the designer which carries the most weight, in my view.
My current view is that people just don't understand the loading of transformers so get wacky FR. I will admit the only time I have used an MC stepup I was less than impressed BUT didn't bother trying to work out its loading.
Whilst DC to daylight is difficult with a single core, as long as you are not passing DC you should be able to get a line transformer which can get to nearly 100kHz. I dare say given we are no longer in the 600 Ohm world with 10K input impedances you might even be able to put in a smaller core and go even higher.
Would certainly deal with one big chunk of Common mode grot. Still wont help for Richards Pathological DVD player .
I don't think RNM's CD player is atypical. Controlling the EMI inside a piece of gear and keeping it from leaking out is a bit of an art. If you are using an audio spectrum analyzer to look at the output during the design phase you will miss it entirely.
Then there is the issue of actual filter design. Getting the signal in the stop band down by say 90 dB is quite difficult. So as a CD standard is 2 VRMS out depending on the logic level and the number of sources, the EMI can start out many dB above the signal.
So in addition to the filter you need good design to keep the EMI sources shielded from the output section. To keep costs down there often is one PCB. Now logic levels are dropping not to keep EMI down as much as to speed things up. Lower voltages with a fixed maximum slew rate require less switching time and less power dissipation.
Then there is the issue of actual filter design. Getting the signal in the stop band down by say 90 dB is quite difficult. So as a CD standard is 2 VRMS out depending on the logic level and the number of sources, the EMI can start out many dB above the signal.
So in addition to the filter you need good design to keep the EMI sources shielded from the output section. To keep costs down there often is one PCB. Now logic levels are dropping not to keep EMI down as much as to speed things up. Lower voltages with a fixed maximum slew rate require less switching time and less power dissipation.
Seems fair enough. If you use ears to listen to the CD player they will miss it entirely as well.
Not at all. These days you enter the desired in-band, transition band, and stop band parameters into one of several software packages and out comes the parameters for your digital filter.
The largest peak digital signal voltage you'll find in modern digital audio gear will generally be less than that 2 V RMS analog signal which has a peak value of 5.6 volts. That's basic electronics that apparently you don't know. As if it mattered.
Basic electronics (which may indeed have escaped you) says its 5.6V peak to peak.
Bill, if we could get you to stick to technology and abandon the personal attacks, maybe there would be an actual discussion.
Trying disparately to get you back on topic Bill, please tell us what part of a modern music player actually implements the 90 dB attenuation in the stop band that is likely to be audible?
Hair splitting noted. I say peak, you say Peak-to-peak don't you know their similarity? Ever wonder how one makes an undistorted 2 VRMS signal with a single-ended power supply?
- Status
- Not open for further replies.