Made in Milwaukee, the summer job to die for was to work in the chrome plating shop. Big bucks just don't inhale.
Only circa 1940?
Here is one estimated to be "pre flood". Wow. I wonder if Jesus ever drove a Landrover?
The Coso Artifact: Mystery from the Depths of Time
Last edited:
It's already becoming available, starting at the high end of the market. It's called the "42 volt" system, actually 36 volts but called "42 volt" because the normal on-charge running voltage is (14x3 = 42) volts.
The incentive is partly lighter weight but also the sheer impracticality of 12 volts for things such as catalytic converter pre-heating. Some of the high-end BMW models draw over 11 KW of power for the first few minutes after starting. Imagine a cold morning, you have catalytic pre-heat, electric seat heaters, window defrosters, electric power steering...
I remember reading an article years ago about the potential to go to 48V systems for auto. Weight savings due to the lower current requirements is one advantage, but with the recent cost of copper the cost savings would be substantial as well. The main problem if I recall correctly was that at 48V the alternators could not be air cooled as they are currently and that added significant expense, and also reliability concerns. Another minor problem was that all existing automotive circuitry (for cars at least) is based on 12V 
Tony.
Tony.
Only circa 1940?
Here is one estimated to be "pre flood". Wow. I wonder if Jesus ever drove a Landrover?
The Coso Artifact: Mystery from the Depths of Time
Nice story Glen!
Old Post 377
Actually SLS is no longer the descriptor, today it is "Sounds like A__" please try to keep up to date on the slang.
However I proposed to place a relay across the magical devices at loudspeaker level. In low level audio my measurements (not yet complete) show punch through (previously called tunneling by some) occurs around -130dbv or so. As I expect to listen at a level of 70dba average (1M measurement) this would be a peak level of 90dba or just under 1 volt peak. (Yes my loudspeakers are efficient.) My listening area has a background noise level of 31dba slow weighting (B&K 2203 SN4245458 Capsule 1" 4145 #425194 Pistonphone 4228 #1618643. I use this model for these measurement as it is what was most commonly used for the original research I do own more recent equipment).
Now the 90dba peak minus the 31dba would seem to limit the listening range to just under 60db except for a few very important considerations.
It has been often demonstrated that it is possible to perceive a distortion or pure tone even when it is 10 to 30 db below the music level.
For those who wish to try this the easy way is to either create a wave file with a fixed tone or to use an MLS white noise generator mixed with a sine wave oscillator and then passed through a pinking filter.
Then there is the issue of the actual masking. I use a model of music having peak energy around the 100 to 200hz octave and rolling off at 3db per octave above and below this.
So if there is say seventh order distortion of 200hz I would expect this to be heard even if it is 38db below the music level. Of course there is another factor to make this even worse! Our hearing is more sensitive to midrange frequencies than to lower ones. (Fletcher Munson) Good practice would have me throw in another 10-20db to correct for this and as a "safety factor".Thus I expect the -130db of the relay to be just below my possible distortion hearing range of 117 db. Of course that is opinion so I should also try just a resistor of the same value as the Bybee device for a fair comparison.
Of course if you run these numbers for a relay at preamp levels you would find they predict you could hear the distortion.
So we actually do not disagree if it would have been worded as relays often "SLA!"
There are three reasons why the distortion that you hear of the Burson opamp actually goes down as you increase the gain a bit.
This design of this opamp is sensitive to the source impedance. As it rises or becomes unbalanced the even order harmonics rise, as to if this then masks other less pleasant distortions is subject to debate, but one should be aware that it exists.
The second reason is that the opamp has a finite current output and to match the low source resistance to the non inverting input in my test setup would require about a 500 ohm load ohm to the output for the feedback network. This is low enough to cause distortion to rise as shown in the article.
Finally is the confusion about how increasing the feedback around an opamp improves the gain linearity. The linearity is improved by the open loop gain that is available for feedback. That is well known. What is not appreciated is what really happens. If you have an amplifier that before feedback creates 10% third harmonic distortion after feedback this will be reduced to .1% or less depending on the gain. In creating this cancellation signal there will be another 10% distortion of the correction. Or in other words 1% of ninth order distortion! As can be seen above ninth order distortion stands out much more than third by almost 30 to 40 db. Thus the reduction of a measured 20db of distortion may actually sound worse. So more feedback is not always better sounding. Careful design does allow one to optimize global feedback to allow for known hearing parameters.
But the answer to the question of would people believe such a test of the Bybee devices seems to be a clear "No."
anyone can! Relays sound s**t, we know this since decades!
Regards
Actually SLS is no longer the descriptor, today it is "Sounds like A__" please try to keep up to date on the slang.
However I proposed to place a relay across the magical devices at loudspeaker level. In low level audio my measurements (not yet complete) show punch through (previously called tunneling by some) occurs around -130dbv or so. As I expect to listen at a level of 70dba average (1M measurement) this would be a peak level of 90dba or just under 1 volt peak. (Yes my loudspeakers are efficient.) My listening area has a background noise level of 31dba slow weighting (B&K 2203 SN4245458 Capsule 1" 4145 #425194 Pistonphone 4228 #1618643. I use this model for these measurement as it is what was most commonly used for the original research I do own more recent equipment).
Now the 90dba peak minus the 31dba would seem to limit the listening range to just under 60db except for a few very important considerations.
It has been often demonstrated that it is possible to perceive a distortion or pure tone even when it is 10 to 30 db below the music level.
For those who wish to try this the easy way is to either create a wave file with a fixed tone or to use an MLS white noise generator mixed with a sine wave oscillator and then passed through a pinking filter.
Then there is the issue of the actual masking. I use a model of music having peak energy around the 100 to 200hz octave and rolling off at 3db per octave above and below this.
So if there is say seventh order distortion of 200hz I would expect this to be heard even if it is 38db below the music level. Of course there is another factor to make this even worse! Our hearing is more sensitive to midrange frequencies than to lower ones. (Fletcher Munson) Good practice would have me throw in another 10-20db to correct for this and as a "safety factor".Thus I expect the -130db of the relay to be just below my possible distortion hearing range of 117 db. Of course that is opinion so I should also try just a resistor of the same value as the Bybee device for a fair comparison.
Of course if you run these numbers for a relay at preamp levels you would find they predict you could hear the distortion.
So we actually do not disagree if it would have been worded as relays often "SLA!"
There are three reasons why the distortion that you hear of the Burson opamp actually goes down as you increase the gain a bit.
This design of this opamp is sensitive to the source impedance. As it rises or becomes unbalanced the even order harmonics rise, as to if this then masks other less pleasant distortions is subject to debate, but one should be aware that it exists.
The second reason is that the opamp has a finite current output and to match the low source resistance to the non inverting input in my test setup would require about a 500 ohm load ohm to the output for the feedback network. This is low enough to cause distortion to rise as shown in the article.
Finally is the confusion about how increasing the feedback around an opamp improves the gain linearity. The linearity is improved by the open loop gain that is available for feedback. That is well known. What is not appreciated is what really happens. If you have an amplifier that before feedback creates 10% third harmonic distortion after feedback this will be reduced to .1% or less depending on the gain. In creating this cancellation signal there will be another 10% distortion of the correction. Or in other words 1% of ninth order distortion! As can be seen above ninth order distortion stands out much more than third by almost 30 to 40 db. Thus the reduction of a measured 20db of distortion may actually sound worse. So more feedback is not always better sounding. Careful design does allow one to optimize global feedback to allow for known hearing parameters.
But the answer to the question of would people believe such a test of the Bybee devices seems to be a clear "No."
Actually a pure compressive third order distortion creates all odd harmonics in a geometrically decreasing magnitude when put inside the loop.
This is an excellent discussion which uses explicit mathematical solutions and not a simulator.
http://www.its.caltech.edu/~musiclab/feedback-paper-acrobat.pdf
BTW Jerry Sussman is one of the sharpest minds your likely to come across and he honestly gives credibility to >100kHz hearing, etc. gives me pause.
Last edited:
Yes that's true. Lots of people here say 'feedback is bad' without giving proper credit to this complexity.
jd
Wow! Scott that is quite a paper.
Manfred Schroeder did a simple experiment where he created waveforms using DSP with controlled amounts of phase shift on the harmonics but keeping the peak values the same. His conclusion as I recall was that 5 degrees at 20K could be heard. That is the same conclusion that R Neve got from playing with phase shift in transformers.
I often turn on a 30K tone (70db ish) and ask people if they hear it. They almost always say "No." After turning it off they recognize that it is now gone and can then determine when it is on or off. It is not perceived the way lower frequencies are but it is possible to detect it.
Then there are also other factors that affect our perception even when we normally do not notice them. One friend used to do research on separating oil from water using RF energy to change the zeta potential. Turns out it also had effects on humans at levels of a few mw!
It is different to say: Feedback is bad, that is based on subjective appraisal, and feedback is complex, which is an established fact. It just so happens that because feedback is complex, it most often does as much damage as it fixes. This is why people say: 'Feedback is bad' Usually they are right.
To say that Feedback is good, is a lot like saying: Digitized to improve the original performance. This farce goes on, all the time, as well.
To say that Feedback is good, is a lot like saying: Digitized to improve the original performance. This farce goes on, all the time, as well.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.