There is more to it than just temperature coefficient. Carbon Composition resistors have harmonics well out into the 20s! (7th and higher harmonics really stand out and -100 on the high harmonics may not mask them!)
What color jellybeans ? I hope you only used red jelly bellys after all those in the know do but you did not hear it from me.
I thought it was common knowledge:
Use Blue beans for the Blues music
Black for Heavy Metal
etc.
An unsupported claim is not "data." Sorry, go back to the voltage divider equation.
Take a look at the carbon composition distortion plot shown earlier. Now take the spectrum of music multiply that times an inverse Fletcher Munson curve multiply that times the distortion products multiply that by the loudspeaker distortion and then sum the result. As there is coherent phase between the distortion components the result is going to be less than 20 dB.
Voltage divider has nothing to do with a resistor in parallel with the input. If you want you can even drop the thermal third harmonic you will still sum to "brighter".
Yes there is magic involved as you can make something appear that wasn't there!
It is not a Kodak moment, much more Picasso.
Sorry to interupt the collective monologue, but has anyone figured out what coresta's talking about yet?
😕
😕
Coresta appears to believe, if I understand him correctly, that thicker conductors produce less noise. At one time this appeared to be because he believed that they produced less thermal noise because of their lower resistance, but now it seems he believes this is because they have fewer (or 'bypassed'?) boundaries. I hope I have not misrepresented his view.
No evidence or physics-based explanation has been offered, apart from an anecdote (in another thread?) about a thick FM antenna picking up a better signal than a thin FM antenna - which can be easily explained using conventional ideas like radio theory and antenna theory.
No evidence or physics-based explanation has been offered, apart from an anecdote (in another thread?) about a thick FM antenna picking up a better signal than a thin FM antenna - which can be easily explained using conventional ideas like radio theory and antenna theory.
#85:
JUST TRY IT .... 
Johnson noise have something to do with that : when you cool a microwave antenna (very big ratio wavelength/size) , it becomes more silent ... http://http://www.setileague.org/askdr/cooling.htm
Hmmm 🙂 I'm not alone ...
JUST TRY IT .... Johnson noise have something to do with that : when you cool a microwave antenna (very big ratio wavelength/size) , it becomes more silent ... http://http://www.setileague.org/askdr/cooling.htm
Last edited:
Thanks for demonstrating that you don't understand the issue. Let me explain.coresta said:
At microwave frequencies the ambient noise can be extremely low - even down to the cosmic background noise temperature of about 3K. Antennas may be large horns which require an incoming signal to reflect and bounce back and forth a number of times, each time picking up a little thermal noise as the antenna metal is not perfectly reflective but very slightly lossy. Antenna thermal noise can then be important.
At VHF the antenna noise temperature is typically around 200-500K and antennas have very low losses so the conductor thermal noise is irrelevant.
Unless you understand antenna theory and radio theory it is best not to invoke them to support a false idea in audio circuit theory. Of course, if you did understand them then you would be much less likely to have the false audio idea.
Why bother and make someone else waste time trying to get something useful out of you? No test conditions or setup, then the unspecified and vague data applied to an inappropriate analysis, followed by an irrelevant analogy; it's complete nonsense top to bottom.
This is a trivially simple voltage divider problem. Best not to look at your dust clouds, but look at the actual issue, unless one's goal is to mystify and mislead rather than illuminate.
This is a trivially simple voltage divider problem. Best not to look at your dust clouds, but look at the actual issue, unless one's goal is to mystify and mislead rather than illuminate.
If we have an audio signal voltage of 200mVac (no DC) across a 10k attenuator that is made up from two 600mW 1% 100ppm resistors, what temperature variation might we have on each "resistive element" in the attenuator? Let's assume 9k & 1k for -20dB of attenuation.
What resistance change could there be for that estimated temperature change.
To save you all working it out, the PEAK dissipation across the whole attenuator is 8microwatts.
What resistance change could there be for that estimated temperature change.
To save you all working it out, the PEAK dissipation across the whole attenuator is 8microwatts.
Last edited:
The graphs are straight from the Linear Audio article as is the test method. Why you try to confuse things is beyond me.
The voltage divider has no effect on the distortion. It also drops the signal so the % stays the same. The third harmonic distortion is the only one that increases with the square of the voltage. The other harmonics have much less or no voltage sensitivity. So the thermal PPM rating is of no importance. It is the other distortions that come into play. The thermal third by itself is not even particularly objectionable.
BTY there is an advantage to silver wire for antennas! It has to do with skin effect and that silver oxide is actually a better conductor than silver itself. Not so for copper. That is why where P.I.M. and other RF issues matter silver plated coax is used. (Google Headend Cable Belden 9167)
The point is that sometimes people really do make changes that seem trivial but do result in real measurable significant changes. The explanations are sometimes quite amusing.
Now why you diss someone and then try to turn the argument around is another issue.
Last edited:
Nonsense. And since the test conditions are unrelated to the application at hand here, double nonsense. And since you're throwing in Fletcher-Munsen while ignoring the fundamental (this seems to be a recurring trick of yours), triple nonsense.
If I'm to believe you, you don't know basic voltage divider theory. Since I know you do, I remained puzzled about why you're deliberately kicking up so much dirt to obscure a simple problem.
SY run your own test. Carbon Composition resistors have distortion at all signal levels. It is the nature of the beast. In the 60's when they last standardized resistor distortion measurements they couldn't use the bridge technique as the resistors were not uniform enough. So they settled for using a current source and just looking at third harmonic. There ain't nothing new there.
As to the perception of distortion, even you know that the resonant frequency of the ear canal is around 3-5K resulting in the most sensitivity there.
Now when you play music it does not have a flat spectrum. The normal model used is that the energy peaks around 150 hertz and tolls off at 3 dB per octave. This does not account for extreme cases.
So just looking at 150 hertz vs 4800 hertz the music spectrum has a 15 db boost for uncorrelated signals. Fletcher Munson at normal listening levels gives another 16 dB boost. (Note when you look at the curves they are for pure tone and what you listen to is not so be sure to pick the right level curve for your personal listening volume.)
So allowing for masking sensitivity of 30 dB. F-M add of 16 db and 15 dB of music slope You should be able to perceive 5th harmonic distortion at a level of 61 dB below the music level. That comes out to .09% distortion just for the fifth harmonic distortion. There are lots of others. Now here is the kicker, distortion is not uncorrelated. So it adds up faster than noise.
Now what the before modification distortion was and the after is really the only question.
No hand waving, no magic, just the perception basics. Just actual resistor measurements. Lets talk again after you have run a test or two.
As to the perception of distortion, even you know that the resonant frequency of the ear canal is around 3-5K resulting in the most sensitivity there.
Now when you play music it does not have a flat spectrum. The normal model used is that the energy peaks around 150 hertz and tolls off at 3 dB per octave. This does not account for extreme cases.
So just looking at 150 hertz vs 4800 hertz the music spectrum has a 15 db boost for uncorrelated signals. Fletcher Munson at normal listening levels gives another 16 dB boost. (Note when you look at the curves they are for pure tone and what you listen to is not so be sure to pick the right level curve for your personal listening volume.)
So allowing for masking sensitivity of 30 dB. F-M add of 16 db and 15 dB of music slope You should be able to perceive 5th harmonic distortion at a level of 61 dB below the music level. That comes out to .09% distortion just for the fifth harmonic distortion. There are lots of others. Now here is the kicker, distortion is not uncorrelated. So it adds up faster than noise.
Now what the before modification distortion was and the after is really the only question.
No hand waving, no magic, just the perception basics. Just actual resistor measurements. Lets talk again after you have run a test or two.
At the risk of being shot, here's how I look at it.
A resistor that distorts can be represented by a resistor that dynamically changes its value with the signal, allright for the discussion?
A grid leak resistor can distort (modulate its value) all it wants, but since there's no current trough it there's no signal modulation so no distortion.
Enter the input volume control.
NOW the grid sits at the node of a voltage divider (vol. stuff and grid leakage resistor) and NOW the resistor distorting is distorting the signal because it is modulating the divider ratio.
But only because of the vol. control stuff - the voltage division.
So if the resistor distorts say 0.01% and the ratio between vol control stuff and grid leak is 10k/400k = 1/40, the resultant distortion is 0.01/40 = 0.00025%.
Can I go play outside now?
Jan
A resistor that distorts can be represented by a resistor that dynamically changes its value with the signal, allright for the discussion?
A grid leak resistor can distort (modulate its value) all it wants, but since there's no current trough it there's no signal modulation so no distortion.
Enter the input volume control.
NOW the grid sits at the node of a voltage divider (vol. stuff and grid leakage resistor) and NOW the resistor distorting is distorting the signal because it is modulating the divider ratio.
But only because of the vol. control stuff - the voltage division.
So if the resistor distorts say 0.01% and the ratio between vol control stuff and grid leak is 10k/400k = 1/40, the resultant distortion is 0.01/40 = 0.00025%.
Can I go play outside now?
Jan
You should be punished for doing the basic calculation that Ed has studiously ignored. There IS current through it, but it's easy to calculate how the variation in resistance affects the voltage across it, which you have demonstrated. And the current is about 3 orders of magnitude (60dB) below Ed's test condition, so his ridiculously low numbers (which themselves have come under criticism for basic testing errors) become lost in the noise of the Universe.
This is not trivial, as the usual NFB network is a voltage divider with significant voltage across the resistor connected to the output.
This would be thermally modulated in value, and unlikely to track the associated resistor to ground, because of the considerable power difference in the two (other than at a very low gain).
I've always used considerable overkill for the power rating on the resistor connected to the output for this reason.
- Status
- Not open for further replies.