Resistor Distortion
If you can detail your measurement method I would be very interested.
I have two Radiometer Copenhagen CLT-1's that I use for checking resistors, pots and other assorted passive components. They have a floor in the -170 dB range, but work at one frequency only. It can drive most resistors to 1/2W which is enough to see a lot. The long article on cap distortion that dismissed them didn't accurately describe how they work . . .
Most premium resistors get close to the floor unless they are defective or soon will be. Pots are all across the map with the premium Japanese and US products being much better than the generic products.
These measurements are very sensitive to external noise. The CLT-1 has a really narrow bandpass that makes it more tolerant but it still can be a problem. A thought- perhaps different passive components are more sensitive to external magnetic & e-m fields and more likely to pick up everything from transformer harmonics to cell phone radiation to fields from Zobel networks on the output.
I think -165 to -170 dB distortion is not an issue, at that level you won't hear distortion products. You will be hearing other things. Other factors may be more significant like frequency response, voltage coefficient, temp coefficient, thermal response time etc. These are hard to characterize.
If you can detail your measurement method I would be very interested.
I have two Radiometer Copenhagen CLT-1's that I use for checking resistors, pots and other assorted passive components. They have a floor in the -170 dB range, but work at one frequency only. It can drive most resistors to 1/2W which is enough to see a lot. The long article on cap distortion that dismissed them didn't accurately describe how they work . . .
Most premium resistors get close to the floor unless they are defective or soon will be. Pots are all across the map with the premium Japanese and US products being much better than the generic products.
These measurements are very sensitive to external noise. The CLT-1 has a really narrow bandpass that makes it more tolerant but it still can be a problem. A thought- perhaps different passive components are more sensitive to external magnetic & e-m fields and more likely to pick up everything from transformer harmonics to cell phone radiation to fields from Zobel networks on the output.
I think -165 to -170 dB distortion is not an issue, at that level you won't hear distortion products. You will be hearing other things. Other factors may be more significant like frequency response, voltage coefficient, temp coefficient, thermal response time etc. These are hard to characterize.
Scott
As I look at what should be non reactive passive components, there are two kinds of distortion that may be present.
One is the deviation from linear I.E. non ohmic, (or maybe from whatever the desired response should be) the second is sort of a step function discontinuity. Which is in many ways similar to missing code steps in a DAC.
The normal assumption is that the step discontinuities are basically just barrier break-over at very very low voltages. I have not yet seen anything to believe otherwise. Although if a discontinuity does show up at a specific non-zero voltage or current level that would be most interesting!
As most people accept that the threshold of pain is around 130dba and the perception threshold is 0dba for adults and may go as low as -10dba for some children and others. (Not to be confused with those who behave as children.)
This would indicate that a 140 db range should cover all situations and probably less as most of us do not run our sound systems at the threshold of pain.
However there are a few other issues such as level mismatching. As a typical example I tend to raise my volume control to hear soft passages or poor recordings and then to lower it for louder ones or recordings that make different use of the dynamic range available to them for any given recording method. A typical volume control range is 45db, although I find most of my movement is less than 20db. So worst case would be 185db of range which I suspect we both would consider ridiculous. However it does make a point for at least 120 to 140db of range in the reproduction system.
At the lower ranges and with more than normal amplification the step discontinuity distortions begin to become dominant. At higher levels my feel is that we have assumed for too long that superposition holds even though it is limited to linear systems. Why many presume two nonlinear systems in series should follow superposition strikes me as silly. Although there can be good math to strike an optimal gain structure to minimizes this.
So 23 bits would begin to seem a reasonable range. My feel says to go a factor of 10 better to be sure you are not missing anything. So 26 bits or 160db is what in my OPINION is a valid design goal. As this is not currently reached, more information is required as to what are the limiting factors.
Now we probably both agree that the vast majority of users do not require this kind of resolution, but there certainly those who think they do.
Superposition is not a law, nor is there a first law applicable. This is not an undergraduate linear algebra course.
Math is used to model the real world, it may give insights into what happens allow us to manipulate formulas and predict results, but it is not a controlling force. It is only one type of modeling.
The principle of superposition is what allows us to use Norton and Thevenin's THEOREMS to simplify and solve problems.
There are not just primary, but also secondary, tertiary etc. effects on real world behavior.
The skill in engineering is knowing when you can take short cuts and when you must do the work the hard way.
I certainly can place a discontinuous or discrete process in series with a continuous one. There is also the issue of hysteresis that may occur in some systems.
I don't mind answering questions, but do not wish to spend tome correcting misconceptions.
You're using the term "superposition" in a restricted sense.
If you violate continuity, you violate the First Law. Simple as that.
edit: I am using "continuity" in a physics sense, not a mathematical one- perhaps that's the reason for the misunderstanding?
From Wikipedia:
"A continuity equation in physics is a differential equation that describes the transport of some kind of conserved quantity. Since mass, energy, momentum, electric charge and other natural quantities are conserved, a vast variety of physics may be described with continuity equations.
Continuity equations are the (stronger) local form of conservation laws. All the examples of continuity equations below express the same idea, which is roughly that: the total amount (of the conserved quantity) inside any region can only change by the amount that passes in or out of the region through the boundary. A conserved quantity cannot increase or decrease, it can only move from place to place."
If you violate continuity, you violate the First Law. Simple as that.
edit: I am using "continuity" in a physics sense, not a mathematical one- perhaps that's the reason for the misunderstanding?
From Wikipedia:
"A continuity equation in physics is a differential equation that describes the transport of some kind of conserved quantity. Since mass, energy, momentum, electric charge and other natural quantities are conserved, a vast variety of physics may be described with continuity equations.
Continuity equations are the (stronger) local form of conservation laws. All the examples of continuity equations below express the same idea, which is roughly that: the total amount (of the conserved quantity) inside any region can only change by the amount that passes in or out of the region through the boundary. A conserved quantity cannot increase or decrease, it can only move from place to place."
There have been some interesting quantum mechanical experiments using gold wire drawn until only a few atoms support conduction. I have seen no credible audio related references to these step function discontinuities. We looked for the so call "micro-diodes" in multi-strand wire 20yr. ago, the resolution floor was <-150db and found nothing.
I also beg to differ on one thing, put two signals through a squaring circuit separately then combined. The results do not superpose and no one I know "has assumed too long" that they did.
I think you are mixing high level and low level effect too freely. Resistors have well known thermal and voltage co-efficient effects, not only are the non-linearities limited to very low order, they scale as the square of the voltage (same is true of excess noise). Picking resistors for preamp circuits by stressing them at 10's of volts makes little sense. The effects at .1 to 1V levels are an additional 40 to 60dB down. So a Dale RN55 that measures -120dB at 10V rms meets your -180db down in your preamp and about -240dB as a cartridge termination.
Sy
When I write 1+1 = 10 I expect the reader to be on the same page and not to tell me that is wrong.
I used superposition principle as it is used in this engineering situation. The use of superposition for wave equations is not the same. Wiki also has a nice section on the use of superposition in the right sense.
Please note that not all phenomena can be modeled by continuity or differential equations.
But you already knew that I suspect.
Simon, I know this is a slight departure, but have you read about quantum resistors? There was an important paper written about 12 years ago that I have been able to get with 'quantum resistor noise' on google. These devices violate Johnson noise predictions and show an alternate way to make resistors. It might provide you some insight.
Scott
Thanks for answering. I appreciate your taking the time.
This is one of those we are not communicating the same thing. I was referring to how distortion and or gain mechanisms combine. The common method for audio is to specify something such as .01% distortion at 100 watts 1khz. Which is accurate, but really does not tell you what it will scale to at a useful range of 1 or .1 watt. A discontinuity distortion (if present) is important at low levels but is lost at higher ones.
And I looked for it at lower levels and even heated the sample with a torch and did not find any evidence of micro-diodes either. Let us be polite and presume neither of us lend support to that distracting theory.
I would agree that I did not write that clearly enough. I don't think I am thermally over-stressing resistors. Since I test at 1/8 rated power or less. I understand other have measured them closer to the wattage ratings, but I am more interested in how they are really used. (John mentioned higher voltages so I think that may be where that idea came from.) I wish to use as low a test voltage as possible and still get results above the noise. If I wish to add heat I can use a DC or low frequency bias.
In your analysis you are taking the -120db at 10 vrms and scaling it down to .1v, that is where I am curious to see if superposition holds. In my experiments I am seeing some data that may indicated a low level distortion mechanism that is often confused with or lost in the noise. I am interested in being able to produce a repeatable measurement of this.
Just for fun, I have measured vibration induced noise in some resistors.
So I think the experiment is worth doing.
The nice thing is that a few on this thread have help me clarify what I should be looking for, so thanks!
Simon et al, the specific paper that I am recommending is: 'Carbon Nanotube Quantum Resistors' by Frank, der Heer, et al 1998. It is relatively old, but it IS declassified and it is relatively clear. It shows a completely different way of electron conduction, based on first quantum principles. It shows that some here should 'get up to speed' before telling the rest of us 'where the bear sits'. ;-)
A totally worthy experiment, modern test gear should get you down to resolution bandwidths where something might show up. I do understand how a casual intuitive look at something like two wires twisted together or mechanically touching whould say there has to be some kind of barrier to overcome. Just in my experience the effects were far below any noise masking threshhold.
How do you vibrate ONLY the resistor?
.
Physicists vs. EEs
.Superposition principle (corolar of the local form of a conservation law) vs. circuit superposition theorem (requiring circuit linearity).
John, you answered your own question "completely different method of conduction", means Nyquist (Johnson) might not apply. BTW quantum noise is commonly used to include shot and Nyquist noise not some other quantum mechanical effect. You might read Carver Mead's excellent argument for them being the same, it might shed some light on this.
Relatively clear? Unless you are up on Fermi fluids and lots of other arcane stuff, this paper is good for throwing irrelevant distractions into the disscusion.
Last edited:
I think not. Bulk nanotube "powder" is not what they are talking about and I know where this is heading.
-240 dB is about the path loss of a VHF signal from earth to moon and
scattered back to earth.
hi Mr. wurcer,A totally worthy experiment, modern test gear should get you down to resolution bandwidths where something might show up. I do understand how a casual intuitive look at something like two wires twisted together or mechanically touching whould say there has to be some kind of barrier to overcome. Just in my experience the effects were far below any noise masking threshhold.
How do you vibrate ONLY the resistor?
in my experience the effects were far below any noise masking threshhold.
what bandwidth do you use for audio noise masking and why?
Last edited:
- Status
- Not open for further replies.