Right now I have a voltage divider for each channel in my amp bypassing gain pots--fixed voltage divider of (0, 20k ohm). I am using a Metal Film 1% tolerance resistor right now for each channel--each measuring exactly 20.0k ohms on my multimeter.
Would I hear an audible improvement if I replaced those 25 cent metal film resistors with 20k ohm Vishay bulk foil resisters? They cost $20 for each resistor--so it would be a $40 "upgrade".
Would I hear an audible improvement if I replaced those 25 cent metal film resistors with 20k ohm Vishay bulk foil resisters? They cost $20 for each resistor--so it would be a $40 "upgrade".
Audible improvement will depend on how good your system (electronics & speakers) is in fine resolution.
There are also different types in Vishay BFs, and prices. Have a look at Percy's, that stocks them all and has good prices:
Michael Percy Audio Ordering Information
There are also different types in Vishay BFs, and prices. Have a look at Percy's, that stocks them all and has good prices:
Michael Percy Audio Ordering Information
I very much doubt it. Metal films have good linearity and excess noise properties already. I think the only types of resistors with problematic linearity are carbon composition and thick film.
EDIT: Reading up, it seems carbon films aren't exactly the very definition of linearity either (though less bad than carbon comps), but I think they were used in lots of audio equipment in the past and few people ever complained. The voltage coefficient of metal films is only about 1% as high as with these.
Last edited:
NO, you are wrong, Bone. The best resistors will measure the 'best' as well for the most part. For example, $10 Vishay bulk metal does not sound better (to my ears and many others) as a Roderstein Resista ($0.10). The measurements show similar performance. So 100:1 difference in price is not that different. Trust me, I use both for serious designs.
Just came across this interesting post on Metal Film vs. Bulk Foil resistors from 2011 and thought I'd add my 2 cents:
Noise in electronic circuits is generally measured in nano-volts and is highly temperature and frequency dependent. For reference, one nano-volt is 0.000000001 volts, an extremely small value. In electronic circuits that use resistors and capacitors it is the resistors that contribute the substantial portion of noise and for the most part with audio circuits in the 20 - 20 khz, which is typically assigned to human hearing, you can completely ignore capacitors. In electronics the general formula for noise is E = 7.43 X SqrRoot(R X T X DeltaF) where E is in nano-volts, T is the temperature (in Kelvin), R is in K-Ohms, and DeltaF is in is the frequency bandwidth in K-htz.
In general, the formula says that noise increases as the resistance value goes up, the temperature goes up, or the frequency (or bandwidth) goes up or any combinations goes up. In reality all electronic components contribute noise to the system in which they used. And to be more specific, the noise that is attributed to different compositions of resistors is actually referred to as Johnson Noise, which is the random noise created in the material due to atoms moving even in the absence of an electrical signal.
Without getting into some rather complicated theory and associated math, I'll conclude by saying that when comparing certain types of resistors such as metal film, metal foil, carbon film, carbon composite, and wire-wound, for audio metal foil has been found to produce the least measurable noise contribution. However, whether one can hear the contribution of a reduction in noise as a result of using metal foil resistors is highly dependent on the specifics of the audio equipment being used and what, or who, is listening or measuring the noise levels. Since noise levels are highly affected by the voltages used in a circuit, it's questionable whether a person with average hearing will be able to discern noise levels at the nano-volt, or even micro-volt (which is an order of magnitude higher than nano-volt) levels.
In the early 60's the well regarded and famous Harmon Hardon Citation I Stereo pre-amp used carbon composition resistors in its circuitry and you could clearly hear the increase in "noise", usually a "hiss", when you turned up the treble levels beyond their flat settings. The Citation I used lots of feedback and was also known as the Feedback Pre-amp in its day. So circuit topology can contribute to noisy design as well. The Citation I used tubes rather than transistors and some of the plate voltages used were in excess of 400 volts, a level you do not see in today's transistorized and even more modern tube designs.
So to close this comment, I'd say that unless the audio gear has carbon composition (as in the Harmon Kardon Citation I pre-amp) which are the noisiest and/or is a tube design, then replacing the current resistors with more expensive resistors is unlikely to result in an audible improvement IMO. Also, keep in mind that resistor value accuracy has nothing to do with their inherent thermal and noise characteristics. For example, a 1000 ohms resistor rated at 5%, regardless of its composition, simply means that its value - and not its noise - can vary between 950 ohms and 1050 ohms and still be considered a 1000 ohm resistor. Similarly a 1000 ohm resistor rated at 1% simply means its value can be between 990 ohms and 1010 ohms and still be considered a 1000 ohm resistor.
So replacing 5% resistors with 1% resistors is not always going to result in a lower overall noise level - unless you know the circuit design requires more precise resistance values to achieve a more precise results. But that generally has nothing to do with noise levels. Now it is also true that 1% resistors are generally not carbon film and so were of a less noisy composition type, but that's just a byproduct of the tighter tolerance and not directly associated with it. In any case, interesting discussion.
Noise in electronic circuits is generally measured in nano-volts and is highly temperature and frequency dependent. For reference, one nano-volt is 0.000000001 volts, an extremely small value. In electronic circuits that use resistors and capacitors it is the resistors that contribute the substantial portion of noise and for the most part with audio circuits in the 20 - 20 khz, which is typically assigned to human hearing, you can completely ignore capacitors. In electronics the general formula for noise is E = 7.43 X SqrRoot(R X T X DeltaF) where E is in nano-volts, T is the temperature (in Kelvin), R is in K-Ohms, and DeltaF is in is the frequency bandwidth in K-htz.
In general, the formula says that noise increases as the resistance value goes up, the temperature goes up, or the frequency (or bandwidth) goes up or any combinations goes up. In reality all electronic components contribute noise to the system in which they used. And to be more specific, the noise that is attributed to different compositions of resistors is actually referred to as Johnson Noise, which is the random noise created in the material due to atoms moving even in the absence of an electrical signal.
Without getting into some rather complicated theory and associated math, I'll conclude by saying that when comparing certain types of resistors such as metal film, metal foil, carbon film, carbon composite, and wire-wound, for audio metal foil has been found to produce the least measurable noise contribution. However, whether one can hear the contribution of a reduction in noise as a result of using metal foil resistors is highly dependent on the specifics of the audio equipment being used and what, or who, is listening or measuring the noise levels. Since noise levels are highly affected by the voltages used in a circuit, it's questionable whether a person with average hearing will be able to discern noise levels at the nano-volt, or even micro-volt (which is an order of magnitude higher than nano-volt) levels.
In the early 60's the well regarded and famous Harmon Hardon Citation I Stereo pre-amp used carbon composition resistors in its circuitry and you could clearly hear the increase in "noise", usually a "hiss", when you turned up the treble levels beyond their flat settings. The Citation I used lots of feedback and was also known as the Feedback Pre-amp in its day. So circuit topology can contribute to noisy design as well. The Citation I used tubes rather than transistors and some of the plate voltages used were in excess of 400 volts, a level you do not see in today's transistorized and even more modern tube designs.
So to close this comment, I'd say that unless the audio gear has carbon composition (as in the Harmon Kardon Citation I pre-amp) which are the noisiest and/or is a tube design, then replacing the current resistors with more expensive resistors is unlikely to result in an audible improvement IMO. Also, keep in mind that resistor value accuracy has nothing to do with their inherent thermal and noise characteristics. For example, a 1000 ohms resistor rated at 5%, regardless of its composition, simply means that its value - and not its noise - can vary between 950 ohms and 1050 ohms and still be considered a 1000 ohm resistor. Similarly a 1000 ohm resistor rated at 1% simply means its value can be between 990 ohms and 1010 ohms and still be considered a 1000 ohm resistor.
So replacing 5% resistors with 1% resistors is not always going to result in a lower overall noise level - unless you know the circuit design requires more precise resistance values to achieve a more precise results. But that generally has nothing to do with noise levels. Now it is also true that 1% resistors are generally not carbon film and so were of a less noisy composition type, but that's just a byproduct of the tighter tolerance and not directly associated with it. In any case, interesting discussion.
Last edited by a moderator:
FWIW, I have compared Takman metal film and carbon film, Shinkoh tantalum and Texas Components TX2575 naked foil resistors* as I/V converters in a Teradak Chameleon DAC. The naked foils were a clear winner on sound quality, followed by the Shinkohs. I don't know if this is relevant here.
* They make the Vishays, I believe.
* They make the Vishays, I believe.
All resistors produce Johnson noise.
All resistors produce the same Johnson noise as predicted by the standard formula.
All resistors produce some excess noise.
A few types produce very low excess noise that is often less than the Johnson noise. Metal element resistors fall into this category. There may be others: Any candidates?
Some types produce excess noise that can often be a bit higher than Johnson noise.
And a few types produce levels of excess noise that can be enormous. Carbon resistors fall into this category.
Excess noise is generally avoidable by using metal element resistors.
Once you have accounted for the Johnson Noise and chosen a low excess noise type, I think you will find that avoidable noise has been minimised.
All that is left is to choose a metal element type that suits the OTHER parameters that need to be met, such as accuracy, temperature coeficient, voltage coefficient, low/high parasitics, suitable operating temperature range, etc.
All resistors produce the same Johnson noise as predicted by the standard formula.
All resistors produce some excess noise.
A few types produce very low excess noise that is often less than the Johnson noise. Metal element resistors fall into this category. There may be others: Any candidates?
Some types produce excess noise that can often be a bit higher than Johnson noise.
And a few types produce levels of excess noise that can be enormous. Carbon resistors fall into this category.
Excess noise is generally avoidable by using metal element resistors.
Once you have accounted for the Johnson Noise and chosen a low excess noise type, I think you will find that avoidable noise has been minimised.
All that is left is to choose a metal element type that suits the OTHER parameters that need to be met, such as accuracy, temperature coeficient, voltage coefficient, low/high parasitics, suitable operating temperature range, etc.
Last edited:
As I previously mentioned, carbon composition resistors are the worst in terms of overall noise offenders.
But in audio we're not dealing with Radar or trying to pick up signals on the order of exa watts (10 to the -18th power) where Johnson noise and other forms of noise do indeed matter.
The real issue is whether one can hear the noise contribution different resistor types will make when dealing with audio gear.
Johnson noise is prevalent in all electronic components whose temperature is above absolute zero so effects not just resistors, and Johnson noise is only one part of the electronic noise equation; there are other forms of noise such as shot noise and parasitic noise associated with particular types of resistors, and this is where the different material types come into this topic.
However, measuring any form of noise and particularly Johnson noise, is not a trivial task as it requires that the noise floor of the measuring equipment and overall methodology be less than the Johnson noise or other noise trying to be measured.
In audio so much is subjective even when you can measure a difference. This subjective aspect is not unlike the topic of sampling rate in the digital domain. The current sample rate for CD audio is 44.1 Khz. This was chosen mainly on the basis that human hearing does not exceed 20 Khz at the high end (indeed, if you can hear over 20 Khz you're likely to be non-human). And so a sample rate of 44.1 Khz ensures that the Nyquist Sample Rate, which dictates that the sample rate must be at minimum 2Fmax, preferably more, where Fmax is the maximum frequency to be sampled, is met in order to adequately capture the signal.
Many individuals, and that would include myself feel, based on listening, that digital audio sounds better when captured using higher sample rates, while others say they can't tell the difference.
The use of low-noise resistors in audio ultimately boils down to economics: Using more expensive resistors may indeed improve on lower measured noise levels, but in the end can the majority of individuals that are attuned to hi-fidelity sound actually hear a difference when subjected to gear using common metal-film vs bulk metal foil using a double-blind comparison?
I'm not making a judgement call on not using low-noise resistors though I will likely garner negative comment by stating that the practicality of using relatively expensive low-noise resistors in audio designs needs to be considered carefully.
But in audio we're not dealing with Radar or trying to pick up signals on the order of exa watts (10 to the -18th power) where Johnson noise and other forms of noise do indeed matter.
The real issue is whether one can hear the noise contribution different resistor types will make when dealing with audio gear.
Johnson noise is prevalent in all electronic components whose temperature is above absolute zero so effects not just resistors, and Johnson noise is only one part of the electronic noise equation; there are other forms of noise such as shot noise and parasitic noise associated with particular types of resistors, and this is where the different material types come into this topic.
However, measuring any form of noise and particularly Johnson noise, is not a trivial task as it requires that the noise floor of the measuring equipment and overall methodology be less than the Johnson noise or other noise trying to be measured.
In audio so much is subjective even when you can measure a difference. This subjective aspect is not unlike the topic of sampling rate in the digital domain. The current sample rate for CD audio is 44.1 Khz. This was chosen mainly on the basis that human hearing does not exceed 20 Khz at the high end (indeed, if you can hear over 20 Khz you're likely to be non-human). And so a sample rate of 44.1 Khz ensures that the Nyquist Sample Rate, which dictates that the sample rate must be at minimum 2Fmax, preferably more, where Fmax is the maximum frequency to be sampled, is met in order to adequately capture the signal.
Many individuals, and that would include myself feel, based on listening, that digital audio sounds better when captured using higher sample rates, while others say they can't tell the difference.
The use of low-noise resistors in audio ultimately boils down to economics: Using more expensive resistors may indeed improve on lower measured noise levels, but in the end can the majority of individuals that are attuned to hi-fidelity sound actually hear a difference when subjected to gear using common metal-film vs bulk metal foil using a double-blind comparison?
I'm not making a judgement call on not using low-noise resistors though I will likely garner negative comment by stating that the practicality of using relatively expensive low-noise resistors in audio designs needs to be considered carefully.
I think you missed my point.
Adopt a resistor with low excess noise and you have now minimised the noise contribution of the resistor.
Noise minimisation is not what costs the money.
It's the other parameters that cost money to minimise and it's these other parameters that change the performance of the circuit into which you place the varying resistors.
Noise is not the problem, except in very high gain circuits. Power amplifiers generally do not suffer from noise issues if you select appropriate resistor values and use low excess noise types.
The biggest offender might be signal-correlated excess current noise because it tends to smear the signal, whereas uncorrelated noise does not (it just add some constant, uncorrelated noise which is much more benign even when higher in level). Thick film SMT resistors, notably smaller sizes, are the type of resistors you want to avoid when aiming for transparency. One cannot measure that type of noise in an audio circuit without signal (or DC) so it often gets missed.
When doing very sensitive electronics the most important thing is to de-rate the resistors to a minimum of 50% of their power rating and preferably less. Also the layout should take into account critical resistors and these should be placed to avoid heating from other components and also have thermal sinking to avoid the resistor getting hot (the cooler the better). Doing this and other techniques will give you a better circuit than just resistor type alone (though carbon are a no no). Also look for automotive parts as they tend to have better temp. specs and performance.
Voltage rating is often overlooked. These days both thin and thick film 0603 smd parts are typically 75V max, before worrying about the meager 100mW power rating.
Putting two resistors in series for critical functions like the feedback resistor reduces voltage swing on each part and any voltage non-linarity
Putting two resistors in series for critical functions like the feedback resistor reduces voltage swing on each part and any voltage non-linarity
I think that all the comments on the noise issue with resistors are valid.
In terms of costs, however, replacing commonly available resistors with say all Vishay resistors can be costly. Of course, if one is doing a manufacturing run then Vishay can provide price breaks but for a single unit.
And certainly, choosing the least noisy type would be preferable to known noisy type resistors - after all, a designer does not want to intentionally use noisy components. But often, design compromises are taken for practical considerations. Where costs are no barrier - to the purchaser - then there's nothing to prevent use of all low-noise resistors.
However, taking cost out of the picture for the moment, I still believe that there needs to be a double-blind test performed with a couple of audio pre-amp/amps where one set has commonly used resistors (other than carbon composition) and the other set has bulk metal foil resistors, and ascertain the results through objective listening tests.
In terms of costs, however, replacing commonly available resistors with say all Vishay resistors can be costly. Of course, if one is doing a manufacturing run then Vishay can provide price breaks but for a single unit.
And certainly, choosing the least noisy type would be preferable to known noisy type resistors - after all, a designer does not want to intentionally use noisy components. But often, design compromises are taken for practical considerations. Where costs are no barrier - to the purchaser - then there's nothing to prevent use of all low-noise resistors.
However, taking cost out of the picture for the moment, I still believe that there needs to be a double-blind test performed with a couple of audio pre-amp/amps where one set has commonly used resistors (other than carbon composition) and the other set has bulk metal foil resistors, and ascertain the results through objective listening tests.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.