Christer said:
All resistors are voltage dependent. Quoting some figures from
the fact pages in the Elfa catalogue we have:
carbon comp. 200 - 500 ppm/V
carbon film <100 ppm/V
metal film c:a 1 ppm/V
metal oxide < 10 ppm/V
wire wound c:a 1 ppm/V
They also say (not surprisingly) that this dependency
causes distorsion and that it is often stated as a ratio in dB
between the signal and the third overtone. I have yet to see
such a figure in a datasheet though. (Elfa don't even have
datasheets for the resistors they sell.).
Hardly surprising it causes distortion - I'd no idea the figures for carbon were that bad.
To put some rough numbers on it - supposing we use this as the gain-setting resistor in a feedback network in an amp delivering 10V peak output. If the error is 100ppm/V, on signal peaks the gain could be wrong by 1000ppm - in other words the distortion is something like 0.1% !
Maybe in real life it's not as bad as that - e.g. if the variance only gets to the quoted figure at or near the max working voltage of the resistor. Still, it's certainly something not to be ignored.
Cheers
IH
Don't forget that the figures I quoted was for the direct
voltage dependency only. In many cases we also get an
indirect voltage/current dependecy through heating.
Otherwise, I have also been thinking along your lines that
the feedback resistor must be one of the most crucial ones
in this respect. It could perhaps be interesting to build two
identical amps with carbon resistors, but use a metal film for
the feedback resistor in one of them and then measure the
distorsion.
It may also be interesting to note that when looking for info
on the voltage dependency, I also checked my old introductory
electronics textbook. There I was only told that the voltage
dependency in ordinary resistors is negligible!!! Maybe some
author conisders 500 ppm/V negligible, which could explain
why this phenomenon is seldom observed.
voltage dependency only. In many cases we also get an
indirect voltage/current dependecy through heating.
Otherwise, I have also been thinking along your lines that
the feedback resistor must be one of the most crucial ones
in this respect. It could perhaps be interesting to build two
identical amps with carbon resistors, but use a metal film for
the feedback resistor in one of them and then measure the
distorsion.
It may also be interesting to note that when looking for info
on the voltage dependency, I also checked my old introductory
electronics textbook. There I was only told that the voltage
dependency in ordinary resistors is negligible!!! Maybe some
author conisders 500 ppm/V negligible, which could explain
why this phenomenon is seldom observed.
Hi,
QUOTE]Maybe some
author conisders 500 ppm/V negligible, which could explain
why this phenomenon is seldom observed.[/QUOTE]
This may well be true for industrial applications.
For audio use we should take this voltage depency into account.
For high voltage circuits such as tubed designs it is wise to oversize resistors ( wattage and voltage) as this lowers distortion and noise.
As for NFB resistors, you'd be surprised how audible different resistors become when you try them out in such a loop.
Cheers,
QUOTE]Maybe some
author conisders 500 ppm/V negligible, which could explain
why this phenomenon is seldom observed.[/QUOTE]
This may well be true for industrial applications.
For audio use we should take this voltage depency into account.
For high voltage circuits such as tubed designs it is wise to oversize resistors ( wattage and voltage) as this lowers distortion and noise.
As for NFB resistors, you'd be surprised how audible different resistors become when you try them out in such a loop.
Cheers,
IanHarvey said:
does the above assume that the change in resistance is an instant function of voltage? If for example the change in resistance is mostly driven by heat dissipation, the distortion will happen over a period of time. In this way, distortion wouldn't be nearly as bad.
Maybe Christer can clarify that.
Christer said:
it probably depends on the mechnism under which the voltage / resistance relationship works.
I suspect that the study you had quoted measures resistance changes after a voltage has been applied to the resistor for a period of time. So the source of that would be the temporature coefficient of the resistive material. If that's the case, it should have not caused measurable audio distortions (unless we are talking about really small-sized resistors and slow frequencies).
but it is just my conjecture.
fdegrove said:As for NFB resistors, you'd be surprised how audible different resistors become when you try them out in such a loop.
Cheers,
Jonathan's experiment indicated that it is about 0.003%. Unless you find facts to support otherwise, I tend to think you need some ultra-high resulotion golden ears to hear that kind of distortion.
On a 2nd thought, if you have guys who can hear mica insulation pads from sil-pads, who's to say that they cannot hear 0.003% thd distortion?
fdegrove said:Mind you, not all carbon resistors are the same so I think it's best not to make blanket statements about this.
Cheers,
what carbon resistors lose to metal resistors in the noise department is the contact noise. As contact noise is proportional to the current going through the resistor, sizing up the resistor helps (kind of observed by Janathan's experiment).
However, contact noise is inversely proportional to frequency, and Janathan did not report seeing that so we don't really know what happened there.
Again, some people like that noise characteristics of carbon resistors and actually think carbon resistors sound better. so who knows?
millwood said:
Yes, I did assume that. If the voltage-dependent effects have a time constant much larger than the audio frequency concerned, you are right that the distortion would be lower.
I assumed it wasn't caused by thermal effects, on the grounds that applying say, 1V across a 2Ohm resistor heats it up by a dramatically different degree compared to a 2 Meg resistor, and therefore giving a ppm/V figure without specifying the resistance value is likely to be pointless.
Cheers
IH
I am afraid I have no more details on the voltage dependancy.
I assumed it is instantaneeous and independent of the thermal
effects, but neither of my sources give any further details to
clarify the exact nature of this phenomenon. Come to think of
it, I guess it has to be more or less instantaneous since the
source I quoted from actually said it causes distorsion and is
measured as the ratio between signal and third overtone.
I assumed it is instantaneeous and independent of the thermal
effects, but neither of my sources give any further details to
clarify the exact nature of this phenomenon. Come to think of
it, I guess it has to be more or less instantaneous since the
source I quoted from actually said it causes distorsion and is
measured as the ratio between signal and third overtone.
To answer the "thermal effects" question (I hope) definitively, to quote from http://www.ohmcraft.com/pdf/howToSpecifyaResistor.pdf :
"The Voltage Coefficient of Resistance is the change in resistance with applied voltage. This is entirely different and in addition to the effects of self-heating when power is applied."
Cheers
IH
"The Voltage Coefficient of Resistance is the change in resistance with applied voltage. This is entirely different and in addition to the effects of self-heating when power is applied."
Cheers
IH
More info...
If you're really not bored about voltage coefficients, I've dug out MIL-STD-202:
http://nepp.nasa.gov/DocUploads/1F6AB74B-4517-4AD0-A34813268E75B8EB/MIL-STD-202.pdf
which (if you plough onto Method 309) describes a standard procedure for measuring voltage coefficient.
It specifies the two test voltages should be 0.1x and 1.0x the rated working voltage, and says it shouldn't be applied for more than 0.5 seconds out of every 5 second period.
So we still don't know whether there is a significant time constant to the VCR effect, but we do know that it's neither a thermal effect nor a long-term drift effect.
Cheers
IH
If you're really not bored about voltage coefficients, I've dug out MIL-STD-202:
http://nepp.nasa.gov/DocUploads/1F6AB74B-4517-4AD0-A34813268E75B8EB/MIL-STD-202.pdf
which (if you plough onto Method 309) describes a standard procedure for measuring voltage coefficient.
It specifies the two test voltages should be 0.1x and 1.0x the rated working voltage, and says it shouldn't be applied for more than 0.5 seconds out of every 5 second period.
So we still don't know whether there is a significant time constant to the VCR effect, but we do know that it's neither a thermal effect nor a long-term drift effect.
Cheers
IH
Re: More info...
it seems to me that the intent of the test is to limit the "thermal effect".
BTW, typical thermal effects, per Vishay/Dale, are about 20ppm/v. That's quite small, isn't it?
BTW2, current noise between carbon and metal resistors are roughly the same, from 0.1uv/u to 1uv/u.
IanHarvey said:
it seems to me that the intent of the test is to limit the "thermal effect".
BTW, typical thermal effects, per Vishay/Dale, are about 20ppm/v. That's quite small, isn't it?
BTW2, current noise between carbon and metal resistors are roughly the same, from 0.1uv/u to 1uv/u.
Re: Noise
does your "most all of your statement" include the following?
"BTW2, current noise between carbon and metal resistors are roughly the same, from 0.1uv/u to 1uv/u."
If so, Motchenbacher and Fitchen should really take that issue up with Vishay / Dale as that's what Vishay has published in its datasheets.
jewilson said:
does your "most all of your statement" include the following?
"BTW2, current noise between carbon and metal resistors are roughly the same, from 0.1uv/u to 1uv/u."
If so, Motchenbacher and Fitchen should really take that issue up with Vishay / Dale as that's what Vishay has published in its datasheets.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.