Pjotr said:
He must be making his own pigments then. I am an amatuer painter
and I can assure you there is no pigment called titanium blue,
at least not for artist use (maybe there is some industrial pigment
called so). I could give you a long list of other blue pigments
that are used, if you wish. 🙂
Christer said:
Absolutely right Christer 🙂
Tianium dioxide is a common known white pigment. But there is a commercial available oil paint (don’t know the brand) with a blue colour called “titanium blue” which is a very deep shiny blue. Probably they do not use titanium for the blue pigment but just named the colour “Titanium blue”. That oil paint is rather expensive and they keep the real pigment(s) secret. Hmm… looks like audio 😀
Cheers 😉
millwood said:
We could do that, but I am sure somebody (me? no!) will immediately point out that electrons move the ooposite way of the current, so the electrons are actually send forward. Huh? Why do we then call it feedback? The confusion....😱
Jan Didden
cfb op-amp makers are right
Folks,
There hasn't been, as far as I know, a universal consensus on what constitutes "current feedback", or CFB. In 1988 I acquired a 1964 General Electric transistor catalog/handbook from a colleague who retired. I kept it for historical reference. Another oldie but goodie textbook in my collection is "Pulse, Digital, and Switching Circuits", by Millman and Taub, c. 1965. The 1964 GE manual describes CFB as what I would call "series feedback". The 1965 Pulse & Digital text describes CFB as a feedback signal derived by sensing the output current with a low valued resistor in series with the load. Only written a year apart, these two references don't even agree.
As far as "low impedance VFB" goes, I beg to differ. The reason that the semiconductor industry chose the term "CFB" makes sense to me. With a conventional VFB (high inverting node input impedance), the closed loop gain and the bandwidth are determined by the feedback *voltage*, not the feedback current. In the non-inverting configuration with input resistance Rin equal to feedback resistance Rfb, the closed loop gain is 2. This value of 2 is the reciprocal of "beta", the feedback factor, which is the fraction of the output voltage that is returned to the input, or "fed back". Forgive me for throwing this in, but "beta" also is used as the BJT forward current gain factor. Should we be upset about this double usage? Anyway, if we reduce Rfb to zero, and/or Rin to an open, we now have a follower (closed loop gain = 1). The bandwidth is now about twice what it was with a gain of 2. The beta for a folower is 1. With Rfb = 4*Rin, closed loop gain is 5 since beta is 0.20, and the bandwidth is 0.20 that of a follower. Suppose we changed the values of Rin & Rfb, but kept the ratio the same. The bandwidth, BW, is still proportional to beta, or the ratio of the resistors. The feedback current changes with value changes, but not the voltage. If Rin = 10 kohm = Rfb, Vin = 1.0 volt, then Vout = 2 volts, and the feedback current is 0.10 mA. Now with Rin & Rfb reduced to 1.0 kohm, the feedback current increases from 0.10 to 1.0 mA, but the gain is still 2, and the BW is unchanged. Clearly, the gain-bandwidth product is proportional to feedback *voltage*, NOT the feedback current, hence the semi industry calls these op-amps "VFB", rightly so.
Now looking at the "CFB" variety of op-amp (low inverting node input impedance), let Rin = 10 kohm = Rfb. The gain is 2, with a specific bandwidth BW, and beta = 0.50, with 0.10 mA feedback current. Now, change Rin & Rfb to 1.0 kohm. The gain remains at 2 , but the BW is 10 times greater due to the 10 times lower feedback resistance. What changed, feedback voltage or feedback current? The feedback voltage is identical in both cases. The feedback current increased from 0.1 to 1.0 mA. In the case with lower Rfb the feedback current increased.
The bandwidth is directly proportional to the feedback current, NOT the feedback voltage. So the semi industry correctly identified this topology as "CFB", which makes perfect sense to me. Holding Rfb at 1.0 kohm, and opening Rin results in a follower (gain of 1). The feedback current, however is still 1.0 mA, and the feedback voltage increased form 0.50 to 1.0 V, but the BW remains unchanged. Again, the BW is proportional to feedback current, not feedback voltage. The gain changes with feedback voltage, by design. Hence, we obtain high BW independent of gain, the whole idea behind CFB.
The CFB nomenclature specifically describes the relation between bandwidth and feedback current. It does not mean that the feedback quantity itself is the actual output current, or a sensed facsimile of the same. I see no reason to argue with or slam the semi industry. I understand what is meant by the phrase "CFB" by examining the context where it is used. Their app notes explain this in great detail, and I find them very informative.
As far as transfer functions go, a resistive feedback network (or R-C) has four possible transfer functions, Iout/Iin, Iout/Vin, Vout/Iin, and Vout/Vin. Two are dimensionless ratios, one is transimpedance, one is transadmittance. A transfer function is a ratio, and cannot be simply a "voltage", or a "current".
Those of you in the minority who insist that the op-amp circuit designers and field application engineers for the semi industry have it all wrong, and that you know better, please forgive me for asking these two rather pointed questions.
Why aren't YOU the ones that are designing the op-amps and writing the app notes???!!! Have the op-amp producers hired the WRONG people???!!!
Best regards to all.
Folks,
There hasn't been, as far as I know, a universal consensus on what constitutes "current feedback", or CFB. In 1988 I acquired a 1964 General Electric transistor catalog/handbook from a colleague who retired. I kept it for historical reference. Another oldie but goodie textbook in my collection is "Pulse, Digital, and Switching Circuits", by Millman and Taub, c. 1965. The 1964 GE manual describes CFB as what I would call "series feedback". The 1965 Pulse & Digital text describes CFB as a feedback signal derived by sensing the output current with a low valued resistor in series with the load. Only written a year apart, these two references don't even agree.
As far as "low impedance VFB" goes, I beg to differ. The reason that the semiconductor industry chose the term "CFB" makes sense to me. With a conventional VFB (high inverting node input impedance), the closed loop gain and the bandwidth are determined by the feedback *voltage*, not the feedback current. In the non-inverting configuration with input resistance Rin equal to feedback resistance Rfb, the closed loop gain is 2. This value of 2 is the reciprocal of "beta", the feedback factor, which is the fraction of the output voltage that is returned to the input, or "fed back". Forgive me for throwing this in, but "beta" also is used as the BJT forward current gain factor. Should we be upset about this double usage? Anyway, if we reduce Rfb to zero, and/or Rin to an open, we now have a follower (closed loop gain = 1). The bandwidth is now about twice what it was with a gain of 2. The beta for a folower is 1. With Rfb = 4*Rin, closed loop gain is 5 since beta is 0.20, and the bandwidth is 0.20 that of a follower. Suppose we changed the values of Rin & Rfb, but kept the ratio the same. The bandwidth, BW, is still proportional to beta, or the ratio of the resistors. The feedback current changes with value changes, but not the voltage. If Rin = 10 kohm = Rfb, Vin = 1.0 volt, then Vout = 2 volts, and the feedback current is 0.10 mA. Now with Rin & Rfb reduced to 1.0 kohm, the feedback current increases from 0.10 to 1.0 mA, but the gain is still 2, and the BW is unchanged. Clearly, the gain-bandwidth product is proportional to feedback *voltage*, NOT the feedback current, hence the semi industry calls these op-amps "VFB", rightly so.
Now looking at the "CFB" variety of op-amp (low inverting node input impedance), let Rin = 10 kohm = Rfb. The gain is 2, with a specific bandwidth BW, and beta = 0.50, with 0.10 mA feedback current. Now, change Rin & Rfb to 1.0 kohm. The gain remains at 2 , but the BW is 10 times greater due to the 10 times lower feedback resistance. What changed, feedback voltage or feedback current? The feedback voltage is identical in both cases. The feedback current increased from 0.1 to 1.0 mA. In the case with lower Rfb the feedback current increased.
The bandwidth is directly proportional to the feedback current, NOT the feedback voltage. So the semi industry correctly identified this topology as "CFB", which makes perfect sense to me. Holding Rfb at 1.0 kohm, and opening Rin results in a follower (gain of 1). The feedback current, however is still 1.0 mA, and the feedback voltage increased form 0.50 to 1.0 V, but the BW remains unchanged. Again, the BW is proportional to feedback current, not feedback voltage. The gain changes with feedback voltage, by design. Hence, we obtain high BW independent of gain, the whole idea behind CFB.
The CFB nomenclature specifically describes the relation between bandwidth and feedback current. It does not mean that the feedback quantity itself is the actual output current, or a sensed facsimile of the same. I see no reason to argue with or slam the semi industry. I understand what is meant by the phrase "CFB" by examining the context where it is used. Their app notes explain this in great detail, and I find them very informative.
As far as transfer functions go, a resistive feedback network (or R-C) has four possible transfer functions, Iout/Iin, Iout/Vin, Vout/Iin, and Vout/Vin. Two are dimensionless ratios, one is transimpedance, one is transadmittance. A transfer function is a ratio, and cannot be simply a "voltage", or a "current".
Those of you in the minority who insist that the op-amp circuit designers and field application engineers for the semi industry have it all wrong, and that you know better, please forgive me for asking these two rather pointed questions.
Why aren't YOU the ones that are designing the op-amps and writing the app notes???!!! Have the op-amp producers hired the WRONG people???!!!
Best regards to all.
mikeks said:
Yes. By your logic I could argue that electrical current cannot
be called current, because the word "current" already has
another meaning synonymous or similar to the word "now".
Surely you wouldn't agree with that conclusion?
Pjotr said:
I assume you are not referring to a paint for artist use. I have
not seen any such thing as titanium blue from any of the major
manufacturers and it is not mentioned in any of my handbooks.
If you actually know of such a paint, I would be curious to know
the brand since it must be some small and obscure brand that
not many artists use. Alternatively, it is some cheap brand
or studio quality series paint. They sometimes make up
non-standard names for such paint. Perhaps it is just a mix
of titanium white and some blue pirgment, like phtalo blue.
For instance, Winsor & Newton even has an artist quality
colour called cadmium green, but it is a made up name and is
really a mix of cadmium yellow and phtalo green.
If it is an industrial pigment, then it is another thing. It might
then be a pigment that is not suitable for artist use for one
reason or another.
Thanks Claude!
What a detailed but very understandable post explaining the difference between current feedback and voltage feedback amplifiers. Almost all of the data sheets on current feedback op amps outline the optimum the relative independence of bandwidth and voltage gain when compared with voltage feedback amps. The designations make perfect sense when looking the terminology from the input terminals of the amp where the amp feedback signal is sensed. The VFB amp's open loop voltage gain creates a small error voltage between the noninverting input and the inverting input; while the CFB amp amp's current to voltage gain or transimpedance (given in ohms) creates a small error current between the noninverting input and the inverting input. I seems clear the anyone that knows that looking
into the base of a transistor with the emitter grounded, the impedance is very much greater than looking into the emitter of a transistor with the base grounded.
Knowing the inverting input is trying to follow the voltage at the noninverting input through the action of the open loop gain and feedback, the feedback terminal sees a low impedance open loop as well closed loop and the error signal is a current. The same principle applies with tubes and FETs. I am really perplexed that this whole thing became an issue. The only reasons I can think of are a deliberate effort to cause confusion for fun (or ego), or a lack of understanding two principles above. This should not be confusing and I think your explanation is clear to people with a working knowledge of electronics and doesn't required an Electrical Engineering degree to understand. Thank you for your contribution to the forum and I hope it doesn't invite the wrath of those who don't want others to understand the different topologies, and the rational for their naming. Of couse it probably will.
http://www.analog.com/UploadedFiles/Obsolete_Data_Sheets/270284AD846.pdf
http://www.analog.com/UploadedFiles/Data_Sheets/372662099AD811_d.pdf
What a detailed but very understandable post explaining the difference between current feedback and voltage feedback amplifiers. Almost all of the data sheets on current feedback op amps outline the optimum the relative independence of bandwidth and voltage gain when compared with voltage feedback amps. The designations make perfect sense when looking the terminology from the input terminals of the amp where the amp feedback signal is sensed. The VFB amp's open loop voltage gain creates a small error voltage between the noninverting input and the inverting input; while the CFB amp amp's current to voltage gain or transimpedance (given in ohms) creates a small error current between the noninverting input and the inverting input. I seems clear the anyone that knows that looking
into the base of a transistor with the emitter grounded, the impedance is very much greater than looking into the emitter of a transistor with the base grounded.
Knowing the inverting input is trying to follow the voltage at the noninverting input through the action of the open loop gain and feedback, the feedback terminal sees a low impedance open loop as well closed loop and the error signal is a current. The same principle applies with tubes and FETs. I am really perplexed that this whole thing became an issue. The only reasons I can think of are a deliberate effort to cause confusion for fun (or ego), or a lack of understanding two principles above. This should not be confusing and I think your explanation is clear to people with a working knowledge of electronics and doesn't required an Electrical Engineering degree to understand. Thank you for your contribution to the forum and I hope it doesn't invite the wrath of those who don't want others to understand the different topologies, and the rational for their naming. Of couse it probably will.
http://www.analog.com/UploadedFiles/Obsolete_Data_Sheets/270284AD846.pdf
http://www.analog.com/UploadedFiles/Data_Sheets/372662099AD811_d.pdf
Once we have arrived at a definitive understanding of current v. voltage feedback, what next?
Mars? Alpha Centauri? Betelgeuse?
Can meaningful connections be drawn between the topology and sonics?
Why?
Indeed, How?
Cheers,
Hugh
Mars? Alpha Centauri? Betelgeuse?
Can meaningful connections be drawn between the topology and sonics?
Why?
Indeed, How?
Cheers,
Hugh
review of Dr. Cherry paper
Dear friends,
I already have the Cherry paper, thanks to Jan for that. It is a very impressive piece of work, thoroughly examining the four classical amplifier/feedback topologies. I will keep it as a reference and revisit it from time to time. However, I found a couple issues with it that bear mentioning.
On p337-338, he writes:
"Stripped of its input level shifters and complementary first stage, and with the current mirrors and voltage follower replaced by the simplest inverting amplifier (a single common-emitter transistor), the current-feedback amplifier reduces to Fig. 6. A generation ago this circuit was called a voltage-feedback pair."
The following sentence is where Dr. Cherry and I differ:
"It differs from Fig. 2a only in that the current flowing into the left-hand side of the feedback network is the input current multiplied by the gain of the first transistor, as distinct from the input current itself."
Folks, that is a pretty BIG difference. The common emitter amplifier with its current gain sinks and sources current to the feedback resistors, Rf1 and Rf2. In a conventional VFB op-amp, only the op-amp output sinks/sources the feedback resistor current. The input current provides very little of the resistor current. Dr. Cherry acknowledges this on p336, left-hand side, center of page:
"The second [bracket] in eqn. 14a includes forward leakage of signal through a real feedback network. This can almost always be neglected, because the current amplification factor B of the amplifier without feedback is likely to be a large number whereas the feedback factor Rf1/(Rf1+Rf2) cannot be greater than unity."
I totally concur with the above statement. Therefore the two circuits greatly differ. Dr. Cherry firmly acknowledges (I agree with him on this point) that a conventional VFB input stage cannot source/sink substantial currents to the feedback resistors. As we know, of course, a CFB input stage can and does source/sink current via the CE amp. Dr. Cherry said that the two circuits are identical EXCEPT for the CE. But it is the CE that provides the current for the feedback resistor pair which gives this topology its exceptional speed! The CE is extremely important! The operation of a CFB differs greatly from a VFB. A CFB CANNOT be modeled as a "low-impedance" VFB, or "loaded down" VFB. The VFB has but one source, the output stage, to provide current to the feedback resistor pair, whereas the CFB has two sources, the CE-buffered input stage, AND the output stage.
On p338, Dr. Cherry writes:
"If the feedback resistors are far removed from the optimum, loop gain cannot remain constant as overall gain is varied. Fundamentally this is the explanation for circuits (such as the so-called current-feedback amplifier) in which bandwidth remains constant as gain is varied. However, the loop gain is less than it could have been for the same combination of transistors and overall gain."
I agree with the above, since we can't get something for nothing, tradeoffs being inevitable. However some at this forum have intepreted the above as meaning that the CFB is a "greatly compromised" VFB, and that the bandwidth is constant vs. gain, because it has already been degraded.
If that was the case, how do they explain how the CFB is FASTER at all gains vs. the VFB. If the CFB bandwidth is degraded by low-valued, less than optimum feedback resistors (loaded down), then the original bandwidth would have to be even faster yet. How do you degrade a VFB by loading it down, and end up with MORE BANDWIDTH at all gains??? It makes no sense.
Every argument that attempts to model a CFB as merely a "low-impedance" or "loaded down" VFB, collapses under scrutiny.
It looks like the folks at TI, NS, LT, AD, etc. who design op-amps are quite capable, and know what they are talking about. Their app notes explain it very well, and those who diss them have so far shown absolutely nothing definitive to back up their criticism. Best wishes.
Dear friends,
I already have the Cherry paper, thanks to Jan for that. It is a very impressive piece of work, thoroughly examining the four classical amplifier/feedback topologies. I will keep it as a reference and revisit it from time to time. However, I found a couple issues with it that bear mentioning.
On p337-338, he writes:
"Stripped of its input level shifters and complementary first stage, and with the current mirrors and voltage follower replaced by the simplest inverting amplifier (a single common-emitter transistor), the current-feedback amplifier reduces to Fig. 6. A generation ago this circuit was called a voltage-feedback pair."
The following sentence is where Dr. Cherry and I differ:
"It differs from Fig. 2a only in that the current flowing into the left-hand side of the feedback network is the input current multiplied by the gain of the first transistor, as distinct from the input current itself."
Folks, that is a pretty BIG difference. The common emitter amplifier with its current gain sinks and sources current to the feedback resistors, Rf1 and Rf2. In a conventional VFB op-amp, only the op-amp output sinks/sources the feedback resistor current. The input current provides very little of the resistor current. Dr. Cherry acknowledges this on p336, left-hand side, center of page:
"The second [bracket] in eqn. 14a includes forward leakage of signal through a real feedback network. This can almost always be neglected, because the current amplification factor B of the amplifier without feedback is likely to be a large number whereas the feedback factor Rf1/(Rf1+Rf2) cannot be greater than unity."
I totally concur with the above statement. Therefore the two circuits greatly differ. Dr. Cherry firmly acknowledges (I agree with him on this point) that a conventional VFB input stage cannot source/sink substantial currents to the feedback resistors. As we know, of course, a CFB input stage can and does source/sink current via the CE amp. Dr. Cherry said that the two circuits are identical EXCEPT for the CE. But it is the CE that provides the current for the feedback resistor pair which gives this topology its exceptional speed! The CE is extremely important! The operation of a CFB differs greatly from a VFB. A CFB CANNOT be modeled as a "low-impedance" VFB, or "loaded down" VFB. The VFB has but one source, the output stage, to provide current to the feedback resistor pair, whereas the CFB has two sources, the CE-buffered input stage, AND the output stage.
On p338, Dr. Cherry writes:
"If the feedback resistors are far removed from the optimum, loop gain cannot remain constant as overall gain is varied. Fundamentally this is the explanation for circuits (such as the so-called current-feedback amplifier) in which bandwidth remains constant as gain is varied. However, the loop gain is less than it could have been for the same combination of transistors and overall gain."
I agree with the above, since we can't get something for nothing, tradeoffs being inevitable. However some at this forum have intepreted the above as meaning that the CFB is a "greatly compromised" VFB, and that the bandwidth is constant vs. gain, because it has already been degraded.
If that was the case, how do they explain how the CFB is FASTER at all gains vs. the VFB. If the CFB bandwidth is degraded by low-valued, less than optimum feedback resistors (loaded down), then the original bandwidth would have to be even faster yet. How do you degrade a VFB by loading it down, and end up with MORE BANDWIDTH at all gains??? It makes no sense.
Every argument that attempts to model a CFB as merely a "low-impedance" or "loaded down" VFB, collapses under scrutiny.
It looks like the folks at TI, NS, LT, AD, etc. who design op-amps are quite capable, and know what they are talking about. Their app notes explain it very well, and those who diss them have so far shown absolutely nothing definitive to back up their criticism. Best wishes.
Claude Abraham said:
Class-AB operation at the feedback node......
Claude Abraham said:
I do not recall saying this.... 🙄 ....
In so-called 'current' feedback, 3-Db bandwidth appears constant with changes in closed loop gain only because first stage gain falls (together with loop transmission!!) with falling closed loop gain.......
Carefull reading of Cherry should reveal that this effect is somewhat dubious, as the overall bandwidth defined by non-dominant singularities remains unchanged......(section 5)
...hardly surprising, as one is merely redistributing the location of these singularities by making foward path gain a function of the feedback factor......
Claude Abraham said:
Well claude, you do not end up with more bandwidth at all gains with so-called current feedback either.....
It's not so much the loading down, as the reduction in the first stage's contribution to foward path gain that is the key.....
I can design a voltage amplifier in two minutes with a foward path bandwidth extending right across the audio band, without using the feedback network to effect this result, as is the case with so-called 'current' feedback....
Merely use a resistively loaded diff. stage with adequate degeneration to reduce Gm, and use double-pole compensation....
The key difference here of course, is that with the later you can not vary first stage gain with the feedback network, while with so-called 'current' feedback, this is key....
MalichiConstant said:
No...on all counts...This is actually very...very elementary linear electronics... 🙄
....input stimulus applied to base of transistor, with attenuated, in-phase copy of output voltage applied to its emitter....
Result? The transistor amplifies the difference between the input voltage and feedback voltage...
Result? This difference is expressed as the transistors base-emitter voltage, which provokes a change in said transistor's collector current.........etc...etc....
....said transistor DOES NOT amplify a difference in current...Period.
This is true of both voltage, and so-called 'current' feedback...the later is really just low impedance voltage feedback...
MalichiConstant said:
For your information, the base of the transistor to which feedback is applied is not grounded.....
http://diyaudio.com/forums/showthread.php?postid=423668#post423668
MalichiConstant said:
speak for yourself...in respect of 'confusion', 'ego'...etc....
http://diyaudio.com/forums/showthread.php?postid=422735#post422735
http://diyaudio.com/forums/showthread.php?postid=422607#post422607
Moreover..
...don't ignore this...:
http://www.anasoft.co.uk/EE/currentfeedbackmyth/currentfeedbackmyth.html
...don't ignore this...:
http://www.anasoft.co.uk/EE/currentfeedbackmyth/currentfeedbackmyth.html
We saw it the first time, bub..........
You must enjoy seeing your name in lights by mentioning it again.
But enough of you........
Claude:
I will never utter another obscene word again when I drive through Ohio. Even though the gasoline (petrol in "Mikey-speak") prices are obscene.
Thanks...........
Jocko
You must enjoy seeing your name in lights by mentioning it again.
But enough of you........
Claude:
I will never utter another obscene word again when I drive through Ohio. Even though the gasoline (petrol in "Mikey-speak") prices are obscene.
Thanks...........
Jocko
Re: Moreover..
Mike,
do you agree with the whole content of that page, eg. that the mentioned/drawed CFB schematic is a CFB amplifier topology?
Cheers 😉
mikeks said:
Mike,
do you agree with the whole content of that page, eg. that the mentioned/drawed CFB schematic is a CFB amplifier topology?
Cheers 😉
- Status
- Not open for further replies.