Great! Now lets move on to current-mode operation --- and topologies?
Current mode circuits | Vishwam Gupta - Academia.edu
THx-RNMarsh
Yes, isn't it nice
@Banat, like i wrote,
Voltage to Current FeedBack and
Current to Current FeedBack
Mona
The current mirror does not belong to the input stage.
It is only a load for the current output by the input stage in most CF-opamps, but it is absolutely not mandatory.
It is your opinion.
The base is the non-inverting input.
This emitter is the inverting input which receives the feedback signal, not the signal of the source.
Simulation shows that buffering the signal from the feedback network, just like it happens with a long tail pair input stage, makes the circuit an indisputable VFA. You can see that it has very little effects on the current and the voltage involved in the feedback signals around the device in charge of the subtraction process.
To summarize, a VFA tends to look like a buffered CFA, if we admit that these acronyms have any meaning.
Attachments
Last edited:
Sorry for one more post, but in my opinion that is because of the holy cows and all the ********.
I think that they hatched some weeks ago.
BTW: A so called "Current Mode Circuit" has nothing to do about CFA.
S
"Simulation shows that buffering the signal from the feedback network, just like it happens with a long tail pair input stage, makes the circuit an indisputable VFA. You can see that it has very little effects on the current and the voltage involved in the feedback signals around the device in charge of the subtraction process."
At DC I would expect little or no difference whatsoever. You have in inverting input, a non-inverting input, an amplifier with gain and a feedback network.
What is the point you are trying to make here?
At DC I would expect little or no difference whatsoever. You have in inverting input, a non-inverting input, an amplifier with gain and a feedback network.
What is the point you are trying to make here?
Last edited:
Don't confuse current feedback amplifiers with current output amplifiers - they are not the same thing.
Voltage Output Amplifiers
VFA: Voltage output>voltage sampled>voltage feedback (feedback into hi-z - input)
CFA: Voltage output>voltage sampled>current feedback (feedback into lo-z 'current sink' - input)
Current Output Amplifiers
VFA: Current output>current sampled>voltage feedback
CFA: Current output>current sampled>current output
Voltage Output Amplifiers
VFA: Voltage output>voltage sampled>voltage feedback (feedback into hi-z - input)
CFA: Voltage output>voltage sampled>current feedback (feedback into lo-z 'current sink' - input)
Current Output Amplifiers
VFA: Current output>current sampled>voltage feedback
CFA: Current output>current sampled>current output
Attachments
Yes, it shunts out nested local feedback in the 1'St stage through resistance of the main feedback loop.
I tried to make clear :
- that, whether the feedback network be buffered or not, the device of the input stage in charge of converting to current the difference betwen the signals present at the non-inverting and inverting inputs is governed by one principle only, which is transconductance.
- and, as a consequence, that, M. Kiwanuka is right in contesting the specious concept of Current Feedback term applied to a topology of amplifier input stages.
forr,
instead of delivering the displacement current via the lo z feedback resistor network, you are simply delivering it from the opamp output. The - input is not the feedback buffer + input. It is still the lo z input port at the transistor emitter.
This does not prove that CFA = VFA or that there is no such thing as CFA.
Mikeks is smart and argues pursuasively but his premise is wrong.
instead of delivering the displacement current via the lo z feedback resistor network, you are simply delivering it from the opamp output. The - input is not the feedback buffer + input. It is still the lo z input port at the transistor emitter.
This does not prove that CFA = VFA or that there is no such thing as CFA.
Mikeks is smart and argues pursuasively but his premise is wrong.
Why not? To me forr's circuit including the buffer looks like a valid nullor implementation. Of course you can also look at it as an amplifier with an extra buffer, whatever you find most convenient.
As I pointed out before, in the "current" mode the 1'St stage has a local feedback by current due to a resistance of a feedback loop, in the "voltage" case it is simply shunted. I.e. it is a nested feedback that has different values, while the global feedback by voltage stays the same. Well, almost the same, since local feedback decreases OL gain.
You have not changed the fundamental operation of the input stage compared to a VFA in my view - you have simply hidden it behind a buffer.
Now you see why I view the H bridge as a CFA and not a VFA.
CFA: lo z - input port, current on demand into TIS/TAS input, BW does not change with CLG
VFA: both input ports hi z, peak current into TIS/TAS = Tail current. Constant gain BW in its simplest guise assuming MC
However, as mentioned in a earlier post, in high OLG/LG exemplars of either topology, the behaviors are similar I.e big performance and behavior overlap. But fundamental operation is different - please see CLG gain derivations TI, Intersil et al
There has been a discussion about CFA's input topologies hindering a really low noise approach.
Ovidiu Popa has done it nevertheless:
"The HPS 4.x series had a voltage equivalent noise of well under 0.3nV/rtHz, while the equivalent input current noise was measured under 5pA/rtHz, making them ideal for the most demanding MC cartridges (having low impedances)"
Schematic: HPS 4.2 Schematic
Site: http://www.synaesthesia.ca/LNschematics.html
Ovidiu Popa has done it nevertheless:
"The HPS 4.x series had a voltage equivalent noise of well under 0.3nV/rtHz, while the equivalent input current noise was measured under 5pA/rtHz, making them ideal for the most demanding MC cartridges (having low impedances)"
Schematic: HPS 4.2 Schematic
Site: http://www.synaesthesia.ca/LNschematics.html
Yes - Ovidiu used parallel transistors and a very low gain setting resistor (1 Ohm IIRC) and a c. 40 Ohm feedback resistor. So the net feedback resistance was under 1 Ohm and this made for an exceptionally quiet MC amplifier. Since the MC cart source resistance is in the region of 10 ohms, noise voltage is the dominant noise mechanism, not noise current.
When the application is considered and all the potential noise sources etc, a CFA can do a good job in low noise applications as well - great for MC but I would not use them for MM. Its a case of using a solution that fits the application.
When the application is considered and all the potential noise sources etc, a CFA can do a good job in low noise applications as well - great for MC but I would not use them for MM. Its a case of using a solution that fits the application.
Last edited:
Mind you, that's totally atypical and extreme. Input stage uses 4 ultra low noise discrete devices, biased at the optimum (voltage noise wise) collector current of 12mA each. At this collector current, each transistor has a beta of about 600, which pretty much covers the rather low input current noise. Then the negative input impedance is 1 ohm, which makes the emitter current noise contribution small. I suspect this is though the largest component of the input referred noise.
Completely out of reach for any IC in an usual process. Otherwise, it's a typical "because I can" monster implementation, without any advantages over much simpler, jfet based, approaches.