A CFA (open or closed loop) is already a current conveyor, CCII
Current conveyor - Wikipedia, the free encyclopedia
Well, the cap will be in the part of the feedback bridge shunting the signal to the ground, for the gain will be 1 at DC. It *increase* the feedback current at DC in the emitter of the input transistor.
It is voltage mode operation , otherwise why the need
of a voltage divider from output to inverting output.?
Do you need two resistors to divide a current ? no , all
you have to do is to increase the FB resistor , the shunt
resistor connected to the inverting intput would have
no role if it was a genuinely current controled input.
The diamond can measure and sound good.
My experience says :
1) that the bias resistors around the buffer should be sub 20 Ohm.
2) bias generator should be low noise and very stable
3) Bias current should be ~1 - 4mA
4) couple the transistors in the diamond buffer very close => SMT
5) The bias generator sets the current in all stages of the amp. Make sure it is ultra stable and does not drift.
6) Avoid the HDAM version (Some calls it a CFP version). It has a peak at ~100MHz. -> affects the stability.
Bonsai said:
From day #1 when this designation appeared for the scheme which the current discussion refers to, CFA was denounced as incongruous.
Well seen (in french, "bien vu").
This goes in parallel with another very convincing argument which is that the difference of a current from a voltage has no meaning.
For a hugely compromised (let's do not forget : an economical) design, it behaves remarkably well.
For the standard blameless, the usual number quoted is 34 dB at 20 kHz.
Can CFAs provide more ?
Is there a fundamental need for having symmetrical maximal slew-rates ? However, if thought to be inadequate, the situation of the standard blameless can be improved, as Samuel Groner has shown.
Like point 1.
Is there a fundamental need for having symmetrical PSRRs ?
Are PSRRs of CFAs better with identical supplies ?
My blameless in use since 1994 has no power supply decoupling of the low power stages. It does not output any hum.
At what value does a maximal slew-rate become decent ?
Lets not get into this again. Easy to explain in a really practical way:
With a VFA, within reason, if you change the total value of the feedback resistor network, but keep the ratio the same, you will not change the fundamental operation of the VFA - response, rise fall time and even distortion will not change.
If you conduct the same experiment with the CFA, the above parameters change, just as theory predicts.
The fact that CFA input is sampling a voltage and that the output voltage is sensed does not make the fundamental operation within the feedback loop voltage controlled, as is the case with a VFA.
Think also about post #57 - the CFA input stage Io is expansive while the LTP is not - it's limited by the tail current.
In their classic guises, you are dealing with two fundamentally different topologies.
its fairly simple but you have to keep the idea that 2 things are being done somewhat "orthogonally" in "CFA" feedback
the feedback divider provides a sample of the output V
and the feedback Z changes the gm of the input stage by adding in series to the -input internal R
changing the gm of the input stage changes the high frequency gain intercept, GBW product - lower feedback Z effectively decompensates the GFA
monolithic VFA op amps today mostly have fixed internal compensation - often for unity gain - only a few fast VFA op amps are offered in internally decompensated versions
the feedback divider provides a sample of the output V
and the feedback Z changes the gm of the input stage by adding in series to the -input internal R
changing the gm of the input stage changes the high frequency gain intercept, GBW product - lower feedback Z effectively decompensates the GFA
monolithic VFA op amps today mostly have fixed internal compensation - often for unity gain - only a few fast VFA op amps are offered in internally decompensated versions
Last edited:
The need for greater precision operation -
The need for greater precision and predictability (and monitization) quickly lead to the adoption of current-mode operation using current conveyors (mirrors). True compl processes still were in their infancy so matching was still an issue;
Translinear circuits have been developed as a next step. it isnt practcal to match devices and IC processes lacked the ability to make closely matched compliments. However, we (the DIY mob/gangs which roam here) can do it and make extreamly low distortion.
Further developments moved to the MOSFET devices. Without an input current, beta is a non-factor. And, with ever lower voltage operation and higher speeds, cancelling techniques have keep the current-mode performance characteristics while increasing speeds/bandwidth.
Thx-RNMarsh
The need for greater precision and predictability (and monitization) quickly lead to the adoption of current-mode operation using current conveyors (mirrors). True compl processes still were in their infancy so matching was still an issue;
Translinear circuits have been developed as a next step. it isnt practcal to match devices and IC processes lacked the ability to make closely matched compliments. However, we (the DIY mob/gangs which roam here) can do it and make extreamly low distortion.
Further developments moved to the MOSFET devices. Without an input current, beta is a non-factor. And, with ever lower voltage operation and higher speeds, cancelling techniques have keep the current-mode performance characteristics while increasing speeds/bandwidth.
Thx-RNMarsh
Last edited:
Richard, you asked what kind of PS to use, I suggested what kind I thik it's suitable and you did not comment it. What is your suggestion?
BR Damir
BR Damir
If matching/complementary is such an issue for CFAs, why not try a different solution? Instead of looking for perfectly matched NPN, or perfectly complementary NPN/PNP devices, why not using "compound" transistors that are more perfect by nature? I mean CFP/Sziklai or super-pair transistors.
dadod --
The absolute value of the supply voltages is not important. However, a power supply design which keeps the pos and neg voltages the same/equal value under dynamic loading conditions relaxes the demand on the amp for rejection characteristics; A zero delta between the pos-neg voltage value under dynamic conditions.
See #8,9 of "References and further reading". I introduced the audio community to A Complementary Push-Pull power supply (TAA 1988). This means and perhaps others (Master-Slave et al) works well.
Thx-RNMarsh
The absolute value of the supply voltages is not important. However, a power supply design which keeps the pos and neg voltages the same/equal value under dynamic loading conditions relaxes the demand on the amp for rejection characteristics; A zero delta between the pos-neg voltage value under dynamic conditions.
See #8,9 of "References and further reading". I introduced the audio community to A Complementary Push-Pull power supply (TAA 1988). This means and perhaps others (Master-Slave et al) works well.
Thx-RNMarsh
Last edited:
SMPS. One well known in this forum is not regulated but offers very good symmetry.
They always win (i know no opposite opinion from people having made the test) , in listening challenges vs Linear ones, even oversized. May-be the RFI/RMI better rejection too ?
possible and without feedback to the driver with very low distortion. See the attached file. Such a driver is enough to add highly linear output stage, such as is used in some models of Denon amp or error correction distortion (ECH) Hawksford.
RM_2013_07_Superdriver.pdf
best regards
Petr
RM_2013_07_Superdriver.pdf
best regards
Petr
The problem of unequal supply voltages is pronounced with power amps with thier unregulated supplies. Anything which can tend to keep the voltages more equal under dynamic conditions will be a benefit to the amp.
There is always a down side.... emi/rfi from smps can be high and difficult to keep out of the amplifier circuit. CMAmps may be more immune to field-coupled noise because of their low Z circuitry.
THx-RNMarsh
On my (limited) experience, SMPS seems to me more immune to RFI/RMI from AC lines: The switching transistor makes one stage more isolation, and the transformer has less coupling capacitor due to its little size. One of the evils of liner PSU is the switching noise of the rectifier. It seems the FET used to switch the AC behave better on this point too ?
Christophe,
I may be wrong on this point but I think what Richard is saying about the EMI/RFI of a smps is not its own inherent immunity to those things passing through them but to their own tendency to produce those things. The radiation of EMI and RFI that is subsequently picked up by the amplifier circuitry through radiation from the smps itself producing those interferences from their own circuit even with no input from an external source.
Richard correct me if I am wrong on that point.
I may be wrong on this point but I think what Richard is saying about the EMI/RFI of a smps is not its own inherent immunity to those things passing through them but to their own tendency to produce those things. The radiation of EMI and RFI that is subsequently picked up by the amplifier circuitry through radiation from the smps itself producing those interferences from their own circuit even with no input from an external source.
Richard correct me if I am wrong on that point.