Can you, please, elaborate this a little, John ?
My thoughts:
To reduce internal TIM, the perfect way is to set all stages faster than the previous one. In order they never run out of current at HF. And reduce the speed of the input signal with a low pass filter before the first stage.
Now, we have to tune the harmonic distortion of each stage, playing with the GM of the input and VAS one in order to get the lowest value with the closed loop. That is easier nowadays with simulations. Of course, those concern are to be kept in mind when you begin with design decisions about the topology of each stage, like the use of perfect current sources, or/and Cascodes. At the end, we can have a nice very simple schematic finishing in a very complex schematic, with a big numbers of active devices. Simple is beautiful is not always true
I believe the major advantage of Current feedback is that the feedback is subtracted from the input signal in the same transistor (less resultant level in this one) and applied in a common base configuration (low impedance and higher bandwidth).
At the end, we are still far from perfection with our actual devices, yet too slow, despite the numbers. Looking at the response curve at the output of the first stage, in closed loop, we can see the signal level start to increase above 1-3 Khz by 6dB/Oct.
Current feedback topologies show ~ one octave more of linearity in the highs (with the same devices and internal topologies) , reason why some like the 'fluid' way they sound ?
Well, CFA had one disadvantage: while you don't use a differential input, you have less PSRR/CMR. But we can keep an input differential stage, a nice way to get a symmetrical input and keep the VFB advantages on the rejection issue. Anyway, both keep advantages of perfect PSU filtration on each stages, and ideal current sources where needed.
Last edited:
Question: What causes the need for high PSRR ?
Q: Would it be best to solve the problem at it's source... the PS?
Thx-RNMarsh
About Wideband Amplifiers...CFA...VFA...Folded Casc..
Hi all,
this book:
Wideband Amplifiers - Peter Starič, Erik Margan - Google Livros
I know that many of you do not need beyond what is in this book but it's tip here it seems to treat these topologies with their pros and cons. i hope was helpful somehow.
Regards
Hi all,
this book:
Wideband Amplifiers - Peter Starič, Erik Margan - Google Livros
I know that many of you do not need beyond what is in this book but it's tip here it seems to treat these topologies with their pros and cons. i hope was helpful somehow.
Regards
An excellent book, even gives a good introduction to the mathematics involved. A lot of wisdom from the good old days at Tektronix is between the covers, among other things. Not an audio-oriented book per se, and don't anticipate construction project details
Brad
Richard, i love the way you ask questions while you know the responses, may-be better than everybody else
So I'll try to answer for others.
First any ripple of PSU will create a parasitic signal in the output, modulating the signal. This noise will be canceled partially by the feedback, but we don't need this unwanted signal in the feedback loop, where it will add distortion products, did we ?
About canceling the ripple at the source, how to answer ? Of course, we have advantages to have a very flat and noiseless rails from power supply.
But the internal impedance is not 0, at least because the inductances of the PSU rails/wires. Even if there is few ripple residue from AC plug, the rails will be modulated by the current counsuption of the power stages while they follow the input signal. Specially during high transients (HF) where the efficiency of the power amp is reduced.
Filtering high currents is expensive (big caps etc). You can have more efficiency in a more economical way by filtering each stage individually.
As habit, the response to reach perfection will be: Do the best (means the two together).
Now, a personal remark. Most of audiophiles think that separated power supply (Dual mono amps) is better. I think the contrary. During transient in one channel, the output level can be reduced, because the PSU will not provide enough current. This will move the stereo image if the other channel does not suffer the same.. Keeping this equal to the both channel by the use of a single supply to the both channels, and you stereo image will stay fixed. Of course cares have to be kept to avoid ground loops.
Any comment or contradictory argues are welcome, of course.
Richard, i love the way you ask questions while you know the responses, may-be better than everybody else
About canceling the ripple at the source, how to answer ? Of course, we have advantages to have a very flat and noiseless rails from power supply.
But the internal impedance is not 0, at least because the inductances of the PSU rails/wires. Even if there is few ripple residue from AC plug, the rails will be modulated by the current counsuption of the power stages while they follow the input signal. Specially during high transients (HF) where the efficiency of the power amp is reduced.
Question:
For high-end, local stage regulation is cheap to do.... rather than distributed buss arrangment (with high Ls and cross-talk).
However for output stages, in particular:
... what form does the modulated (noise?) signals on the PS rails have? Asumming a + and - PS rails.
[is it mostly common-mode or differential-mode]
Thx-RNMarsh
Last edited:
Here, the advantage of advanced fast current sources, as they both reduce distortion and add PSRR.
Well any change of the middle point will modulate the output (Noise). Any change of the total V will add distortion, as the PSRR is not infinite in a emitter follower stage ? Same thing for the transconductance of a FET ?
Anybody who had played with battery/ACpower on an headphone amp had notified how much the psu can have influence on the sound. (May-bee ground noises issues too)
Last edited:
Not so fast! John, do you realise you have the same birthay date as Ed Dell, the founder of Audio Amateur, now AudioXpress? So, Happy Birthday to both of you!
He's making 90 even, so you'r pretty junior to him!
jan
Erik Margan is a pretty smart guy. Keeps the LHC at Geneva running, among others. And has an article about balanced audio stages in the upcoming Linear Audio Vol 5.
jan
Peter is as well, but has not gotten round to learning much English I gather.
Well, lets say the answere is - differential-mode, primarily. Then what do you do at the PS to cancel D-M?
Thx-RNMarsh
I can understand why a faster input stage can compensate a little the lack of speed of the output stage, allowing higher HF levels to compensate the output slowliness, but, obviously, it will increase high order distortions thus not address TIM. The medicine is more effective as the disease is small, don't you think ?
Sorry, my poor English+ my allergy to acronyms, what means D-M ?
Last edited:
- Status
- Not open for further replies.