I am not convinced that it is so much that people "like colour", in the sense we are inherently pre-disposed to it in preference to accuracy (ignoring, for a moment, what that is supposed to be). I think that most people in this hobby would agree that it takes more than casual money to approach good sound quality while at the same time the best gear remains less than perfect (even if only marginally less).
Given the hard limit of the wallet even the objective listener (assuming such a beast exists) has no choice but to choose their poison based on personal taste. Nor is this an unreasonable approach to the problem.
And since the mind prefers easy answers to complex dilemmas (hard thinking actually is tiring, so our own biological systems are designed to avoid it) it's difficult to consciously over-ride what the brain has already decided is best for us. You are never going to convince the general public that choosing an amplifier demands they work at it, lighting up the MRI image like Christmas in Time's Square.
Critical listening is hard work, whether the subject is good at it or not. How much energy does it take to go "yep, that mp3 sounds the same as the CD to me. Can I go now? I need to pick up the kids from daycare."
The "lunatic fringe audiophile" might be willing to invest more in the exercise, but if she owns a system, she spends most of her time listening to that system. How can that not shape her unconscious preference for certain sound qualities? The brain is lazy and not normally in the business of making us miserable by pointing out flaws in the things we choose to enjoy unprompted. We have to do that ourselves ;-)
Last edited:
That a very nice schematic but I don't understand how it works.
The input difference stage have + and - input, the upper and lower mosfet pair are biased by constant current source. It is without feedback loop, so I think the output from this stage should be peak square wave when sine wave as the input.
Can somebody give me explanation how it works.
Hi,
It is easy to make such claim. It is of course totally indefensible, as local feedback (more precisely degeneration) operates very different to global looped feedback and Error Correction does not need to include any feedback component at all, it can be pure and completely feed forward.
Hence the results obtained from any of said methods vary substantially.
To simply lump in one bucket as defence against suggestions that global looped feedback may not be all that good an idea lacks not only entertainment value and ingenuity. Worst of it, it is completely lacking of any proof for the contention.
Whatever mathematical proof I have seen presented invariably reduces the equations to reductio ad absurdum, simply declaring any number of significant terms irrelevant (again sans any proof that they in actuality are irelevant) and points to said resulting reduced equations as "proof" that all these different forms of distortion reduction are the same.
By the same method of course one may provide provide that a square, triangle and circle of identical area are really the same, as the area is the same and the rest of the differences may be safely ignored...
Ciao T
It is easy to make such claim. It is of course totally indefensible, as local feedback (more precisely degeneration) operates very different to global looped feedback and Error Correction does not need to include any feedback component at all, it can be pure and completely feed forward.
Hence the results obtained from any of said methods vary substantially.
To simply lump in one bucket as defence against suggestions that global looped feedback may not be all that good an idea lacks not only entertainment value and ingenuity. Worst of it, it is completely lacking of any proof for the contention.
Whatever mathematical proof I have seen presented invariably reduces the equations to reductio ad absurdum, simply declaring any number of significant terms irrelevant (again sans any proof that they in actuality are irelevant) and points to said resulting reduced equations as "proof" that all these different forms of distortion reduction are the same.
By the same method of course one may provide provide that a square, triangle and circle of identical area are really the same, as the area is the same and the rest of the differences may be safely ignored...
Ciao T
Hi,
Well, I am not going to do the PCB (maybe AlexMM can?), but here is a topology that performs very well in (mostly) open loop operation:
Report page
As can be seen, 0.1% THD at full power and 0.004% THD at 10 Watt as well as < 0.1 Ohm Output impedance are excellent for an Amplifier that omits global feedback loops.
Yet it is fairly trivial to adjust the circuit for nearly any desired bandwith and gain and to apply a feedback loop. I estimate the frontend GM at 9mA/V if operated with a 100 ohm source (lower feedback) resistor instead of 270 Ohm and the circuit can likely attain a DC gain of over 90dB if the VAS load is removed. So up to 64dB global loop feedback may be applied in this case, giving the same 26dB Gain as the original open loop circuit.
As it stands, it should be noted that the "inverted darlington" (aka FetSziklai) output stage is essentially a Pass F5 padded to unity gain and thus having around 42dB NFB with an 8 Ohm load and around 36dB NFB with a 4 Ohm Load, with an estimated open loop bandwidth of around 200KHz.
So the Amplifiers output stage alone contains enough negative feedback to satisfy the Bruno Putzney Requirement and thus the whole Amplifier may not be considered "Zero Feeedback" in the classic sense, yet it can operate without the Input and VAS stages enclosed within a a feedback loop...
It should make for an interesting experiment to build this amplifier and to compare it in "open loop" and "closed loop" mode. I have something similar in mind, BTW...
Ciao T
Well, I am not going to do the PCB (maybe AlexMM can?), but here is a topology that performs very well in (mostly) open loop operation:
Report page
As can be seen, 0.1% THD at full power and 0.004% THD at 10 Watt as well as < 0.1 Ohm Output impedance are excellent for an Amplifier that omits global feedback loops.
Yet it is fairly trivial to adjust the circuit for nearly any desired bandwith and gain and to apply a feedback loop. I estimate the frontend GM at 9mA/V if operated with a 100 ohm source (lower feedback) resistor instead of 270 Ohm and the circuit can likely attain a DC gain of over 90dB if the VAS load is removed. So up to 64dB global loop feedback may be applied in this case, giving the same 26dB Gain as the original open loop circuit.
As it stands, it should be noted that the "inverted darlington" (aka FetSziklai) output stage is essentially a Pass F5 padded to unity gain and thus having around 42dB NFB with an 8 Ohm load and around 36dB NFB with a 4 Ohm Load, with an estimated open loop bandwidth of around 200KHz.
So the Amplifiers output stage alone contains enough negative feedback to satisfy the Bruno Putzney Requirement and thus the whole Amplifier may not be considered "Zero Feeedback" in the classic sense, yet it can operate without the Input and VAS stages enclosed within a a feedback loop...
It should make for an interesting experiment to build this amplifier and to compare it in "open loop" and "closed loop" mode. I have something similar in mind, BTW...
Ciao T
Just subjective practical observation. All schematics, does not matter No GNFB or with deep GNFB, if an output signal current goes through the PS electrolytics, aquire rather strong sound colour of those electrolytics. So, the experiment No GNFB - deep GNFB will be strongly contaminated.
Substantial sound differences among SS amps are produced by PS, snubbers etc.
Good sounding MBL 9007 in monoblock mode has very special PS, rather than especially low THD. IMHO.
Substantial sound differences among SS amps are produced by PS, snubbers etc.
Good sounding MBL 9007 in monoblock mode has very special PS, rather than especially low THD. IMHO.
Hi,
Yes, the problem is greater if the same electrolytic capacitor must both close the signal current loop AND at the same acts as the main PSU Reservoir Cap, cross-contamination of the currents is unavoidable.
For my latest Solid State Amp Project I had made a batch of very special chokes, with a very high Al (self inductance) Manganese/Iron Core and 10A DC rating and around 1mH Inductance.
The intention is to use a pair of PSU Capacitors (Elna "For Audio" 18,000uF/71V) for each rail separated by a pair of these chokes with the "ground" only established by joining the two fully separate supplies at the nominal star ground.
There will also be snubbers using Elna Silmic Cap's and very large value (90uF in 15uF/160V Electrocube Miniature Cap's) film bypass capacitor on the second PSU capacitor.
Together with an all lateral fet compound feedback pair output stage (much better PSRR than common outputs) this should go a long way addressing the problem.
In other similar applications I have instead banked on "semi-regulated" supplies and RC filter chains, which can also help a lot to clean up this set of problems...
Yet if I compare these approaches with the power supplies commonly found in Solid State Amplifiers then we have indeed much left to go with the common designs.
The Amplifiers front-end incidentally will use CCS fed shunt regulators and of course CLC filtered supplies prior to the Shunts.
I'd love to make a Class A Amp with > 100W output and a fully shunt regulated, current source fede supply (the CCS we may be able to approximate by using a choke input supply), but realistically such a beast is just not practical.
Ciao T
Yes, the problem is greater if the same electrolytic capacitor must both close the signal current loop AND at the same acts as the main PSU Reservoir Cap, cross-contamination of the currents is unavoidable.
For my latest Solid State Amp Project I had made a batch of very special chokes, with a very high Al (self inductance) Manganese/Iron Core and 10A DC rating and around 1mH Inductance.
The intention is to use a pair of PSU Capacitors (Elna "For Audio" 18,000uF/71V) for each rail separated by a pair of these chokes with the "ground" only established by joining the two fully separate supplies at the nominal star ground.
There will also be snubbers using Elna Silmic Cap's and very large value (90uF in 15uF/160V Electrocube Miniature Cap's) film bypass capacitor on the second PSU capacitor.
Together with an all lateral fet compound feedback pair output stage (much better PSRR than common outputs) this should go a long way addressing the problem.
In other similar applications I have instead banked on "semi-regulated" supplies and RC filter chains, which can also help a lot to clean up this set of problems...
Yet if I compare these approaches with the power supplies commonly found in Solid State Amplifiers then we have indeed much left to go with the common designs.
The Amplifiers front-end incidentally will use CCS fed shunt regulators and of course CLC filtered supplies prior to the Shunts.
I'd love to make a Class A Amp with > 100W output and a fully shunt regulated, current source fede supply (the CCS we may be able to approximate by using a choke input supply), but realistically such a beast is just not practical.
Ciao T
Exactly. Thus it is necessary to create an ultimative outdoor power supply with a large inductor (choke) between two big cap groups - for exactly listening results between zero- and strong GNFB. Otherwise you will not hear only the parasitic effects of the caps. Mainly you will hear the consequences of all summing effects include spread capacitances and much more other through mostly not ideal wiring- and GND- management (Therefore, it rarely sounds good, regardless of the topology).
Thus by listening comparable tests between various commercial top class products nobody knows the reasons for the often to observe clearly audible differences.
Last edited:
Therefore, if one observes most of best sounding solutions, the resulting amp is almost inevitable becomes muscul-less low-wattage toy. The only hope remains that speaker manufacturers will rectify situation.
Hi,
So, we are back in the 1930's...
Don't hold your breath. Consumers want something that is smaller than Bo$e and sounds as good or a little better as Bo$e and is cheaper than Bo$e...
Ciao T
So, we are back in the 1930's...
Don't hold your breath. Consumers want something that is smaller than Bo$e and sounds as good or a little better as Bo$e and is cheaper than Bo$e...
Ciao T
Stuart is incorrect. When you operate with a HIGH open loop bandwith, then his argument may be correct, but as we know, most feedback amplifiers have a LOW open loop bandwidth. This adds a 90 degree phase shift inside the amplifier. This is what modulates, if the open loop linearity isn't just about perfect.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.