So, for a stable circuit, we must to compensate. A little capacitor over the R2 will cure the oscillations, but, we sacrifice slew-rate of the entire circuit. Cloosed loop gain will not be "perfect", in a high frecvency will still remain a potential of trouble and here we dont have negative feedback . The problems at the HF will goes down , in a audio range by IMD distorsions. If you "eradicate" all HF comportament of the opamp, will obtain a slow circuit.
In my opinion, the negative feedback it is not the guilty, the first guilty it is the whole circuit, how you implement it in a real schematic.
In my opinion, the negative feedback it is not the guilty, the first guilty it is the whole circuit, how you implement it in a real schematic.
50 pF from the inverting input to ground is no realistic value,
so the demonstration has limited value.
That said, the LT1028 seems rather oscillations friendly
according to the models provided by the manufacturer.
NE5532 on the other hand is very stable compared
to any so called up to date op amp wich often are mediocre
circuits whoses questionnables better perfs are due to way
better and recent silicon foundries processes but surely
not to designs breakthroughs.
Using the archaic processes available at the 5532 launch,
all current hyped op amp would be average at best..
so the demonstration has limited value.
That said, the LT1028 seems rather oscillations friendly
according to the models provided by the manufacturer.
NE5532 on the other hand is very stable compared
to any so called up to date op amp wich often are mediocre
circuits whoses questionnables better perfs are due to way
better and recent silicon foundries processes but surely
not to designs breakthroughs.
Using the archaic processes available at the 5532 launch,
all current hyped op amp would be average at best..
May be I showing the slowness of OPA2134 for buffering DAC not so sistematic.
Ok, lets start with most people agree that OLBW is unused, right? (and hard to change this religion). Then in this case an opamp used for DAC buffer to prevent DAC drop when filtering.
Attached image is DAC output waveform that show at least it needs buffer with good 'speed'. As I know here, that low OLBW is slow and higher OLG is higher and much overshoot and less stable.
OLBW may also important in other side of audio signal amplification, like driving loudspeaker when output is so unlinearly reactive and 'feedback applied', since no one could show about it, it is just also a myth.
Attachments
I conclude, as I heard here of Hilbert transformation, it implies that if you have a flat response the phase is also flat. But neither loudspeaker nor amps have a flat phase response. They are both not minimal phase. We have had a discussion at the german Visaton forum about it.
That's an incorrect understanding of the Hilbert transform. The phase curves on the datasheets you referenced are minimum phase. If the opamps don't deliberately contain a delay line or an allpass filter, they are forced to be minimum phase.
In its simplest aspect, minimum phase means that the phase is proportional to the derivative of amplitude with respect to frequency.
Indeed ? 10pF is ok ? Looking at the graph ..... LT1028 is not good ?
Ok, try with another high performance opamp. The problem comming from the source, badly design digital source in special. This have too much HF energy which can not be correctly treaty and then that HF energy will be pushed down and up in frecvency . I was listening same music live, with analog mic and pre, and then the same music recorded on a weak digital record sistem. The amplifier and loudspeakers was the same, but the differences in sound was teribble.
Theoretical speaking, digital recorder it is much better then performances of pre+mic combo. In reality HF energy(over audio range) playing an important role.
Attachments
Indeed ? 10pF is ok ? Looking at the graph ..... LT1028 is not good ?
10pF is not realistic , as well as the output coil that would
be equivalent to almost a 10mm / 10 turns wirewounded coil.
As pointed by Janneman, it s just a resonnant circuit
that can be build on whatever provide gain , but is in no way
an exemple of how op amps are used in audio amplification.
SY,
Refering to the underlined (by me) part of your post here
I would like to know if there is a consensus that the underlined (by me) part of the following quoted post is accepted as applicable.
It striked me as most important when I read it (that parallel connection can make a system as non minimum phase) but I find it very surprising that no one yet has commented on this post of Charles (phase_accurate)
Regards
George
Refering to the underlined (by me) part of your post here
I would like to know if there is a consensus that the underlined (by me) part of the following quoted post is accepted as applicable.
It striked me as most important when I read it (that parallel connection can make a system as non minimum phase) but I find it very surprising that no one yet has commented on this post of Charles (phase_accurate)
Regards
George
Last edited:
Your circuit is unrealistic - nobody in his right mind puts 50pF on that input.
jan didden
I don't have time to address all your points but this is not how it is.First of all, your circuit is unrealistic - this way I can 'proof' anything.
Secondly, you confuse slew rate and freq response.
jan didden
This is so uncorrect that it isn't even wrong. How many people have tried to explain these issues here, and still the myths live on. I'm getting disappointed.
jan didden
originally posted by SY
Thanks
So, what Charles has pointed out refers only to loudspeakers but not to amplifying blocks ?
And this “only to loudspeakers” may imply that series type cross-over is preferable to parallel type cross-over, for the reason that the former (series) has more chances to form a minimum phase system? (Heavily off-topic)
Regards
George
Thanks
So, what Charles has pointed out refers only to loudspeakers but not to amplifying blocks ?
And this “only to loudspeakers” may imply that series type cross-over is preferable to parallel type cross-over, for the reason that the former (series) has more chances to form a minimum phase system? (Heavily off-topic)
Regards
George
Last edited:
The discussions in this thread are really important, and demonstrate the controversial nature of negative feedback. In some cases they also demonstrate some common misunderstandings about NFB. I tried really hard in my book, Designing Audio Power Amplifiers, to cover these topics thoroughly and in a clear way, explaining why NFB, when used properly, is an effective tool that improves sound quality.
For those of you who are interested in the pros and cons and proper usage of NFB, and who are skeptical about NFB, I urge you to get a copy of my book and read the sections on NFB. They explain how feedback reduces distortion, how it is compensated, how slew rate is established and improved, pros and cons of NFB and many other related topics. Much of the coverage is in Chapters 2, 3, 4, 9, 22, and 24. You can find the detailed table of contents on the Amazon book page or on my website at CordellAudio.com - Home for further detail on how NFB is covered.
Many of you have already read those sections in my book (and many here don’t need to). Some of you may disagree with what I have said, felt that I have left some important aspect or issue out, or saw some of my explanations as unclear. One of these days I’ll be writing a second edition (a few years) and I would deeply appreciate you pointing out those areas where I can correct these deficiencies. It is my goal to be complete, thorough and clear in covering topics on negative feedback – not just to convince you that NFB is a good thing. Indeed, for those who choose the no-NFB route, I have a chapter on amplifiers that do not use negative feedback.
It is especially important to understand the role of open-loop bandwidth on an apples-to-apples basis, where the apples are the gain crossover frequency (~closed-loop bandwidth) and stability. If we just increase Cdom in a conventional Miller-compensated amplifier and keep everything else the same, we know that the open-loop bandwidth will decrease, the slew rate will decrease, and the amplifier will become “slower”. But the gain crossover frequency has decreased, so this is not apples-apples. If we instead increase input stage current and transconductance by the same factor we increased Cdom, then gain crossover frequency will remain the same and we have apples-apples. AND, although the open-loop bandwidth has decreased, the slew rate and “speed” of the amplifier has remained unchanged from the original Cdm case. This is an example of why apples-apples comparisons are so very important in discussing issues concerning open-loop bandwidth.
Cheers,
Bob
For those of you who are interested in the pros and cons and proper usage of NFB, and who are skeptical about NFB, I urge you to get a copy of my book and read the sections on NFB. They explain how feedback reduces distortion, how it is compensated, how slew rate is established and improved, pros and cons of NFB and many other related topics. Much of the coverage is in Chapters 2, 3, 4, 9, 22, and 24. You can find the detailed table of contents on the Amazon book page or on my website at CordellAudio.com - Home for further detail on how NFB is covered.
Many of you have already read those sections in my book (and many here don’t need to). Some of you may disagree with what I have said, felt that I have left some important aspect or issue out, or saw some of my explanations as unclear. One of these days I’ll be writing a second edition (a few years) and I would deeply appreciate you pointing out those areas where I can correct these deficiencies. It is my goal to be complete, thorough and clear in covering topics on negative feedback – not just to convince you that NFB is a good thing. Indeed, for those who choose the no-NFB route, I have a chapter on amplifiers that do not use negative feedback.
It is especially important to understand the role of open-loop bandwidth on an apples-to-apples basis, where the apples are the gain crossover frequency (~closed-loop bandwidth) and stability. If we just increase Cdom in a conventional Miller-compensated amplifier and keep everything else the same, we know that the open-loop bandwidth will decrease, the slew rate will decrease, and the amplifier will become “slower”. But the gain crossover frequency has decreased, so this is not apples-apples. If we instead increase input stage current and transconductance by the same factor we increased Cdom, then gain crossover frequency will remain the same and we have apples-apples. AND, although the open-loop bandwidth has decreased, the slew rate and “speed” of the amplifier has remained unchanged from the original Cdm case. This is an example of why apples-apples comparisons are so very important in discussing issues concerning open-loop bandwidth.
Cheers,
Bob
Wahab, in modern IC's you cannot really separate the process from the circuit. So, i dont think yo ur last statement would hold There are things you can do in silicon that are just not practical in discrete form. One of the reasons you are not likely to see the problem above that much on a 5532 is it's lower open loop gain. Modern op amps are doing 40 or 50 dB more.
That said, the 5532 is not bad for a 30 year old design.
Yes, I realize that now. My statement is incorrect according single pole amplifier model, that higher amp OLG with same dominant pole will be faster and bandwidth plus slewrate increased but overshoot and oscillation never happened.
Also it isn't even wrong, because in reality all amplifier has the same limitation, same silicon, same electron speed, etc, so the designer have to choose fast and low OLG or slow and high OLG. Otherwise, overshoot and oscillation is coming.
Edit: Please look at their second pole, most amplifiers has same cutoff ~10MHz. Also there is 3rd, 4th, and more.
May be it is on bode plot, when there is only flat(0deg), 20dB/decade(90deg), 40dB/decade(180deg), ....and so.
In real there are too complex for bode plot transformation.
Last edited:
"Modern" op amps have the benefit of better performing transistors
due to way more refined processes that allow higher DC OLG mainly..
If you look at the 5532 OLG/frequency diagram, you ll see that modern
circuits have just extended the low frequency gain below the 5532
1khz OLG , not for the full audio range.
Most modern circuits will have a nearby OLG above 1KHZ , just look
at the famed OPA2134 OLG at 20khz , it is barely better if not worse
than the one of the ancestor while the OP627 doesnt do wonders as well
in this respect, so the 40/50db figure is quite discutable on an audio POW.
I never realised that, this is quite interesting. What happened to progress??
jan didden
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.