The challenge for the designer is to have high OL BW AND high OL gain. But remember, if one opamp has 10Hz OLBW and 40dB OLG at 5kHz, and the other has 1kHz OLBW and the same 40dB OLG at 5kHz, they have the same performance at 5kHz. They even have the same performance at 1kHz. Only below, say 100Hz, does the first one have better performance.
jan didden
Are u sure about all modern opamps ? LME49860 has almost 80dB at 10khz, the same AD797.
The difference betwen 5532 and OPA2134 reside from the input stage and not from the OLG at 10khz (60dB).
No, i don't confuse them. The voltage in my example was constant, so ,in that situation, the slew-rate-ul and the frecvency are directly related between them .
If my example is unrealistic, can you present us a realistic version ?
The only diference between my example and reality is the parasitic capacitance over the feedback resistor, which help you to have a stable circuit in real life. But..... it is posible to have a few resistors in series in a feedback path, and the parasitic capacitance to lowering at the level where u will have oscillations or ringing.
Ah.. again. Correct for single pole models, where no overshoot nor oscillations.
There is no challenge in single pole theory.
Then in real it also not like that if they have same 2nd pole at 10MHz, the one will be more stable than the other, except one has its 2nd pole moved to 1Ghz with SOI technology or similar.
Please accept the reality. Other reality is SOI opamp isn't cheap and BJT opamp is still very good for audio.
As an exemple...
Yes, like that. Higher OLG is higher overshoot, because from that graph they have the same 2nd pole at 10MHz. They aren't the same performance, while low THD not always coming from higher NFB, but also its own linearity.
Ok, I confused it.
I know that the phase is propotional to the derivative of amplitude with respect to freqency. Therefore where the response is flat the phase declines with a constant rate.
Ok, I have given the OPA2134 as an example of minimum phase. As a contrary the OPA2111, that shows no minimum phase. There is a peak in the phase between 100k and 1MHz.
Does it still follow a Hilbert transformation?
By the way the OPA2111 is rumored to have a "good or big" (I guess not realistic) soundstage or imaging.
Have you seen the circuit especailly the inner feedback of the NJM4565D?
There is a diode in the feedback.
What does it do there?
According to decay:
I have found something interesting about ringing artifacts:
Ringing artifacts - Wikipedia, the free encyclopedia
What about that?
OOps.., I am simulating it again with amplifier and I can't reach the same BW after increasing OLG except compensate it. The result is it slower, same as the first result, and overshoot reduced after compensated.
But something very strange here. When I try with only opamp (by modify its model) instead of with amplifier, it is very slow even for 10kHz, and same performance result.
Attachments
I think you'll find that modern op amps are inherently more linear at HF as well - and this has nothing to do with just more feedback. Scott Wurcer mentioned op-amps and buffers that are IIRC at -60dB THD at 100MHz. This is only possible with some of the advanced input stage topologies and processes. So, I'd say there has been quite some progress in this field over the last 10 years.
Amazingly, in the audio range they are not automaticaly better
than the old school 5532 and can be far worse.
Bandwith is not all.
http://www.analog.com/static/imported-files/data_sheets/AD8036_8037.pdf
http://www.datasheetcatalog.org/datasheet/analogdevices/AD9618AQ.pdf
Last edited:
Member
Joined 2009
Paid Member
For me, one of the greatest gifts of this hobby is a thread such as this with such passion and discussion.
Another gift is that it is a hobby, as I am freed from having to think and work like an engineer/physicist. I enjoy physics and engineering, but they require great self-discipline to be good, and to be great you have to leave your ego behind and be open to question. It's all very tiring
As a hobby we can indulge in a bit of subjectiveness too. And I believe some of the best amplifier designers do just that. You know who you are
And now I get to add my 'hunch' to the debate.
I think there is much to be explored under the subject of IM. I think this is responsible for many things about how amplifiers sound and why it's hard to pin down the factors in amplifier design that affect sound. I don't think Nelson was underestimating the situation when he called IM the 'elephant on the dance floor'.
Negative feedback affects IM distortion so we notice it has a big impact on sound. It's not as simple as nfb reducing IM distortion - there are many faces to nfb and how it's implemented. It's difficult to do nfb well.
Power supply also affects IM. And PSRR is connected intimately with nfb. The psu introduces noise from the mains supply, it generates noise in the rectifiers and it adds distortion via the rail decoupling capacitors. This is why shunt regulators sound good - they attack these sources of noise that contribute to IM.
Load interaction at the output affects distortion and sound in complex ways. Again nfb plays a part here. I don't know how non-linear loads impact IM, but I assume they affect phase relationships which in turn affects IM amplitudes.
As an aside, the IM of ripple from filament heater supply with cathode signal in a directly heated triode is of legendary importance. Many people have removed this IM using a dc filament supply, often to their regret. In this instance, the IM products have a beneficial affect on their subjective enjoyment of the sound.
edit: Do I hear somebody thinking "how do I get some 60Hz IM into my amp, like a DHT ? - ah, it's already there if I remove gnf from the output??"
Another gift is that it is a hobby, as I am freed from having to think and work like an engineer/physicist. I enjoy physics and engineering, but they require great self-discipline to be good, and to be great you have to leave your ego behind and be open to question. It's all very tiring
As a hobby we can indulge in a bit of subjectiveness too. And I believe some of the best amplifier designers do just that. You know who you are
And now I get to add my 'hunch' to the debate.
I think there is much to be explored under the subject of IM. I think this is responsible for many things about how amplifiers sound and why it's hard to pin down the factors in amplifier design that affect sound. I don't think Nelson was underestimating the situation when he called IM the 'elephant on the dance floor'.
Negative feedback affects IM distortion so we notice it has a big impact on sound. It's not as simple as nfb reducing IM distortion - there are many faces to nfb and how it's implemented. It's difficult to do nfb well.
Power supply also affects IM. And PSRR is connected intimately with nfb. The psu introduces noise from the mains supply, it generates noise in the rectifiers and it adds distortion via the rail decoupling capacitors. This is why shunt regulators sound good - they attack these sources of noise that contribute to IM.
Load interaction at the output affects distortion and sound in complex ways. Again nfb plays a part here. I don't know how non-linear loads impact IM, but I assume they affect phase relationships which in turn affects IM amplitudes.
As an aside, the IM of ripple from filament heater supply with cathode signal in a directly heated triode is of legendary importance. Many people have removed this IM using a dc filament supply, often to their regret. In this instance, the IM products have a beneficial affect on their subjective enjoyment of the sound.
edit: Do I hear somebody thinking "how do I get some 60Hz IM into my amp, like a DHT ? - ah, it's already there if I remove gnf from the output??"
Last edited:
Hi wahab,
There is actually more to it than that. One of the most significant advances in process since the 5532/34 that some op amp manufacturers rely on is fully complementary PNP devices. These were not generally available back in the days of the 5534. Anyway, what makes moedern op amps better is not just more OLG at low frequencies.
Cheers,
Bob
Fairly right, but it has terrible frequency response overshoot at low gain.
Curiously ,since its noise figure is not that good, this lead to
THD+N no better than the NE553X.
There s no free lunch , really..
http://cds.linear.com/docs/Datasheet/1468fb.pdf
The AD797 is another example. You are right, the LT1468 must be compensated at low gains. Modern Opamps do it a bit different with higher open loop linearity like the OPA1461 or the LME49990. Anyway i build many circuits where the NE5534 worked very well.
Notably in a subsonic filter i designed and in my FPS phono stage. But there i used them inverted to avoid common mode distortion. Common mode distortion and DC offset can be a problem in circuits that use it though. If that is solved it is still a wonderful bargain. I have a sweet spot for it anyway because in 1983 i build my first successful preamp with it that worked as well as an expensive Burmester of the day. I demonstrated my subsonic filter in an expensive system at CES 2 years ago and many people like the sound more when it was in the system. It was not a normal filter though but was exactly tunes to the arm-cartridge resonance. JC was in the room too.
The good thing with the NE is also that some pins can access the internals, so the input stage can be switched off and the output stage can be bypassed. I build low noise phono stages and headphone amps with good performance on the cheep that way. In general i think creativity and competence are more important then then physical material anyway. Jimmy Hendrix played big stadiums with a simple tube stack and when the people did not listen he burned the guitar.
Notably in a subsonic filter i designed and in my FPS phono stage. But there i used them inverted to avoid common mode distortion. Common mode distortion and DC offset can be a problem in circuits that use it though. If that is solved it is still a wonderful bargain. I have a sweet spot for it anyway because in 1983 i build my first successful preamp with it that worked as well as an expensive Burmester of the day. I demonstrated my subsonic filter in an expensive system at CES 2 years ago and many people like the sound more when it was in the system. It was not a normal filter though but was exactly tunes to the arm-cartridge resonance. JC was in the room too.
The good thing with the NE is also that some pins can access the internals, so the input stage can be switched off and the output stage can be bypassed. I build low noise phono stages and headphone amps with good performance on the cheep that way. In general i think creativity and competence are more important then then physical material anyway. Jimmy Hendrix played big stadiums with a simple tube stack and when the people did not listen he burned the guitar.
Quite some experiences, Gerhard...
What rendered me skeptical of these DIL8 packages is that for audio
purposes , they have some difficulties in matter of linearity, wich should
be their stronghold , compared to discrete solutions , although these latters
are difficult to implement layout wise.
What rendered me skeptical of these DIL8 packages is that for audio
purposes , they have some difficulties in matter of linearity, wich should
be their stronghold , compared to discrete solutions , although these latters
are difficult to implement layout wise.
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.