-
Welcome to ASR. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Why don't we care about TIM (transient intermodulation distortion) anymore?
- Thread starter Music1969
- Start date
A simple question: the unity gain bandwidth of the ubiquitous NE5532 is typically about 10 MHz. How much is the feedback at 20 kHz if I use it as a buffer?
Not if you ensure good phase margin with such loads. This is, of course, more difficult to do as the amount of NFB increases, but Putzeys certainly isn't advocating blind application of more and more NFB without concern for stability.
Good phase margins depend on the amount of feedback and respect for the safety margins we already know
since Nyquist and fellow researchers explained how to limit feedback.
NFB is en excellent method to make designs less dependent of the parameters of the components used.
Especially with vacuum tube designs these can vary over time.
I also designed a tube pre amplifier that showed remarkable improvement against solid state designs during
listening tests. I did not bother to spend time why there was such a huge difference in quality.
For me hardware is only a tool to make recording and good reproduction of music possible.
In those days even a poorly designed tube amplifier gave acceptable reproduction quality.
The design of solid state amplifiers with more attention to design inspired me to go further with the design
of tube amplifiers. That approach paid off.
since Nyquist and fellow researchers explained how to limit feedback.
NFB is en excellent method to make designs less dependent of the parameters of the components used.
Especially with vacuum tube designs these can vary over time.
I also designed a tube pre amplifier that showed remarkable improvement against solid state designs during
listening tests. I did not bother to spend time why there was such a huge difference in quality.
For me hardware is only a tool to make recording and good reproduction of music possible.
In those days even a poorly designed tube amplifier gave acceptable reproduction quality.
The design of solid state amplifiers with more attention to design inspired me to go further with the design
of tube amplifiers. That approach paid off.
Last edited:
sceptical1
Active Member
- Joined
- Mar 10, 2025
- Messages
- 170
- Likes
- 141
When was it difficult to deal with unity gain and phase shift, are you serious?
sceptical1
Active Member
- Joined
- Mar 10, 2025
- Messages
- 170
- Likes
- 141
I can't imagine that anything you have devoted any time to was not already dealt with years before and scientifically settled.
Look at the constant objectivist goings on between the objectivist members here. Nobody can agree on anything.
Look at the constant objectivist goings on between the objectivist members here. Nobody can agree on anything.
Theres so many variables that effect amp stability an "old rule" for a feedback limit dosnt make sense.
From B Putzey:
The F-word or, why there is no such thing as too much feedback.
solderdude
Grand Contributor
"Theres so many variables that effect amp stability an "old rule" for a feedback limit dosnt make sense. "
There are actually only two variables: phase margin and amount of feedback.
I prefer the old rule about feedback. That rule has proved itself time and time again.
Many years professional experience in the music industry made that clear.
I wonder what experience made you challenge the 36 dB rule?
There are actually only two variables: phase margin and amount of feedback.
I prefer the old rule about feedback. That rule has proved itself time and time again.
Many years professional experience in the music industry made that clear.
I wonder what experience made you challenge the 36 dB rule?
Haha! Appreciate the response very much, but not quite100dB below 1kHz and from there on 24dB/decade so about about 70dB at 20kHz.
For the NE5532, the unity gain bandwidth is the same as the gain bandwidth product assuming it has a one-pole open loop gain curve (the TI datasheet specifies only the unity gain bandwidth and no open loop gain curve is given). For GBW product = 10 Mhz, open loop gain at 20 kHz = 10 000 000 / 20 000 = 500.
For a unity gain buffer, output voltage = input voltage and therefore, for equation (2), y / x = 1, which means (1 + b A) = A. Since A = 500, 20 log(1 + b A) = 20 log(500) = 54. Therefore, we have 54 dB feedback at 20 kHz (and 74 dB at 2 kHz, etc. given the -20 dB / decade slope of the open-loop gain curve, and when frequency is above that of the pole).
So, if we go with PaulK's 36 dB rule, we surely can't use NE5532 in a buffer for audio, and we have to use opamps that have less open-loop gain! Clearly insanity!
[Edit] It is NE5532, not 5322
Last edited:
I think that is good advise, use opamps with less open loop gain.Haha! Appreciate the response very much, but not quite
For the NE5322, the unity gain bandwidth is the same as the gain bandwidth product assuming it has a one-pole open loop gain curve (the TI datasheet specifies only the unity gain bandwidth and no open loop gain curve is given). For GBW = 10 Mhz, open loop gain at 20 kHz = 10 000 000 / 20 000 = 500.
For a unity gain buffer, output voltage = input voltage and therefore, for equation (2), y / x = 1, which means (1 + b A) = A. Since A = 500, 20 log(1 + b A) = 20 log(500) = 54. Therefore, we have 54 dB feedback at 20 kHz (and 74 dB at 2 kHz, etc. given the -20 dB / decade slope of the open-loop gain curve, and when frequency is above that of the pole).
View attachment 474256
So, if we go with PaulK's 36 dB rule, we surely can't use NE5322 in a buffer for audio, and we have to use opamps that have less open-loop gain! Clearly insanity!
Huge open loop gain with opamps caused several cases of TIM that could have been avoided.
Do you know how many NE5532 are used in devices for audio applications?
Last edited:
solderdude
Grand Contributor
Haha! Appreciate the response very much, but not quite
20 log(500) = 54
54dB, 64dB, what's an order of magnitude among friends? (edit, not quite but I like to parrot my old advisor!) The 2009 Fairchild data sheet gives a different number too.
solderdude
Grand Contributor
Not all 5532 (data sheets) are created equal.
Which depend on open loop gain and the pole locations which are decided by the many variables in an amp design. One example. If you replace the resistor in the voltage gain stage with a current source you get more gain and you can usually use more feedback.
EE graduate courses in analogue circuit design and
https://www.diyaudio.com/community/attachments/the-f-word-bruno-putzeys-pdf.1140285/
Similar threads
