G.Kleinschmidt said:
Hi Glen,
I don't think the effective output inductance of a small MOSFET power amplifier is nearly as high as you might think.
Take a MOSFET power amplifier where the output stage is biased at 200 mA idle. The effective Rs (=1/gm) of each output transistor at crossover is about 1.25 ohms. The parallel combination of the P and N devices brings the open loop output impedance down to 0.6 ohms.
Even with only 20 dB of NFB at 20 kHz, the effective output impedance will be about 1/10 this after the loop is closed, or about 0.06 ohms.
A reactive output impedance of 0.06 ohms at 20 kHz corresponds to approximately 0.5 uH, which is quite adequately low.
Keep in mind that this is for only 20 dB of NFB and no error correction. This example would not be a very high-performing example of a MOSFET power amplifier.
Cheers,
Bob
john curl said:
John, I think that your forte is the spreading of unsubstantiated fertilizer.
You are the one that keeps bringing this up. I is a pity that you cannot get past something that happened over 25 years ago. You are living in the past. You couldn't make MOSFEts work for you 20 years ago, so you never used them again and told other people they are no good. You were disappointed in ECAD as a simulator 40 years ago, so you dismiss SPICE as a useful audio design tool.
Let it go and start dealing in the present. You'll find that you have a lot more credibility if you do that.
Bob
john curl said:
Hi John,
I stand by what I said there about Otala's writings. I am, of course, very sorry to hear that he had a stroke. I was unaware of that.
Here's the thing about TIM, PIM IIM, etc. Matti generally blamed these on high values of negative feedback often accompanied by low open loop bandwidth. Unlike many others, Matti was kind enough to define ways of objectively measuring these distortions. All I did was to demonstrate, with real amplifiers, that TIM, PIM and IIM were just as low when large feedback factors, low open loop bandwidth, and high open loop output imedance were involved. I just showed that, using Otala's own measurement techniques, his generalizations about large amounts of negative feeback being a culprit were unjustified generalizations that were wrong most of the time.
I was certainly not alone in criticizing Otala's conclusions, nor was I the first. Just look at the literature and references. I think I was in pretty good company.
For those who are interested and want to make up their own minds on this, just go to my website at www.cordellaudio.com and read the relevant papers and the many, many included references.
It seems that every time you mention that old 27-year-old "rebuttal", the approximate number of pages increases! Now it is up to 25 pages! Holy cr@p! Why don't you dig it out and scan it, with the signatures, and email it to me and I'll publish it right on my web site. All I remember is a pretty limp letter to the editor in Audio that I responded to. It certainly did not "trash" what I had written.
I fully expect that you will not be able to provide this, as it is often the case when you are challenged you suddenly cannot come up with the goods. However, in the event that I am wrong, I'll look forward to seeing it.
Cheers,
Bob
Folks, I do not go around and try to find problems in Bob's existing design.
The ONLY point that I have ever made here on this website, is that EXTRAPOLATION of Bob's original design to a much more powerful amplifier, would probably not work very well. In other words raising the power supply voltage from +/- 35V to +/- 80V or so, and just changing the fet part numbers to a higher voltage version of the same part isn't going to do it. That is the ONLY concern that I have about all-fet amps, that I have already designed in many examples.
My largest amp operates at +/- 90V, so I some experience in this area.
Now, when it comes to SPICE, why would I generate a Spice simulation AFTER the amp is already in production? I have a pretty good idea how the amp works, and I have real amps at my disposal to test, if I have a question. Why would I need a Spice simulation at this late date? Now, does this imply that I am afraid of Spice? I don't know how you get that idea. I just know that it has limitations. It is good to know this.
In 1966, I was partially hired to my first professional job, because of my enthusiasm for computer circuit emulation. I was even put in charge of modeling the components, emulating the circuits to be simulated, and running the mainframe computer myself. I had technicians who worked under me to help keypunch the cards needed and a seasoned engineer over me (who subsequently went to Lawrence Livermore Labs). I was gung-ho for simulation, UNTIL another engineer found that the computer was giving us FAIL results when it could be proven by a simple slide rule calculation that it was impossible. At least, I LEARNED from this situation.
New and improved SPICE has the same potential limitations that we had with ECAP 40 years ago, and seasoned engineers are always laughing at those who completely depend on Spice for circuit modeling. Is this so difficult to accept?
The ONLY point that I have ever made here on this website, is that EXTRAPOLATION of Bob's original design to a much more powerful amplifier, would probably not work very well. In other words raising the power supply voltage from +/- 35V to +/- 80V or so, and just changing the fet part numbers to a higher voltage version of the same part isn't going to do it. That is the ONLY concern that I have about all-fet amps, that I have already designed in many examples.
My largest amp operates at +/- 90V, so I some experience in this area.
Now, when it comes to SPICE, why would I generate a Spice simulation AFTER the amp is already in production? I have a pretty good idea how the amp works, and I have real amps at my disposal to test, if I have a question. Why would I need a Spice simulation at this late date? Now, does this imply that I am afraid of Spice? I don't know how you get that idea. I just know that it has limitations. It is good to know this.
In 1966, I was partially hired to my first professional job, because of my enthusiasm for computer circuit emulation. I was even put in charge of modeling the components, emulating the circuits to be simulated, and running the mainframe computer myself. I had technicians who worked under me to help keypunch the cards needed and a seasoned engineer over me (who subsequently went to Lawrence Livermore Labs). I was gung-ho for simulation, UNTIL another engineer found that the computer was giving us FAIL results when it could be proven by a simple slide rule calculation that it was impossible. At least, I LEARNED from this situation.
New and improved SPICE has the same potential limitations that we had with ECAP 40 years ago, and seasoned engineers are always laughing at those who completely depend on Spice for circuit modeling. Is this so difficult to accept?
john curl said:
Here we go again. Are you in some kind of time warp? You act like you are the only one who learned from the past. I have got news for you: you are not alone. You insinuate that anyone who uses SPICE nowadays does so recklessly, with total dependence on it, without consideration of its limitations. HOGWASH.
BTW, wasn't it actually Bongiorno who was first to publish and commercialize the full-complementary input differential pair?
Bob
No Bob, it was NOT Bongiorno, who first developed the complementary differential input stage. First, there was Jon Iverson of ElectroResearch and me (then designing in the Ampex audio department) who did it in 1968, quite independently of each other. We did NOT publish our efforts, because, in my case, Ampex would have patented it, and in any case, it was a trade secret.
You will find that the FIRST mention of the complementary differential input stage as an ad by Southwest Technical Products in 'The Audio Amateur' about 1972, fully 4 years after I got my amp running, and Jon Iverson has his design developed as well.
I was told by Daniel Meyer, of Southwest Technical Products, who independently developed the complementary differential input stage after we did, that HE showed it to Jim Borgiorno, when Jim Borgiorno showed him the DYNA 400 circuit that he had worked on. You should contact Daniel Meyer about this, if you wish to question this statement.
I went on to develop the self biasing complementary differential JFET input stage in 1971, and that is was I have used since 1971, to this day. I built a 2000W complementary differential power amp with balanced current outputs for Ampex research in 1969, many prototype complementary differential bipolar input power amp prototypes at Alembic Inc in 1970 for the Grateful Dead, and later in 1973, I built more power amp prototypes for the Grateful Dead using by complementary differential jfet input stages and single sided complementary jefet input stages. This is when the Levinson JC-2 was designed as well, and the patented pre-preamp input stage, the JC-1 was first introduced at the NY-AES in Oct 1973.
You will find that the FIRST mention of the complementary differential input stage as an ad by Southwest Technical Products in 'The Audio Amateur' about 1972, fully 4 years after I got my amp running, and Jon Iverson has his design developed as well.
I was told by Daniel Meyer, of Southwest Technical Products, who independently developed the complementary differential input stage after we did, that HE showed it to Jim Borgiorno, when Jim Borgiorno showed him the DYNA 400 circuit that he had worked on. You should contact Daniel Meyer about this, if you wish to question this statement.
I went on to develop the self biasing complementary differential JFET input stage in 1971, and that is was I have used since 1971, to this day. I built a 2000W complementary differential power amp with balanced current outputs for Ampex research in 1969, many prototype complementary differential bipolar input power amp prototypes at Alembic Inc in 1970 for the Grateful Dead, and later in 1973, I built more power amp prototypes for the Grateful Dead using by complementary differential jfet input stages and single sided complementary jefet input stages. This is when the Levinson JC-2 was designed as well, and the patented pre-preamp input stage, the JC-1 was first introduced at the NY-AES in Oct 1973.
This thread has caused the moderators far too much work and our patience has run out. The following actions have been taken:
This thread is being closed until further notice.
John Curl is being Sinbinned for two weeks.
Other protagonists on this thread are being investigated.
The forum is for civilized discussion and transgressions will be dealt with, regardless of who you are.
Attack ideas, not people.
- Status
- Not open for further replies.