My comment was aimed at those who claim an amplifier with GFB inherits a sonic signiture from this alone, almost no matter how it is achieved. Often going on to sort amps they like by their harmonic structure. Please tell me what other conclusion can I make but that these amplifiers have distortions that are audible and they were designed that way?
Ignored usually are the carefully engineered designs that pop up here from time to time that don't exhibit any of these "distortions" and still use "bad" design practices.
By now the class-B biased op-amp on steroids is a straw man. Some folks have presented much better here, discussing them would be more productive.
Hi,
Yes, it states this, but does that make it true?
Either that, or halve as hard... Think about a bit more.
Ciao T
Yes, it states this, but does that make it true?
Either that, or halve as hard... Think about a bit more.
Ciao T
Are you sure Walt's comment on in band phase errors is defensible? What phase errors, the closed loop gain/phase of the complete amplifier? You will certainly find the "PIM" on the AD823 virtually unmeasurable with or without this. It's not a 741 afterall.
Joachim,
Did you ever perchance look at an HK Citation II Schematic?
Harman Kardon Citation II Amplifier
Multiple loops where not uncommon (but not exactly common either) in Tube circuitry. I have always been surprised that we do not see them more often in Transistor Circuitry, where the obvious benefits are much larger (as the problems too are plain massive).
Ciao T
Did you ever perchance look at an HK Citation II Schematic?
Harman Kardon Citation II Amplifier
Multiple loops where not uncommon (but not exactly common either) in Tube circuitry. I have always been surprised that we do not see them more often in Transistor Circuitry, where the obvious benefits are much larger (as the problems too are plain massive).
Ciao T
Hi,
Sorry, I must have missed all of them.
Really? If it is a strawman it does not exist. If that is the case the class-B biased op-amp clearly cannot be found inside audio gear. Ahhhm, what is the average output stage bias for an Op-Amp output stage?
Ciao T
Sorry, I must have missed all of them.
Really? If it is a strawman it does not exist. If that is the case the class-B biased op-amp clearly cannot be found inside audio gear. Ahhhm, what is the average output stage bias for an Op-Amp output stage?
Ciao T
I thought we were talking PA's. You know the classic 1970's style basicly a big power op-amp with lots of feedback. If it's your opinion that this is all you have seen here, so be it. Folks have presented their stuff here in good faith (mostly) for friendly (sometimes) comment and discussion. And believe it or not some of them don't have something to sell. Yet still some have low open-loop BW and GNF, but I don't find all this "mindless" design methodology.
I still see criticism based on what the schematic looks like. I guess we're left with nothing but op-amp rolling. Let's see how fast we can find ten gearhead sites with several dozen op-amp "listening" tests. What do you know they don't agree. Do you think the faves will line up on output bias?
EDIT - Bob Cordell's MOSFET power amp for instance, save for the error correction, a big op-amp with thoughtful updating. Discussed many times or did I miss what you meant by missed.
Last edited:
Hi,
It is still commonly encountered.
Most solid state design I see here is still the exact same, with minor variations. There are exception, they are mostly over at Nelson's part of the Neighbourhood.
Even multiloop designs are rare as hens teeth and rocking horse droppings, more common is to see the local degeneration removed from sensible designs in order to get more loop gain.
This is DIY-Audio. While some here also play on the commercial side of the fence, there is not much of a sense to even try to sell these wares here.
Really? Well, there is not enough data here, but generally I would consider them mindless, unless other unusual methods are used in the design.
I normally criticise the design techniques that are embodied in these schematics, not the way the look, given that quite a few of mine are badly hand-drawn on napkins and the like...
No, there is tube rolling. Or we could make amplifiers which operate open loop for a change. Oh, I forgot, I did many of these, so I guess for me it is time to play with feedback for a change.
In many cases yes - the most preferred ones tend to be discrete designs that have pathetic levels of open loop gain (but very wide open loop bandwidth), high levels of output bias and have comparably high measured THD (though in reality still past good and bad).
Ciao T
It is still commonly encountered.
Most solid state design I see here is still the exact same, with minor variations. There are exception, they are mostly over at Nelson's part of the Neighbourhood.
Even multiloop designs are rare as hens teeth and rocking horse droppings, more common is to see the local degeneration removed from sensible designs in order to get more loop gain.
This is DIY-Audio. While some here also play on the commercial side of the fence, there is not much of a sense to even try to sell these wares here.
Really? Well, there is not enough data here, but generally I would consider them mindless, unless other unusual methods are used in the design.
I normally criticise the design techniques that are embodied in these schematics, not the way the look, given that quite a few of mine are badly hand-drawn on napkins and the like...
No, there is tube rolling. Or we could make amplifiers which operate open loop for a change. Oh, I forgot, I did many of these, so I guess for me it is time to play with feedback for a change.
In many cases yes - the most preferred ones tend to be discrete designs that have pathetic levels of open loop gain (but very wide open loop bandwidth), high levels of output bias and have comparably high measured THD (though in reality still past good and bad).
Ciao T
Did you see Marcel van de Gevel's comment on Bruno's article? (Jan you really need to figure out how to compress those pdf's). The argument that wide OLB is "good" in and of itself has no technical merit. IMNSHO of course.
You have made your opinion clear, again, what you think of (most/many/some??) of the folks here or at least their designs.
EDIT - I forgot putting a resistor to load the VAS on an ordinary op-amp to "improve" it is also "mindless" design.
Last edited:
Hi,
Yes I read it.
I have not. I merely told you what those geek sites you told me to go say and what the circuits they say it about are like.
My personal opinion does happen that high slew-rate and high output bias sound better, which is probably why I like Video Op-Amp's. But I like open loop tube circuits even better, strictly to listening to.
It is. Applying local (loop) feedback is a better choice, as it actually both widens bandwidth and reduced distortion over the circuit without it, while VAS loading widens bandwidth but increases distortion over the unloaded cases.
So in one case we lessen distortion and overall loop gain (like John Curl's JC-3 or the Citation II) which I would suggest is smart, in the other we increase distortion and lower loop gain, so linearity will take a major hit, but heck, we can make it all up with more feedback, right?
Ciao T
Yes I read it.
I have not. I merely told you what those geek sites you told me to go say and what the circuits they say it about are like.
My personal opinion does happen that high slew-rate and high output bias sound better, which is probably why I like Video Op-Amp's. But I like open loop tube circuits even better, strictly to listening to.
It is. Applying local (loop) feedback is a better choice, as it actually both widens bandwidth and reduced distortion over the circuit without it, while VAS loading widens bandwidth but increases distortion over the unloaded cases.
So in one case we lessen distortion and overall loop gain (like John Curl's JC-3 or the Citation II) which I would suggest is smart, in the other we increase distortion and lower loop gain, so linearity will take a major hit, but heck, we can make it all up with more feedback, right?
Ciao T
The idea behind the JC-3 (double feedback loop) was to reduce the OVERALL global feedback to about 20dB, and NOT throw away the initial gain generated by the driver stage, but to use it to linearize the input and driver stage, as well as LOWER the drive impedance to the output stage. Typically the same thing is done with resistor loading of the driver stage, but I hoped for an improvement to this approach. It worked.
Question - Without the two 1Meg resistors do you need to add a comp cap or is the stability about the same.
I feel at a disadvantage, like others, building it virtually in simulation and presenting results or even building it and measuring results are simply discounted. When the purely technical claims don't hold water the fall back position involves market success, happy customers, the "sound", yadda yadda.
I got Samuel's thesis. He found several brands of resistors below the noise threshold and most polystyrene or PP film caps too.
The fuzzbox guys swear by two watt NOS carbon resistors the cruftier the better.
PMA, the design is over 35 years old. What were YOU doing, when I designed this amp? '-) However, you are right, AND I have never designed an inverting design like this since, because I now CASCODE the input stage to reduce the non-linear input capacitance, AND I SERVO to get the offset down to virtually nothing.
In its day, it was very advanced, because it allowed a film cap to be used instead of a much higher value tantalum or aluminum cap for DC offset reduction, and very low capacitance non-linearity addition, or common mode distortion. I was competing with the ELECTROCOMPANIET power amp, you know Lostroh (sp) and Otala, who also used a 5K input resistor, or so with a bipolar input, or essentially the same thing.
In truth, 10K input resistive noise is almost trivial in a power amp input. Especially with jfet inputs. I now use approximately 7.5K in series with the input of the JC-1 power amp, in order to get better common mode cancellation, and I still get a good rating in the industry, except from Bob Cordell, who called me on it, once.
In its day, it was very advanced, because it allowed a film cap to be used instead of a much higher value tantalum or aluminum cap for DC offset reduction, and very low capacitance non-linearity addition, or common mode distortion. I was competing with the ELECTROCOMPANIET power amp, you know Lostroh (sp) and Otala, who also used a 5K input resistor, or so with a bipolar input, or essentially the same thing.
In truth, 10K input resistive noise is almost trivial in a power amp input. Especially with jfet inputs. I now use approximately 7.5K in series with the input of the JC-1 power amp, in order to get better common mode cancellation, and I still get a good rating in the industry, except from Bob Cordell, who called me on it, once.
I don't expect him to be sending me a copy! Yes there have been good resistors available for a long time. They were developed for strain gauge use long before even vacuum tube amplifiers!
It just took a while till audio designs found the common carbon comp versions were part of the limits to performance.
I find it quite interesting that some carbon comps actually would limit A/D performance to less than 16 bits!
You have recalled when monolithic circuits only used lateral PNP BJT's.
There actually is progress, (Side joke does anyone know what the opposite of progress is?)
The last two files I posted since no one has come forth were .1% 9th harmonic of 150 HZ and 0% deliberate distortion.
As the energy in music may be 20 db higher at 150 compared to the 9th harmonic those that can hear .1% would do well to design for that harmonic to be more than 80 db down! (That leaves open the issue of at what actual operating level!)
- Status
- Not open for further replies.