Hi Michael
What models have you used for the output FETs?
Often people use NMOS and PMOS models and this can make the distortion numbers pleasantly but unrealistically low.
I want to learn more about "CFA"s but like to be confident about the details.
Best wishes
David
I like it just fine, I just calculate the Class A region instead of worrying about the conduction angle. It's only a general term, it's not for very specific condition.
Put in in another words, If people define Class B as what self defined.....that his Class B is just Class AB push pull that SATISFIES OLIVER'S CONDITION. It means NOTHING for other type of amplifiers. Oliver's condition is very specific for push pull by complementary transistors. For RF that using a common emitter for BJT or common source for FETs, this so called "self Class B" will lead to more confusion.There is no Oliver's condition with those amplifiers.....and believe me, they are a whole lot more common than stereo amps. For us here, this is bigger than life. for the rest of the electronic field, this is really really tiny. Tiny to the point that semi conductor companies keep discontinuing their products for audio and we have to hunt for obsolete parts.
The conventional Classes definition are for all different amp configurations, push pull is just one of the many configurations. You cannot just stay in this very little corner and attempt to change the world just to make it more convenient for this small group of people.
Even in stereo world, this so called Self Class B does not even mean much in tube amps .
Last edited:
I have used the lat-fet models on the Cordel site (thanks).
The interesting thing about the OPS is how the folded cascode is driven with the output and how that creates a short and thight corrective feedback loop around the most problematic devices.
The interesting thing about the OPS is how the folded cascode is driven with the output and how that creates a short and thight corrective feedback loop around the most problematic devices.
I have waited in vain for a real physicist so I'll have a try.
Heat conduction in the metal header is predominantly mediated by free electrons, not lattice vibrations, and any transmission delay will be essentially zero compared to the thermal time constant for the transistor, let alone the heat-sink.
You need a quantum mechanical analysis even to be approximately correct, "Introduction to Solid State Physics" by Kittel was my textbook and has sufficient detail for a starter headache.
The first order model for heat conduction has no actual "delay" term at all. Obviously that can't be completely accurate but it should be practically perfect for conditions relevant to audio amplifier applications.
Best wishes
David
I can't see any ECX10P16 or N models at the Bob's site.
?
That's a problem that interests me too.
Best wishes
David
HarryD..... very few circuits run right AT the defining angle so I would not worry about it.
360 degreees is Class A.
Class AB is >180 and <360.
[Class C is much less than 180... typically 90 degrees.]
Keep it simple as it has always been.
THx-RNMarsh
360 degreees is Class A.
Class AB is >180 and <360.
[Class C is much less than 180... typically 90 degrees.]
Keep it simple as it has always been.
THx-RNMarsh
Last edited:
Ohh I use the renesas/toshiba models, just with a new name as to match those in the drawer. I believe all lat-gets are from the same process.
Hi MiiB,
Interesting circuit. have you breadboarded it?
I wouldn't use the term folded-cascode here. It isn't. I would say the input is more like a normal long-tailed pair with a twist, but with the OP feeded back to the BJT input to provide unity-gain.
At least that is how I see it.
Mogens
Interesting circuit. have you breadboarded it?
I wouldn't use the term folded-cascode here. It isn't. I would say the input is more like a normal long-tailed pair with a twist, but with the OP feeded back to the BJT input to provide unity-gain.
At least that is how I see it.
Mogens
I have to agree with this. This is what I was taught at college, and it is the definition that appeared in recognised texts.
By using the AB definition to mean optimal bias for 'todays' power amplifiers muddies the waters. Does a 5532 AB output stage run in optimum bias ?
Stick to the industry wide meaning of the terms.
I always tend to use Bobs 2SK1056 and 2SJ162 models for any lateral fets I need.
Would it possible to alter the models to reflect the 'double die' devices available simply by doubling up the appropriate parameters in the file ? Or are such models already out there ?
Would it possible to alter the models to reflect the 'double die' devices available simply by doubling up the appropriate parameters in the file ? Or are such models already out there ?
Or maybe make a subcircuit with 2 // devices, possibly with a parameter the user can set to as many // devices as needed?
That's how I would do it.
Jan
In fact you could have an additional parameter specifying the device making it universally useable for any device.
And to go over the top you could also specify Re and Rb - a single symbol output stage.
How's that for uncluttering your schematics! 😉
jan
And to go over the top you could also specify Re and Rb - a single symbol output stage.
How's that for uncluttering your schematics! 😉
jan
🙂 Jan, there are not enough days in the year for me to suss that one out. I shall have to study my big LTSpice IV simulator book, but thanks, you've given me some great ideas.
I’m trying to make the point that when it comes to “class-B”, “as it has always been” is absolutely not “simple”, and you have ignored or not understood what I am saying. So, I used % instead of angle, but it comes to the same thing.
You say "very few circuits run right AT the defining angle”. What do you mean by this? If I say “I’ve built a class-B output stage” and the conduction angle is 180.1, is that class-B or AB because 180.1 is >180? How about 180.01 or 180.4, 181, 182? Where is your cutoff for class-B vs AB?
The Semelab/Alfet/Exicon 20P20 is just a dual-die 10P20. Improved models for either and their complements are here:
http://www.diyaudio.com/forums/soft...ower-mosfet-models-ltspice-2.html#post4365913
There are several ways we could characterize the threshold from class A to class B:
We can set the threshold at the Oliver point, if we are concerned only about distortion behavior.
For an acoustic method, we can set it at the point where signals fitting within the class A region sit at the threshold of audibility. Optionally, argue about audibility metrics.
Consider anything falling between complete class A and complete class B to be neither, and consider this region to be class AB.
Define class B by the presence of a dead zone.
I'm sure there are others to think of. The main thing though is that the threshold is never absolute, and it depends on what the end application will be, or why it matters.
Motor controller designers probably would consider any bias at all to be class A and not suitable for the vast majority of controller designs.
As I said before in replying to HarryDymond, The Classes of amplifiers are defined for ALL different amplifiers, not just push-pull with complementary BJT or MOSFET. This so called Self Class B does not mean anything for amps that is common emitter/source or common base/gate amplifies so common in RF world that are broadly used in the modern days. Self's Class B is just Class AB push pull that conforms to Oliver's condition.
Audiophile is everything for us here, but in the overall electronic field, this is such a small sector that the major semi conductor companies keep discontinuing devices for audiophile. It is ridiculous for people in this very small sector to try to change the definition for everybody and it will never fly. It's a waste of time to even talk about this.
Audiophile is everything for us here, but in the overall electronic field, this is such a small sector that the major semi conductor companies keep discontinuing devices for audiophile. It is ridiculous for people in this very small sector to try to change the definition for everybody and it will never fly. It's a waste of time to even talk about this.