Re: Re: Re: Re: Re: Re: Re: parametrized open-loop EF dist./harm. vs bias graphs
Andy_c
Thank you. I will start my spice education first and try these exercices. I beleive that a great and unique use of simulation models is to understand individual level of influences of different sources of distortion. This would be impossible with real experiments.
A few questions:
What kind of readings should I do to get used to LTSpice?
Are there ways to introduce thermal effects like time constants and hysteresis on the specific parameter under test of these specially made models.
Do you mind if I ask you questions during my progress with LTspice and should I do it by direct email with you.
Thanks for your help
Jean-Pierre
Andy_c
Thank you. I will start my spice education first and try these exercices. I beleive that a great and unique use of simulation models is to understand individual level of influences of different sources of distortion. This would be impossible with real experiments.
A few questions:
What kind of readings should I do to get used to LTSpice?
Are there ways to introduce thermal effects like time constants and hysteresis on the specific parameter under test of these specially made models.
Do you mind if I ask you questions during my progress with LTspice and should I do it by direct email with you.
Thanks for your help
Jean-Pierre
Just one example. My class A output stage, mentioned here, uses BD139-16/BD140-16 drivers and MJL21193/MJL21194 output devices. In case I make distortion simulation in MC8 with models of these transistors, the result is ABSOLUTE NONSENSE. In case I change models to MJE15030/MJE15031 drivers and MJL1302/MJL3281 output devices, the FFT result fits to measurements, both THD and CCIF IMD, within several dB. WHY?
.PMA said:
Measuring the reality is one thing, understanding the origin or the influence of some phenomenon is something else.
One common approach is via modeling with a knowledge of the model limitation. It is very often a more flexible approach. The results have to be confirmed by measurements of course.
When the mecanisms are better understood, better designs are then possible and then again real measurements are mandatory.
Jean-Pierre
JPV said:
This is a very good point. SPICE can often help us understand the subtleties going on in the innards better, and may lead to two things: better designs in the first place by allowing fast architectural exploration and/or by better component value choices; and secondly, can actually help guide us in the measurement process by perhaps hinting at where to make measurements and what kind might be best.
I think we all agree that SPICE is not a replacement for measurements, just like measurements are not a replacement for listening.
Cheers,
Bob
PMA said:
Hi Pavel,
Just in case you are using a pspice model for the BD139 or the BD140 that came from OnSemi, please note at least one of them does not work correctly.
Use the BD139 and BD140 pspice models from Fairchild's website, at:
http://www.fairchildsemi.com/models/
By the way, Fairchild's pspice model files for those do state that they are based on the BD139-16 and BD140-16 versions.
I first saw this model problem mentioned here:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=86810&highlight=
I once wasted almost an entire day because of the problem with the OnSemi BD139 and BD140 pspice models. But those are great little transistors. I use them in a small "power booster" amplifier that is inside the feedback loop(s) of an opamp that is configured as a Howland current source. I can flip a switch to change it into a voltage source. It drives the calibrated base/gate current/voltage "stairsteps" output, for my Curve Tracer product.
- Tom Gootee
http://www.fullnet.com/~tomg/index.html
PMA said:
myhrrhleine said:
Hi Myhrrhleine and Pavel,
Admittedly, I was a bit joking. On the other hand, I do trust my measurements (THD and IMD) more than my subjective ears. My trouble is that always when I'm listening to (technically) perfect amplifiers, I never get exited, it sounded okay but, but nothing special about it, it sounded even a little dull, as if something were missing. The same experience I had with professional monitor loud speakers. Within a few dB flat from 20Hz to 20kHz, and a very low distortion. But again, these speakers didn't impressed me.
In the end I concluded that nothing is wrong with equipment showing perfect specs, but that I my ears were 'looking' for something that cannot be delivered by true hi-fi systems, just because these equipments are not supposed to reproduce sounds that did not exist in the original recordings.
Cheers, Edmond.
David,
I basically agree but wouldn'd call it distortion. As Stan Curtis once explained, the subtle interaction of an amp and speaker can give just that little bit coloration, or interaction, that makes it just right to enjoy. And because it is so subtle and dependent on many things, it is very hard to nail down with numbers.
Jan Didden
I basically agree but wouldn'd call it distortion. As Stan Curtis once explained, the subtle interaction of an amp and speaker can give just that little bit coloration, or interaction, that makes it just right to enjoy. And because it is so subtle and dependent on many things, it is very hard to nail down with numbers.
Jan Didden
Hi Jan,
I believe, maybe similarly as John Curl, in sound of topologies, that we are not able to measure by our THD and IMD stationary FFT analysis methods, performed into resistive load.
What is my reason for saying so. 5 years ago I first time built linear class A stage with low output impedance, tried it inside and outside GNFB and compared to several other GNFB designs. All of those GNFB designs have had something common in their sound, quite different from the output stage outside GNFB. It cannot be blamed on THD or IMD, because some of the compared amps measured better and some worse than the non-GNFB amp. I have retrieved and improved this class A amp recently and again, I am very surprised by its sound. As mentioned, it cannot be blamed on THD, IMD or damping factor.
BUT I also want to have my THD and IMD low, and only harmonic components of very low order.
I believe, maybe similarly as John Curl, in sound of topologies, that we are not able to measure by our THD and IMD stationary FFT analysis methods, performed into resistive load.
What is my reason for saying so. 5 years ago I first time built linear class A stage with low output impedance, tried it inside and outside GNFB and compared to several other GNFB designs. All of those GNFB designs have had something common in their sound, quite different from the output stage outside GNFB. It cannot be blamed on THD or IMD, because some of the compared amps measured better and some worse than the non-GNFB amp. I have retrieved and improved this class A amp recently and again, I am very surprised by its sound. As mentioned, it cannot be blamed on THD, IMD or damping factor.
BUT I also want to have my THD and IMD low, and only harmonic components of very low order.