the thing is that a -3db roll off doesn t mean the same thing that
in an OPA...this latter has usually very low bandwith at -3db with very
high gain at the start of the roll off point, whenever a bjt sustain it s current
gain in a wide bandwith and then start a roll off at very high frequency..
the -3db roll off frequency of a bjt somewhat define what is the
upper usable frequency of the device if linear operation is the concern...
in an OPA...this latter has usually very low bandwith at -3db with very
high gain at the start of the roll off point, whenever a bjt sustain it s current
gain in a wide bandwith and then start a roll off at very high frequency..
the -3db roll off frequency of a bjt somewhat define what is the
upper usable frequency of the device if linear operation is the concern...
Member
Joined 2009
Paid Member
I wouldn't make that extension. In fact quite the opposite. Clearly the universe works the way it does and I'm content with that.
But yes, the SET you built today may sound bad tomorrow because there is indeed some uncertainty as to how it will perform tomorrow and how I perceive sound will depend on how I feel tomorrow.
Incorrect. The two variables must be conjugate.
Much closer. Position and momentum (not velocity) are canonically conjugate. Same with energy and time. Also angular momentum and angle...
My previous comments about Heisenberg being the audio equivalent of Godwin come to mind.
It's not intended to have a simple answer. Should we care only about fT when considering BJTs for a NFB amplifier?
Claude is arguing that the fbeta (knee frequency) is higher for the BJT with lower beta and this will result in the phase shift being less at high f that the BJT with higher beta. If we're defining speed by size of current gain phase shift then the beta is very important. But you may think another speed measure is more relevant.
One thing that is of interest, but perhaps not relevant to audio application of BJTs is that, a long, long time ago, in the days of TTL logic (before Shottky TTL), Gold doping was used to knock down the minority carrier lifetime so that logic switching involving transistor saturation was faster. This, of course, knocked down the beta.
So, in this very limited context, the low-beta transistors were "faster".
Cheers,
Bob
Hi Bob,
That is a good piece of info. Also worth mentioning is that ft, beta, & Vcebr (collector-emitter breakdown voltage), form what I call a "tradeoff triangle". Improving one of those parameters usually happens at the expense of one of the other two, or both. Likewise, improving two of them degrades the third.
Regarding traderbam's puzzle, it comes down to what is meant by "faster". I used a simple small signal 1-pole approach, which may not be suitable for all applications. With digital logic, I'm sure that my 1-pole model is an oversimplification. Again, "faster", "speed", and/or "bandwidth", have to be defined in order to even discuss anything further.
The freq where the transfer function is down 3 dB is not always used as the BW measure. But for a simple 1-pole network, it typically is defined as such. The bjt has internal time delay parameters as well. I wasn't sure how deep we were to delve into semiconductor physics.
Obviously there is some benefit to bjt device "a)". For the same ft, a larger lf (low freq) beta value certainly offers tangible benefits. So it comes down to how we define "faster". In control systems theory, 2 systems with differing break freq exhibit differing rise times. So the higher break freq is usually considered to be a faster system. Did I omit something?
It all comes down to definitions, and considering all relevant facts involved. If I've missed something, hopefully someone will bring it to our attention.
Member
Joined 2009
Paid Member
And there's the rub with many a specification stated as a single number (i.e. Ft) - the reality is always a curve and interdependencies.
We are probably over analyzing this.
Hi,
just shortly,
the Fourier transform (provides a linear time-frequency representation, expressing one-dimensional signals with a single formula, in which frequencies are constant over time) is based on the uncertainty (Heisenberg inequality) principle that states: the more precisely the position of a particle is determined, the less precisely the momentum is known in this instant, and vice versa. The processed signal can be either analyzed with good time resolution or good frequency resolution. Again, music signals are extremely complex, totally varying in nature (frequency, amplitude and phase, full of short-duration transients and bursts containing an infinitely dense packing of an infinite number of frequencies; an impenetrable jungle, a nightmare for any analyzing method. In my opinion, the degree of inaccuracy renders the analysis worthless.
just shortly,
the Fourier transform (provides a linear time-frequency representation, expressing one-dimensional signals with a single formula, in which frequencies are constant over time) is based on the uncertainty (Heisenberg inequality) principle that states: the more precisely the position of a particle is determined, the less precisely the momentum is known in this instant, and vice versa. The processed signal can be either analyzed with good time resolution or good frequency resolution. Again, music signals are extremely complex, totally varying in nature (frequency, amplitude and phase, full of short-duration transients and bursts containing an infinitely dense packing of an infinite number of frequencies; an impenetrable jungle, a nightmare for any analyzing method. In my opinion, the degree of inaccuracy renders the analysis worthless.
traderbam,
the beta/fT relationship has a device-specific characteristic, it strongly depends on biasing (chiefly on Ic), topology and other factors (source, load) in a particular circuit. Beta is a function of Ic, temperature and Vce, falling with increasing frequency, lowering the input impedance at these frequencies. Beta is anything but a constant value. Beta and fT have a different meaning in different topologies.
the beta/fT relationship has a device-specific characteristic, it strongly depends on biasing (chiefly on Ic), topology and other factors (source, load) in a particular circuit. Beta is a function of Ic, temperature and Vce, falling with increasing frequency, lowering the input impedance at these frequencies. Beta is anything but a constant value. Beta and fT have a different meaning in different topologies.
Look at the value of Cob that has a huge impact on high frequency performance and fT versus Ic in the used range. Data sheets do not tell everything about linearity, especially not in terms of sonic properties.
it s your post that is non sense...
the mathematic formulas that are used to describe
electromagnetism and quantum physics did exist
well before the scientists even heard of the existence
of electrons...
you make confusion between the subject (physics)
and the tool (mathematics) that is used to describe the
former...
It has nothing whatever to do with Heisenberg. That's plain and simply wrong.
Wahab, Fourier transforms, a well-developed formulation which are useful for describing classical phenomena, cannot possibly be based on a quantum formulation developed a hundred years after Fourier's death. The other way around, maybe, possibly, in the sense that both involve conjugate variables. However, the commutator between frequency and time in a classical system is zero, they are NOT (in this context) canonical variables in a QM sense.
HUP is irrelevant. It's brought into audio discussions to make them sound more profound than they really are. Fourier uncertainty is really what you guys are talking about, but it doesn't sound as cool.
But our hearing is based in time to frequency domain conversion, so we suffer all that lack of precision too
btw: I never expected the non-sense to go so far