I would really like to understand what youre talking about
but thanks to guys like you we have got all the musical nuances
)
Sorry for the long posts, and I could be wrong about some of this, but I still hear the same things I described. I don't know if any of my posts will make any difference in the future firmware, but I feel compelled to share my perceptions and hope to get the perceptions of others and maybe a patch that clearly demonstrates I'm wrong. I have no problem being wrong.
Until FW 6.00 comes out all of my observations are moot to an extent, but if it helps provoke an improvement in future FW even if it's 7.0 or 8.0 then it was worth it.
If I had my druthers I would reduce the range on the Xformer Match control.
0.1 to 10.0 is a range of 100, and if that represents the turns ratio then it represents an impedance range of 10,000:1! It doesn't sound to my ears like it's changing the impedance ratio by a factor of 10,000. I'm guessing the parameter values must represent the impedance ratio and not the turns ratio based on what I'm hearing. Even an impedance ratio of 100:1 (the range of the values of the control) is a huge range.
Assuming it DOES represent the turns ratio, if the range were changed to values between 0.5 to 2 with 1.0 being the nominal ratio, that would represent a change in the turns ratio of 4, and an impedance range of 16:1 instead of 10,000:1.
Changing the impedance ratio of the OT by a factor of 16 is still huge and would have a dramatic effect on the sound in a real amp. I just don't get the 0.1 to 10 range - it seems like something that would be impossible to model and have work properly. If the delta in the turns ratio were limited to 4:1 it seems to me that a realistic model could be developed that reflected the impedance of the tubes back to the secondary which drives the speaker and there could be realistic interaction between the output impedance of the secondary and the impedance curve of the speaker.
As it stands now, it functions, and
having the control is infinitely better than not having the control, even if it doesn't behave like a real amp. But the distortion produced by pushing too much current through the power tubes into a low impedance primary of an OT is not the same as the distortion produced by a low power tube current into a high impedance OT to create voltage clipping since the voltage is limited by B+, and
these two different ways of creating distortion operate on a different part of the non-linear curve of the power tubes in a real amp, but to my ears the XFormer Match sounds the same as whatever the master volume control is doing which in a real amp is increasing the current in the power tubes into a fixed impedance.
Again, not a complaint, just an observation.
To go off topic for a moment, I've been studying the Analog Devices website and reading their introductory literature, and also started learning C++. Analog Devices has something called DSP++ that runs on a PC and can be used to simulate and develop DSP applications that will run on their products including the TigerSharc. All I can say is this stuff gets complicated!!! DSP++ is also extremely expensive although you can get a 90 day evaluation for free. I'm not even close to being able to use it though.
All the theory associated with this picks up just a little bit past where I left off in school, and I think it will take years to get to a point where I could create any sort of useful design of even a simple device. But you have to start somewhere. I was taught Fortran, and I was quite good at it, but it's a dinosaur and useless for DSP. Just learning C++ at an intermediate level is a challenge.
Anyway,
the suggestion to change the range of the Xformer Match should be taken with my humility in not understanding the process by which this is all being modeled. I'm not trying to be a know it all because I'm far from it by my own recognition and admission. I wish I knew more about this stuff but
will be slowly acquiring knowledge pertaining to how DSP works. I have a lot of high level math to learn (and some to relearn) while I'm at it. I'm giving myself about 5 years to come up to speed, longer if necessary. I'm an analog circuit designer, always have been, and now I want to learn how to simulate the stuff I've designed and sometimes built using DSP instead of etching custom boards and soldering analog parts onto them.
I have one more thought that I have to spit out. I was wondering if it would be worth it to create a
low res amp block model that used all the power of one processor. This would (theoretically I think) allow twice the complexity of the circuit model used in hi res mode and perhaps allow
a complete component level model of an amp with no shortcuts or simulations of circuit behavior. The model would work exactly like a real tube amp since it would have nothing but discrete circuit elements. Just a thought, again submitted with the humility that
I have no idea what this would entail or the degree to which the existing hi res amps are modeled at the discrete component level.
The
fact that a speaker with an impedance that varies greatly over it's frequency range doesn't have a changing frequency response when the turns ratio of the output transformer is radically altered
is proof beyond any doubt that the power amp's output transformer or adjacent components are not discrete circuit models but rather artificial simulations. There's nothing wrong with that other than the fact the controls (which in all fairness don't exist on real amps) don't behave like they would if they actually did exist on real amps.
Thanks.