I believe you not remember, but it was you who took the test:
http://www.diyaudio.com/forums/solid-state/125679-function-output-inductor-25.html
Not ones I have much confidence in. I posted some here:
http://www.diyaudio.com/forums/soli...one-seen-front-end-before-23.html#post3478096
http://www.diyaudio.com/forums/soli...one-seen-front-end-before-23.html#post3478096
This sort of thing always fascinates me. It kind of goes against our engineering intuition. When a small change makes a big sonic difference, my engineering intuition tens to make me wonder if some misbehavior was "on the edge" and that the small change improved the sound markedly by bringing the design back away from the edge of misbehavior.
But this may not always be the case, since we need to allow for some things we do not explicitly understand or measure.
The other thing that usually comes up in my mind is "how did you know to make the change in that particular component?".
Cheers,
Bob
Bob, you make sense to me.
If Kean's schematic is so sensitive to very tight tolerances then there is probably something else going on.
Whereas, when JLH tells us to trim the R of the R+C feedback, he shows us precisely what to look for, rather than "it sounds better because I told you so".
For me the changes usually involve topology rather than specific component values, except when a manual adjustment is necessary for the topology. I've found that without changing topology, components change the sound on a continuum but don't affect the fundamental quality of the circuit. If the topology can't be changed but the result still disappoints, then I will try tweaking components as a last resort, but this is usually much less effective.
I use the word topology with a very specific meaning. Adding or removing a component is a change in topology. In a limited way you can change topology without adding or removing components. For instance, if you change the relationship of poles to each other, this counts as a change in topology. Likewise, changing harmonic structure can be treated as a change in topology because the relation of one transfer curve to another is flipping. A harmonic null would also count as an important topology aspect.
Usually if the problem is fatigue or harshness, I will adjust compensation or put a ferrite bead somewhere. For instance right now I have a highly improvised IRF250 class A amp. Harshness was eradicated by placing a ferrite bead at the output. A 500nH/6R network didn't work - but the ferrite bead did. Unfortunately I can't find the info on my beads to know what L/R network they are equivalent to.
If the amp doesn't have basic imaging or realism, I work on compensation details and distortion. Usually after messing with these I find an option that works. With every improvement the sound is liberated from the speakers and fills the room more. I will note that in practice, I have found that lower distortion is usually better and if not, it's not because distortion is needed but because something else is wrong. After I fix that, lower distortion always seems to sound better.
I use the word topology with a very specific meaning. Adding or removing a component is a change in topology. In a limited way you can change topology without adding or removing components. For instance, if you change the relationship of poles to each other, this counts as a change in topology. Likewise, changing harmonic structure can be treated as a change in topology because the relation of one transfer curve to another is flipping. A harmonic null would also count as an important topology aspect.
Usually if the problem is fatigue or harshness, I will adjust compensation or put a ferrite bead somewhere. For instance right now I have a highly improvised IRF250 class A amp. Harshness was eradicated by placing a ferrite bead at the output. A 500nH/6R network didn't work - but the ferrite bead did. Unfortunately I can't find the info on my beads to know what L/R network they are equivalent to.
If the amp doesn't have basic imaging or realism, I work on compensation details and distortion. Usually after messing with these I find an option that works. With every improvement the sound is liberated from the speakers and fills the room more. I will note that in practice, I have found that lower distortion is usually better and if not, it's not because distortion is needed but because something else is wrong. After I fix that, lower distortion always seems to sound better.
Hi Andrew
Sorry for the confusion, I was talking about another issue about the use and nonuse of output inductor a few posts ago.
http://www.diyaudio.com/forums/soli...lls-power-amplifier-book-296.html#post3483601
There may be some potentially confusing semantics at work here. I would normally consider removing a component a change in topology. I do understand that gradually moving the value of the component to zero migh result in the same thing in the limit, but I'd consider that a special case.
If you put in a ferrite bead and the sound becomes better, I'm guessing that the ferrite bead eliminated an instability or oscillation somewhere. This can especially be the case with vertical MOSFETs, which can be very fast devices with high equivalent ft.
Cheers,
Bob
Right on, and that's why I bought a bunch of ferrite beads. They are really useful for troubleshooting harshness issues. I came to the point where I wanted to put L/R networks at the bases of BJTs and then realized ferrite beads would save me a lot of work. If you buy ones with high R value, you can parallel them with a discrete resistor or even trimmer and experiment.
When I describe "ringing" I didn't mean to imply that the ringing is heard directly since it is at RF; but it seems to be the dominant cause of certain effects that are heard. I didn't think someone might think I meant otherwise.
I probably shouldn't have used the word topology. It seems to have multiple definitions and a vague meaning for many people. But what word would work better for what I describe? What I describe as a topological change is basically a change in the mathematical construct of the amp even if the physical construct of the amp doesn't change. For instance if an EQ network had the same physical arrangement but two poles were transposed, by changing values, I would consider this a change in "topology" - because the pole construct is just as real as the physical construct.
When I describe "ringing" I didn't mean to imply that the ringing is heard directly since it is at RF; but it seems to be the dominant cause of certain effects that are heard. I didn't think someone might think I meant otherwise.
I probably shouldn't have used the word topology. It seems to have multiple definitions and a vague meaning for many people. But what word would work better for what I describe? What I describe as a topological change is basically a change in the mathematical construct of the amp even if the physical construct of the amp doesn't change. For instance if an EQ network had the same physical arrangement but two poles were transposed, by changing values, I would consider this a change in "topology" - because the pole construct is just as real as the physical construct.
Hi Bob, I've seen this concept also in your book. What exactly is the equivalent Ft for a mosfet? For a bipolar, ft is the current gain - bandwidth product, which has no equivalent for a mosfet.
One suggestion would be to define ft for mosfets as ~ gm/Cgs (following the bipolar approximation of ft ~ gm/Cbe), this is what you had in mind? This approximation holds for small signal analysis, but would fall flat for large signal (in particular for vertical mosfets used in power stages). So...?
Some in the high end eschew the use of ferrite beads, and even the use of resistors with plated steel leads. Their concern is that anything with electromagnetic properties can be a source of nonlinearity. I tend to stay away from them for this reason, but I must say that I have difficulty justifying it based on numbers put to theory. There are many things in high-end audio like this that are not well-understood.
As for the semantics of topology, I don't think I can come up with a good word for what you are describing.
Cheers,
Bob
Hi Waly,
You are exactly right. I define the ft for a MOSFET in the same way as for a BJT - gm/Cbe. You are right that it largely applies in a small-signal sense, partly because gm and Cbe change differently from those of a BJT as a function of current. Until the BJT hits the ft droop current region, ft is somewhat flat with current, since both gm and Cbe increase in proportion to collector current. ft for a vertical MOSFET is a function of drain current.
Cheers,
Bob
One concern is that if the core saturates, instability will result. I don't know if any vendors specify the saturation current on their datasheets - perhaps the beads are designed so that they will not fit around a wire large enough to carry a saturating current.
In any case, if a ferrite bead causes so much trouble then the speaker magnet must be terrible!
They are very helpful for prototyping, but I still tend to prefer a more elegant solution if available. After all, ferrites don't come in E12 values.
If it's a useful concept, we can make up our own word.
In any case, if a ferrite bead causes so much trouble then the speaker magnet must be terrible!
They are very helpful for prototyping, but I still tend to prefer a more elegant solution if available. After all, ferrites don't come in E12 values.
If it's a useful concept, we can make up our own word.
They ARE terrible!
But not as bad as Neodynium. This must be the worst material commonly used for speaker magnets. The only advantages of Neo is 'cheep en small'.
The best magnet material for speakers for performance is Alnico. Samarium Cobalt is 2nd.
If I was making the ultimate speaker, I'd use Alnico and deal with the manufacturing problems.
There's an old JBL paper in JAES that explains the evils of ferrites and how to design them out to get nearly Alnico performance. The sins enumerated also apply to Neo but to much greater degree.
