|
|
我google 左兩下"slew ratio" 係咩野發覺原來好重要!
我吾明旦盒點可以加快slew rate, 更加吾識點量 ...
Lester 發表於 2012-9-19 23:40 
Nelson Pass 
One of the most fascinating tests I ever ran was one of those "let's go see how many teeth the horse really has," as opposed to discussing it, kind of things. By the way, Peter Walker of Quad did exactly the same thing at the same time—two minds, at least, thinking alike. We decided to find out in real life what kind of slew rates were really occurring. Then, operating off that basis, you can come up with your 10:1 figure that you need to achieve. So I built a differentiating circuit—this was pre-digital--and using a very high-quality moving-coil cartridge and step-up system that was certainly not compromised with regards to bandwidth, I simply played music into loudspeakers near the capacity of the amplifier, which was 200Wpc on various types of recorded material.
On record ticks and pops I could register slew rates as high as 20–30V/µs—real signals that were being processed through the differentiating network and then onto a storage 'scope. So it came time to play some music. Again, if you clipped the amplifier you could see 40 and 50V/µs recovery; that is to say, the amplifier, in catching up from clipping to get to where the signal was, would travel that fast to catch up. So clearly the system—the amplifier and the measuring system—were capable of resolving 40 and 50V/µs transients. In playing music we were looking at piano, cymbals, violins—a large range of instruments all played loud, near the power capacity of the amplifier. And we were getting values like 0.1V/µs, 0.5V/µs. In fact, at that time in that series of tests, the highest transient we were able to achieve was about 1V/µs. So the 10:1 ratio would dictate that you would need at least 10V/µs performance. That was somewhat faster than some of the slow amps that were being criticized at that time, which were in the 4 and 5V/µs range. But clearly, I didn't see where slew was overloading them as such.
Norton: You mention that the slew rate of record ticks and pops was quite a bit faster. Was their level so low that it didn't cause problems?
Pass: No, it was a factor. Subsequent to that, though, the Sheffield Drum Record was used in a test and we started getting figures into the 5V/µs ballpark. And those were rim shots. In fact, as far as I can make out, rim shots present the fastest transient that I've seen on a recording. The Sheffield Drum Record was the fastest example that stands out. I haven't seen anything like it since. But given that you're on the 5V/µs rate, we ultimately settled on approximately 50V/µs as a speed beyond which was fine, but it became a minimal figure that we deal with in our designs.
There are other design elements where we depart from the mainstream. One is that we use single-ended input stages and initial gain stages. The output stages are complementary; push-pull. That's a necessity: a single-ended stage in that area would make class-A look efficient. Push-pull class-A has a 2:1 efficiency, a 50W output requiring an idle of 100W. But a single-ended 50W output requires an idle of 200W; not only that, it current-limits at exactly that point, no more, whereas a push-pull can at least extend in higher currents into class-AB.
So there is every reason in the world why nobody's building single-ended output stages. But those conditions don't constrain the rest of the front-end circuitry which doesn't have to deliver a lot of current. The overriding consideration here—and this relates to people's perception of absolute phase; audiophiles who believe in the importance of absolute phase are, I believe, essentially correct—is that air is single-ended and is not symmetric with regard to plus and minus. True, the nonlinearities and the asymmetry are quite small in the level that we normally like to think of, but in fact, when you're in the high end you're dealing with subtleties. This is a subtle thing, but it's real.
It so happens that if you vary pressure around, say, a static 14psi pressure, you can't go any lower than zero, but you can certainly go much higher than 28. The characteristic of air is not symmetric in both directions. This means that there'll be a different characteristic off of say, a, bass drum pulse, positively phased vs negatively. I think that that's one of the areas where absolute phase has some measure of importance. And because air is single-ended (I think that the ear is also a single-ended mechanical structure), single-ended distortions don't stand out as much to the human ear and brain as symmetric distortions. And so I prefer even-ordered harmonics: second-order over third-order, fourth-order over fifth. That's consistent with a slight asymmetry.
We always set up the amplifiers so that the asymmetry of the front end is the same polarity as what we expect that of the signal to be. And what we expect to see in the case of the way the atmosphere will behave responding to the loudspeaker. Toward that end, that's why we use single-ended gain stages in the earlier stages of an amplifier. This is in contrast to quite a few other people who go with push-pull symmetry, which has a way of lowering distortion but also pushes the distortion energy into the more-discernible odd-order harmonics. |
|
[複製鏈接]
發表於 2012-9-20 00:57
爆炸.....
樓主
, 有result 我會POST出嚟.
