Re: Tube E or not Tube E (Long as hell)

From Mark Pulver
Sent Wed, Dec 4th 1996, 15:36

>Ok I know I forgot to mention limited slew rates. I was kind of leaving it
>till I'd chewed it over a little more and forgot to mention it entirely. The
>reason I needed to think about that one a little is thus. A Slew rate
>limiter is a first cousin to a low-pass filter. Only slight differences but
>the same over all effect. Fast transients get slowed down. Or their
>amplitude rolls off. A device for limiting slew rate is usually made with an
>integrator.

But... You're missing the main issue with tubes... There is NOTHING linear
about them, and very few aspects are totally predictable.

You can sit with an analyzer and watch a response on a tubed circuit for a
given input, but what if that input signal is just a hair hotter, or colder
for that matter?

In a tranny circuit, there is a good amount of pure math that can reliably
predict the outcome. In a tubed circuit, math is tossed aside when it comes
to the finer points. There's no common saturation point even across a set
of matched tubes... There's the issue of the plate and grid voltage being
MUCH high than the base/emiiter potential of a tranny. If I have a .1%
fluctuation of base voltage in a transistor circut, while signifigant, it's
nothing compared to a .1% flucuation in the plate voltage of a tube
circuit. How does this voltage change effect the output of the stage? Very
little in a transistor based circuit, quite a bit in a tube circuit.

You also talk about simulating the randomness of a tube circuit digitally...

>From being a low level coder for eons, I can tell you that there is no such
thing as a random digital circuit. I can make something pseudo random, but
not truely random in the sense that we're talking about here. Sure, I can
introduce bit errors into a signal, but keep in mind that we're talking
about (at 16 bits) 65536 steps to describe a 1-2v signal. If you're gonna
only flip the lower two bits, then you're only introducing a .00003v change
in a 2v signal. A  tube circuit is MUCH hotter than this, and wildly
whipping level changes of 100mv or more is much more common.

To take a digital circuit up to that level of change, you need to start
flipping 7-8 bits, and that's gonna seriously cause problems elsewhere down
the line.

A tube will also age, often in wonderful ways, over time. Heat and current
are great aging factors on the tube. Heat and current will kill a TTL circuit.


>It also reminds me of when I use to work in the television industry. I think
>everyone's probably noticed that Cine film looks far superior to video. Why?
>Well cine film has thousands of times more resolution. The equivalent of
>hundreds of thousands of pixels per line.

Wait... You're talking analog and digital here...

In video, these days, you're shooting through a CCD device, pure digital.
The CCD will have a 16/24bit resolution on a "scan line" as it digitizes
the image.

In film, you have (basically) infinite resolution since you don't just have
65536 levels for a scan line, you have all the wonderful stuff in-between.
Plus, there are no "scan lines" at all! The film will react chemically to
the light reflected from the image, a full frame at a time, all in one
split second.

You go on to say this in industry terms, but using film as an analogy shows
the same reasoning on the tube/TLL/digital issue.


Good thread...

Mark

---------------------------------------------------------------------------
 Mark Pulver/Chicagoish, IL             The home of "Son of The MIDI Wall"
 mpulver@wwa.com                           http://shoga.wwa.com/~mpulver
 The Gear Wanted List is online! http://shoga.wwa.com/~mpulver/gwlist.html