On 2/8/2011 8:48 AM, NL7VL wrote:
Jerry,
That's quite a bit to digest, with a lot of it over my head, but I do
see things I agree with and some things I had not realized. I'd like to
make sure I get this if you don't mind.
There are many factors in sound equipment, some due to active device
nonlinearities, some to clipping characteristics, and some to
impedance matching.
Solid state circuits tend to swing almost all the way to the supply
and ground or + and - supplies and clip hard. Tubes tend to go
nonlinear before reaching the supply limits and so clip more gently,
from voltage rise on one extreme to current saturation on the other
extreme.
OK, so that's one reason why a solid state amp sounds muddy when it
clips and a tube amp will gradually distort. Turning a preamp up -
whether it be solid state or tube, will gently push the tube into its
nonlinear region.
Without massive feedback tubes, transistors, and fets are nonlinear in
different ways. Different tubes have different transfer curves, some
less linear than others.
There tends to be a whole lot more feedback in solid state circuits,
especially in low level stages using OP amps with open loop gains
100,000 times the closed loop gain. While that takes out almost all
of the nonlinearities but retains the supply voltage hard clipping
limits. Sometimes there's some feedback around tube output stages and
output transformers, but not 99.9% like an solid state output stage.
Then the passive devices that couple the tube stages and contribute to
a tube amp's distortion characteristics? (Also at your last point,
too.)
Yes. Even resistors can be a bit voltage sensitive changing value
slightly as the voltage changes. Capacitors have storage effects and
leakage that depends on the voltage across them, more of a factor with
electrolytics.
With that high level of feedback, the solid state power amp's output
impedance is very low, far lower than the rated speaker impedance.
With a tube PA, the load impedance is matched by the source impedance
of the amplifier, in a solid state PA, the load impedance is the
lowest impedance that doesn't overload the PA devices or the power
supply. Its not the impedance that would have half the peak to peak
voltage that the amp did open circuit like it would be on a matched
tube PA. That low impedance adds much speaker damping so that the
speaker cone position is controlled closely by the AC voltage. Which
keeps a speaker with poor acoustics from ringing from transients as
much.
As I think about this, the low end of the solid state tends to
reproduce more cleanly than the tube, and this is because low impedance
of the output transistors?
Yes, the speaker has a fundamental resonance at LF and if the enclosure
doesn't damp that resonance you get a boomy bass that's always the same
pitch. The Q of the enclosure and its resonant frequency have to match
the speaker. So the low impedance from the amp can damp that speaker
depending on the impedance of the connecting wires. There's where
monster cables may have an effect. But to counter all the golden ear
audio ideas there are the whole house systems with speakers stuck in
walls and ceilings without enclosures wired with #22 through resistive L
pads for local level control.
As for power rating, the advent of solid state has created new vistas
of audio power rating specmanship. One of the amps I liked the best
was good for a whole 0.2 watt but was class A solid state. It was
clean. But most musical instrument and home stereo amps are rated at
50 to a few hundred watts. Speaker efficiency hasn't gone down,
ratings have grown with "inspired" techniques. Most are peak power,
not average, but peak to peak voltage squared divided by the load
(which gives a number conveniently 4 times "RMS Power"), and if
measured are measured with the supply the amp would be delivered with
replaced by substantial voltage regulated supplies so the power out is
what it would be on the leading edge of a key closing tone before the
power supply drooped (leading to lower power and clipping). I suspect
some is rated with a much lower load impedance than the amp is rated,
but since the solid state amp is a constant voltage source with very
low output impedance the amp does more power the lower the load
impedance, until its power supply croaks or the output devices melt
off their emitter lead inside the device package. The result of these
subterfuges is that an amp that probably really delivers 10 watts,
might be rated at 200 watts. Of course without the regulated supplies
and extra output device cooling not shipped with the amp, it can only
do that power if measured as peak to peak voltage squared divided by a
load impedance of a fraction of an ohm (where normally rated for 4
ohms minimum load Z) over one or two cycles of a 10 kHz keyed tone.
After a very short time, power supply droop and output device heat
takes over to limit the useful real output power.
I wonder if some of the unbelievably high solid state ratings are
taking into consideration the slewing power of the circuit - its
ability to handle very sharp transients like the crack of a snare drum
etc. Yesteryears speakers could not handle what is expected of them
these days. Even back a few years, it wasn't that difficult to fry the
voice coil with highs or separate the paper from the coil with lows.
The new speaker engineering really impresses me.
The power rating is inflated to show those transient capabilities then
rated at the peak to peak voltage instead of the RMS voltage on those
peaks and since the amp is a voltage source, I suspect rated while
running a low impedance load that would fry the amp if run at that
impedance for more than a split second.
All it takes for feedback to a guitar's strings is delay from
circuits and the acoustics path between the speaker and the guitar to
have one cycle or an integer multiple of one cycle time delay. It
shouldn't matter much what the active devices are.
Hmm. I'll have to look into that one. I believe you, I've just never
considered the concept before.
Fundamental for any oscillator from ULF to ultra violet, output has to
reach the input through a delay loop that makes the loop phase delay
(including the gain stage(s)) an integral multiple of 360 degrees and
the loop gain has to be at least 1.0. And sound travels something like 5
seconds per mile velocity. About 1000 feet per second. So the wavelength
at 1 KHz is 1 foot. At 100 Hz is 10 feet. Sometimes in a guitar amp
there's a phase inverting switch just to cure unwanted feedback or to
enable it. And there's attenuation because the wave from the speaker
spreads in a semicircle so the energy density is dropping with the
square of the distance from the speaker.
Tube preamps tend to not have the feedback of the solid state op amp
based preamp because its too hard to get excess gain in a tube amp
without getting hum and noise. To hard to get excess gain so there's
no serious feedback beyond a partly unbypassed cathode resistor. But
that makes the stage more susceptible to heater cathode leakage
injection 60 Hz hum.
I see. But what about microphonics in tubes? I swear at times I can tell
when someone is running tubes, whether it's CW or SSB. I can't imagine a
solid state amp with those "ringing" attributes. And to me, it has the
same quality that my Hammond M-3 would have when I would get a little
ruff with it - or a guitar pick plucking strings over a really hot
humbucker pickup.
We tried to ignore microphonics, but replaced many a tube because of them.
The there may be come effects from the coupling and bypass
capacitors, tubes using paper or mylar caps while solid state with
inherent low impedance uses miniature electrolytics of dubious
quality and not quite perfect approximations of real capacitors.
So this get back to the point above - that some (of the older?) passive
components can contribute to tube feedback effects through unintended
coupling - yes?
Even more so with the poorer quality electrolytic caps used in solid
state amps. Old electrolytics as power supply bypasses contribute
unwanted feed back, eventually noticed as radio or amplifier
motorboating, or in the worst case oscillation at mid audio frequencies.
73, Jerry, K0CQ
|