[Ardour-Users] New tube amps

[Gordonjcp]

On Tue, Dec 26, 2017 at 11:34:44AM +0000, Will J Godfrey wrote:
> <some snippage>
> 
> >> http://gjcp.net/solidstatevalve.jpg
> 
> <and some more>
> 
> First, let's be clear. I am not suggesting a valve amp can't be emulated, but
> your drawing is, to put it mildly, 'basic'.
> 

Yup!  Not quite as basic as it comes, but pretty damn basic.  I wouldn't
say it's exactly an emulation of a valve preamp but it does roughly get
the shape of a variable-mu pentode right and I'd challenge anyone to
tell the difference from the audience side of the monitor wedges :-D

> A friend of mine retired recently (well into his 70s) from a company that does
> indeed make amps with extremely good emulation, and if it's good enough for
> him, having cut his teeth on valve stuff, it's certainly good enough for me -
> a mere refugee from the valve TV days.

Ah, I technically missed the valve TV days but while I was in highschool
I and many of my friends had "third-hand" black-and-white  tellies tha
had done duty in the living room, then migrated to our parent's bedrooms,
then into ours as nice new colour ones replaced them.  Living up north,
the good old BRC 1400 chassis was a favourite and I can still just about
remember some 30-odd years later which 220k resistor you needed to
change to fix the uncontrollable brightness fault - but the exact
details of the Teletext Killer that improved the flyback blanking so you
didn't get a scatter of dotty lines over the top third of the screen
escapes me.

> Your drawing looks similar to one I've heard in operation, and where it failed
> was on long slowly decaying single notes. As the volume dropped you could
> clearly hear the steps in the harmonic content change. This of course doesn't
> happen in a real amp where the non-linearity is a continuous sweep until actual
> hard limiting is reached (quite difficult to get to).
> 
> A silicon diode is the worst possible device for doing this. It starts to
> conduct at about 0.4V and flattens off at 0.6. I've heard much better results
> from the now totally obsolete OA81. Being germanium, these start to conduct at
> 0.1V and level off around 0.5V (by which time you're at their disipation
> limit). Therefore a chain of these will give a closer analog to the real thing.

I somewhat agree, but in this case you have such an unholy amount of
current through the diode chain (making it unsuitable for battery
operation, unless you build a pedal that can take two car batteries)
that they transition between stages fairly smoothly.  The idea is that
you're using the nonlinear portion of the curve to affect the amount of
feedback, and compressing the gain the hotter the signal gets.

> You'll get even better results using an FET, but by then you might as well use
> a JFET wired in a similar way to a triode! Wire two in cascode configuration
> and you'll get somewhere near pentode behaviour.
> 

A long-tailed pair has actually got quite a nice saturation curve too,
possibly even better than a FET for this sort of application.

> Talking about pentodes, why on earth emulate an EF86? They were horrible
> valves. Fragile, microphonic and frequently suffering from heater-cathode
> leakage - everything you *don't* want in an AC heated input valve... in kit
> that's going to be rattling around in the back of an old Bedford van. The only
> ones I saw were in amps people were desperately trying to get rid of!

Because I've had a soft spot for them ever since I built my first valve
guitar amp when I was a spotty teenager, out of the modulator chassis
from an old Murphy AM VHF transmitter.  It developed about 50W from two
EL34s, with an ECC83 gain stage, ECC82 buffer and phase splitter and -
tada! - an EF86 for the mike preamp.  It also used an EB91 wired as
antiparallel diodes as a clipper, and I tweaked some component values to
make it flatter since it had "radio comms" emphasis on it.  I needed to
rewind the secondary of the output transformer to drive an 8 ohm speaker
since the original one was intended to go in the anode circuit of the
valve RF power amp.

> Back to subject, your drawing seems to try to combine the features of the input
> stage (creating mostly even order distortion) with the output stage (mostly odd
> order). This takes away control from the guitarist who may want to overdrive
> the input for consonant harmonics with just a little compresssion, then with a
> quick flick of the wrist change to clean input and heavy overdrive of the
> output.
> 
> Added to this your need for EQ, if it was adequate to tame the sound for open
> strings, would completely flatten the 'edge' when working high up the
> fretboard. Watch the best Rock guitarists. They are all over the fretboard!

It works well for a fairly bluesy rocky style of playing.  You can coax
some fairly convincing Santana-like tones out of it if you give it a
good whack of signal.

Again, it really needs a good powerful output stage that you can keep
clean, keeping all the clipping in that stage.  The EQ doesn't really
need to be that special, just whatever you'd use in any other amp.

Give it a shot!  It's easy enough to knock up and you definitely have
the bits in your junkbox :-)

-- 
Gordonjcp