"Patrick Turner" wrote

I wonder how these guys would get on if there wasn't any spice
simulation programs at all.

I guess they'd have to use a ruler and pencil to work it all out,
like they always did and some of us still do.


Making a new effort in the new year, I won't say that Patrick is
deluding himself for his habitual purpose of belittling others.

Simulation is additional to what you do, Patrick, not instead of.

You may have noticed that Robert starts with the curves of a real
valve, then produces a model that reproduces them in simulation.
They are not exact, and so he must make judgements about what
aspects of the fit are most important, so he can minimise the
practical error and state limitations to use.

That takes exactly the kind of understanding of data that you are
talking about. So if he were denied a simulator, he couldn't do what
you do instead. He already does that. He would be just denied a
simulator, and very upset no doubt that he had invested so much
effort into something that got taken away.

What we actually use simulation for is in designing circuits, not
valves. It can save the kind of calculation that you can't do
anyway, because you don't even know what feedback is. Oops...forgot
about new year.

For example, you have an idea for an amp design, and you have a
specific objective...let's say you want to minimise the maximum rate
of change of odd harmonic distortion with respect to amplitude.

That would be a tall order for your back-of-an-envelope arithmetic.

Being honest, and critical, it would be a tall order for my
simulation too. It is possible, but calls into question the true
value of simulation.

When Robert, or anyone else I can find produces a model, he adjusts
parameters until he gets a reasonable fit. If he is wise, he will
load the inevitable error where it least matters for the intended
purpose. But when doing so by eye, it is impossible I claim to make
a reasonable judgement about what kind of distortion the error will
produce in a typical circuit. In fact using Robert's method (which
is fairly universal), it is impossible to accurately reproduce the
distortion characteristics of real valves anyway.

The example I gave would require that the distortion would be
correct over a wide area of the characteristic curves. Not much
chance of that.

The genuine limitation of simulation is the models. The common
approach is based on the same pair of functions you will find in
RDH, with one set of parameters applied across the entire range of
operation. A better approach would be interpolative. Different
parameters could be used for each of a set of representative anode
curves, and then run-time interpolation could be used. Trouble I
have here is getting interpolation to work in spice 3. It also gets
very slow to run.

And difficulties in modelling valves pale in comparison to those in
modelling inductors anyway...that's where the real log-jam is.

Then I'd like some built in filter programs, they'd be real handy
for speaker crossovers etc.....

Now that *would* be trying to replace your calculations, which
rather misses the point, again. There are plenty of crossover design
programs if you need them.

I use the back of an envelope. Then I make a model. Then I use the
model for some purpose, perhaps as part of a simulated load for
testing a simulated amp. There are some areas of overlap. For
example, if I wished to know the peak current in the inductors of
such a crossover, I would use my simulation whereas you use the
envelope. But that's only because I have made the model already and
it's handy.

I have not found any program which is a free download, and that
treats
me like a
complete idiot, and attempts to talk me right through the basics
step by idiotic step starting with No 1, and finishing with
No 3,427, if that's how many steps there are to get a single
transistor or tube gain stage working.

If you can drag and drop, and pick from menus, you can use a decent
simulator. You are being silly. A single transistor doesn't work
does it? Not without something else attached. Simulating a circuit
takes as many steps as wiring it up and testing it in reality, each
step is the same as if you were working on your bench. You are
perhaps confusing modelling with simulating. That is like confusing
making valves with making circuits with valves in.

Good simulators are expensive unless you work for a corporation in
that line of business. Cheap spice is very limited.

I could teach you my simulator in a day Patrick, if you were willing
to learn. Free stuff (including circuitmaker) is quirky and/or
unreliable and often difficult to install. If you bought one you
would get a manual. You need to be a geek to use an application
without a manual. I got mine given by a friend.

cheers, Ian