Advice: Amp building
 

"Keith G" wrote


Oh, sure....

You'll be telling people to get their HiFi from Argos next.....


Or even Lidl's:

http://www.mysilvercrest.de/en/artikel.php?a=45


Oh ****! I blogged again....

(Now I'll get into trouble.....)


:-(

Advice: Amp building
 

Jim Lesurf wrote:
In article , Don Pearce
wrote:
On 27 Jul 2006 06:01:27 -0700, "Andy Evans"
wrote:

c) Check his laptimes - if they are better, assume he was right....

But that assumption would be faith based, since you can't be sure what
to attribute the change to

Once you are at the stage of fine tuning, you know exactly what to
attribute changes to - I don't think anybody in motor racing employs the
Taguchi method. You make one change and measure - if it is better, go a
bit further until you peak. Then you start on the next parameter and do
it again. Once everything is as good as it gets, you go back to the
beginning and start again because everything interacts. Some parameters
you adjust in pairs because the interaction is first order A bit like
amplifier development in some ways.

You also try to build models, based on an understanding of the physics
involved. This helps you to assess the effects of interactions and guides
you in making modifications or changes and expliot the interactions rather
than find them a frustration.

The 'analogy' Andy gave was, actually, quite useful - it does show his lack
of grasp of the scientific method and how it can be applied in such cases.
The key point is, indeed, that you would *measure the results*. However it
is more than that.


What is 'the' scientific method - intrigued!

If you wanted to check if what the driver said was reliable, you would
repeatedly make small changes *without the driver knowing the details* and
then observe the results. You would do this enough times in a 'randomised'
manner to get results whose sigificance could be assessed for reliability.

This would mean that not only you could determine which 'changes' or
arrangements were 'best' for the task. It would also tell you if the
driver's opinions were any guide to this, or not.


Maybe, but you wouldn't necessarily end up with a very good, or
consistently quick, car.

Rob

Advice: Amp building
 

"Jim Lesurf" wrote in message
...
In article , Don Pearce
wrote:
On 27 Jul 2006 06:01:27 -0700, "Andy Evans"
wrote:

c) Check his laptimes - if they are better, assume he was right....

But that assumption would be faith based, since you can't be sure what
to attribute the change to

Once you are at the stage of fine tuning, you know exactly what to
attribute changes to...

Well, you've got a short list.

- I don't think anybody in motor racing employs the
Taguchi method.

I wouldn't bet on that. Racing is a big bucks business, and many of its
leading practitioners are very sophisticated managers.

You make one change and measure - if it is better, go a
bit further until you peak.

That's unecessarily slow. The next step of sophistication is to make
multiple changes that are as disconnected from each other as possible. Then,
you make interacting changes and use sophisticated analytical techniques to
unwind the interactions.

Then you start on the next parameter and do
it again. Once everything is as good as it gets, you go back to the
beginning and start again because everything interacts. Some parameters
you adjust in pairs because the interaction is first order A bit like
amplifier development in some ways.

Definately old school. Thing is, optimizing amplifiers is a chump game. We
have low cost high volume production amps that exceed any reasonable need.
Amps are only useful and interesting as components of more complex systems,
they are almost completely disinteresting as ends in themselves.

The 'analogy' Andy gave was, actually, quite useful - it does show his
lack
of grasp of the scientific method and how it can be applied in such cases.

It also shows his naivate about how the art of engineering is practiced
today.

The key point is, indeed, that you would *measure the results*. However it
is more than that.

The big plus to measurements is that they are relatively fast and reliable.
The down side is that their relevance to customer satisfaction can be far
more complex.

If you wanted to check if what the driver said was reliable, you would
repeatedly make small changes *without the driver knowing the details* and
then observe the results.

In fact there are two extremes of race drivers - those that are intimately
involved with the details of optimizing the car for a given race, and those
who are almost completely disconnected. Both kinds of drivers can be highly
effective and win races.

You would do this enough times in a 'randomised'
manner to get results whose sigificance could be assessed for reliability.

Race cars have had every imaginable sensor and telemetry device attached to
them. I worked for a summer in the early 1970s in a shop that among other
things built Indy cars. I still remember the busted block of an Offenhauser
4 leaning up against a back wall. I also remember the telemetry equipment
and the sensor arrays. That was all simple junk compared to what is done
today.

This would mean that not only you could determine which 'changes' or
arrangements were 'best' for the task. It would also tell you if the
driver's opinions were any guide to this, or not.

In the end its about gathering, analyzing and making decisions based on
information gathered as many ways as is reasonbly possible. The words of the
driver is just one stream of information of many.

Advice: Amp building
 

"Rob" wrote in message
...
Jim Lesurf wrote:

Maybe, but you wouldn't necessarily end up with a very good, or
consistently quick, car.

That's a false criteria, because many people use similar techniques and
tools to optimize race cars and drivers, and some end up with good ones and
some end up with also-rans. Furthermore, the winners and losers often
interchange roles on a weekly, monthly, and yearly basis.

For example, Jeff Gordon last year and this year. For example Jeff Gordon
last Sunday and about a month ago at the same track - Pocono.

When a whole industry changes from tubes to SS, it reflects positively on
the benefits of the new technology.

When the whole art of racing changes from hihgly subjective evaluation
techniques to the current heavy reliance on objective evaluation techniques,
it reflects positively on the benefits of the new technology.

People who don't study and learn from history just keep making the same
mistakes over and over again.

Advice: Amp building
 

On Fri, 28 Jul 2006 10:08:38 +0100, Jim Lesurf
wrote:

In article , Don Pearce
wrote:
On 27 Jul 2006 06:01:27 -0700, "Andy Evans"
wrote:

c) Check his laptimes - if they are better, assume he was right....

But that assumption would be faith based, since you can't be sure what
to attribute the change to

Once you are at the stage of fine tuning, you know exactly what to
attribute changes to - I don't think anybody in motor racing employs the
Taguchi method. You make one change and measure - if it is better, go a
bit further until you peak. Then you start on the next parameter and do
it again. Once everything is as good as it gets, you go back to the
beginning and start again because everything interacts. Some parameters
you adjust in pairs because the interaction is first order A bit like
amplifier development in some ways.

You also try to build models, based on an understanding of the physics
involved. This helps you to assess the effects of interactions and guides
you in making modifications or changes and expliot the interactions rather
than find them a frustration.


True - and these would be both physical models of bits of the
amplifier and computer models using Spice, J-Omega, MDS or whatever
suits the application. The usefulness of these kinds of model depends
on how well you have derived and applied the models.

The 'analogy' Andy gave was, actually, quite useful - it does show his lack
of grasp of the scientific method and how it can be applied in such cases.
The key point is, indeed, that you would *measure the results*. However it
is more than that.

If you wanted to check if what the driver said was reliable, you would
repeatedly make small changes *without the driver knowing the details* and
then observe the results. You would do this enough times in a 'randomised'
manner to get results whose sigificance could be assessed for reliability.


I suspect that Ferrari wouldn't try this with the likes of Michael
Schumacher - he is clearly a real "golden ears" when it comes to
racing cars, and he would easily tell the difference between what was
real and what wasn't. Also at about ten grand a lap that would be an
expensive game to play.

This would mean that not only you could determine which 'changes' or
arrangements were 'best' for the task. It would also tell you if the
driver's opinions were any guide to this, or not.


This might be a useful part of the interview for a new test driver -
maybe a little unfair though, because it would take time to attune
oneself to the finer subtleties of a car. I guess the expensive
drivers are the ones who can do that quickly.

d

--
Pearce Consulting
http://www.pearce.uk.com

Advice: Amp building
 

In article , Rob
wrote:
Jim Lesurf wrote:
In article , Don Pearce
wrote:


The 'analogy' Andy gave was, actually, quite useful - it does show his
lack of grasp of the scientific method and how it can be applied in
such cases. The key point is, indeed, that you would *measure the
results*. However it is more than that.


What is 'the' scientific method - intrigued!

Pretty much, it what I summarised. The key point is that you set out to
collect data in a way that allows you to assess the results for relevance,
reliability, etc. In particular, you test in ways where the outcome can
'falsify' the ideas you before you started.

The purpose of collecting a fair amount of data/results is to enable you to
see if the results aren't likely to be due to 'chance'. It also is to allow
you to 'randomise' other factors which may be altering the individual
results as a result of an uncontrolled variable.

The purpose of trying to exclude some factors (e.g. not telling the
'driver' what arrangement/chance is being tried) is to see if their
opinions on what is going on mean anything.

The idea is also to see if you can 'catch out' the ideas people have and
find they have flaws, or are making an error. Not simply to find 'support
for a belief'.

The actual experimental protocol would vary according to the aims of the
specific area being investigated. So for a specific case we'd have to lay
out a more detailed set of methods.

If you wanted to check if what the driver said was reliable, you would
repeatedly make small changes *without the driver knowing the details*
and then observe the results. You would do this enough times in a
'randomised' manner to get results whose sigificance could be assessed
for reliability.

This would mean that not only you could determine which 'changes' or
arrangements were 'best' for the task. It would also tell you if the
driver's opinions were any guide to this, or not.


Maybe, but you wouldn't necessarily end up with a very good, or
consistently quick, car.

Indeed. There is no guarantee in advance that ones ideas or experiments
will lead to any specific desired 'result' in terms of performance. The
process does stand a good chance of doing so, however, if the participant's
ideas have any real relevance and worth. But all the scientific method
promises is that you can test and assess the results, and that others can
also do so for themselves and *not* simply have to accept your conclusions
simply because you assert them.

Slainte,

Jim

--
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Audio Misc http://www.st-and.demon.co.uk/AudioMisc/index.html
Armstrong Audio http://www.st-and.demon.co.uk/Audio/armstrong.html
Barbirolli Soc. http://www.st-and.demon.co.uk/JBSoc/JBSoc.html

Advice: Amp building
 

Dave Plowman (News) wrote:

Then perhaps you'd use universally accepted technical terms when you
next describe the sound of your latest speaker/amplifier?

Why should he?


--
Wally
www.wally.myby.co.uk
If it ain't broke, fix it until it is.

Advice: Amp building
 

Keith G wrote:

Brown would be more accurate.

Yuh, I think I see where you're coming frarm - kinda *warm, earthy and
natural sounding*...??

Eddie Van Halen has what he calls a 'brown' sound. I think it started with a
Marshall amp on one of those voltage dropper resistor thingies (variac?).
It's quite distinctive - a dirty, distorted sound, but not harsh. A guy on a
gituar group spent time (and money, no doubt) fine-tuning his set up to get
that sound - when he posted an example of his playing, the similarity to Van
Halen's sound was striking.


--
Wally
www.wally.myby.co.uk
Things are always clearer in the cold, post-upload light.

Advice: Amp building
 

On Fri, 28 Jul 2006 21:07:01 +0100, "Wally" wrote:

Dave Plowman (News) wrote:

Then perhaps you'd use universally accepted technical terms when you
next describe the sound of your latest speaker/amplifier?

Why should he?

Chipolata ink the must being of.

Clear now?

d

--
Pearce Consulting
http://www.pearce.uk.com

Advice: Amp building
 

Don Pearce wrote:

Why should he?

Chipolata ink the must being of.

Clear now?

Slope fallacy slippery you present - while semiotic tolerance neglecting.

Try again: Why should Keith use universally accepted technical terms when he
next describes the sound of his latest speaker/amplifier?


--
Wally
www.wally.myby.co.uk
Things are always clearer in the cold, post-upload light.