Re: NFB101 part 3
- From: "Ian Iveson" <IanIveson.home@xxxxxxxxxxxxxxxx>
- Date: Wed, 4 Jun 2008 00:57:57 +0100
Can't be bothered with all these empty insults, below. I
have made several valid and legitimate points which you have
either ignored, deliberately misconstrued in a ludicrously
contrived manner, or erroneously rejected without
foundation. From your last-but-one comment I gather you
don't even know what thread you're in.
It should be plain to you from your conversation with
Patrick that it is possible to understand enough about
feedback for the purpose of valve amp design without using
what you say is the standard EE categorisation and
associated jargon. I have myself in the past suggested,
probably to Patrick and certainly here somewhere, that
precisely your categorisation and terminology be adopted,
but with no success AFAIK. I wouldn't even then have done so
if the purpose had been to help newcomers. The discipline of
formal categorisation just isn't a good place to start, and
is of dubious value to non-EEs anyway, because much of the
discipline of engineering is about the need for
standardisation in a safety-critical market. This is a
recreational group.
It should be plain that, when sourced from a purely
resistive load, there is no difference between series- and
shunt-derived feedback. I don't understand why you keep
insisting there is, and you haven't bothered explaining so
it's just an empty assertion. And a wrong one, to boot, I
contend. You may wish to argue that the cat is alive, but I
won't be drawn into saying it's dead. It makes no
difference, either way. I make the point because it relates
to questions wider than electronics, that educated newcomers
may be aware of. The more links to prior knowledge the
better.
My examples, of the SMPS terminology and the anode follower
with unbypassed cathode resistor, were made to illustrate a
third legitimate point: the danger of blithely following
convention without properly considering the particular
circuit. I have seen it said many times that the output
impedance of such an anode follower is increased due to the
series-derived, voltage-applied nature of the feedback.
True, but not the whole story, and a naive application of
the rules for calculating the ensuing output impedance will
give the wrong result. As an adjunct to this call for
caution, I point out that, if the circuit must be inspected
and understood in order to decide what category its feedback
should be properly placed into and named, then the
categorisation and naming cannot be a transport to that
understanding. I could be wrong, but the point is legitimate
nonetheless.
You may care to look up "argue" in a dictionary.
Ian
"flipper" <flipper@xxxxxxxx> wrote in message
news:h37b44ljh18asteuoclq7j0mtrbedg2l6r@xxxxxxxxxx
On Tue, 3 Jun 2008 16:36:46 +0100, "Ian Iveson"
<IanIveson.home@xxxxxxxxxxxxxxxx> wrote:
flipper wrote:
In a CF the feedback is taken from directly across
the
load and is
therefore shunt derived. Remember shunt derived
feedback
lowers output
impedance which is why a CF has a low output
impedance.
RDH4 does not agree with what you are saying.
The difference is easily misinterpreted. An unbypassed
cathode resistor provides current-derived feedback if
the
output is from the anode, but not if it is from the
cathode.
At least I think that's what these folk mean. I think
it
is
simpler, if the feedback is a voltage, to describe
where
it
comes from in each case, rather than rely on a general
term
that is easily misunderstood.
That's because RDH4 is old, apparently , but modern
texts
will specify
both ends just as he says.
The feedback in a standard Williamson or a cathode
follower
is known as series voltage negative feedback.
RDH4 is apparently speaking of the input side only
but
look back at my
headphone amp thread. There are two ways to get a
signal
from the
output end. One is what most would likely consider
'normal': across
the load. That's shunt derived
The other would be to put a current sense R under
the
load
and feed
that signal back just as the other was. That's
series
derived.
RDH4 doesn't provide any means to describe the two
but
they will
behave quite differently.
Which is where I rather gave up on that thread. You
are
trying to have your cake and eat it with that
headphone
amp.
I have no idea what the hell you're talking about.
Well, so it seemed then.
You should try actually saying something other than
broad
generalizations ending with "It should be possible to
find
the data."
On the one hand, you argue that headphones are
near-as-dammit resistive loads such that a 200ohm series
resistance is not an impediment in relation to 32ohm
phones,
You're the one who 'argued'
It was you who argued the case for the series resistance.
Telling you where it came from is not an 'argument',
When asked why I used a series resistor on the 'PC
Speaker' amp I said
I had seen it done before and upon further research to
'refine' the
value I discovered there's an official IEC specification
(61938) for
headphone interfaces.
Which has sod all to do with fidelity.
That's your claim. Now it's incumbent upon you to prove it
and, so
far, you haven't made a case.
You also attempted to
justify the high output impedance of your design on other
grounds related to the characteristics of headphones.
Wrong. My 'design' was conventional series-shunt feedback
with a 120
ohm series resistance per IEC but, in conversation, I
noted there was
'another way' to do the feedback and mentioned I've seen
papers that
claimed current drive was 'better' for speakers.
But I didn't 'argue' anything nor did I even bother to
post a
schematic of the 'other way' on my web site.
Btw, it's 120 ohm, not 200.
Thanks.
You wanted to 'argue' with the IEC and made
generalizations, ending
with "It should be possible to find the data." Well, did
you find it?
A long time ago, thanks.
Apparently not or else you'd have whipped out your 'secret
data' to
support the claims you were making.
I was encouraging you to do the
same, since it is you who is designing the amp.
IEC specifies the interface and that's what I did. I also
told you if
you don't like IEC then to remove the resistor because it
works either
way.
But all you did was jump up and down about 'fidelity'
without a shred
of anything to support the rant.
As for your 'argument' that "In reality, unless you are
designing
headphones, you don't need to know how they work. We
just
need
performance data," that's what I did. I used the IEC
'data' for the
interface.
Now you are being deliberately silly.
There' an IEC spec. I followed it.
What you do is a mystery.
and on the other that series- and shunt-derived feedback
behave quite differently.
They do. How the device under control 'behaves' depends
on
the device.
Not always, as I have pointed out.
No, you pointed out how one particular device behaves
while dismissing
the NFB impedance effect.
How can you have it both ways? If
feedback is from a resistive load, there is no
difference
at
all between series- and shunt-derived.
In the first place, I never said headphones were a
'resistive load',
that was your leap.
Not how I remember it,
Not my problem.
but if they are not resistive then
how come it's OK to drive them with a comparitively high
source resistance?
You said we don't need to know 'how they work', just the
'data'. And
the 'data' is IEC specifies they are to work with a 120
ohm source
impedance.
And, secondly, the nature of headphones has
nothing to do with the characteristics of the 4 basic
feedback
topologies: series-shunt, series-series, shunt-series,
and
shunt-shunt.
To the extent that they are resistive, there is no
difference between series- and shunt-derived feedback.
Yes there is. The output impedance is different.
Maybe you don't care if series-series feedback increases
the effective
output impedance if you're going into a grid but maybe
you
care a lot
if it's going into a reactive load. But, in either case,
it's useful
to know what the feedback does so you can make an
intelligent
decision.
That can be done by inspection in any particular case.
If you want to reinvent the wheel for each 'particular
case' that's
your problem.
Deciding whether it falls into this or that conventional
category is an extra complication that a novice doesn't
need
to face.
Nope, it's simpler than reinventing the wheel each time.
A generalised distinction between
series/current-derived
and
shunt/voltage-derived feedback is a contrivance best
avoided.
Along with how they behave differently too I suppose.
There's an awful lot of universities that will need to
scrap their
E.E. texts to meet your criteria.
There's a whole lot of universities these days that
don't
deserve the title. That aside, I rather guessed that it
must
be standard fare on EE courses...
Did the gratuitous poke in the eye of un-named
'universities' boost
your ego?
Why should it? Neither was it gratuitous:
It was an almost textbook case of a gratuitous insult as
testified by
you setting it "aside" in the very next breath.
I was making the
point that following the practice of universities may not
be
the best choice. If I named all the universities here in
the
UK not worthy of the title, it would be a long list of no
value to anyone here. We used to have universities,
polytechnics and colleges offering, more or less
respectively, academia, practical analytical skills, and
training. Some years ago, polytechnics were given
university
charters, and more recently many colleges have also begun
calling themselves universities.
I'm not even going to get into that nonsense and it is not
a 'point'
when set aside.
Consequently many large employers of graduates have
adopted
their own entrance tests and, more widely many degrees are
discounted because they have been confered by institutions
of ill repute.
Wouldn't do them any good with you because you'd declare
their test to
be nonsense and that you have a better way of doing
things.
it smacks of received
wisdom ill-applied.
Your comment smacks of more self aggrandizing.
More? It was a jab at you, the rampant EEs, and other
self-proclaimed experts here who keep being horrid for no
good reason. There is nothing grand about independent
thought, which is commonly derided and admittedly on
occasion a hostage to fortune.
Let's see. the 'noble you' against "self-proclaimed (by
"institutions
of ill repute".granting them EE degrees) experts" who,
according to
your "independent thought," are all wrong is not self
aggrandizement.
Not to mention I've never even said whether I have a
degree or not, or
in what subject, much less waved it as 'authority' and
said over and
over in the 'headphone' discussion I was merely passing on
what I had
found from doing research on it.
IIRC Ian is specifically not writing for
that audience, so doesn't need to complicate matters
with
unnecessary and apparently misleading conventions
contrived
presumably for some wider historical purpose. You are
not
alone in the field, I realise that: Morgan Jones, for
example, also makes the distinction. I just said I think
it's best avoided, and I meant in this context.
Since you "guessed that it must be standard fare on EE
courses" you
apparently haven't read any and perhaps you should
before
declaring
universal wisdom on what is "ill-applied."
Possibly in other contexts too. I've got this book on
SMPS
design that makes the same distinction, equally
confusingly
because by current-derived it means from sensing switch
current, whereas by voltage-derived it means from
sensing
load voltage. That's two differences, so the single
distinction made by the jargon isn't enough. I had to
leaf
through looking for a circuit diagram before I could be
sure
what the text was on about.
Maybe if you read some of those EE texts you wouldn't be
so confused.
Cheap enough I suppose,
It's not a cheap shot at all. I'm suggesting there's a
reason for all
the stuff in those EE texts.
but you're still missing my point.
I don't need an EE degree,
I didn't say you did.
so why should I?
To learn it.
That's another
point I'm making.
What? That ignorance is knowledge?
The word 'shunt' does not appear, because the NFB
does
not
try to shunt the input signal.
The word shunt doesn't appear because they ignore
how
the
signal is
derived.
Don't they say where it comes from? Isn't that enough?
They say where it goes too. Isn't that enough? So why
mention shunt
or series on that end?.
It's a simpler, more straightforward distinction at that
end.
It's the same 'straightforward distinction' on both
ends.
No, as I have intimated above with the SMPS example. I
could
have used the example of an unbypassed cathode resistor
used
in an anode follower. The input is unambiguous as the
destination of feedback, but the source of the feedback is
not always the output. This means that the rules you apply
to your categories of feedback must be used with
particular
caution with respect to its derivation.
Your confusion on how to apply the basics does not mean
there's a
'problem' with the basics. It means you have a problem
applying them.
That's not particularly unusual and is why EE isn't a one
semester
course.
I don't particularly want to champion RDH4 though,
I gave RDH4 another look and while I didn't do an
extensive
investigation for consistency it seems they tend to use
the terms
'current feedback' and 'voltage feedback' for the modern
shunt/series
'derived' (from the output) terminology and series/shunt
for the
input. However, RDH4 also throws in a couple of 'if
then,
unless, but
not, or else' kind of caveats into their 'what they do'
table so the
modern texts are more 'universal' and consistent.
It's the ifs and buts that make received wisdom
potentially
dangerous. I doubt RDH4 was wrong, so perhaps the only
difference is that EE conventions have evolved towards
greater cohesion, so the caveats may be in another chapter
of whatever book you read. Perhaps it's that cohesion that
has led to its current insularity.
The 'problem' that causes the ifs ands or buts is they
don't first
establish a set of basic principles but tend to do it,
like you favor,
'on a case by case basis' with assumptions, as Patrick
opined, "you're
supposed to know."
Forgivable given the time period with the restricted
devices and
topologies possible.
and
if I'd been history I would have tried not to create
such
awkward categories, or attached to them such misleading
jargon.
In particular, it contributes nothing to understanding
for
the purpose of designing and building valve audio
equipment.
Just a thought but perhaps 50 years worth of experts.
e.g.
designers,
authors, universities, et al, know something about it
you've missed.
Eh? How does knowing more about it make them more easily
misled?
One wonders how you can take a statement of expertise and
'interpret'
it as being 'misled'.
Or help them to understand more how someone outside
their world might be misled
Because, for one, they all started off deaf dumb and
blind, just like
everyone else, and had to learn it and then have
considerably more
experience than you in the job of teaching it.
by the careless bandying about
of EE rules and conventions?
Your problem is in insisting that whatever you don't like
is "careless
bandying about of EE rules and conventions."
I'm right in the middle of Ian's target market, you
aren't.
I haven't even read Ian's paper so I don't know what his
'market' is,
whether he wrote it well, or whether he used the proper
terminology.
The customer is always right, and you're not.
That's a lovely thought for the customer but it's also
B.S.
Ian
.
- Follow-Ups:
- Re: NFB101 part 3
- From: Ian Thompson-Bell
- Re: NFB101 part 3
- References:
- Re: NFB101 part 3
- From: Ian Iveson
- Re: NFB101 part 3
- From: Ian Iveson
- Re: NFB101 part 3
- Prev by Date: Re: Suggestions for a soldering iron
- Next by Date: Re: NFB-301?
- Previous by thread: Re: NFB101 part 3
- Next by thread: Re: NFB101 part 3
- Index(es):
Relevant Pages
|
