On Thu, 15 May 2008 01:26:18 +0100, "Ian Iveson"
<IanIveson.home@[EMAIL PROTECTED]
> wrote:
>flipper wrote:
>
>>>> >> Btw, the OTL doesn't 'require' the 120 ohm resistor
>>>> >> and it works just
>>>> >> fine as a 'low impedance' output.
>>>> >>
>>>> >> You could just remove the 120 ohm, and then adjust
>>>> >> the FB resistor for
>>>> >> the changed Vo, but there's also a nice trick can be
>>>> >> played. Instead
>>>> >> of the FB R and blocking cap. as shown, put a 12ohm
>>>> >> under the existing
>>>> >> Rk on V2A and take the headphone return to that
>>>> >> junction. That
>>>> >> eliminates the DC blocking cap.
>>>> >
>>>> >If I am following your description of this mod
>>>> >correctly, doesn't that
>>>> >convert the negative feedback from voltage feedback to
>>>> >current feedback,
>>>>
>>>> Yep
>>>>
>>>> >with the expected effect on source impedance?
>>>>
>>>> Yes, but then a moving coil device is actually current
>>>> driven so I'm
>>>> not sure what the 'net effect' would be.
>>>
>>>The common "dynamic" loudspeaker is also a moving coil
>>>device
>>
>> Quite right, and it's current driven too.
>
>What does current driven mean, exactly?
It means the cone is made to move by magnetic field and the field
generated by a coil is pro****tional to nI.
> They are generally
>designed to be voltage controlled.
As I said, we tend to think of voltage as 'the signal'.
>
>>> and people
>>>seem to feel they need to be driven from a voltage source.
>>
>> It's certainly the most common means, but then that's the
>> most common
>> means for most things.
>>
>> We seem to conceptually favor voltage as the 'signal
>> thing'.
>
>The 'signal thing' *is* voltage, as we receive it, in most
>cases.
Because we design with that concept. But, for example, the 'signal'
from a photo diode is current, the output of a tube is current and
bipolar transistors are both current in and current out.
> If it is convenient, it is not uncommon to see the
>signal transferred to current within circuits.
The other way, actually. A vacuum tube's output current is converted
to voltage by the load but with bipolars that not need be as current
can be the input as well..
> Surely
>voltage is the obvious first choice for encoding a signal
>intended for a speaker, because it relates directly to cone
>position, or at least it would if it wasn't moving about so
>quickly...
Current drives the cone and your 'if it weren't for' is precisely one
of the 'problems' with using constant voltage drive.
>> But I've seen papers suggesting 'current drive' would be
>> 'better' for
>> speakers as well. Whether they're 'correct', or whether
>> different coil
>> techniques would/should go with it, I don't know.
>>
>> Interestingly, when they plotted speaker response it
>> turned what was a
>> response dip with constant voltage into an equal but
>> opposite response
>> peak with constant current. Is that worse? or better? or
>> just
>> different?
>>
>> That wasn't necessarily the only effect but it's one I
>> remember.
>>
>>> They talk of
>>>"damping factor" and such.
>>
>> By comparison, speakers have lots of mass and move lots of
>> air, not to
>> mention the coil electrical characteristics are different.
>
>Well, moving lots of air and having lots of mass work in
>opposite directions as far as the need for damping is
>concerned.
They both oppose cone movement requiring more power aggravating the
problem when your 'signal is a voltage that gets reduced by source
impedance in series with coil impedance.
>They also have more area, and much more power available to
>overcome the mass and move the air.
Power that drops across the source impedance when your 'signal' is
constant voltage.
> The crucial point is
>that the scaling of all these parameters is not equally
>pro****tional, and the result favours quieter things in
>smaller spaces.
So why go through all that only to arrive at the same conclusion?
>
>>> Can I assume that "damping" isn't a factor
>>>with headphones, at least if they are driven from a 120
>>>Ohm source?
>>
>> I've seen professionals debate both sides of the issue
>> but, by the
>> same token, they were talking 'ultimate' and not that 120
>> ohms was
>> 'crap'. Unfortunately, neither side was able to put things
>> like cone
>> mass, etc., into numbers and, near as I can tell, the talk
>> about
>> damping factors is simply a direct extrapolation from
>> speakers which,
>> as noted above, are a whole different animal than
>> headphones.
>>
>
>No, the same principles apply.
I didn't say the 'principles' were different.
> The differences arise only
>from scale,
Well, the 'only difference of scale' is why large animals don't have
exoskeletons like insects.
'Scale' breaks a lot of things, 'same principles' notwithstanding.
> because some parameters are related to area,
>some to volume, and others to distance. Same reason
>elephants need such thick legs, and still can't
>jump...er...that's not a good analogy considering I'm
>arguing it's not another animal...all I mean is that you
>would use the same equations for speakers and headphones,
>but the results would be different because the parameters
>scale in several different ways.
While they could all be derived from 'principles', down to quantum
mechanics to be pedantic, but 'equations' are usually reduced to
'significant' variables with the "10 to 1" damping factor we're
talking about as not necessarily applicable being an excellent example
seeing as how there's not a hint of cone mass, excursion, or anything
else even obliquely referenced in it.
>In reality, unless you are designing headphones, you don't
>need to know how they work. We just need performance data.
You do if you're trying to make a case that speaker damping factor
rules of thumb are applicable to proper headphone operation.
>Now, possibly at the same site you have been to, it is said
>that a pair of Grado 32 ohm phones vary by 4 ohms over the
>range 20Hz to 20kHz. It also suggests that Grados may be
>well-behaved in this respect compared to other phones. Some
>more data would be good.
>
>But for the moment, how significant is that 4 ohms, when
>driven by a 120 ohm source? Not very, many may argue, when
>they've worked out how many dBs power variation ensues.
>
>Considering the load they present is so nearly resistive,
>there would be correspondingly little difference between
>current- and voltage-derived feedback. Normally,
>current-derived feedback arising from the reactive part of a
>speaker load would approximate to velocity feedback rather
>than the usual approximation to position feedback, for a
>sine wave. For headphones, none of this applies if their
>deviation from a resistive load is not significant. Neither
>would damping factor be an issue, because the dynamic
>behaviour of the machine is obviously not apparent to the
>source, and so cannot be damped by it. Presumably mechanical
>damping is sufficient.
Why do you go through all that only to arrive at the same conclusion
you're supposedly 'debating'?
>Somewhere, last time I was looking, I remember seeing a
>comparison between the ideal response for headphones (not
>flat, apparently) and the response of some real phones, with
>a low impedance and a high impedance drive. Can't find it
>now, but such info must surely be available from somewhere?
>
>It should be possible to find the data. The real argument
>then concerns the threshold of significance.
Depends on what you mean by 'high impedance drive' and the proposed
series-series FB circuit was not a constant voltage drive through
'high impedance' but a constant current drive that has high impedance.
They both have high impedance but in one case the voltage doesn't
change while it does in the other. Or vice versa, depending on which
one likes to call 'the signal'.
>> But I don't have the 'answer' either. I'm just relaying
>> what I've read
>> and discussing it.
>>
>> The circuit can be done either way. Well, 4 ways so far.
>> Series-shunt
>> or series-series FB and both either with or without the
>> 120 ohm.
>
>
>Cheers, Ian
>
|
