Terrestrial Switchoff - sorry to labour the point but...

On 01/04/2013 13:06, J. P. Gilliver (John) wrote:
In message , Jim Lesurf
writes:
In article , J. P. Gilliver (John)
wrote:
In message , Jim Lesurf
writes:
In article , Johny B Good
wrote:


[1] Alternatively the other analogy commonly used was that it was
akin to the 'crossover distortion' that would ensue in a class B
amplifier lacking the necessary bias required to avoid the 'dead
zone'.

That's an interesting analogy as *Perfect* class-B would not give any
crossover distortion as described by theory. But reality tends to
defeat this hope once you go beyond any kind of lab demo that is
tweaked with insane levels of care and attention. :-)

I didn't know perfect class B actually existed; I thought it was always,
in practice, class AB.

Indeed. In practice Class B is too evanescent and fragile to have any
useful existence. Hard enough to balance a pencil on its tip without it
being held or supported. Impossible when you then want then the pencil to
write a message...

Slainte,

Jim

I didn't know it could be created at all, in an amplification sense: I
thought all practical circuits had some distortion near the cutoff
(where the signal crosses zero). [I've said "in an amplification sense"
as I think precision rectifiers can be constructed for instrumentation
purposes.]

A pefect class B amplifier can be made in theory, if you have
transistors with a zero switch-on time, and other components with an
infinite bandwidth. The distortion in practical circuits is caused by
the time it takes for the transistors to switch on in response to the
signal, and by non-linearities in the frequency response of the required
feedback loop.

Normally, the best compromise is class AB, where the transistors are
biased to just start conducting. Good efficiency, and reasonable quality.

A lot of modern amplifiers are class D, which has the transistors
operating as HF switches, and not as variable resistances.

--
Tciao for Now!

John.

In article , J. P. Gilliver (John)
wrote:
In message , Jim Lesurf
writes:


That's an interesting analogy as *Perfect* class-B would not give
any crossover distortion as described by theory. But reality tends
to defeat this hope once you go beyond any kind of lab demo that is
tweaked with insane levels of care and attention. :-)

I didn't know perfect class B actually existed; I thought it was
always, in practice, class AB.

Indeed. In practice Class B is too evanescent and fragile to have any
useful existence. Hard enough to balance a pencil on its tip without it
being held or supported. Impossible when you then want then the pencil
to write a message...

Slainte,

Jim

I didn't know it could be created at all, in an amplification sense: I
thought all practical circuits had some distortion near the cutoff
(where the signal crosses zero).

You've over-extended my use of 'perfect'. :-) No amplifier - class A, AB,
etc, that I know of has *no* distortion in practice. Despite simplified
models (like that for the orginal 'current dumping') that seem to show
otherwise.

The distinction between B and AB is that the no-signal condition bias
current in the output devices is *just* reduced to zero, and any signal at
all causes some current in one device, or the other, but never both of
them.

Not that either B or AB would have 'no distortion'.

You can set up a lab demo and tweak the circuit to get arbitrarily close to
class B when there is little or no signal. But even in this situation the
required conditions tend to wander about. Hence the analogy with trying to
balance a pencil on its tip. Add a significant signal level and things get
worse...

In practice therefore even 'class B' amps often tended to be AB when in
actual use. But with such poorly controlled behaviour as to give poor
results.

Slainte,

Jim

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

In article , John Williamson
wrote:
On 01/04/2013 13:06, J. P. Gilliver (John) wrote:
In message , Jim Lesurf
writes:


A pefect class B amplifier can be made in theory,

Yes, in theory all kinds of things are possible... which turn out to be
impractical in reality.

if you have transistors with a zero switch-on time, and other components
with an infinite bandwidth. The distortion in practical circuits is
caused by the time it takes for the transistors to switch on in response
to the signal, and by non-linearities in the frequency response of the
required feedback loop.

Slight modification: You'd need zero response time. Or *exactly* the same
response time for both output devices of a pair. Otherwise infinite
bandwidth won't save you. :-)

And, alas, a response time of zero isn't sufficient. You also have the
problem that in real-world devices the devices tend to change behaviour
with temperature or as a result of their recent history of behaviour. This
is particularly significant with real-world audio signals which often are
far from symmetric. So the device temperatures do not remain equal over a
range of timescales.

Plus any variations in, say, the rail voltages that aren't common will tend
to affect the two devices in slightly different ways.

Etc....

A lot of modern amplifiers are class D, which has the transistors
operating as HF switches, and not as variable resistances.

A confusing factor here is that we now have many 'switching' amps that use
different logical topologies/methods. So there are actually a number of
different designs which are only covered by being the same 'class' by
making that classes definition sufficiently vague and sweeping.

Slainte,

Jim

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

In article , Roderick Stewart
wrote:
On Sat, 30 Mar 2013 20:22:26 +0000, tony sayer wrote:

One of the ones that does sound good is a steam driven fairground organ
sometime mid to late 60's.

Its very very "real" sounding, and never fails to impress. When I tell
people how and when it was recorded they tend not to believe it was
done at that time on analogue equipment;!...

I have a few stereo classical CDs made from recordings made in the 1950s
that stand comparison with anything made today. They must have been made
on analogue tape, but don't sound like "historical" recordings. Such
equipment was evidently capable of superb results if it was used
properly.

Same here. I'm particularly a fan of the 'Decca 50' box set released last
year as it has a number of such recordings. I'd also recommend the new
mastering of the Britten War Requiem for a quite amazing sound.

The real advantage back then was the approach of using a few mics,
carefully placed, in a recording location with a good (for the purpose)
acoustic. The equipment was technically poorer in various measurable ways.
But those involved often got good results by the application of due care
and skill.

One of the best EMI recordings I have on CD is a 1950s recording of
Prokofiev symphonies, etc. Made as the first few stereo takes intended for
LP release. Some loud passages are distorted. Some quiet ones show
background noise. But in between, the sound is really convincing and
impressive for its natural imaging, etc.

And it does help that the performances are superb. :-)

Slainte,

Jim

--
Please use the address on the audiomisc page if you wish to email me.
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Armstrong Audio http://www.audiomisc.co.uk/Armstrong/armstrong.html
Audio Misc http://www.audiomisc.co.uk/index.html

Jim Lesurf wrote:

In practice therefore even 'class B' amps often tended to be AB when
in actual use.

Indeed. In fact, I'm surprised that anybody thinks this discussion
needs to take place. When I learnt this stuff, it was well known that
'class B' was only a theoretical concept - a kind of 'logical design'
that was interesting and informative to study, whilst acknowledging
that it couldn't be made to work satisfactorily in the real world due
to the real-world limitations in component performance.

Having acknowledged that obvious fact, the discussion then proceeded to
the far more relevant investigation of how far a class B design should
be biased towards 'class A' to get a satisfactory performance /
efficiency trade-off.

This stuff is so obvious, so widely taught, and so well known, it is
surely not a debatable matter, is it?

--
SteveT

In message , Johny B Good
writes:
[]
I still have a 3 inch reel of triple play tape with a recording of a
late night ghost story broadcast by Radio Caroline sometime in 1967.
The tape recorder it was made with has long since been consigned to
Land Fill so I can only audition it at a standard speed where it
either sounds very flat at 1 7/8 ips or quite obviously chipmunky at 3
3/4 ips. My best guess at the original recording speed would be a
surprisingly consistent approximation to 2 1/2 ips, give or take.
[]
I was hoping that there would be a trace of 4 or 5 KHz heterodyne to
help in re-establishing the original speed but of that there was
absolutley no trace (I guess my homebrewed superhet transistor radio
was just too selective to allow such interference). It's going to be a
case of "Adjusting by Ear" when I do get hold of the Round Tuit.

There might be a trace of mains hum (or second or third or fourth
harmonic thereof) - even if your recorder was battery-powered (was it?),
there'd probably be some pickup from the house, or even on the original
signal, which might have been 60 Hz if Caroline used US generators and
equipment - and not necessarily that stable, but probably enough so to
help establish the speed. The spectrum analyser ("waterfall" I think
it's called) display in GoldWave (or probably most other sound
processing software) would I suspect show it.
--
J. P. Gilliver. UMRA: 1960/1985 MB++G()AL-IS-Ch++(p)[email protected]+H+Sh0!:`)DNAf

Everything in moderation. Including moderation. - Billy Connolly('s website,
according to Radio Times, 14-20 February 2009)

On Mon, 1 Apr 2013 22:14:11 +0100, "J. P. Gilliver (John)"
wrote:

In message , Johny B Good
writes:
[]
I still have a 3 inch reel of triple play tape with a recording of a
late night ghost story broadcast by Radio Caroline sometime in 1967.
The tape recorder it was made with has long since been consigned to
Land Fill so I can only audition it at a standard speed where it
either sounds very flat at 1 7/8 ips or quite obviously chipmunky at 3
3/4 ips. My best guess at the original recording speed would be a
surprisingly consistent approximation to 2 1/2 ips, give or take.
[]
I was hoping that there would be a trace of 4 or 5 KHz heterodyne to
help in re-establishing the original speed but of that there was
absolutley no trace (I guess my homebrewed superhet transistor radio
was just too selective to allow such interference). It's going to be a
case of "Adjusting by Ear" when I do get hold of the Round Tuit.

There might be a trace of mains hum (or second or third or fourth
harmonic thereof) - even if your recorder was battery-powered (was it?),
there'd probably be some pickup from the house, or even on the original
signal, which might have been 60 Hz if Caroline used US generators and
equipment - and not necessarily that stable, but probably enough so to
help establish the speed. The spectrum analyser ("waterfall" I think
it's called) display in GoldWave (or probably most other sound
processing software) would I suspect show it.

I use CoolEdit Pro which can display a real time spectrum as well as
show a more detailed spectral display gathered from an arbitary length
sample of the wav which I plan to put to such use.

The tape recorder was a battery powered model which is good news from
the point of view of picking out any modulated hum (60 or 50Hz) or,
less likely, since the portable radio was also battery powered, 50Hz
from the local mains supply.

When I auditioned the tape, I could hear no trace of '9Khz'
heterodyne, not even when it was downshifted by the slow playback
speed. This comes as no surpise since my homebrewed transistor
superhet used a 3 stage IF amplifier with adjustable regenerative
feedback employed to improve selectivity on weak signals (not that
Radio Caroline, whether North or South, needed much gain at that time
of night[1]).

If I'm going to detect any constant background signal to use as a cue
to the original speed of the recording, it's most likely to be mains
hum sourced rather than a 9KHz heterodyne.

[1] I often used to listen to Caroline South in the middle of the day
from the Liverpool area just because I could (nothwithstanding the LO
sideband noise effect on such weak signals as I was picking up on the
built in ferrite rod antenna).
--
Regards, J B Good

On Mon, 1 Apr 2013 13:06:06 +0100, "J. P. Gilliver (John)"
wrote:

In message , Jim Lesurf
writes:
In article , J. P. Gilliver (John)
wrote:
In message , Jim Lesurf
writes:
In article , Johny B Good
wrote:


[1] Alternatively the other analogy commonly used was that it was
akin to the 'crossover distortion' that would ensue in a class B
amplifier lacking the necessary bias required to avoid the 'dead
zone'.

That's an interesting analogy as *Perfect* class-B would not give any
crossover distortion as described by theory. But reality tends to
defeat this hope once you go beyond any kind of lab demo that is
tweaked with insane levels of care and attention. :-)

I didn't know perfect class B actually existed; I thought it was always,
in practice, class AB.

Indeed. In practice Class B is too evanescent and fragile to have any
useful existence. Hard enough to balance a pencil on its tip without it
being held or supported. Impossible when you then want then the pencil to
write a message...

Slainte,

Jim

I didn't know it could be created at all, in an amplification sense: I
thought all practical circuits had some distortion near the cutoff
(where the signal crosses zero). [I've said "in an amplification sense"
as I think precision rectifiers can be constructed for instrumentation
purposes.]

Indeed! I used a couple of op-amps to improve the VU meters on the
GX630DB to get rid of the dead band just below the -20db mark. I had
to adjust the end stop at the bottom of the scale to recalibrate the
-20db point making the original zero point approximate to -30db.

It's wonderful to still be able to meter sub -30db activity instead
of watching a totally dead meter. It nicely shows the recorded disk
warp induced subsonics on pretty well all of my vinyl transfers.

The circuit is pretty simple, the meter, fed by a fullwave bridge
rectifier, is simply placed in series with a suitably chosen feedback
resistor between the output and the inverting input with a suitable
value of resistor going from there to the ground reference point
(non-inverting Hi Z input version). The massive op-amp voltage gain
overcomes the rectifier drop when such rectifier volt drop is within
the feedback network.

This modification was just one of many improvements I made to the
GX630DB
--
Regards, J B Good

In article , Johny B Good
wrote:


Indeed! I used a couple of op-amps to improve the VU meters on the
GX630DB to get rid of the dead band just below the -20db mark. I had to
adjust the end stop at the bottom of the scale to recalibrate the -20db
point making the original zero point approximate to -30db.

It's wonderful to still be able to meter sub -30db activity instead of
watching a totally dead meter. It nicely shows the recorded disk warp
induced subsonics on pretty well all of my vinyl transfers.

I'm slightly surprised you get wow subsonics above -30dB - although I
obviously don't know your reference level.

I've transferred a number of my LPs to digital copies using a Tascam
digital recorder. That has a linear bar level indicator with a 60dB span.
The subsonics for LP tend to be down at more like the -50dB level than
anywhere near -30dB.

That said, I'm probably setting the gain so that the RIAA 0dB reference is
well below 0dB on the meters to ensure I record peaks without clipping. [1]
So the subsonics may well approach -30dB in RIAA reference terms. Even
though official 'rumble' values for turntables, etc, tend to be below
-40dB!

One of the advantages of the digital recorder over any analogue deck I've
owned is the excellent metering. The Nak I use does have good meters, but
not as good as the digital recorder. And the digital recorder lets me tweak
settings like decay time for the mean power level and hold time for peak
indicators on a 'project by project' basis.

Slainte,

Jim

[1] I did check this early on using some test LPs. But I've now forgotten
what level I chose!

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

On Tue, 02 Apr 2013 10:05:54 +0100, Jim Lesurf
wrote:

In article , Johny B Good
wrote:


Indeed! I used a couple of op-amps to improve the VU meters on the
GX630DB to get rid of the dead band just below the -20db mark. I had to
adjust the end stop at the bottom of the scale to recalibrate the -20db
point making the original zero point approximate to -30db.

It's wonderful to still be able to meter sub -30db activity instead of
watching a totally dead meter. It nicely shows the recorded disk warp
induced subsonics on pretty well all of my vinyl transfers.

I'm slightly surprised you get wow subsonics above -30dB - although I
obviously don't know your reference level.

Not wow or deck rumble,just the inevitable 2 or 3Hz warp induced
subsonic from the slight warp of the vinyl disk (which is at its
maximum at the start of the groove, fading in level as the stylus
tracks to the innermost part of the groove).


I've transferred a number of my LPs to digital copies using a Tascam
digital recorder. That has a linear bar level indicator with a 60dB span.
The subsonics for LP tend to be down at more like the -50dB level than
anywhere near -30dB.

That said, I'm probably setting the gain so that the RIAA 0dB reference is
well below 0dB on the meters to ensure I record peaks without clipping. [1]
So the subsonics may well approach -30dB in RIAA reference terms. Even
though official 'rumble' values for turntables, etc, tend to be below
-40dB!

I set the recording level to just below the 0db FSD level for the
maximum peak on a case by case basis for each disk. As long as the
peak level captured is somewhere in the region of -2db and 0db FSD
after removing the obvious loud clicks and pops, I'm done.

If the music content shows evidence of being clipped on some of the
peaks, I'll reduce the level and repeat the capture process unless
it's literally just one or two peaks that suffered a mild amount of
clipping (i.e two or less samples 'pegged' to the FSD clip level). I
very rarely find myself having to do a repeat take.


One of the advantages of the digital recorder over any analogue deck I've
owned is the excellent metering. The Nak I use does have good meters, but
not as good as the digital recorder. And the digital recorder lets me tweak
settings like decay time for the mean power level and hold time for peak
indicators on a 'project by project' basis.


[1] I did check this early on using some test LPs. But I've now forgotten
what level I chose!

That's the beauty of using a software meter, no under or over shoot
whatsoever. CoolEdit Pro's metering also allows you to use different
scale ranges (30 to 90db in 15 db increments) has dynamic peak and
valley markers as well as a clip indicator which is very useful (but,
sadly, not for all those PCI and on-board soundcards that were
clipping the line input at -3.5db FSD up until about 5 or 6 years
back).
--
Regards, J B Good