About Us

I read in alt.binaries.schematics.electronic that John Stewart
wrote (in )
about 'Who needs NFB when there is error correction?', on Thu, 9 Sep
2004:

Well, that's nice, but sometimes people need to be reminded.

You presented it as if it was a result new to you.

Seems like
that may have worked for you as well!!!

No, I am continually aware of the effects of negative feedback on
harmonic and intermodulation distortion. In fact, I make sure I think of
it specifically for five minutes every day. (;-)
--
Regards, John Woodgate, OOO - Own Opinions Only.
The good news is that nothing is compulsory.
The bad news is that everything is prohibited.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk

On Thu, 09 Sep 2004 12:51:15 +0200, Sander deWaal
wrote:

Patrick Turner said:

What a pathetic set of analogies.

And patheticism, or pathos, has been around for so long now....

Perhaps mr. Thompson needs some humor correction? ;-)

I find no humor in technologically unsound pontifications.

...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona Voice

480)460-2350 | |
| E-mail Address at Website Fax

John Stewart wrote:

John Byrns wrote:

Could someone please explain how this so called "error correction" system
is any different from ordinary negative feedback? It still seems to
depend on an error in the output signal, just as with negative feedback to
generate the "correction" signal, unlike for example a feed forward signal
which can theoretically cancel the entire error signal in the output.
Looking at the schematics that have been posted before all I see is
slightly complicated implementations of ordinary negative feedback, what
am I missing?

Regards,

John Byrns

A few days ago if my memory has not failed someone made
the point in this thread that the SE guys would not tolerate NFB.
That may very well be with good reason.

A while ago I made some measurements of THD while simultaneously
measuring the resulting distortion spectrum. The results showed THD
was usually reduced with NFB as we expect but the spectral
components in the result included several of higher order, all
of which had not appeared with no FB.

Higher order distortion products are far more audible
than those of say the 2nd or 3rd. Another result I observed
was more IM products. PP does not solve the problem.
Local FB in the output stage is a great help.

Cheers, John Stewart

Well, your'e not telling us anything I didn't know already.
probably 1/4 the dudes who use tubes for hi-fi are users of SE
amps of one sort or another, and the those I know
use them with sensitive speakers, say 95 dB/w/m, and so
they need only 1/3 of a watt of average power levels for 90%
of their listening, and 8 watt SET amps without a stitch of NFB
loops are quite acceptable, because the thd is below 1%
and is nearly all 2H.

This being the case, some 10 dB of NFB isn't unusual, and it generally reduces
the
1% of 2H to 0.3%, and the amount of created 3H and other higher no harmonics
produced under such circumstances is quite negligible, even though
the awfulness of higher order H and the IMD
is proportional to N squared divided by 4 where N is the harmonic,
and this calc gives us the equivalent amount of 2H.
So 0.1% of 5H is equal to 0.1 x 5 x 5 / 4 = 0.625% of 2H
0.1% of 7H would be = 1.225% of 2H, and so on

Its lucky that SE amps and class A PP amps have extremely small
amounts of 3,4,5,6,7,8,9,10th harmonics at the normal listening level.
But increase the volume level to the point where occasional clipping takes
place,
and the SE amp begins to sound ragged much like any other amp.
But we don't do that if we know how to use such gear, and if we value our
hearing.

If the amp has 10% of thd with only 10 dB of NFB applied,
then the situation is far far worse, and a large mix of products appear.
But with only 1% to start with, some NFB will often improve the sound,
with crisper cleaner highs and more precise sounding bass and transients.

PP, especially class A PP triodes are more linear than the same tubes used as
SET.
The thd at a few watts with PP KT88 will be at least 1/4 of the SE case.
And thus IMD will be less, since IMD simply results from the non linear
amp behaviour which is quantified by THD.

I have built a few SEUL amps which give around 4% thd with no NFB at
a dB below clip; the 13E1 gives this at 22 watts/8 ohms/13.2vrms.
The NFB I use reduces this to 0.5%,
and at 2 watts which is 4 vrms/8ohms, the thd is only 0.13%,
of which 9/10 is 2H.
The folks I know find they have no objection to the NFB, and one
guy said after a long listen, " this prooves you can use
NFB with SE amps."
This man said this at a demo of my SEUL 22 watters to 30 gathered audiophiles
in their old library venue, using 89 dB/w speakers which are 4 ohms at 400 Hz,
which meant the thd was 3 times the amount for 8+ohms.
They were very happy listeners.
On another occasion I proved a PP 5050 ULAB1 amp using 2 x 6550 per channel
with 16 dB/global NFB was streets ahead of another guy's all trioded PP amp with
4 x 6550 per channel
with only 6 dB/global NFB.

The NFB didn't stop the sound being "good tube sound"

Heaven forbid that any of my amps ever sound like some
variety of sound under the label of "tube sound".
Some I have heard is typically lazy sounding,
and sandpapery, and plain harsh, and unmusical.

The RDH4 gives a lot of valuable info about listening levels
and the effect of distortion on listeners, and it spells out
the amounts of IMD you will get for a given amount of THD,
based on using a two tone test signal with a 4:1 voltage ratio
of bass to HF, say 70 Hz to 5 kHz.

The main thing about NFB is to not use too much.
Choose a circuit that could be used without NFB.
Use it with best practices of critical damping in mind,
and I suggest one reason why some DIY types loathe NFB
is that they have no clue how to apply it properly,
since they can only think and feel subjectively, and any boring rationalization
with numbers reaks to them of compromise.
Feedback equations, load line analysis, distortion measurements,
ohms law, filter theory are all mysteries to such ppl.
Its probably better they struggle to build a nice simple SET amp,
because anything else is quite bamboozling.

I have shared some very fine musical experiences with such delightfully
ignorant people.
When you listen to what they have made, the noise is low,
the distortion is below threshold, output impedance is sufficently low,
their room is treated right, their speakers have a sufficiently flat response,
and the music conveys emotion and intimacy of the recorded moment
and the intentions of the composer.
So they know enough.
And its the rule of "enoughness" at work here which we can admire.

Patrick Turner.

Patrick Turner wrote:

John Stewart wrote:

John Byrns wrote:

Could someone please explain how this so called "error correction" system
is any different from ordinary negative feedback? It still seems to
depend on an error in the output signal, just as with negative feedback to
generate the "correction" signal, unlike for example a feed forward signal
which can theoretically cancel the entire error signal in the output.
Looking at the schematics that have been posted before all I see is
slightly complicated implementations of ordinary negative feedback, what
am I missing?

Regards,

John Byrns

A few days ago if my memory has not failed someone made
the point in this thread that the SE guys would not tolerate NFB.
That may very well be with good reason.

A while ago I made some measurements of THD while simultaneously
measuring the resulting distortion spectrum. The results showed THD
was usually reduced with NFB as we expect but the spectral
components in the result included several of higher order, all
of which had not appeared with no FB.

Higher order distortion products are far more audible
than those of say the 2nd or 3rd. Another result I observed
was more IM products. PP does not solve the problem.
Local FB in the output stage is a great help.

Cheers, John Stewart

Well, your'e not telling us anything I didn't know already.
probably 1/4 the dudes who use tubes for hi-fi are users of SE
amps of one sort or another, and the those I know
use them with sensitive speakers, say 95 dB/w/m, and so
they need only 1/3 of a watt of average power levels for 90%
of their listening, and 8 watt SET amps without a stitch of NFB
loops are quite acceptable, because the thd is below 1%
and is nearly all 2H.

This being the case, some 10 dB of NFB isn't unusual, and it generally reduces
the
1% of 2H to 0.3%, and the amount of created 3H and other higher no harmonics
produced under such circumstances is quite negligible, even though
the awfulness of higher order H and the IMD
is proportional to N squared divided by 4 where N is the harmonic,
and this calc gives us the equivalent amount of 2H.
So 0.1% of 5H is equal to 0.1 x 5 x 5 / 4 = 0.625% of 2H
0.1% of 7H would be = 1.225% of 2H, and so on

Its lucky that SE amps and class A PP amps have extremely small
amounts of 3,4,5,6,7,8,9,10th harmonics at the normal listening level.
But increase the volume level to the point where occasional clipping takes
place,
and the SE amp begins to sound ragged much like any other amp.
But we don't do that if we know how to use such gear, and if we value our
hearing.

If the amp has 10% of thd with only 10 dB of NFB applied,
then the situation is far far worse, and a large mix of products appear.
But with only 1% to start with, some NFB will often improve the sound,
with crisper cleaner highs and more precise sounding bass and transients.

PP, especially class A PP triodes are more linear than the same tubes used as
SET.
The thd at a few watts with PP KT88 will be at least 1/4 of the SE case.
And thus IMD will be less, since IMD simply results from the non linear
amp behaviour which is quantified by THD.

I have built a few SEUL amps which give around 4% thd with no NFB at
a dB below clip; the 13E1 gives this at 22 watts/8 ohms/13.2vrms.
The NFB I use reduces this to 0.5%,
and at 2 watts which is 4 vrms/8ohms, the thd is only 0.13%,
of which 9/10 is 2H.
The folks I know find they have no objection to the NFB, and one
guy said after a long listen, " this prooves you can use
NFB with SE amps."
This man said this at a demo of my SEUL 22 watters to 30 gathered audiophiles
in their old library venue, using 89 dB/w speakers which are 4 ohms at 400 Hz,
which meant the thd was 3 times the amount for 8+ohms.
They were very happy listeners.
On another occasion I proved a PP 5050 ULAB1 amp using 2 x 6550 per channel
with 16 dB/global NFB was streets ahead of another guy's all trioded PP amp with
4 x 6550 per channel
with only 6 dB/global NFB.

The NFB didn't stop the sound being "good tube sound"

Heaven forbid that any of my amps ever sound like some
variety of sound under the label of "tube sound".
Some I have heard is typically lazy sounding,
and sandpapery, and plain harsh, and unmusical.

The RDH4 gives a lot of valuable info about listening levels
and the effect of distortion on listeners, and it spells out
the amounts of IMD you will get for a given amount of THD,
based on using a two tone test signal with a 4:1 voltage ratio
of bass to HF, say 70 Hz to 5 kHz.

The main thing about NFB is to not use too much.
Choose a circuit that could be used without NFB.
Use it with best practices of critical damping in mind,
and I suggest one reason why some DIY types loathe NFB
is that they have no clue how to apply it properly,
since they can only think and feel subjectively, and any boring rationalization
with numbers reaks to them of compromise.
Feedback equations, load line analysis, distortion measurements,
ohms law, filter theory are all mysteries to such ppl.
Its probably better they struggle to build a nice simple SET amp,
because anything else is quite bamboozling.

I have shared some very fine musical experiences with such delightfully
ignorant people.
When you listen to what they have made, the noise is low,
the distortion is below threshold, output impedance is sufficently low,
their room is treated right, their speakers have a sufficiently flat response,
and the music conveys emotion and intimacy of the recorded moment
and the intentions of the composer.
So they know enough.
And its the rule of "enoughness" at work here which we can admire.

Patrick Turner.

Please Patrick, why can't you say it in 100 words or less? JLS

snip a pile of my stuff,

Patrick Turner.

Please Patrick, why can't you say it in 100 words or less?

I read all your posts, and all your words, and I never complain.

Perhaps you could consider affording me the same consideration.

I will never refrain from promoting challenging discussions here
if and when I feel like it. If the attention span of the group's members are
somewhat restricted, I make no apologies.
It must be the Irish in me ;-)

Anyway, how many darn words did I use?
Do you count them?
100 words is a lot anyway.
I might as well be hung for using 300 words as for 100.

Just as well there isn't a word tax; or
do they tax the time ppl spend reading in the US?

Patrick Turner.

JLS

"Patrick Turner" wrote in message

Jim is like the guy who would never enjoy owning a yacht.
He'd always say that the sailors in the harbour were brainless,
and hadn't they heard of ferries, or the bus route around the bay.

But they don't say that the ferry is a boat and the yacht isn't.

Patrick Turner wrote in message ...
smoking-amp wrote:

I posted the vacuum tube current mirror circuit on ABSE. Also is a
thread on this at the diyAudio tube forum too, in case someone can't
get ABSE. I use an assortment of fixtures with many diode tubes
connected in series, with selectable series tap points for
experimenting and designing with this idea. 6JU8 and 9006 diode tubes
are decent choices. Smaller diodes and higher gm pentodes in general
give higher current gain.

I was able to view the basic schematic of the diodes plus pentode
at abse, but what is the schematic for their applied use?

I just posted an example circuit on ABSE using the Vacuum Tube
Current Mirrors. The current sources shown in the schematic could be
high value resistors or inductors with a little B+ modding. Nothing
sacred about the circuit, just showing how CMs can do DC coupling to
get rid of the usual caps. But putting in caps between each stage
does complicate matters some, since the CMs do need an input bias
current to establish operating condition for the pentode/diodes. Keep
in mind that CMs want a current input signal rather than a voltage
input. The coupling resistors between stages are calculated to drop
the appropriate voltage from plate to grid/diode using the input bias
current. In current mode coupling, the coupling resistors do not cause
any signal loss, since current in is same as current out. Could put
caps across the resistors for better HF response. Another circuit
sequence more amenable to cap/AC coupling would be: LTP to CMs to Load
resistors to LTP to CMs ....

Don

Patrick Turner wrote:

snip a pile of my stuff,

Patrick Turner.

Please Patrick, why can't you say it in 100 words or less?

I read all your posts, and all your words, and I never complain.

Perhaps you could consider affording me the same consideration.

Now you are making me feel bad!!

However, one of the most important lessons in life we can learn is to be
brief & concise. Lets not confuse the issue, whatever it might be, with a
'fog of bull****', something you as a salesperson are probably aware of.

If you were to respond to your amp customers as you sometimes do
here you would never be able to hear much about what they want to buy.
You should be able to explain most of what you intend without hiding
behind a lot of mathematics.

My comments in respect to THD & IMD were directed at the NG in general,
although your response seems to indicate you felt they were for you. Perhaps
it was you who had commented in respect to the SE guys. On that I don't
know but your reply hints at it. I recall sharing some of my test results with
you & the NG a while back. BTW, how do you measure IMD? What do
you use as your TE for that purpose?

Since I had been curious I set up a simulation today of the driver you use
for the 300W Output Stage which you recently referred too. I can tell you
the driver on it's own has far to much phase shift while there is still gain
so that it would only be conditionally stable in a full loop NFB hookup
with the OP stage connected.

The driver appears to be festooned with HF & LF limiting
time constants, probably in an attempt to achieve stability.
Needless to say, my approach to amp design is quite different,
since I do not agree at all with the 'brute force' approach I see
in your designs. But that is your choice.

Didn't count, but I think I exceeded the 100 word limit!!!

Cheers, John Stewart

In article , Patrick Turner
wrote:

John Byrns wrote:

Could someone please explain how this so called "error correction" system
is any different from ordinary negative feedback? It still seems to
depend on an error in the output signal, just as with negative feedback to
generate the "correction" signal, unlike for example a feed forward signal
which can theoretically cancel the entire error signal in the output.
Looking at the schematics that have been posted before all I see is
slightly complicated implementations of ordinary negative feedback, what
am I missing?

In a true error correction method, no signal voltage is fed back
at one of the two available ports at the input of an amp.

What is applied at the spare input is the the distortion,
after extracting it from a network of 2 resistors, R1 and R2, where R1 is from
a low distortion
source of oppositely phased voltage signal to the output, to which R2
connects.
If the values of R1&R2 are adjusted carefully, the signal voltage is nulled,
leaving
only the fraction of output distortion according to the ratio R1 / ( R1 + R2
).

This "error signal", really a fraction of the distortion which occurs when the
amp has the
the distortion reduction circuitry in place, can be also fed into the same
input
as the input voltage is the source impedance of the error signal is high,
say from the plate of a pentode tube. It has to be phased correctly, like the
error signal contained within a normal feedback signal voltage.

I suggest you carefully re-read my posts all over again to save me having to
yet again spell
out and spoon feed the info.
The schematics I have posted can and should be carefully analysed for the
intantaneous
working voltages and include distortion voltages and currents.

A careful analysis of the schematics you have posted thus far show this
"error correction" scheme to be nothing more than plain old NFB disguised
with a few convoluted circuits. It appears to me that this emperor has no
clothes.

Have you built an amplifier based on your new circuit, how did it perform
vs. the same amplifier with a less convoluted application of the same
amount of negative feedback?

Someone else mentioned SE amplifiers, your "error correction" circuit
ought to work with an SE amplifier as well as the push pull examples you
have posted so far. It would be instructive if you could post a SE
version of the circuit which would eliminate some of the extra circuit
complexity caused by the push pull feature.

Regards,

John Byrns

Surf my web pages at, http://users.rcn.com/jbyrns/

There was an error in the schematic of the 30 watt class A1 amp with balanced
error correction.

Two days ago I posted a schematic of a balanced error correction
amp at abse which applied the similar ideas of an earlier error correction amp
which had unbalanced EC.

The schematic lingered at ABSE, and almost nobody bothered to read or comment
upon it.
I guess it either was too boring an activity or they just don't know WTF I am
on about at all, and skip to the simple one liner posts which they can
understand,
or they cannot understand why anyone would ever bother to use vacuum tubes at
all.

But one thing nobody did was to correct my initial explantion of the way the PP
amp allegedly worked.

For those actually interested in circuits which are beyond the normal run of the
mill,
and slightly beyond the concepts in am ST70 et all,
I suggest you open the file where you have stored the schematic in your
PC so what I am going to say now will make sense.
The title on the image is
"30 watt class A1 with balanced error correction sept 2004"

Take a look at the R network of R1A, R2A and R3A.

I said R1 and R2 form a divider to null the +240 vrms anode signal voltage
against the
drive voltage of -24 vrms at the anode of V2, so the distortion
voltage of the output is reduced to 0.083vrms at the junction of R1-R2,
and this Dv is fed into V4, and amplified to become 0.66Dv at V2 anode, and then
this is applied to
the g1/V6, which generates a load current to cancel the open loop load
distortion load current,
and thus purify the signal, and reduce Ro of the amp.
Thus a calculated 17.6 dB of error correction is obtained.

But what I didn't remember to tell anyone was that the Dv which appears at g1/V4

will be much higher than 0.083v because there is a substantial amount of Dv at
V4 anode,
and also at V6 anode, and therefore there would be a lot lot more than 0.083v at
g1/V4
which is connected to the junction of R1A and R2A.
If indeed 0.66Dv appears at V2 anode, and Dv appears at V6anode,
how could 0.083 dV appear at R1-R2 junction?
The V4 grid circuit is a high resistance circuit.

In fact as the circuit is drawn, the gain of V2&3 would cause the circuit to
oscillate very badly
because there is a positive feedback effect going on.

Luckily there is a simple answer.

The resistor R3 which is a bias R to V4 needs to be reduced to
somewhere between 5k and 12k ohms, so that indeed the Dv which appears at g1/V4
is at a low level such as not to drive the V2&V4 into oscillation.

The 2 uF coupling cap would then perhaps need to be increased to about 10 uF.

when R3 is smaller than 5k, the value of Dv appearing at the anode of V2
will be smaller than the value of Dv at the anode of V6, and all is well,
because
once the Dv applied to g1/V6 becomes as big as Dv at the anode of V6, it means
you theoretically have 20 dB of error correction, and stability issues
raise their head in a similar manner as with 20 or more Db of NFB application.

As R3 is increased beyond 5k, the value of DV at anode of V2 increases beyond
1 x Dv, so when its applied to the g1/V6, the amount of correction
becomes too great for stability to be maintained.

Basically, R1 / R3 should be less than the differential gain of the LTP formed
by V2/V4.

In the earlier first schematic where a separate stand alone triode has its
output voltage
in opposition with the OPT secondary voltage, there is no such issue of PFB
involved.

Patrick Turner.

I read in alt.binaries.schematics.electronic that Patrick Turner
wrote (in
u) about 'Who needs NFB when there is error correction?', on Fri, 10
Sep 2004:

The schematic lingered at ABSE, and almost nobody bothered to read or
comment upon it.

No-one is under any obligation to do so.
--
Regards, John Woodgate, OOO - Own Opinions Only.
The good news is that nothing is compulsory.
The bad news is that everything is prohibited.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk

Patrick Turner wrote:
Its lucky that SE amps and class A PP amps have extremely small
amounts of 3,4,5,6,7,8,9,10th harmonics at the normal listening level.
But increase the volume level to the point where occasional clipping
takes place,

Have you ever asked yourself, why the second order distortion is so much
less disturbing? Maybe it is because the original sound covers around
40dB/oct, so a 1% 2nd harmonic cannot even be heard. Not so with the third
harmonic and upwards.

The main thing about NFB is to not use too much.
Choose a circuit that could be used without NFB.
Use it with best practices of critical damping in mind,
and I suggest one reason why some DIY types loathe NFB
is that they have no clue how to apply it properly,
since they can only think and feel subjectively, and any boring
rationalization with numbers reaks to them of compromise.
Feedback equations, load line analysis, distortion measurements,
ohms law, filter theory are all mysteries to such ppl.
Its probably better they struggle to build a nice simple SET amp,
because anything else is quite bamboozling.

I have shared some very fine musical experiences with such
delightfully ignorant people.
When you listen to what they have made, the noise is low,
the distortion is below threshold, output impedance is sufficently
low,
their room is treated right, their speakers have a sufficiently flat
response, and the music conveys emotion and intimacy of the recorded
moment
and the intentions of the composer.
So they know enough.
And its the rule of "enoughness" at work here which we can admire.

Patrick Turner.

So you consider ourselfs smart and informed engineers, who doesn't agree?

Hey you should look closer at speaker chassis. They do have quite high
levels of distortion at already 90dB SPL, but some have more 3rd harmonics
whereas some others have very low 3rd (0.1%) but quite high 2nd
harmonics(1%) these should have a similar effect as your tube amp. Most high
efficiency speakers(Lowther) have quite higher THD values. That is a turnoff
in my ears, so I doubt your anecdote here.
Lately I use the Usher GT10, because I like the low distortion of this
woofer.

Could you not drop feedback altogether if you use a "predistortion" which
compensates for the transfer curve nonlinearities? Should be possible with a
DSP and some adaptive filtering by measuring output current and -voltage.
This way also the loudspeaker could be linearized (and protected as a
byproduct).

Maybe an extra control could then deliberatly add 2nd harmonics for the tube
affectionados.
--
ciao Ban
Bordighera, Italy

John Stewart wrote:

Patrick Turner wrote:

snip a pile of my stuff,

Patrick Turner.

Please Patrick, why can't you say it in 100 words or less?

I read all your posts, and all your words, and I never complain.

Perhaps you could consider affording me the same consideration.

Now you are making me feel bad!!

However, one of the most important lessons in life we can learn is to be
brief & concise. Lets not confuse the issue, whatever it might be, with a
'fog of bull****', something you as a salesperson are probably aware of.

Was I ever a salesman?
Goodness me, I don't recall ever having to work at selling what I make,
the gear speaks for itself, I just stand back and let them listen, and if they
buy,
its the soldering iron that gets the credit.

If you were to respond to your amp customers as you sometimes do
here you would never be able to hear much about what they want to buy.
You should be able to explain most of what you intend without hiding
behind a lot of mathematics.

Gees, I thought I was the ultimate minimalist when it comes to
using math to describe electronic functions.

My comments in respect to THD & IMD were directed at the NG in general,
although your response seems to indicate you felt they were for you. Perhaps
it was you who had commented in respect to the SE guys. On that I don't
know but your reply hints at it. I recall sharing some of my test results with
you & the NG a while back. BTW, how do you measure IMD? What do
you use as your TE for that purpose?

I used to measure IMD using the 4v of LF to 1v of HF ratio,
50 Hz to about 7 kHz, but since I always only ever measured what one is supposed
to measure
as laid out in RDH4, I simply measure THD at 1 kHz and use empirical reasoning
to establish the rest. Since all my amps exceed best standard practice as laid
out in RDH4, I have never got a surprisingly poor test result.

The IMD gear is elegantly simple.
It has a passive high pass filter to remove all the LF, leaving only the
products over about 3 kHz, and this is displayed on the CRO
and the 50 Hz AM modulation of the 7kHz is plain to see if its a high value.

The 7 kHz is run through a diode and RC detector, and amplitude of the AM
measured.
If I have 4 volts of 7 kHz, and 0.05volts of AM, the IMD% can be worked out.
The shape of the AM can be displayed on the CRO against the shape of the LF wave
causing the modulation of the HF, The IMD gets real bad at clipping.

I doubt I need to analyse what exact products get produced.
I have not got a filter to display the 6,950 Hz and 7,050 Hz products,
ie the two side bands typically produced by a large LF wave working
to change the gain of the amp during LF signal swings.

All IMD is bad, some badder than others, and I guess the IMD from SE amps
is not as bad as the IMD from a PP amp, if the measurements are the same.

Since I had been curious I set up a simulation today of the driver you use
for the 300W Output Stage which you recently referred too. I can tell you
the driver on it's own has far to much phase shift while there is still gain
so that it would only be conditionally stable in a full loop NFB hookup
with the OP stage connected.

The driver for the 300 watt amp is one of the best.
The use of the series R between the ends of the CT choke and the anodes of the
6BQ5 in triode isolates the effects of the choke, since Ra is 2k2
and the series R is 9.4k.
So the choke's normal phase shift causing tendencies at extreme LF and HF are
not possible.
I now have dropped the Rk network for this driver, so
its got even better phase performance.

The OPTs are good from 18 Hz ( saturation ) to 270 kHz at 320 watts.
The bandwidth was limited to 65 kHz like all my amps by means of
critical damping.
The amp was easy to stabilise, with a large NFB safety margin.

I now am working on a pair where there is 20% of cathode FB
rather similar to a Quad II circuit, with a regulated G2 supply.
This amounts to about 9 dB of FB in the output stage.
There is another 7 dB of global NFB,
and thd is better than the plain UL was.

The max drive voltage is about 85 vrms to each of the output grids,
and because the drive amp is so darn linear, I don't need a huge amount of global
NFB to linearize its distortions.

The driver appears to be festooned with HF & LF limiting
time constants, probably in an attempt to achieve stability.

There is some partial cathode bypassing of the 100 ohm Rk of each 6BQ5
with a 100 plus 0.0082 uF in series k-k, to slightly boost
the phase of the LTP.
Also typical RC HF gain/phase shift reducer networks to control
extreme LF and HF response.

The last thing you ever want in a tube amp is a huge amount of NF applied at
100 kHz or 20 Hz; there should only be a few dB at most.

No matter how hard you try, the combined effect of a capacitor load
with the leakage inductance will try to cause a phase shift which will cause
oscillations at HF.

The LL of an OPT is like a series inductance fitted to the output of most
SS amps, with the difference being that in a tube amp the FB is usually taken from

the secondary, where the LL is *included* in the FB loop, wheras
in an SS amp, the NFB is always taken from the output of the devices which is
*before*
the series L zobel network.
If you try to take the FB from the speaker terminal of an SS amp with
LR-C zobel netork included in the FB loop, all the problems one has with
applying lots of NFB in a tube amp are then just the same.

Needless to say, my approach to amp design is quite different,
since I do not agree at all with the 'brute force' approach I see
in your designs. But that is your choice.

Brute force?

I make amps from 10 to 300 watts, and all are sophisticated
examples of tube craft. Their behaviour is that of the capable
gentleman with panache and verve.
There isn't anything troglodyte about my amps.
Nothing barbaric.
But any 300 watt amp could be a brute in the wrong hands.

Some ppl want 300 watts.

And these 300 watters are not mainly class B amps like most 300 watt amps are.
The first 50 watts is class A, with 4 ohms, and 100 watts with 8 ohms.

Using the 300 watter is like having 6 x 50 watt class AB amps in parallel,
where a pair of KT88/6550 make 53 watts into 4 ohms, and 34 watts into 8 ohms,
the latter being mostly class A.
One guy had speakers rated for 80 dB/W/M, so I figured
he needed at least 6 times the power I need with my 89DB/W/M speakers.
The maker recommended 200 watt amps, so I gave him
300+ watts of class AB, with 220 watts of AB into 8 ohms
with 100W of class A.
And each tube is set up to idle at 490v x 50 mA, nice and easy,
at 24.5 watts per 6550.
The total plate input power is 294 watts at idle, max class A efficiency is 42%,
so max class A is 123 watts.
Load matching allows high class A% into 2 ohm loads.

I know a guy who ran his amp with 6550 at 40 watts each,
but that's too hot, and the tubes don't last, but he got
much more class A.
Now he has gone to 50 watt SE 805 amps.

ARC made a Reference 600W with 16 x 6550 within,
and that's 75 watts per pair of output tubes, like McIntosh, etc.

I'd rather reduce the power produced per tube, and I don't have any intention
to set power records with tubes.
If I did, I might use transmitting tubes to get the kW+ with just a few tubes,
but then rarely do such high power jobs ever have a lot of class A.
A Quad of 833 in PP might be interesting though, with about 1500 v supply.

Didn't count, but I think I exceeded the 100 word limit!!!

Have a great weekend,

Patrick Turner.

Cheers, John Stewart

I read in alt.binaries.schematics.electronic that Ban
wrote (in ) about 'Who needs NFB
when there is error correction?', on Fri, 10 Sep 2004:

Have you ever asked yourself, why the second order distortion is so much
less disturbing? Maybe it is because the original sound covers

'masks', not covers, and the figure is nearer 35 dB/octave I think.
around
40dB/oct, so a 1% 2nd harmonic cannot even be heard. Not so with the
third harmonic and upwards.

I suspect that in a direct comparison, many people could hear 1% second
harmonic (if a loudspeaker with sufficiently low distortion is
available!). I know that quite small percentages of third are very
audible on tone signals; this came up due to the use of old BFO audio
generators for squeaking TV loudspeakers for coil rub.
--
Regards, John Woodgate, OOO - Own Opinions Only.
The good news is that nothing is compulsory.
The bad news is that everything is prohibited.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk

John Byrns wrote:

In article , Patrick Turner
wrote:

John Byrns wrote:

Could someone please explain how this so called "error correction" system
is any different from ordinary negative feedback? It still seems to
depend on an error in the output signal, just as with negative feedback to
generate the "correction" signal, unlike for example a feed forward signal
which can theoretically cancel the entire error signal in the output.
Looking at the schematics that have been posted before all I see is
slightly complicated implementations of ordinary negative feedback, what
am I missing?

In a true error correction method, no signal voltage is fed back
at one of the two available ports at the input of an amp.

What is applied at the spare input is the the distortion,
after extracting it from a network of 2 resistors, R1 and R2, where R1 is from
a low distortion
source of oppositely phased voltage signal to the output, to which R2
connects.
If the values of R1&R2 are adjusted carefully, the signal voltage is nulled,
leaving
only the fraction of output distortion according to the ratio R1 / ( R1 + R2
).

This "error signal", really a fraction of the distortion which occurs when the
amp has the
the distortion reduction circuitry in place, can be also fed into the same
input
as the input voltage is the source impedance of the error signal is high,
say from the plate of a pentode tube. It has to be phased correctly, like the
error signal contained within a normal feedback signal voltage.

I suggest you carefully re-read my posts all over again to save me having to
yet again spell
out and spoon feed the info.
The schematics I have posted can and should be carefully analysed for the
intantaneous
working voltages and include distortion voltages and currents.

A careful analysis of the schematics you have posted thus far show this
"error correction" scheme to be nothing more than plain old NFB disguised
with a few convoluted circuits. It appears to me that this emperor has no
clothes.

Its not NFB.
At the rated load, its possible to have no signal applied to the second input,
just the fracton of distortion, and in the case of ther PP balanced amp,
this is amplified by a positive FB loop, if you followed my reasoning...
But the application of the distortion signal to the
right point in the amp does reduce the distortion in much the same way
as NFB.

Everyone should be able to see the distinction I make about NFB and
error correction.

Have you built an amplifier based on your new circuit, how did it perform
vs. the same amplifier with a less convoluted application of the same
amount of negative feedback?

Yes, and the amp tended to oscilate at HF probably due to too much error correction.

I have not used it in commercial designs.

I mainly rasied the issue to share the thoughts with other thinking folks
who wonder like I do whether its worth persuing some other path
because that keeps life interesting.

Someone else mentioned SE amplifiers, your "error correction" circuit
ought to work with an SE amplifier as well as the push pull examples you
have posted so far. It would be instructive if you could post a SE
version of the circuit which would eliminate some of the extra circuit
complexity caused by the push pull feature.

The initial error correction I posted has a stand alone triode at the front end
to generate a linear current/voltage to "buck" the current in a resistor from the
output,
and the amp's distortion is thus left intact while most of the signal is removed,
and this is applied to the spare input of an LTP.
This *is* a very simple means of error reduction;
no more parts are used compared to a standard PP amp.

The second balanced error correction amp uses the existing
front end tubes to drive the outputs *and* to get a signal nulled source of
distortion voltage, see R1A, and R2A, etc.

But R3A has to be about 6k ohms or else the input tubes have far too much
positive FB, and instability results.

The ratio of R1/R3 should be greater than the differential gain of the
LTP of each half of the balanced driver amp.

I have another design up my sleeve which uses a pair of CF off the
R1-R2 junction, then an 18 : 1 transformer to get the distortion voltage at R1-R2
junction reduced
to a small voltage from a very low impedance source from the secondary winding,
which can then be applied
to the cathodes of a normal LTP two triode driver.
The secondary has a CT and CCS to some -ve V, and the
LTP then gets its correction voltage applied to its cathodes from a floating
secondary,
so the LTP can be simply driven into one side.
The usual input triode is thus eliminated.

I will post the schematic tomorrow when I am happy it should work

What is possible intrigues me.

Like NFB, the amount of error correction is dependant on the gain of the input stage

which stays virtually fixed, and the varying gain of the output stage,
so that with low value loads, the gain reduces, and so does the amount of NFB.
Its particularly true of pentode output stages.
Every amp shouldn't oscillate wildly if the output is short circuited.
With no signal, shunting the output should do nothing, but
with too much error correction in the balanced EC amp
I last posted, it would go beserko with a shorted output,
as analysis of the voltages produced will show.

Patrick Turner.

Regards,

John Byrns

Surf my web pages at, http://users.rcn.com/jbyrns/

On Sat, 11 Sep 2004 03:07:36 +1000, Patrick Turner
wrote:

[snip]
Its not NFB.
At the rated load, its possible to have no signal applied to the second input,
just the fracton of distortion, and in the case of ther PP balanced amp,
this is amplified by a positive FB loop, if you followed my reasoning...
But the application of the distortion signal to the
right point in the amp does reduce the distortion in much the same way
as NFB.

Everyone should be able to see the distinction I make about NFB and
error correction.

[snip]

A "distinction" without a difference.

Time to PLONK you... you're more hopeless than Paul Burridge.

...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona Voice

480)460-2350 | |
| E-mail Address at Website Fax

Ban wrote:

Patrick Turner wrote:
Its lucky that SE amps and class A PP amps have extremely small
amounts of 3,4,5,6,7,8,9,10th harmonics at the normal listening level.
But increase the volume level to the point where occasional clipping
takes place,

Have you ever asked yourself, why the second order distortion is so much
less disturbing? Maybe it is because the original sound covers around
40dB/oct, so a 1% 2nd harmonic cannot even be heard. Not so with the third
harmonic and upwards.

Well I like *all* distortion reduced in tube amps to a lot
less than 1%.

Figured what the IMD becomes when 1% 2H is present?

Its probably better sounding than when 1% of 3H is there,
but I'd much rather have 0.2% thd max, regardless of what the harmonics are.

If you have a tone of 200 Hz which makes a sound level of 90 dB,
say 2.83v into 8 ohms, say 1 watt,
and then there is 1% of 400 Hz present, then the 400Hz voltage is 0.0283vrms,
and then the 2H power is 0.0001 watts, which produces an SPL = 50 dB.
Can it be heard? probably, by trained ears,
especially if the 400 Hz is switched on and off.
And don't forget, the ear is more sensitive to an equal power level of 400 Hz
than it is to
200 Hz.
Trey feeding a tone voltage of 28.3 mV to an average speaker.
You betcha you hear it!

The main thing about NFB is to not use too much.
Choose a circuit that could be used without NFB.
Use it with best practices of critical damping in mind,
and I suggest one reason why some DIY types loathe NFB
is that they have no clue how to apply it properly,
since they can only think and feel subjectively, and any boring
rationalization with numbers reaks to them of compromise.
Feedback equations, load line analysis, distortion measurements,
ohms law, filter theory are all mysteries to such ppl.
Its probably better they struggle to build a nice simple SET amp,
because anything else is quite bamboozling.

I have shared some very fine musical experiences with such
delightfully ignorant people.
When you listen to what they have made, the noise is low,
the distortion is below threshold, output impedance is sufficently
low,
their room is treated right, their speakers have a sufficiently flat
response, and the music conveys emotion and intimacy of the recorded
moment
and the intentions of the composer.
So they know enough.
And its the rule of "enoughness" at work here which we can admire.

Patrick Turner.

So you consider ourselfs smart and informed engineers, who doesn't agree?

There's all types in the world, some who know a lot, some who know a little,
some who make fine systems despite their lack of knowledge.
This is mainly a group of audio amateurs, judging by the questions posted here.
There are some very smart cookies lurking here.....

Hey you should look closer at speaker chassis. They do have quite high
levels of distortion at already 90dB SPL, but some have more 3rd harmonics
whereas some others have very low 3rd (0.1%) but quite high 2nd
harmonics(1%) these should have a similar effect as your tube amp.

Indeed.
And depending which way you connect the pair of speaker wires may
make the amp's thd cancel or add to the speaker's thd.

But usually the thd of speakers is a many varied thing since there
is usually more than one driver, and the harmonics vary with F,
and perhaps their phase.
Everything about electro-acoustics and transducers is never simple of perfect.

Most high

efficiency speakers(Lowther) have quite higher THD values.

I thought horn loading reduced thd....

That is a turnoff
in my ears, so I doubt your anecdote here.
Lately I use the Usher GT10, because I like the low distortion of this
woofer.

Could you not drop feedback altogether if you use a "predistortion" which
compensates for the transfer curve nonlinearities?

Many SE amps use this inherently. The 3% of 2H produced by the
driver tube cancels part of the usual 5% 2H of the output tube.
But only when its working hard.

Trying to generate the pre-distortion from some device outside the amp
is far too difficult to reliably achieve to bother with it.
Nobody has succeded yet.

Should be possible with a
DSP and some adaptive filtering by measuring output current and -voltage.
This way also the loudspeaker could be linearized (and protected as a
byproduct).

Filter delays could be a problem, but who knows with DSP...

Maybe an extra control could then deliberatly add 2nd harmonics for the tube
affectionados.

I think the musicians have a range of at least 100 little black boxes they can
plug into
their guitar amps to change the sound by changing harmonic content.

But when I hear a Stradvarus, I don't want the gear to have any more than 0.1%
thd
preferably only a bit of 2H.
It doesn't have to be loud; only as loud as listening to the violin
being played 3 metres away, or 30 metres away in the orchestra.

Patrick Turner.

--
ciao Ban
Bordighera, Italy

Jim Thompson wrote:

On Sat, 11 Sep 2004 03:07:36 +1000, Patrick Turner
wrote:

[snip]
Its not NFB.
At the rated load, its possible to have no signal applied to the second input,
just the fracton of distortion, and in the case of ther PP balanced amp,
this is amplified by a positive FB loop, if you followed my reasoning...
But the application of the distortion signal to the
right point in the amp does reduce the distortion in much the same way
as NFB.

Everyone should be able to see the distinction I make about NFB and
error correction.

[snip]

A "distinction" without a difference.

Time to PLONK you... you're more hopeless than Paul Burridge.

You can use your time avoiding me to buy ferry tickets instead of going sailing
in a real yacht. You wouldn't know the difference between the two experiences,
its all boating, and the same thing.

I can be such a disagreeable person, but that doesn't always make you right :-)

Patrick Turner.

Patrick Turner wrote:

Figured what the IMD becomes when 1% 2H is present?

Its probably better sounding than when 1% of 3H is there,
but I'd much rather have 0.2% thd max, regardless of what the
harmonics are.

If you have a tone of 200 Hz which makes a sound level of 90 dB,
say 2.83v into 8 ohms, say 1 watt,
and then there is 1% of 400 Hz present, then the 400Hz voltage is
0.0283vrms, and then the 2H power is 0.0001 watts, which produces an
SPL = 50 dB.
Can it be heard? probably, by trained ears,
especially if the 400 Hz is switched on and off.
And don't forget, the ear is more sensitive to an equal power level
of 400 Hz than it is to
200 Hz.
Trey feeding a tone voltage of 28.3 mV to an average speaker.
You betcha you hear it!

Yeah, but the *masking* as JW courtiously corrected will make it unhearable.
We do *not* hear power but sound pressure which is proportional to voltage
for a fixed impedance, so power figures can be a bit misleading.

Hey you should look closer at speaker chassis. They do have quite
high levels of distortion at already 90dB SPL, but some have more
3rd harmonics whereas some others have very low 3rd (0.1%) but quite
high 2nd harmonics(1%) these should have a similar effect as your
tube amp.

Indeed.
And depending which way you connect the pair of speaker wires may
make the amp's thd cancel or add to the speaker's thd.

That is wishful thinking IMHO.

But usually the thd of speakers is a many varied thing since there
is usually more than one driver, and the harmonics vary with F,
and perhaps their phase.
Everything about electro-acoustics and transducers is never simple of
perfect.

Most high

efficiency speakers(Lowther) have quite higher THD values.

I thought horn loading reduced thd....

Horn loading reduces the excursion at low frequencies, which is one factor
in a loudspeaker.
Another factor is the breakup of the cone at higher frequencies. Efficiency
increases with bigger radiating surface and less moved mass, both
tendencies increase the distortion.

Trying to generate the pre-distortion from some device outside the amp
is far too difficult to reliably achieve to bother with it.
Nobody has succeded yet.

Now this would be a challenge for you. Outside or inside do not matter if
you can measure certain parameters and digitally process them. I'm sure you
could do that with matlab and your computer. if you have input voltage and
O/P voltage and -current these are already some starting values. Get some
more like filament temperature, supply voltage etc. and maybe a calibration
routine to find initial values.
If a car catalyzer can be controlled to put out less poisonous emissions,
why not a tube amp as well?

Should be possible with a
DSP and some adaptive filtering by measuring output current and
-voltage. This way also the loudspeaker could be linearized (and
protected as a byproduct).

Filter delays could be a problem, but who knows with DSP...

But you do not feed back the output, you analyse the output and find the
proper correction of the transfer function. This process will be influenced
mainly by temperature and supply voltage, so the time constants are not that
high, especially with tubes and power transistors on heatsinks.

But when I hear a Stradvarus, I don't want the gear to have any more
than 0.1% thd
preferably only a bit of 2H.
It doesn't have to be loud; only as loud as listening to the violin
being played 3 metres away, or 30 metres away in the orchestra.

You see, if there should be distortion it is already on the CD, because the
guitar player has used his Mesa-Boogie for the recording with a good mike.
So for what these tube amps SET. How can the added coloration be useful if
we listen to recorded music? And when it is not colouring the sound as the
high quality tube amps, why not use a transistor amp with so much more
performance and power per buck?

--
ciao Ban
Bordighera, Italy

I read in alt.binaries.schematics.electronic that Patrick Turner
wrote (in
u) about 'Who needs NFB when there is error correction?', on Sat, 11
Sep 2004:
Trey feeding a tone voltage of 28.3 mV to an average speaker. You betcha
you hear it!

Of course, in the absence of the fundamental 40 dB louder. But not
easily, and maybe not at all, in its presence.
--
Regards, John Woodgate, OOO - Own Opinions Only.
The good news is that nothing is compulsory.
The bad news is that everything is prohibited.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk

I take it that the error in the schematic, which you don't explicitly
identify was that the value of R3 was too large to prevent the driver
stage from oscillating? I did originally notice the positive feedback
loop around the driver, but positive feedback loops are hardly unknown in
Hi-Fi amps, and I assumed that you had made sure that there was not enough
positive feedback to cause instability.

You say some people "cannot understand why anyone would ever bother to use
vacuum tubes at all." That is a separate issue from your "error
correction" circuit which should not be dependent on vacuum tube
circuitry, and should also work with transistor circuits if it is valid.

I think your first schematic was a better presentation of your idea in its
pure form, and I will take it as the basis of a more generic circuit to
explain why I believe there is no difference at all between your so called
"error correction" circuit and the common negative feedback circuit. You
will have to imagine the block schematic in your mind.

The amplifier schematic titled "Error Correction, Typical UL Amp September
2004" can be broken down into two amplifier modules. The first is a power
amplifier module with differential inputs, and a voltage gain of about 10
times. For simplicity I will assume the differential stage is "blameless"
and all distortion is generated in the output stage. To further simplify
the discussion I will use my own numbers, which only approximate your
actual amplifier. Your amplifier starts by feeding the input signal, say
1 vrms into the positive input and grounding the negative input of the
differential power amplifier. This results in a 10 volt output from the
power amplifier into the 6 Ohm load. To correct the output errors you
then feed the output back to the negative input, but in addition to the
distortion components the output also includes the desired output signal.
Unfortunately feeding back the amplified input along with the errors
greatly reduces the gain of the amplifier, from 10X to about 0.5X. To
restore the gain you then feed the input signal through a second auxiliary
voltage amplifier with a gain of -10X and add this negative amplified
version of the input signal to the signal from the output before feeding
the combination into the negative input of the differential input main
power amplifier in order to cancel the unwanted portion of the output
feedback signal which represents the amplified input. This action
restores the apparent gain of the amplifier to 10X.

Ignoring the details of how we might actually implement a summer for the
output signal and the amplified and inverted version of the input signal
to apply to the negative input of the main amplifier, we have the
following equation for the amplifier gain, where Vin is the input to the
auxiliary voltage amplifier:

(1) Vout = 10 x ( Vin - ( Vout + ( -10 x Vin)))

Which reduces to:

(2) Vout = 10 x Vin

Equation #1 can also be rewritten in this form:

(3) Vout = 10 x ( - ( - Vin + Vout + ( -10 x Vin)))

which is equal to:

(4) Vout = 10 x ( - ( + Vout + ( -11 x Vin)))

Which again reduces to"

(5) Vout = 10 x Vin

Which is identical to equation #2

This is mathematically equivalent to disconnecting the input signal from
the positive input of the main amplifier, and feeding an additional
inverted version of the input signal into the summer feeding the positive
input.

This is equivalent to a common negative feedback amplifier where the
output and input signals are summed at the input of an inverting
amplifier. In this example, as well as your first example schematic, the
main amplifier has less than unity gain with feedback applied,
necessitating the voltage amplifier preceding the main amplifier.

Does this analysis capture the essence of your original "error correction"
circuit if not the exact numbers? If it does, what do you claim is the
difference between your "error correction" circuit, and a simple negative
feedback amplifier where the positive input of the main amplifier is
grounded and the gain of the auxiliary voltage amplifier is increased
slightly to compensate?

Regards,

John Byrns

In article , Patrick Turner
wrote:

There was an error in the schematic of the 30 watt class A1 amp with balanced
error correction.

Two days ago I posted a schematic of a balanced error correction
amp at abse which applied the similar ideas of an earlier error correction amp
which had unbalanced EC.

The schematic lingered at ABSE, and almost nobody bothered to read or comment
upon it.
I guess it either was too boring an activity or they just don't know WTF I am
on about at all, and skip to the simple one liner posts which they can
understand,
or they cannot understand why anyone would ever bother to use vacuum tubes at
all.

But one thing nobody did was to correct my initial explantion of the way
the PP
amp allegedly worked.

For those actually interested in circuits which are beyond the normal
run of the
mill,
and slightly beyond the concepts in am ST70 et all,
I suggest you open the file where you have stored the schematic in your
PC so what I am going to say now will make sense.
The title on the image is
"30 watt class A1 with balanced error correction sept 2004"

Take a look at the R network of R1A, R2A and R3A.

I said R1 and R2 form a divider to null the +240 vrms anode signal voltage
against the
drive voltage of -24 vrms at the anode of V2, so the distortion
voltage of the output is reduced to 0.083vrms at the junction of R1-R2,
and this Dv is fed into V4, and amplified to become 0.66Dv at V2 anode,
and then
this is applied to
the g1/V6, which generates a load current to cancel the open loop load
distortion load current,
and thus purify the signal, and reduce Ro of the amp.
Thus a calculated 17.6 dB of error correction is obtained.

But what I didn't remember to tell anyone was that the Dv which appears
at g1/V4

will be much higher than 0.083v because there is a substantial amount of Dv at
V4 anode,
and also at V6 anode, and therefore there would be a lot lot more than
0.083v at
g1/V4
which is connected to the junction of R1A and R2A.
If indeed 0.66Dv appears at V2 anode, and Dv appears at V6anode,
how could 0.083 dV appear at R1-R2 junction?
The V4 grid circuit is a high resistance circuit.

In fact as the circuit is drawn, the gain of V2&3 would cause the circuit to
oscillate very badly
because there is a positive feedback effect going on.

Luckily there is a simple answer.

The resistor R3 which is a bias R to V4 needs to be reduced to
somewhere between 5k and 12k ohms, so that indeed the Dv which appears
at g1/V4
is at a low level such as not to drive the V2&V4 into oscillation.

The 2 uF coupling cap would then perhaps need to be increased to about 10 uF.

when R3 is smaller than 5k, the value of Dv appearing at the anode of V2
will be smaller than the value of Dv at the anode of V6, and all is well,
because
once the Dv applied to g1/V6 becomes as big as Dv at the anode of V6, it means
you theoretically have 20 dB of error correction, and stability issues
raise their head in a similar manner as with 20 or more Db of NFB application.

As R3 is increased beyond 5k, the value of DV at anode of V2 increases beyond
1 x Dv, so when its applied to the g1/V6, the amount of correction
becomes too great for stability to be maintained.

Basically, R1 / R3 should be less than the differential gain of the LTP formed
by V2/V4.

In the earlier first schematic where a separate stand alone triode has its
output voltage
in opposition with the OPT secondary voltage, there is no such issue of PFB
involved.

Patrick Turner.

Surf my web pages at, http://users.rcn.com/jbyrns/

Why is this continuing on alt.binaries.pictures ?
--
Carl
WA1KPD
Visit My Boatanchor Collection at
http://home.comcast.net/~chnord/wa1kpd.html

"Patrick Turner" wrote in message
...

smoking-amp wrote:

My reply from Sept 07 seems to have disappeared into cyber space. I
will try to remember what I said:

I posted the vacuum tube current mirror circuit on ABSE. Also is a
thread on this at the diyAudio tube forum too, in case someone can't
get ABSE. I use an assortment of fixtures with many diode tubes
connected in series, with selectable series tap points for
experimenting and designing with this idea. 6JU8 and 9006 diode tubes
are decent choices. Smaller diodes and higher gm pentodes in general
give higher current gain.

I was able to view the basic schematic of the dioses plus pentode
at abse, but what is the schematic for their applied use?

I don't try to reach extremely low levels of dist. in my designs
either, my .01% figures were just for illustration.

Saw your new class A error correction circuit on ABSE, similar to what
I had imagined would be required for a non-circlotron P-P with an
error corr. An LTP for each tube since the output xfmr. cannot be so
depended on to accurately cross couple signal between primary halves
in a normal P-P. The cross connected plates is interesting. If this
cross connecting of plates is left out, would it not work for class
AB? Most likely will want to use a trimpot on each error corr.
feedback network to set subtractor loop gain precisely, this is
something I think is needed for all P-P err. corr. designs in general,
since the output tubes will never be exactly matched.

I was thinking over your earlier comment about the cross coupling and
oscillator similarity in the circlotron like design, and maybe missed
your point somewhat in my earlier reply. My comments were on the
cathode outputs fully driving the primary between the sampling points,
so that no cross coupling thru the transformer was necessary or
likely, which is the major worry in a normal P-P err. corr. design.
But there is obvious cross coupling in the LTP error corr. circuitry
itself by design, which does make it look like a typical osc. circuit.
This similarity to an oscillator is actually inherent in all error
corr. designs, some just more obvious than others. The subtractor
circuit must be spot on in accuracy or some residual signal is left in
its output with either + or - polarity. This residual amounts to
additional positive or negative feedback added to a loop that is right
at unity gain, so oscillation is a constant threat. This is a generic
problem with all error corr. type circuits. This is somewhat like the
problem with bootstrapped load resistors, too much gain and its
positive feedback, too little gain and its not quite a current source
load but a high value resistor instead, and just the right gain makes
it look like a current source load as wanted. An err. corr. design
operating at .9999 loop gain for say .01% distortion residual (in
theory anyway) would be susceptible to any .01% variation in the
feedback loop gain, potentially causing oscillation. Hence my comment
about using error corr. circuitry to just lop off the bulk, say 90% of
distortion, but using conventional NFB for pushing the decimal point
beyond that.

We saw this post yesterday I think.
I addressedissues raised.

Patrick Turner.

Don

Ban wrote:

Patrick Turner wrote:

Figured what the IMD becomes when 1% 2H is present?

Its probably better sounding than when 1% of 3H is there,
but I'd much rather have 0.2% thd max, regardless of what the
harmonics are.

If you have a tone of 200 Hz which makes a sound level of 90 dB,
say 2.83v into 8 ohms, say 1 watt,
and then there is 1% of 400 Hz present, then the 400Hz voltage is
0.0283vrms, and then the 2H power is 0.0001 watts, which produces an
SPL = 50 dB.
Can it be heard? probably, by trained ears,
especially if the 400 Hz is switched on and off.
And don't forget, the ear is more sensitive to an equal power level
of 400 Hz than it is to
200 Hz.
Trey feeding a tone voltage of 28.3 mV to an average speaker.
You betcha you hear it!

Yeah, but the *masking* as JW courtiously corrected will make it unhearable.
We do *not* hear power but sound pressure which is proportional to voltage
for a fixed impedance, so power figures can be a bit misleading.

But where you would have 1% thd, the IMD under a standard test
of 4v LF to 1v HF will give several % .
There are many F all present in music, so several % IMD
is a little too much imho.

.....

I thought horn loading reduced thd....

Horn loading reduces the excursion at low frequencies, which is one factor
in a loudspeaker.
Another factor is the breakup of the cone at higher frequencies. Efficiency
increases with bigger radiating surface and less moved mass, both
tendencies increase the distortion.

I like to use 200mm to 300mm dia woofers, 140mm mids,
and dome tweeters to avoid most cone break up.
The seakers I use are inefficient compared to a horn
but they sound linear to me, and free of most resonances
in poorly made enclosures.
I'll start using horns if I hear a pair better than my own dynamics in very
stout enclosures
with very good rear absorbtion.
I've been waiting for 20 years.....

Trying to generate the pre-distortion from some device outside the amp
is far too difficult to reliably achieve to bother with it.
Nobody has succeded yet.

Now this would be a challenge for you. Outside or inside do not matter if
you can measure certain parameters and digitally process them. I'm sure you
could do that with matlab and your computer. if you have input voltage and
O/P voltage and -current these are already some starting values. Get some
more like filament temperature, supply voltage etc. and maybe a calibration
routine to find initial values.
If a car catalyzer can be controlled to put out less poisonous emissions,
why not a tube amp as well?

I don't have the time for that type of R&D.

Listeners never speak of the poisonous sonic emmissions of the equipment
I fabricate.

Should be possible with a
DSP and some adaptive filtering by measuring output current and
-voltage. This way also the loudspeaker could be linearized (and
protected as a byproduct).

Filter delays could be a problem, but who knows with DSP...

But you do not feed back the output, you analyse the output and find the
proper correction of the transfer function.

I analyse nothing. The voltage fed back containing thd/imd phase shift, F
response analomies
is all automatically amplified in such a way to reduce the horrors by around
16 dB which is good enough considering how substantially good
the device is even with no correction by NFB or any other method.

This process will be influenced

mainly by temperature and supply voltage, so the time constants are not that
high, especially with tubes and power transistors on heatsinks.

Time constants in amplifiers are independant of temperatures and heatsinking.
The phase inaccuracies are due to LC&R inter action.

I suggest you read a few books on basic electronics.

But when I hear a Stradvarus, I don't want the gear to have any more
than 0.1% thd
preferably only a bit of 2H.
It doesn't have to be loud; only as loud as listening to the violin
being played 3 metres away, or 30 metres away in the orchestra.

You see, if there should be distortion it is already on the CD, because the
guitar player has used his Mesa-Boogie for the recording with a good mike.

There is very little done by Mesa Boogies that I really prefer when compared to
old violins and cellos about 200 years old.

But I agree, a few more % of grunge added by a crummy hi-fi system
don't some rock and roll music, heavy metal any harm, its 90%
noise made by the noisy, uneducated, untrained, undisciplined, bad habbited and
rebellious youths
who have not matured in any way I can tell.
I am toxically un-cool, and proud of it.

The IMD is already into the 40% figures.....

But when I hear what is real music to me, I like the gear to be as clean as a
whistle.

Do I like wine which is muddy? pictures painted by modern idiots?
sculpture which makes you want to phone up a rubbish collector?
poetry which makes you puke? sound that grates on every sensibility?

I really don't like much of anything, unless whatever it is has had some serious

effort put into it...

So for what these tube amps SET. How can the added coloration be useful if
we listen to recorded music?

The people I know who use SET amps with no FB have put in the hours to
make sure it has very little discernible colourations or distortions.
An example is the guy with the Allessa Vaic 300B triode amps with Tannoy Gods
in 180 litre enclosures with 50mm thick sand filled panels.
Beautiful sound it is, and it sounds very *real*.
Not coloured.
He and I turned it up to very loud, and the peaks never went over a watt.

So what he listens to normally measures very well, because analogue systems
generally have declining
distortion measurements as output reduces to low levels.

And when it is not colouring the sound as the

high quality tube amps, why not use a transistor amp with so much more
performance and power per buck?

Unfortunately, many transistor amps *do* colour the sound, and do it so badly
it has sent many listeners in search of something else, anything else,
and ehen they hear a decent tube set, they pay for it.

I don't have to work hard at selling what I make.

One would think tubes would just fade away like wind up gramophone
record players, but they just won't.

And remember that there are many SS amps which cost a fortune more than
something with tubes.

Each unto his own, I say.

I have seen people turn up the level with a tube set so my ears hurt,
usually with attrociously bad rock and roll, and then they say, "gees, it don't
go very loud".
I can't help them.
I can only please some of the people some of the time.

If you want to reproduce what a grand piano played loudly
sounds like at 2 metres away, then you need some quite expensive
equipment, no matter whether tubed, transistored, horned, or whatever.
I have been on stage while someone pounded away on a large grand,
and its way louder than I like to listen.
At age 20 I didn't like being too close, but I didn't mind.
Now I am an old fart, I do mind,
but I like to sit 10 metres away from the orchestra, to take it all in.
Its what the composer intended that I want.

Patrick Turner.

--
ciao Ban
Bordighera, Italy

John Byrns wrote:

I take it that the error in the schematic, which you don't explicitly
identify was that the value of R3 was too large to prevent the driver
stage from oscillating?

R3A is the biasing resistor to the V4 on one side,
and R3B biases V5 on the other.

I had not allowed for the PFB effect at all.
But there is +0.66Dv at V2 anode, and +1.0Dv at V6 anode.
How could there be 0.083Dv at the junction of R1 and R2?

Its impossible, unless R3 is reduced from 330 k to around 6k.

I am not exactly sure what value, but after working a few numbers,
the amount of Dv at V2 anode should always be less than at V6 anode.

If there was no error correction, perhaps 7.6 x Dv would appear as the open loop
Dn at V6 anode, so if 0.66 Dv were applied to g1/V6, and the gain is -10,
then -6.6Dv is produced at V6 anode, leaving just 1.0Dv left as the closed loop Dn.
If the output stage was a pentode outpit stage with a gain of -20,
then I suggest the amount of Dv applied to g1/V6 would have to be around
0.33 Dv so that you'd still get less than 20 dB of error correction.
Going beyond 16 dB invites instability just like with NFB.

The V2 and V4 tubes act like a paraphase inverter, where
a fraction of the output from V2 is used to create a signal to feed V4,
using a resistance divider dominated bt R1 and R3.

I did originally notice the positive feedback
loop around the driver, but positive feedback loops are hardly unknown in
Hi-Fi amps, and I assumed that you had made sure that there was not enough
positive feedback to cause instability.

One could oerate the circuit without the PFB effect.
If there was another LTP which was set up separately to make a
relatively undistorted +/- 24vrms, and it wasn't used for driving an output stage,
then you could have the same R1/R2 divider set up so that there was no
signal voltage at the R junction, and approx only 0.09 Dv existed.
This could be taken to g1/V4 without any PFB effect.
The diff gain from one input of the V2/V4 diffpair is about 7, so
0.09Dv would appear as +0.63 vrms at V2 anode, which
is then applied to G1/V6 to correct the open loop Dn.

Perhaps its a more stable way to achieve it but an extra pair of triodes is needed.

The schematic as shown is far more complex than the simpler first
unbalanced circuit I proposed, which had a separate V1 to null the signal from the
OPT secondary.

That first circuit could be made to be balanced, but you'd need a secondary on the
OPT
which is accurately centre tapped, and more complex ways of
injecting the error correction signal.

I experimented with these circuits some time ago,
and I have drawn ppl's attention to them as a mental exercize,
and perhaps someone could get them to work better than NFB
if they picked up the ball and ran with it.

You say some people "cannot understand why anyone would ever bother to use
vacuum tubes at all."

There are ppl like this.

That is a separate issue from your "error
correction" circuit which should not be dependent on vacuum tube
circuitry, and should also work with transistor circuits if it is valid.

Of course.

A couple of people have chimed in about error correction being used quite
successfully
in SS designs.
I am not inclined to use it for SS, merely because I am mainly a tube specialist,
and the fact that I find NFB so easy to do with SS, because
the voltage amp in them can so easily have a gain of 20,000,
which is reduced to 20 with NFB, so that the difference between the
signal fed back and the input signal is only
1 mV, and thd is reduced from an open loop amount of say
3% at 200 watts to 0.003% quite routinely.

Perhaps there are other ways to do it all.

With an OPT, the vast amounts of NFB simply cannot be applied
in tube amps because the gain will be too high at F where too huge a phase shift
occurs due to leakage L and stray C effects.

I think your first schematic was a better presentation of your idea in its
pure form, and I will take it as the basis of a more generic circuit to
explain why I believe there is no difference at all between your so called
"error correction" circuit and the common negative feedback circuit. You
will have to imagine the block schematic in your mind.

I agree, the first circuit is very simple, and hopefully more effective.
I don't like the complexity of the balanced circuit.

The amplifier schematic titled "Error Correction, Typical UL Amp September
2004" can be broken down into two amplifier modules. The first is a power
amplifier module with differential inputs, and a voltage gain of about 10
times. For simplicity I will assume the differential stage is "blameless"
and all distortion is generated in the output stage. To further simplify
the discussion I will use my own numbers, which only approximate your
actual amplifier. Your amplifier starts by feeding the input signal, say
1 vrms into the positive input and grounding the negative input of the
differential power amplifier. This results in a 10 volt output from the
power amplifier into the 6 Ohm load. To correct the output errors you
then feed the output back to the negative input, but in addition to the
distortion components the output also includes the desired output signal.

Nope, you still have not read the schematic of 'Error Correction, Typical UL Amp',
quite right.

Sure there is a regular power amp stage with a normal LTP diff pair driver.
3.4 vrms is needed between g1/V2 and g1/V3 to produce 14.4v into 6 ohms at the
output.
Overall open loop gain of the LTPO and output stage is 14.4/3.4 = 4.23.

Let's stick with my figures, and voltages, to make what I say gel
with the schematic, since the other 4 ppl in the world listening in don't want it
explained
in a more confusing way than I have so far managed.

the 22k R from the output is not taken to the input voltage point.

Its taken to the anode of V1, a separate class A SET specially se up
to have an output resistance of 47k.
as the anode of V1 tries to swing -ve, because of +1.7v
g1 input voltage, the speaker output swings +14.4v and tries to
raise the voltage at V1 anode, and the result is that we only get -1.7v swing -ve
at V1 anode.
If the 22k was reduced a little more, we would totally prevent V1 from
producing any signal output at all!

But I settled on 22k, and allowed -1.7 vrms to appear at V1 anode,
so that it could be applied to the g1/V3, thus giving +/- 1.7 to be applied to the
two grids of the V2/V4 LTP, so only 1.7vrms is needed at the input.

Do you see how the 1.7 v input is applied to both the g1/1, and g1/V2?

Any Dv appearing at the speaker output is divided by the
R1 / R2 divider, these being Ro of V1, and 22k,
or 47k : 22k, so 0.68DV appears at the anode of V1.

This is applied only to g1/V3, and amplified by the open loop
gain of -4.2 working from that input at V3, so a signal of -2.86 Dv tries to
appear at the output, but it cannot, because we already have measured 1.0Dv,
this being the Dn with all connected and working as shown.
But what we could say is that +3.86 Dv would have appeared at the output
without correction, but with it we have a signal of -2.86v which subtracts from the
open loop Dn to leave 1.0Dv.

Unfortunately feeding back the amplified input along with the errors
greatly reduces the gain of the amplifier, from 10X to about 0.5X.

No it doesn't.

To
restore the gain you then feed the input signal through a second auxiliary
voltage amplifier with a gain of -10X and add this negative amplified
version of the input signal to the signal from the output before feeding
the combination into the negative input of the differential input main
power amplifier in order to cancel the unwanted portion of the output
feedback signal which represents the amplified input. This action
restores the apparent gain of the amplifier to 10X.

The auxilliary is a phase inverter.
I deliberately set it up to reduce the otherwise too large input voltage of 3.4v.
Its gain is quite low, but its thd and bandwidth is very wide.

Ignoring the details of how we might actually implement a summer for the
output signal and the amplified and inverted version of the input signal
to apply to the negative input of the main amplifier, we have the
following equation for the amplifier gain, where Vin is the input to the
auxiliary voltage amplifier:

(1) Vout = 10 x ( Vin - ( Vout + ( -10 x Vin)))

Which reduces to:

(2) Vout = 10 x Vin

Equation #1 can also be rewritten in this form:

(3) Vout = 10 x ( - ( - Vin + Vout + ( -10 x Vin)))

which is equal to:

(4) Vout = 10 x ( - ( + Vout + ( -11 x Vin)))

Which again reduces to"

(5) Vout = 10 x Vin

Which is identical to equation #2

This is mathematically equivalent to disconnecting the input signal from
the positive input of the main amplifier, and feeding an additional
inverted version of the input signal into the summer feeding the positive
input.

Others may like to comment, but I am completely confused
by your math, since i don't think you are talking about my schematic any more,
and unless you post a schematic of exactly what you are talking about,
I can't comment.

This is equivalent to a common negative feedback amplifier where the
output and input signals are summed at the input of an inverting
amplifier. In this example, as well as your first example schematic, the
main amplifier has less than unity gain with feedback applied,
necessitating the voltage amplifier preceding the main amplifier.

Does this analysis capture the essence of your original "error correction"
circuit if not the exact numbers? If it does, what do you claim is the
difference between your "error correction" circuit, and a simple negative
feedback amplifier where the positive input of the main amplifier is
grounded and the gain of the auxiliary voltage amplifier is increased
slightly to compensate?

I have explained enough, and I will not be battered into explaning yet again
the difference between the circuit I have proposed and
one using normal NFB, which always has two signals applied
to a device, or a pair of them acting as a differential amplifier.

The V1 I have proposed could well be a pentode with a CCS, and then R1
would be a megohm, and thus nearly all the Dv appearing at the output
would appear at V1 anode, and the amount of correction could be increased
to give a dn reduction factor of 1/5.6 instead of the present 1/3.86.
If the LTP used 12AT7 instead of 6SN7, its gain would about double,
and even greater amounts of Dn reduction would be possible.
The overall gain can be trimmed by altering the value of R2, or making the 22k
larger
or smaller.
In my scheme, V1 is not included in a correction network loop.
Only V2,3,&5.
I doubt whether the added Dn from V1 is any worse than that
produced by a typical input triode which is included in a typical NFB loop.

Regards,

John Byrns

Patrick Turner.

Carl WA1KPD wrote:

Why is this continuing on alt.binaries.pictures ?

Because people like yourself continue to cross post your replies.

Have you anything constructive to say about issues raised?

Patrick Turner.

Patrick Turner wrote in
:

Carl WA1KPD wrote:

Why is this continuing on alt.binaries.pictures ?

Because people like yourself continue to cross post your replies.

Or rather - because the OP (whoever that might be) was to lazy (or
ignorant) to set an XFUT in the first post !

--
Regards,
Soeren

* If it puzzles you dear... Reverse engineer *
New forum: URL:http://www.ElektronikTeknolog.dk/cgi-bin/SPEED/

Soeren wrote:

Patrick Turner wrote in
:

Carl WA1KPD wrote:

Why is this continuing on alt.binaries.pictures ?

Because people like yourself continue to cross post your replies.

Or rather - because the OP (whoever that might be) was to lazy (or
ignorant) to set an XFUT in the first post !

I only have an old Nutscrape mailer program.

"XFUT!!!" is the noise a crummy peice of SS gear makes when it dies.

Patrick Turner.

--
Regards,
Soeren

* If it puzzles you dear... Reverse engineer *
New forum: URL:http://www.ElektronikTeknolog.dk/cgi-bin/SPEED/

Patrick Turner said:

"XFUT!!!" is the noise a crummy peice of SS gear makes when it dies.

Crosspost follow up to.

--
Sander deWaal
"SOA of a KT88? Sufficient."

In article , Patrick Turner
wrote:

If the output stage was a pentode outpit stage with a gain of -20,
then I suggest the amount of Dv applied to g1/V6 would have to be around
0.33 Dv so that you'd still get less than 20 dB of error correction.
Going beyond 16 dB invites instability just like with NFB.

That's because it is NFB! There is no fundamental difference between what
you are calling "error correction" and a more straight forward
implementation of NFB. You have yet to address the question of what the
advantage of your "error correction" circuit is over a more straight
forward implementation of NFB.

I experimented with these circuits some time ago,
and I have drawn ppl's attention to them as a mental exercize,
and perhaps someone could get them to work better than NFB
if they picked up the ball and ran with it.

How could they be made to work any better than NFB, when they are NFB circuits?

With an OPT, the vast amounts of NFB simply cannot be applied
in tube amps because the gain will be too high at F where too huge a phase
shift occurs due to leakage L and stray C effects.

How is your circuit any different in this regard, assuming the NFB is
taken from the same point(s) as in your circuit?

Nope, you still have not read the schematic of 'Error Correction, Typical UL
Amp', quite right.

You are quite right, I failed to notice that there was a "-1.7AV" signal
at the grid of V3. I wrongly assumed that you were nulling the desired
signal at the grid of V3, which lead me to assume that the main amplifier
had twice as much gain, 8.47, as the gain of 4.235 the stage actually
has. For my example I rounded gain of 8.47 to a nice round gain of 10.
If I have calculated correctly now the NFB in your amplifier circuit
reduces the gain of the main amplifier from 4.235 to 1.09 for a gain
reduction by NFB of about 11.8 dB.

Its taken to the anode of V1, a separate class A SET specially se up
to have an output resistance of 47k.

Yes, I read your schematic as indicating a 47k source impedance for V1,
but I didn't want to assume that value for a resistive summer in my
example because I was not sure I was interpreting your notes on the
schematic correctly.

My bottom line question remains, why would one want to build your circuit
with its extra complexity instead of a more conventional circuit with the
same 11.8 dB of NFB? There appear to be few, if any, advantages to your
circuit other than a more convoluted circuit topology?

Regards,

John Byrns

Surf my web pages at, http://users.rcn.com/jbyrns/

John Byrns wrote:

In article , Patrick Turner
wrote:

If the output stage was a pentode outpit stage with a gain of -20,
then I suggest the amount of Dv applied to g1/V6 would have to be around
0.33 Dv so that you'd still get less than 20 dB of error correction.
Going beyond 16 dB invites instability just like with NFB.

That's because it is NFB!

But no signal voltage is fed back.
If no signal voltage is fed back, it isn't negative feedback.

Please try to see the distinction.
I know its hard to get you to see such things....

There is no fundamental difference between what
you are calling "error correction" and a more straight forward
implementation of NFB. You have yet to address the question of what the
advantage of your "error correction" circuit is over a more straight
forward implementation of NFB.

I raise no question of advantages.
Take it or leave it.
What you see is what you get.

I experimented with these circuits some time ago,
and I have drawn ppl's attention to them as a mental exercize,
and perhaps someone could get them to work better than NFB
if they picked up the ball and ran with it.

How could they be made to work any better than NFB, when they are NFB circuits?

Again you fail to understand things.

With an OPT, the vast amounts of NFB simply cannot be applied
in tube amps because the gain will be too high at F where too huge a phase
shift occurs due to leakage L and stray C effects.

How is your circuit any different in this regard, assuming the NFB is
taken from the same point(s) as in your circuit?

A NFB amp is arranged very differently.

Nope, you still have not read the schematic of 'Error Correction, Typical UL
Amp', quite right.

You are quite right, I failed to notice that there was a "-1.7AV" signal
at the grid of V3. I wrongly assumed that you were nulling the desired
signal at the grid of V3, which lead me to assume that the main amplifier
had twice as much gain, 8.47, as the gain of 4.235 the stage actually
has. For my example I rounded gain of 8.47 to a nice round gain of 10.
If I have calculated correctly now the NFB in your amplifier circuit
reduces the gain of the main amplifier from 4.235 to 1.09 for a gain
reduction by NFB of about 11.8 dB.

That is an approximately correct amount of error correction,
as distinct from NFB application.

Its taken to the anode of V1, a separate class A SET specially se up
to have an output resistance of 47k.

Yes, I read your schematic as indicating a 47k source impedance for V1,
but I didn't want to assume that value for a resistive summer in my
example because I was not sure I was interpreting your notes on the
schematic correctly.

My bottom line question remains, why would one want to build your circuit
with its extra complexity instead of a more conventional circuit with the
same 11.8 dB of NFB? There appear to be few, if any, advantages to your
circuit other than a more convoluted circuit topology?

Variety is the spice of life John.

Its a dirty old duck that splashes around in the same old water hole.

If ppl can understand my logic in the schemas, their understanding
will generally be broadened imho, and the
group is better off for the exercize.

If someone has some schematic of some prospectively interesting circuit,
then post it, and lets discuss it.

Patrick Turner.

Regards,

John Byrns

Surf my web pages at, http://users.rcn.com/jbyrns/

"Patrick Turner" wrote in message
...
But no signal voltage is fed back.
If no signal voltage is fed back, it isn't negative feedback.

Please try to see the distinction.

Ok, so it's a difference of terminology. But 10+1 = 11 - 10 = 1 * -1 = -1 +
11 = 10, and 10 + 1 = 11 - 1 = 10. Just how you want to go about it.
Fundamentally the circuits behave the same.

How could they be made to work any better than NFB, when they are NFB
circuits?

Again you fail to understand things.

Understand what, a difference in words? Geez...you know what he means!

If someone has some schematic of some prospectively interesting circuit,
then post it, and lets discuss it.

Ok. (Refer to ABPR in reply to this post)

Tim

--
"I've got more trophies than Wayne Gretsky and the Pope combined!"
- Homer Simpson
Website @ http://webpages.charter.net/dawill/tmoranwms

In article , Patrick Turner
wrote:

John Byrns wrote:

In article , Patrick Turner
wrote:

If the output stage was a pentode outpit stage with a gain of -20,
then I suggest the amount of Dv applied to g1/V6 would have to be around
0.33 Dv so that you'd still get less than 20 dB of error correction.
Going beyond 16 dB invites instability just like with NFB.

That's because it is NFB!

But no signal voltage is fed back.
If no signal voltage is fed back, it isn't negative feedback.

Please try to see the distinction.
I know its hard to get you to see such things....

I might justifiably make the same comment, but this idea that "no signal
voltage is fed back" seems to be at the crux of the matter. For the sake
of argument I will accept your assertion that if "signal voltage is fed
back" then we have NFB, and the implication that if no "signal voltage is
fed back" then there is no NFB. What I don't understand is the basis on
which you make the claim that in your "error correction" circuit no
"signal voltage is fed back", while in a conventional NFB amplifier
"signal voltage is fed back"? I don't see any difference between your
amplifier and a conventional NFB amplifier that regard, in both cases
"signal voltage" appears to be fed back. You should be able to explain
why you say no "signal voltage is fed back" in your "error correction"
design without getting into a detailed description, what is the basic idea
that separates your "error correction" circuit from a conventional NFB
amplifier, and causes no "signal voltage" to be fed back? I think if you
really attempt to understand what is going on in the circuit you will see
that your "error correction" circuit is no different qualitatively in
terms of NFB than say an ST-70. It is simply a design, like the "cross
coupled" phase inverter, for people that like to build complex circuits,
to each his own I guess.

With an OPT, the vast amounts of NFB simply cannot be applied
in tube amps because the gain will be too high at F where too huge a phase
shift occurs due to leakage L and stray C effects.

How is your circuit any different in this regard, assuming the NFB is
taken from the same point(s) as in your circuit?

A NFB amp is arranged very differently.

How so, you have not yet explained how your "error correction" circuit is
arranged differently than a conventional NFB amplifier in any fundemental
way? In both cases the output is feedback to the input where it is
subtracted from the input with appropriate scaling.

Regards,

John Byrns

Surf my web pages at, http://users.rcn.com/jbyrns/

On Sunday 12 September 2004 12:31 am, Patrick Turner did deign to grace us
with the following:

Soeren wrote:

Patrick Turner wrote in
:

Carl WA1KPD wrote:

Why is this continuing on alt.binaries.pictures ?

Because people like yourself continue to cross post your replies.

Or rather - because the OP (whoever that might be) was to lazy (or
ignorant) to set an XFUT in the first post !

I only have an old Nutscrape mailer program.

"XFUT!!!" is the noise a crummy peice of SS gear makes when it dies.

No it's not. I work in a shop where they use a lot of Stainless Steel,
and when it dies, it doesn't make that kind of sound at all. ;-)

Tim Williams wrote:

"Patrick Turner" wrote in message
...
But no signal voltage is fed back.
If no signal voltage is fed back, it isn't negative feedback.

Please try to see the distinction.

Ok, so it's a difference of terminology. But 10+1 = 11 - 10 = 1 * -1 = -1 +
11 = 10, and 10 + 1 = 11 - 1 = 10. Just how you want to go about it.
Fundamentally the circuits behave the same.

How could they be made to work any better than NFB, when they are NFB
circuits?

Again you fail to understand things.

Understand what, a difference in words? Geez...you know what he means!

Don't speak in riddles Tim. Geez nothing. And I don't know a lot of what anyone
means unless
they spend the time to spell it out clearly, preferably using a schematic.

I never assume I know anything other than what I spell out clearly.

He is trying to say that my error correction circuit ideas are nothing different
to
negative feedback

I have tried to *avoid* feeding back any *signal* to an available input of an
amp.

I am feeding back a *distortion voltage* which then gets amplified in a way to
oppose its own production by the open loop amp.

NFB feeds back a signal which contains the distortion voltage.

If someone has some schematic of some prospectively interesting circuit,
then post it, and lets discuss it.

Ok. (Refer to ABPR in reply to this post)

I will take a look,
but I don't always get what is posted at abse or abpr.

What is wanted of course is that the rule banning binaries posts
at rec.audio.tubes be dropped immediately.

How the hell can tube audio or radio or other design be well discussed without
us being able to post schematics reliably, and on the very group where
discussions take place?

What is the procedure for getting the system changed?

Patrick Turner.

Tim

--
"I've got more trophies than Wayne Gretsky and the Pope combined!"
- Homer Simpson
Website @ http://webpages.charter.net/dawill/tmoranwms

John Byrns wrote:

In article , Patrick Turner
wrote:

John Byrns wrote:

In article , Patrick Turner
wrote:

If the output stage was a pentode outpit stage with a gain of -20,
then I suggest the amount of Dv applied to g1/V6 would have to be around
0.33 Dv so that you'd still get less than 20 dB of error correction.
Going beyond 16 dB invites instability just like with NFB.

That's because it is NFB!

But no signal voltage is fed back.
If no signal voltage is fed back, it isn't negative feedback.

Please try to see the distinction.
I know its hard to get you to see such things....

I might justifiably make the same comment, but this idea that "no signal
voltage is fed back" seems to be at the crux of the matter. For the sake
of argument I will accept your assertion that if "signal voltage is fed
back" then we have NFB, and the implication that if no "signal voltage is
fed back" then there is no NFB. What I don't understand is the basis on
which you make the claim that in your "error correction" circuit no
"signal voltage is fed back", while in a conventional NFB amplifier
"signal voltage is fed back"? I don't see any difference between your
amplifier and a conventional NFB amplifier that regard, in both cases
"signal voltage" appears to be fed back.

The distinction in words is necessary to preserve the meaning and a menatl picture
of the
use of circuitry covered by the terms.
We all know what NFB is; a fraction of the output signal is fed back
to be in series with the input, or in shunt with it.

Buried within the NFB signal is a distortion voltage which gets amplified to
prevent its production.

In error correction, the distortion voltage gets isolated, and fed back, or fed
forward,
to act in the same way as the distprtion voltage buried in the NFB signal.

You should be able to explain
why you say no "signal voltage is fed back" in your "error correction"
design without getting into a detailed description, what is the basic idea
that separates your "error correction" circuit from a conventional NFB
amplifier, and causes no "signal voltage" to be fed back? I think if you
really attempt to understand what is going on in the circuit you will see
that your "error correction" circuit is no different qualitatively in
terms of NFB than say an ST-70. It is simply a design, like the "cross
coupled" phase inverter, for people that like to build complex circuits,
to each his own I guess.

See above.

With an OPT, the vast amounts of NFB simply cannot be applied
in tube amps because the gain will be too high at F where too huge a phase
shift occurs due to leakage L and stray C effects.

How is your circuit any different in this regard, assuming the NFB is
taken from the same point(s) as in your circuit?

A NFB amp is arranged very differently.

How so, you have not yet explained how your "error correction" circuit is
arranged differently than a conventional NFB amplifier in any fundemental
way? In both cases the output is feedback to the input where it is
subtracted from the input with appropriate scaling.

I am sure I have explained enough.

Patrick Turner.

Regards,

John Byrns

Surf my web pages at, http://users.rcn.com/jbyrns/

Patrick Turner wrote:

I am feeding back a *distortion voltage* which then gets amplified in a way to
oppose its own production by the open loop amp.

NFB feeds back a signal which contains the distortion voltage.

So its not NFB in the most strict sense of words, but its basically the
same thing. So whats the advantage? The distortion voltage *is* the
distortion leaving a clean signal?

What is wanted of course is that the rule banning binaries posts
at rec.audio.tubes be dropped immediately.

How the hell can tube audio or radio or other design be well discussed without
us being able to post schematics reliably, and on the very group where
discussions take place?

What is the procedure for getting the system changed?

http://www.usenet.com/articles/history_of_usenet.htm

Talk to the guys in alt.config

The amount of usenet traffic just for text is mind boggling now. Adding
images is not a good idea. Binaries are not that difficult. If you can't
handle usenet then sign up for some free webspace and paste a link.

Adam

In article , Patrick Turner
wrote:

The distinction in words is necessary to preserve the meaning and a
menatl picture
of the
use of circuitry covered by the terms.

The distinction in words is most likely the idea of the marketing department.

We all know what NFB is; a fraction of the output signal is fed back
to be in series with the input, or in shunt with it.

Buried within the NFB signal is a distortion voltage which gets amplified to
prevent its production.

In error correction, the distortion voltage gets isolated, and fed back,
or fed
forward,
to act in the same way as the distprtion voltage buried in the NFB signal.

The same thing can equally well be said about a NFB amplifier. In a NFB
amplifier the "distortion voltage gets isolated" by exactly the same
mechanism as in your "error correction" amplifier circuit, it is nothing
but the same old circuit dressed up with a fancy new name.

A simple question, how would you classify your original "error correction"
amplifier if the grid of V2 were connected to ground rather than to the
input? Would it remain and "error correction" circuit, or would it become
a NFB circuit, or something else entirely?

Regrads,

John Byrns

Surf my web pages at, http://users.rcn.com/jbyrns/

I read in alt.binaries.schematics.electronic that Patrick Turner
wrote (in
u) about 'Who needs NFB when there is error correction?', on Mon, 13
Sep 2004:

I am sure I have explained enough.

I suggest you re-post the reference to Malcolm Hawksford's AES paper. At
present, you are asking people to believe you, using YOUR arguments,
which are certainly somewhat unconventional. They may find Hawksford
more convincing (if less accessible).
--
Regards, John Woodgate, OOO - Own Opinions Only.
The good news is that nothing is compulsory.
The bad news is that everything is prohibited.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk

I read in alt.binaries.schematics.electronic that John Byrns
wrote (in
ract.com) about 'Who needs NFB when there is error correction?', on
Mon, 13 Sep 2004:

The same thing can equally well be said about a NFB amplifier. In a NFB
amplifier the "distortion voltage gets isolated" by exactly the same
mechanism as in your "error correction" amplifier circuit, it is nothing
but the same old circuit dressed up with a fancy new name.

No; error feedback involves two adders (with one negated input). First,
the input signal is subtracted from an attenuated copy of the output
signal, leaving a difference or error signal. THIS signal is then
subtracted from another copy of the input signal, giving a modified
input signal with an inverted error component designed to oppose the
original error. Of course, the opposition cannot be complete, because
there would then be no error signal generated by the first adder.

I won't get drawn into the question of the merits of the technique. It
has been around for quite a long time, but Hawksford's paper is one of
(possibly the) most recent peer-reviewed treatment.
--
Regards, John Woodgate, OOO - Own Opinions Only.
The good news is that nothing is compulsory.
The bad news is that everything is prohibited.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk

Adam Stouffer wrote:

Patrick Turner wrote:

I am feeding back a *distortion voltage* which then gets amplified in a way to
oppose its own production by the open loop amp.

NFB feeds back a signal which contains the distortion voltage.

So its not NFB in the most strict sense of words, but its basically the
same thing. So whats the advantage? The distortion voltage *is* the
distortion leaving a clean signal?

The advantage if any, over NFB is that less gain stages are needed
in a given amp within the "error correction loop"
Therefore better stability is perhaps possible.
But I have so far not experimented enough to establish a firm truth on the matter.
Like many techniques in electronics, there simply isn't general agreement
amoung all engineers.

Try it, and find your own truth. I don't have the time to find it for you,
and everyone else. My grandpa said if yer don't look yer won't know.

The method of error correction isn't NFB, or "inverse feedback" as it was called
once,
since with NFB the signal fed back in a series voltage NFB amp
is the same phase as the input.
Nothing inverse about it. Perverse maybe.
BUT, if a positive signal is applied to a cathode of V1 of a Williamson,
then its *as if* a negative signal of the same amplitude is applied to the grid.
So the grid signal voltage which is always the same phase as the fed back voltage
has to be greater than the fed back voltage.

Always analyse analog circuits by working out and writing down the
intantaneous VRMS of signals with their positive or negative going trait,
using a schematic to make your notes.

Further simplification is possible by drawing each basic amp stage as a triangle
with two inputs, labeled + and - and assuming a +ve going output.

Quite a few modern text books on basic opamp workings have
the modelling spelled out clearly enough. The ideas are transferable to tubes.

Its different with shunt FB, where the output voltage is always the
oppositely phased to the input signal voltage.

The fraction fed back "in shunt" is R1/ (R1 + R2 ),
where R1 is the R between input and g1, and R2 is the R between output and g1.

What is wanted of course is that the rule banning binaries posts
at rec.audio.tubes be dropped immediately.

How the hell can tube audio or radio or other design be well discussed without
us being able to post schematics reliably, and on the very group where
discussions take place?

What is the procedure for getting the system changed?

http://www.usenet.com/articles/history_of_usenet.htm

Talk to the guys in alt.config

The amount of usenet traffic just for text is mind boggling now.

We live in an ever expanding world with ever expanding capability.
Well, it should be ever expanding capability.
I don't care how much text is transfered.
We want more!

Adding
images is not a good idea.

I disagree; adding binaries would be just great FOR US.
For YOU, and ME, and everyone here.

Binaries are not that difficult. If you can't
handle usenet then sign up for some free webspace and paste a link.

I refuse to agree we just let the status quo stay put for the forseable future.
The guys in charge should realise change to suit demand is the key to a healthy
usenet.
In fact a few binaries we post on our own group will save us posting elsewhere.
And we wouldn't attract the attention of folks who think tubes are ****e,
or that cross posting is rude, etc.

And it isn't as if the binaries stay put after posting; most dissappear after
48 hrs, so I cannot sympathise with folks wanting to complain about
continued restrictions on our communications.
Its a modern age, so let's see more capability.

Patrick Turner.

Adam

Thread Tools
Show Printable Version
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

Menu

About Us