I can't speak for other heads, but the inside of mine finds itself
no further forward. It would help if you were to say what you are
using as a source in your listening test.

Using the battery *instead of* the grid leak gives you a 1ohm shunt
across the lower leg of the attenuator, which is not a good test of
the principle. Perhaps you mean *in series with* the grid leak?

All methods of grid bias in the absence of DC blocking will result
in biasing the source, obviously. Here again it is not clear whether
that in itself may have degraded the sound, so again the test is
inconclusive.

A battery at the foot of the attenuator would not discharge itself
in the absence of a grid leak, and the circuit would be just as
simple as the one you have chosen. The battery could also be charged
through the signal input with the amp off, so recharging could be a
switched function of a dedicated preamp. I dare say there would be
objections to this option, and I wonder what they would be.

What I would really like to know is would it be better, leaving
aside the issue of DC blocking and recharging, to bias via the grid
or the cathode? And would a series or a shunt connection to the grid
be more appropriate?

Do batteries have the same impedance in both directions? Are they
more noisy when discharging, or charging, or in equilibrium? What
are the mechanisms of distortion and will it be less in the
high-impedance low current input or the low impedance high current
cathode circuit. Will it be less by a factor greater than the
relative gain?

I would be surprised if these issues had not already been resolved,
considering you say that the cathode connection is so common. It
seems unlikely that the obvious alternatives were not explored, or
that the merely practical charging and DC blocking issues won the
day with the "ultra fi" fetishists.

And don't forget all of your famous guru friends, who I am sure
would be happy to advise you. I would be interested to hear their
views.

cheers, Ian



"Andre Jute" wrote in message
oups.com...
Uh-uh, Bruce. What you're talking about is the battery in the
place of
the grid leak resistor, between grid and ground. That's a low
impedance
path to ground. I've tried that one with a DACT, a high precision
Swiss
medical switch with SMD resistors and ultrashort paths, with the
tube
in position to amplify the clicks if any, and the amp working into
ultra-sensitive horns, and I heard no clicks.

I republished the T39 KISS Ultrafi circuit with the batteries in
the
third possible position, in the cathode circuit, simply because I
could
hear no difference with the battery in the grid leak position, and
the
cathode circuit position has the very great advantage of being
more
failsafe than either of the other two positions, by virtue of the
trickle charge the tubes can enjoy there.

One final note on the *discussion* circuit I published (battery
between
wiper of attenuator and grid of signal tubes) : A "professional"
implementation would use at least a resistor and a bypass cap in
addition to the battery, as has been pointed out by you and
several
others. But I'm not interested in defending a circuit before
professional production engineers. I'm interested in simplifying
the
circuit to what sounds best, regardless of expense or
"professional
practice", and in this case, because I am publishing the circuit
for
amateurs, in making it reasonably longlasting (there is a
companion
circuit, called the T44 "Populaire", with autobias, recommended
for the
newest aspirants, precisely because it is a largely self-adjusting
circuit). Thus the return to the battery in the cathode...

Thanks to all who contributed valuable insights (even if we were
all
talking about different circuits!). I'm removing the discussion
circuit
because it really doesn't add much to our wisdom; it may be a
viable
alternative (in that it works) but it is a decidedly less
desirable one
than the battery in the cathode circuit. That's the point of
discussion
with one's peers, to choose the best course of action from among
possibilities.

Andre Jute

In article .com,
"Andre Jute" wrote:

Uh-uh, Bruce. What you're talking about is the battery in the place of
the grid leak resistor, between grid and ground. That's a low impedance
path to ground.

Got it. Low impedance, quite! I was looking at the T39 mk VI schematic,
since the "discussion" schematic was refusing to appear.

I've tried that one with a DACT, a high precision Swiss
medical switch with SMD resistors and ultrashort paths, with the tube
in position to amplify the clicks if any, and the amp working into
ultra-sensitive horns, and I heard no clicks.

Well, good on ya, mate, but I'll stand by my observations.

I republished the T39 KISS Ultrafi circuit with the batteries in the
third possible position, in the cathode circuit, simply because I could
hear no difference with the battery in the grid leak position, and the
cathode circuit position has the very great advantage of being more
failsafe than either of the other two positions, by virtue of the
trickle charge the tubes can enjoy there.

I'm glad that the discussion here helped you decide to change that
schematic, since it was, shall we say, a somewhat unhealthy arrangement
of parts. A newbe building an amp based on the T39 circuit might be a
little perturbed to find that his or her batteries would be shorted to
ground whenever the attenuator was turned off. An input attenuator
having a wonky curve because it was loaded down by a 1 ohm load would be
another "surprise".

One final note on the *discussion* circuit I published (battery between
wiper of attenuator and grid of signal tubes) : A "professional"
implementation would use at least a resistor and a bypass cap in
addition to the battery, as has been pointed out by you and several
others.

Just for the record, again, I was making reference to the T39 mk VI.
I've made no comment on the "discussion" circuit.

A cap connected between the wiper and grid, with the batteries in series
with a 100k-500k resistor connected in the grid leak position, is what I
was proposing would work well. No charging, however. Always a compromise.

But I'm not interested in defending a circuit before
professional production engineers.

Which I am. Sorry, can't (and shan't) take the hat off...

I'm interested in simplifying the
circuit to what sounds best, regardless of expense or "professional
practice", and in this case, because I am publishing the circuit for
amateurs, in making it reasonably longlasting (there is a companion
circuit, called the T44 "Populaire", with autobias, recommended for the
newest aspirants, precisely because it is a largely self-adjusting
circuit). Thus the return to the battery in the cathode...

"Professional practice" sounds vaguely derogatory. You rebel, you.

Thanks to all who contributed valuable insights (even if we were all
talking about different circuits!). I'm removing the discussion circuit
because it really doesn't add much to our wisdom; it may be a viable
alternative (in that it works) but it is a decidedly less desirable one
than the battery in the cathode circuit. That's the point of discussion
with one's peers, to choose the best course of action from among
possibilities.

Best of luck.

-bruce seifried
eclair engineering

Seems to me, my dear fellow, that your letter illustrates either that a
Marxist can make a meal out of very little or that you too can be an
"ultrafi fetishist". Those self-charging advantages of the battery in
the cathode circuit you are so keen to dismiss are huge benefits for no
cost. Your better ultrafidelista is not an audiophool. He is imbued
with the conviction that simplicity means better sound. The battery in
the cathode is the simplest safe implementation, requiring no other
components.

Once more, the three circuits considered in these discussions:

1. Battery in cathode circuit. Grid leak is taken care of by leg of
attenuator to ground. This is the common version. The entire circuit is
attenuator, tube, battery and a couple of pieces of wire and maybe a
grid stopper resistor.

2. Battery shunted in grid leak position between grid and ground,
parallel to leg of attenuator to ground.

3. Battery seriesed in line between wiper of attenuator and grid.

Your idea of a battery in series below the attenuator would be a fourth
version and has the disavantage that it would vary the bias with volume
setting. How the battery could be charged by the signal after the amp
is switched off, as you claim, beats me. Maybe you know some magic that
I don't, which doesn't quite seem likely.

My source in all these experiments was a Quad 67 CD player, which has a
capacitively coupled output. I don't have any sources without a cap on
the output. Very few audiophiles can buy such a thing, and no competent
DIYer would build such a thing.

It is true that I already know what the ultrafidelista think. Most of
them think what I think and change their when I change mine; it's like
a ballet. But I was wondering if there is something I overlooked about
battery bias that might be worth some margin of extra silence. Seems
not.

On the subject of noise, you're on a hiding to nothing. A battery is a
more silent component than anything else including a straight piece of
wire. What I was reaching for was a more, even a marginally more,
silent implementation of a battery. The truth is that I couldn't hear
that the battery in the grid circuit was superior to the battery in the
cathode circuit, so therefore the more convenient and conservatively
safer cathode circuit it is.

Of course, by the nature of tubes, where there is very little that is
not already known, most experiments do not lead to improvements. On the
other hand, it became to clear during the course of this discussion
that the vast majority of RATs don't know about battery bias and that
only a few have ever tried it. It was definitely a worthwhile
discussion for me, and I hope it has given those who have not tried
battery bias (in the cathode, gentlemen, in the cathode) something to
do this winter.

HTH.

Andre Jute
Visit Jute on Amps at http://members.lycos.co.uk/fiultra/
"an unbelievably comprehensive web site" -- Hi-Fi News & Record Review



Ian Iveson wrote:
I can't speak for other heads, but the inside of mine finds itself
no further forward. It would help if you were to say what you are
using as a source in your listening test.

Using the battery *instead of* the grid leak gives you a 1ohm shunt
across the lower leg of the attenuator, which is not a good test of
the principle. Perhaps you mean *in series with* the grid leak?

All methods of grid bias in the absence of DC blocking will result
in biasing the source, obviously. Here again it is not clear whether
that in itself may have degraded the sound, so again the test is
inconclusive.

A battery at the foot of the attenuator would not discharge itself
in the absence of a grid leak, and the circuit would be just as
simple as the one you have chosen. The battery could also be charged
through the signal input with the amp off, so recharging could be a
switched function of a dedicated preamp. I dare say there would be
objections to this option, and I wonder what they would be.

What I would really like to know is would it be better, leaving
aside the issue of DC blocking and recharging, to bias via the grid
or the cathode? And would a series or a shunt connection to the grid
be more appropriate?

Do batteries have the same impedance in both directions? Are they
more noisy when discharging, or charging, or in equilibrium? What
are the mechanisms of distortion and will it be less in the
high-impedance low current input or the low impedance high current
cathode circuit. Will it be less by a factor greater than the
relative gain?

I would be surprised if these issues had not already been resolved,
considering you say that the cathode connection is so common. It
seems unlikely that the obvious alternatives were not explored, or
that the merely practical charging and DC blocking issues won the
day with the "ultra fi" fetishists.

And don't forget all of your famous guru friends, who I am sure
would be happy to advise you. I would be interested to hear their
views.

cheers, Ian



"Andre Jute" wrote in message
oups.com...
Uh-uh, Bruce. What you're talking about is the battery in the
place of
the grid leak resistor, between grid and ground. That's a low
impedance
path to ground. I've tried that one with a DACT, a high precision
Swiss
medical switch with SMD resistors and ultrashort paths, with the
tube
in position to amplify the clicks if any, and the amp working into
ultra-sensitive horns, and I heard no clicks.

I republished the T39 KISS Ultrafi circuit with the batteries in
the
third possible position, in the cathode circuit, simply because I
could
hear no difference with the battery in the grid leak position, and
the
cathode circuit position has the very great advantage of being
more
failsafe than either of the other two positions, by virtue of the
trickle charge the tubes can enjoy there.

One final note on the *discussion* circuit I published (battery
between
wiper of attenuator and grid of signal tubes) : A "professional"
implementation would use at least a resistor and a bypass cap in
addition to the battery, as has been pointed out by you and
several
others. But I'm not interested in defending a circuit before
professional production engineers. I'm interested in simplifying
the
circuit to what sounds best, regardless of expense or
"professional
practice", and in this case, because I am publishing the circuit
for
amateurs, in making it reasonably longlasting (there is a
companion
circuit, called the T44 "Populaire", with autobias, recommended
for the
newest aspirants, precisely because it is a largely self-adjusting
circuit). Thus the return to the battery in the cathode...

Thanks to all who contributed valuable insights (even if we were
all
talking about different circuits!). I'm removing the discussion
circuit
because it really doesn't add much to our wisdom; it may be a
viable
alternative (in that it works) but it is a decidedly less
desirable one
than the battery in the cathode circuit. That's the point of
discussion
with one's peers, to choose the best course of action from among
possibilities.

Andre Jute

Andre Jute schreef:

Seems to me, my dear fellow, that your letter illustrates either that a
Marxist can make a meal out of very little or that you too can be an
"ultrafi fetishist". Those self-charging advantages of the battery in
the cathode circuit you are so keen to dismiss are huge benefits for no
cost. Your better ultrafidelista is not an audiophool. He is imbued
with the conviction that simplicity means better sound. The battery in
the cathode is the simplest safe implementation, requiring no other
components.

Once more, the three circuits considered in these discussions:

1. Battery in cathode circuit. Grid leak is taken care of by leg of
attenuator to ground. This is the common version. The entire circuit is
attenuator, tube, battery and a couple of pieces of wire and maybe a
grid stopper resistor.

A quick newbie question about this circuit #1 if I may:
This looks interesting and I would like to try it but I don't know how
for sure how to calculate/implement this. Is the DC bias voltage of the
batteries all there is to it in determining the idle current through
the tube? Or does the value of Ra (in the absence of setting a specific
value for Rk) need to be calculated differently than a normal bypassed
grounded cathode circuit?

Thanks, Wessel

2. Battery shunted in grid leak position between grid and ground,
parallel to leg of attenuator to ground.

3. Battery seriesed in line between wiper of attenuator and grid.

Your idea of a battery in series below the attenuator would be a fourth
version and has the disavantage that it would vary the bias with volume
setting. How the battery could be charged by the signal after the amp
is switched off, as you claim, beats me. Maybe you know some magic that
I don't, which doesn't quite seem likely.

My source in all these experiments was a Quad 67 CD player, which has a
capacitively coupled output. I don't have any sources without a cap on
the output. Very few audiophiles can buy such a thing, and no competent
DIYer would build such a thing.

It is true that I already know what the ultrafidelista think. Most of
them think what I think and change their when I change mine; it's like
a ballet. But I was wondering if there is something I overlooked about
battery bias that might be worth some margin of extra silence. Seems
not.

On the subject of noise, you're on a hiding to nothing. A battery is a
more silent component than anything else including a straight piece of
wire. What I was reaching for was a more, even a marginally more,
silent implementation of a battery. The truth is that I couldn't hear
that the battery in the grid circuit was superior to the battery in the
cathode circuit, so therefore the more convenient and conservatively
safer cathode circuit it is.

Of course, by the nature of tubes, where there is very little that is
not already known, most experiments do not lead to improvements. On the
other hand, it became to clear during the course of this discussion
that the vast majority of RATs don't know about battery bias and that
only a few have ever tried it. It was definitely a worthwhile
discussion for me, and I hope it has given those who have not tried
battery bias (in the cathode, gentlemen, in the cathode) something to
do this winter.

HTH.

Andre Jute
Visit Jute on Amps at http://members.lycos.co.uk/fiultra/
"an unbelievably comprehensive web site" -- Hi-Fi News & Record Review



Ian Iveson wrote:
I can't speak for other heads, but the inside of mine finds itself
no further forward. It would help if you were to say what you are
using as a source in your listening test.

Using the battery *instead of* the grid leak gives you a 1ohm shunt
across the lower leg of the attenuator, which is not a good test of
the principle. Perhaps you mean *in series with* the grid leak?

All methods of grid bias in the absence of DC blocking will result
in biasing the source, obviously. Here again it is not clear whether
that in itself may have degraded the sound, so again the test is
inconclusive.

A battery at the foot of the attenuator would not discharge itself
in the absence of a grid leak, and the circuit would be just as
simple as the one you have chosen. The battery could also be charged
through the signal input with the amp off, so recharging could be a
switched function of a dedicated preamp. I dare say there would be
objections to this option, and I wonder what they would be.

What I would really like to know is would it be better, leaving
aside the issue of DC blocking and recharging, to bias via the grid
or the cathode? And would a series or a shunt connection to the grid
be more appropriate?

Do batteries have the same impedance in both directions? Are they
more noisy when discharging, or charging, or in equilibrium? What
are the mechanisms of distortion and will it be less in the
high-impedance low current input or the low impedance high current
cathode circuit. Will it be less by a factor greater than the
relative gain?

I would be surprised if these issues had not already been resolved,
considering you say that the cathode connection is so common. It
seems unlikely that the obvious alternatives were not explored, or
that the merely practical charging and DC blocking issues won the
day with the "ultra fi" fetishists.

And don't forget all of your famous guru friends, who I am sure
would be happy to advise you. I would be interested to hear their
views.

cheers, Ian



"Andre Jute" wrote in message
oups.com...
Uh-uh, Bruce. What you're talking about is the battery in the
place of
the grid leak resistor, between grid and ground. That's a low
impedance
path to ground. I've tried that one with a DACT, a high precision
Swiss
medical switch with SMD resistors and ultrashort paths, with the
tube
in position to amplify the clicks if any, and the amp working into
ultra-sensitive horns, and I heard no clicks.

I republished the T39 KISS Ultrafi circuit with the batteries in
the
third possible position, in the cathode circuit, simply because I
could
hear no difference with the battery in the grid leak position, and
the
cathode circuit position has the very great advantage of being
more
failsafe than either of the other two positions, by virtue of the
trickle charge the tubes can enjoy there.

One final note on the *discussion* circuit I published (battery
between
wiper of attenuator and grid of signal tubes) : A "professional"
implementation would use at least a resistor and a bypass cap in
addition to the battery, as has been pointed out by you and
several
others. But I'm not interested in defending a circuit before
professional production engineers. I'm interested in simplifying
the
circuit to what sounds best, regardless of expense or
"professional
practice", and in this case, because I am publishing the circuit
for
amateurs, in making it reasonably longlasting (there is a
companion
circuit, called the T44 "Populaire", with autobias, recommended
for the
newest aspirants, precisely because it is a largely self-adjusting
circuit). Thus the return to the battery in the cathode...

Thanks to all who contributed valuable insights (even if we were
all
talking about different circuits!). I'm removing the discussion
circuit
because it really doesn't add much to our wisdom; it may be a
viable
alternative (in that it works) but it is a decidedly less
desirable one
than the battery in the cathode circuit. That's the point of
discussion
with one's peers, to choose the best course of action from among
possibilities.

Andre Jute

bruce seifried wrote:

In article ,
Patrick Turner wrote:

bruce seifried wrote:

In article ,
Patrick Turner wrote:

bruce seifried wrote:

In article ,
Patrick Turner wrote:

But the transients you speak of don't occur any more than with any
other
biasing method.

Patrick Turner.


With the Dact 20K input attenuator, and a +2.4 volt dc source connected
to its wiper, you will see something on the order of a few tens of
millivolts being generated when changing the gain setting at the low
end
of the attenuator. Feeding this to the input of a dc-coupled power amp
will give you low level, audible transients. Not a deal breaker, but
not
good design practice.

Switching the attenuator from its lowest gain setting to off, or vice
versa, will give you a 2.4 volt step signal into your input. I would
call this a rather large transient, and not especially good for the
health of any speaker attached.

-bruce seifried

I don't understand how switching gain settings changes the tube bias;
there is always -2.4V at the grid regardless of the gain setting.

Well, bias really isn't the issue here. Because the battery is a very
low impedance voltage source, no matter what the setting of the DACT, it
will still be pretty close to +2.4 volts... except when the attenuator
is set to 'off', and then it will rudely drop to zero volts. The issue
is the (small) step-change in the dc level, which manifests itself as an
ac signal at the input grid.

The only chance of transients being generated by gain level change is if
the
input grid draws dc current and a small dc voltage exists across the
resistances
of the DACT, so that when switching, you hear the switch steps.

Are we talking about the same circuit?

There is no chance of transient problems with the schematic Andre sent me,
it has a 20k DACT attenuator, then the wiper goes to a battery, then
to the grid.
So at all levels of gain the bias remains and there are no transients.




The battery is constantly supplying current to the attenuator, in fact,
way more than any normal or abnormal grid current would be. I don't have
a 20K DACT series step attenuator here to measure, but roughly speaking,
the resistance to ground at the -60 dB setting (just above off) will be
on the order of 100 ohms. With a pair of 1.2 volt NiCad batteries in
series, you'll see a current of 24 ma flow through the attenuator.

Huh?
I don't follow you at all.

You must be talking about some other circuit.

Patrick Turner.


Aha... apparently we *are* talking about different cicuits.

Until it was changed several days ago, I had been referencing the T39
schematic on this page:

http://members.lycos.co.uk/fiultra/T...trafi-crct.jpg

I was never able to view the other schematic due to unknown browser
demons, but its description sounded like that of the T39 input biasing
arrangement, which I must say was bizarre, but easily fixable (...and I
happen to believe that a good quality capacitor makes a better capacitor
than a battery).

Carry on...

-bruce seifried

Both of Andre's circuits which use battery bias either in the grid or cathode
circuit
are NOT prone to switching transients when the gain is changed.

Patrick Turner.



Assuming that the collective source impedance of the batteries is around
1 ohm, you will see the voltage drop to +2.376 volts. Change the setting
a step higher, and the reistance is roughly around 300 ohms, for a
current flow of 8 ma. The voltage will change to +2.392 volts. You will
see a step-change of roughly 16 millivolts at the input grid. Yes, it is
small, and yes you will hear this.

I have heard this in amps with a DACT. But its a tiny effect compared to
music transients.

Please, wire this up for yourself and see what happens. I did, last
night, just to reaffirm what I already knew. Turn knob, see step-change
in voltge on scope, hear click in headphones.

This would all be just a moot conversation if the original circuit were
designed better: place a resistor in series with the batteries (say
100k-500k) so that that input impedance is somewhat normal ( and not 1
ohm!), and then place a good quality cap after the input pot. This would
work quite well.

cheers,

-bruce seifried

Wessel Dirksen wrote:

Andre Jute schreef:

Seems to me, my dear fellow, that your letter illustrates either that a
Marxist can make a meal out of very little or that you too can be an
"ultrafi fetishist". Those self-charging advantages of the battery in
the cathode circuit you are so keen to dismiss are huge benefits for no
cost. Your better ultrafidelista is not an audiophool. He is imbued
with the conviction that simplicity means better sound. The battery in
the cathode is the simplest safe implementation, requiring no other
components.

Once more, the three circuits considered in these discussions:

1. Battery in cathode circuit. Grid leak is taken care of by leg of
attenuator to ground. This is the common version. The entire circuit is
attenuator, tube, battery and a couple of pieces of wire and maybe a
grid stopper resistor.

A quick newbie question about this circuit #1 if I may:
This looks interesting and I would like to try it but I don't know how
for sure how to calculate/implement this. Is the DC bias voltage of the
batteries all there is to it in determining the idle current through
the tube?

Yes.

Battery bias circuits act just like fixed bias circuits.

Or does the value of Ra (in the absence of setting a specific
value for Rk) need to be calculated differently than a normal bypassed
grounded cathode circuit?

The value of RL ( and not Ra, which is plate resistance )
is chosen so RL = approx 3 x Ra at least, then if you wish you can trim the
B+
supply to suit the best point for working with that RL and that bias current
due to that grid bias.
Most ppl choose the B+, choose the RL, then adjust the applied grid bias for
the wanted dc anode voltage
and best working point, or they bias g1 from 0V, and adjust Rk to get the
wanted
Ia, hence Ek.
Once cathode bias is set up set, this is much more stable than fixed bias
because the cathode bias voltage isn't fixed,
and there is a state of negative current feedback which tries to keep the
tube
current the same.
Should more Ia flow, there is more Ek, to the tube tends to
be cut off because its as if the grid is being driven -ve with a rise in Ia.

However, fixed bias for input tubes usually is a long lasting fix
until the tube wears out or the battery goes flat if there is battery bias.


Patrick Turner.



Thanks, Wessel

2. Battery shunted in grid leak position between grid and ground,
parallel to leg of attenuator to ground.

3. Battery seriesed in line between wiper of attenuator and grid.

Your idea of a battery in series below the attenuator would be a fourth
version and has the disavantage that it would vary the bias with volume
setting. How the battery could be charged by the signal after the amp
is switched off, as you claim, beats me. Maybe you know some magic that
I don't, which doesn't quite seem likely.

My source in all these experiments was a Quad 67 CD player, which has a
capacitively coupled output. I don't have any sources without a cap on
the output. Very few audiophiles can buy such a thing, and no competent
DIYer would build such a thing.

It is true that I already know what the ultrafidelista think. Most of
them think what I think and change their when I change mine; it's like
a ballet. But I was wondering if there is something I overlooked about
battery bias that might be worth some margin of extra silence. Seems
not.

On the subject of noise, you're on a hiding to nothing. A battery is a
more silent component than anything else including a straight piece of
wire. What I was reaching for was a more, even a marginally more,
silent implementation of a battery. The truth is that I couldn't hear
that the battery in the grid circuit was superior to the battery in the
cathode circuit, so therefore the more convenient and conservatively
safer cathode circuit it is.

Of course, by the nature of tubes, where there is very little that is
not already known, most experiments do not lead to improvements. On the
other hand, it became to clear during the course of this discussion
that the vast majority of RATs don't know about battery bias and that
only a few have ever tried it. It was definitely a worthwhile
discussion for me, and I hope it has given those who have not tried
battery bias (in the cathode, gentlemen, in the cathode) something to
do this winter.

HTH.

Andre Jute
Visit Jute on Amps at http://members.lycos.co.uk/fiultra/
"an unbelievably comprehensive web site" -- Hi-Fi News & Record Review



Ian Iveson wrote:
I can't speak for other heads, but the inside of mine finds itself
no further forward. It would help if you were to say what you are
using as a source in your listening test.

Using the battery *instead of* the grid leak gives you a 1ohm shunt
across the lower leg of the attenuator, which is not a good test of
the principle. Perhaps you mean *in series with* the grid leak?

All methods of grid bias in the absence of DC blocking will result
in biasing the source, obviously. Here again it is not clear whether
that in itself may have degraded the sound, so again the test is
inconclusive.

A battery at the foot of the attenuator would not discharge itself
in the absence of a grid leak, and the circuit would be just as
simple as the one you have chosen. The battery could also be charged
through the signal input with the amp off, so recharging could be a
switched function of a dedicated preamp. I dare say there would be
objections to this option, and I wonder what they would be.

What I would really like to know is would it be better, leaving
aside the issue of DC blocking and recharging, to bias via the grid
or the cathode? And would a series or a shunt connection to the grid
be more appropriate?

Do batteries have the same impedance in both directions? Are they
more noisy when discharging, or charging, or in equilibrium? What
are the mechanisms of distortion and will it be less in the
high-impedance low current input or the low impedance high current
cathode circuit. Will it be less by a factor greater than the
relative gain?

I would be surprised if these issues had not already been resolved,
considering you say that the cathode connection is so common. It
seems unlikely that the obvious alternatives were not explored, or
that the merely practical charging and DC blocking issues won the
day with the "ultra fi" fetishists.

And don't forget all of your famous guru friends, who I am sure
would be happy to advise you. I would be interested to hear their
views.

cheers, Ian



"Andre Jute" wrote in message
oups.com...
Uh-uh, Bruce. What you're talking about is the battery in the
place of
the grid leak resistor, between grid and ground. That's a low
impedance
path to ground. I've tried that one with a DACT, a high precision
Swiss
medical switch with SMD resistors and ultrashort paths, with the
tube
in position to amplify the clicks if any, and the amp working into
ultra-sensitive horns, and I heard no clicks.

I republished the T39 KISS Ultrafi circuit with the batteries in
the
third possible position, in the cathode circuit, simply because I
could
hear no difference with the battery in the grid leak position, and
the
cathode circuit position has the very great advantage of being
more
failsafe than either of the other two positions, by virtue of the
trickle charge the tubes can enjoy there.

One final note on the *discussion* circuit I published (battery
between
wiper of attenuator and grid of signal tubes) : A "professional"
implementation would use at least a resistor and a bypass cap in
addition to the battery, as has been pointed out by you and
several
others. But I'm not interested in defending a circuit before
professional production engineers. I'm interested in simplifying
the
circuit to what sounds best, regardless of expense or
"professional
practice", and in this case, because I am publishing the circuit
for
amateurs, in making it reasonably longlasting (there is a
companion
circuit, called the T44 "Populaire", with autobias, recommended
for the
newest aspirants, precisely because it is a largely self-adjusting
circuit). Thus the return to the battery in the cathode...

Thanks to all who contributed valuable insights (even if we were
all
talking about different circuits!). I'm removing the discussion
circuit
because it really doesn't add much to our wisdom; it may be a
viable
alternative (in that it works) but it is a decidedly less
desirable one
than the battery in the cathode circuit. That's the point of
discussion
with one's peers, to choose the best course of action from among
possibilities.

Andre Jute

Patrick Turner schreef:

Wessel Dirksen wrote:

Andre Jute schreef:

Seems to me, my dear fellow, that your letter illustrates either that a
Marxist can make a meal out of very little or that you too can be an
"ultrafi fetishist". Those self-charging advantages of the battery in
the cathode circuit you are so keen to dismiss are huge benefits for no
cost. Your better ultrafidelista is not an audiophool. He is imbued
with the conviction that simplicity means better sound. The battery in
the cathode is the simplest safe implementation, requiring no other
components.

Once more, the three circuits considered in these discussions:

1. Battery in cathode circuit. Grid leak is taken care of by leg of
attenuator to ground. This is the common version. The entire circuit is
attenuator, tube, battery and a couple of pieces of wire and maybe a
grid stopper resistor.

A quick newbie question about this circuit #1 if I may:
This looks interesting and I would like to try it but I don't know how
for sure how to calculate/implement this. Is the DC bias voltage of the
batteries all there is to it in determining the idle current through
the tube?

Yes.

Battery bias circuits act just like fixed bias circuits.

Or does the value of Ra (in the absence of setting a specific
value for Rk) need to be calculated differently than a normal bypassed
grounded cathode circuit?

The value of RL ( and not Ra, which is plate resistance )
is chosen so RL = approx 3 x Ra at least, then if you wish you can trim the
B+
supply to suit the best point for working with that RL and that bias current
due to that grid bias.
Most ppl choose the B+, choose the RL, then adjust the applied grid bias for
the wanted dc anode voltage
and best working point, or they bias g1 from 0V, and adjust Rk to get the
wanted
Ia, hence Ek.
Once cathode bias is set up set, this is much more stable than fixed bias
because the cathode bias voltage isn't fixed,
and there is a state of negative current feedback which tries to keep the
tube
current the same.
Should more Ia flow, there is more Ek, to the tube tends to
be cut off because its as if the grid is being driven -ve with a rise in Ia.

However, fixed bias for input tubes usually is a long lasting fix
until the tube wears out or the battery goes flat if there is battery bias.


Patrick Turner.


Thanks Patrick



Thanks, Wessel

2. Battery shunted in grid leak position between grid and ground,
parallel to leg of attenuator to ground.

3. Battery seriesed in line between wiper of attenuator and grid.

Your idea of a battery in series below the attenuator would be a fourth
version and has the disavantage that it would vary the bias with volume
setting. How the battery could be charged by the signal after the amp
is switched off, as you claim, beats me. Maybe you know some magic that
I don't, which doesn't quite seem likely.

My source in all these experiments was a Quad 67 CD player, which has a
capacitively coupled output. I don't have any sources without a cap on
the output. Very few audiophiles can buy such a thing, and no competent
DIYer would build such a thing.

It is true that I already know what the ultrafidelista think. Most of
them think what I think and change their when I change mine; it's like
a ballet. But I was wondering if there is something I overlooked about
battery bias that might be worth some margin of extra silence. Seems
not.

On the subject of noise, you're on a hiding to nothing. A battery is a
more silent component than anything else including a straight piece of
wire. What I was reaching for was a more, even a marginally more,
silent implementation of a battery. The truth is that I couldn't hear
that the battery in the grid circuit was superior to the battery in the
cathode circuit, so therefore the more convenient and conservatively
safer cathode circuit it is.

Of course, by the nature of tubes, where there is very little that is
not already known, most experiments do not lead to improvements. On the
other hand, it became to clear during the course of this discussion
that the vast majority of RATs don't know about battery bias and that
only a few have ever tried it. It was definitely a worthwhile
discussion for me, and I hope it has given those who have not tried
battery bias (in the cathode, gentlemen, in the cathode) something to
do this winter.

HTH.

Andre Jute
Visit Jute on Amps at http://members.lycos.co.uk/fiultra/
"an unbelievably comprehensive web site" -- Hi-Fi News & Record Review



Ian Iveson wrote:
I can't speak for other heads, but the inside of mine finds itself
no further forward. It would help if you were to say what you are
using as a source in your listening test.

Using the battery *instead of* the grid leak gives you a 1ohm shunt
across the lower leg of the attenuator, which is not a good test of
the principle. Perhaps you mean *in series with* the grid leak?

All methods of grid bias in the absence of DC blocking will result
in biasing the source, obviously. Here again it is not clear whether
that in itself may have degraded the sound, so again the test is
inconclusive.

A battery at the foot of the attenuator would not discharge itself
in the absence of a grid leak, and the circuit would be just as
simple as the one you have chosen. The battery could also be charged
through the signal input with the amp off, so recharging could be a
switched function of a dedicated preamp. I dare say there would be
objections to this option, and I wonder what they would be.

What I would really like to know is would it be better, leaving
aside the issue of DC blocking and recharging, to bias via the grid
or the cathode? And would a series or a shunt connection to the grid
be more appropriate?

Do batteries have the same impedance in both directions? Are they
more noisy when discharging, or charging, or in equilibrium? What
are the mechanisms of distortion and will it be less in the
high-impedance low current input or the low impedance high current
cathode circuit. Will it be less by a factor greater than the
relative gain?

I would be surprised if these issues had not already been resolved,
considering you say that the cathode connection is so common. It
seems unlikely that the obvious alternatives were not explored, or
that the merely practical charging and DC blocking issues won the
day with the "ultra fi" fetishists.

And don't forget all of your famous guru friends, who I am sure
would be happy to advise you. I would be interested to hear their
views.

cheers, Ian



"Andre Jute" wrote in message
oups.com...
Uh-uh, Bruce. What you're talking about is the battery in the
place of
the grid leak resistor, between grid and ground. That's a low
impedance
path to ground. I've tried that one with a DACT, a high precision
Swiss
medical switch with SMD resistors and ultrashort paths, with the
tube
in position to amplify the clicks if any, and the amp working into
ultra-sensitive horns, and I heard no clicks.

I republished the T39 KISS Ultrafi circuit with the batteries in
the
third possible position, in the cathode circuit, simply because I
could
hear no difference with the battery in the grid leak position, and
the
cathode circuit position has the very great advantage of being
more
failsafe than either of the other two positions, by virtue of the
trickle charge the tubes can enjoy there.

One final note on the *discussion* circuit I published (battery
between
wiper of attenuator and grid of signal tubes) : A "professional"
implementation would use at least a resistor and a bypass cap in
addition to the battery, as has been pointed out by you and
several
others. But I'm not interested in defending a circuit before
professional production engineers. I'm interested in simplifying
the
circuit to what sounds best, regardless of expense or
"professional
practice", and in this case, because I am publishing the circuit
for
amateurs, in making it reasonably longlasting (there is a
companion
circuit, called the T44 "Populaire", with autobias, recommended
for the
newest aspirants, precisely because it is a largely self-adjusting
circuit). Thus the return to the battery in the cathode...

Thanks to all who contributed valuable insights (even if we were
all
talking about different circuits!). I'm removing the discussion
circuit
because it really doesn't add much to our wisdom; it may be a
viable
alternative (in that it works) but it is a decidedly less
desirable one
than the battery in the cathode circuit. That's the point of
discussion
with one's peers, to choose the best course of action from among
possibilities.

Andre Jute

In article ,
Patrick Turner wrote:

snip



Aha... apparently we *are* talking about different cicuits.

Until it was changed several days ago, I had been referencing the T39
schematic on this page:

http://members.lycos.co.uk/fiultra/T...trafi-crct.jpg

I was never able to view the other schematic due to unknown browser
demons, but its description sounded like that of the T39 input biasing
arrangement, which I must say was bizarre, but easily fixable (...and I
happen to believe that a good quality capacitor makes a better capacitor
than a battery).

Carry on...

-bruce seifried

Both of Andre's circuits which use battery bias either in the grid or cathode
circuit
are NOT prone to switching transients when the gain is changed.

Patrick Turner.

snip



goddamn, Patrick, you are one stubborn bloke. Please pay attention.

There are at least *three* versions being discussed, not two:

1. batteries in the cathode circuit.
2. batteries *in series* with the grid.
3. batteries *in parallel* with the grid (grid leak position).

Circuit number three is what I have been talking about. The schematic
was locate at the link I referenced above, until a few days ago, when it
was redrawn to show the batteries in the cathode circuit. OK?

I will send you a copy of said schematic if you need to see it.

Can we move on now?

-bruce seifried

Wessel Dirksen wrote:
Andre Jute schreef:

Seems to me, my dear fellow, that your letter illustrates either that a
Marxist can make a meal out of very little or that you too can be an
"ultrafi fetishist". Those self-charging advantages of the battery in
the cathode circuit you are so keen to dismiss are huge benefits for no
cost. Your better ultrafidelista is not an audiophool. He is imbued
with the conviction that simplicity means better sound. The battery in
the cathode is the simplest safe implementation, requiring no other
components.

Once more, the three circuits considered in these discussions:

1. Battery in cathode circuit. Grid leak is taken care of by leg of
attenuator to ground. This is the common version. The entire circuit is
attenuator, tube, battery and a couple of pieces of wire and maybe a
grid stopper resistor.

A quick newbie question about this circuit #1 if I may:

You may, always. Whether anyone will know the answer is another
matter...

This looks interesting and I would like to try it but I don't know how
for sure how to calculate/implement this. Is the DC bias voltage of the
batteries all there is to it in determining the idle current through
the tube?

Yes. The battery fixes negative grid bias, in my example -2.4V.
Negative grid bias, together with your choice of plate voltage, in turn
fixes current drawn at quiescence (in a single-ended stage). Current
and negative grid bias together define the point through which the
transfer function must operate. The load you put on the plate then
defines the slope of the transfer function, which we commonly call the
loadline, through that point where quiescent plate voltage, current and
negative grid bias intersect.

The easiest way to understand these things is always to look at plate
voltage/current/negative grid bias Eb-Ia-Eg transfer curves.

Put a ruler across the curves at the angle of the plate load -RL =
-Eg/Ia. RL is traditionally between 2 and 4 times the plate load Rp,
but modern ultrafidelista design choses -RL somewhere between 4 and 8
times RP for the extra silence, eg. distance from the curves at the
bottom of the -Eg/Ia relationship.

Draw a vertical line from the Eb line at the desired plate voltage.
(Just for the sake of completeness, this line is angled very slightly
by the AC of the signal on top of the DC negative grid bias but in most
conservative designs you can forget this and draw it vertically.) Slide
your ruler, at the angle of -RL, up and down this plate voltage line
until you see an equidistant swing across the negative bias lines to
each side of the central point, or nearest. The idea is that you want
to cross the negatively grid bias lines where they are equidistant or
nearest to it. Low current traditional design down in the nasty curves
at the bottom of the transfer function is not for us. The point where
the vertical power supply line crosses the angled loadline and a
negative grid bias line (which you might have to interpolate) is the
quiescent operating point of an SE stage.

You can now choose a cathode resistor to drop the voltage over the grid
by the amount of negative bias you have determined graphically,
R=-Eg/Ia.

With battery bias you perform the same procedure and then juggle the
numbers until you find a compromise that suits available batteries. Or
you can of course add a multiturn minipot to adjust the voltage from
the batteries, though the most refined designs do without it.
(Ultrafidelista design considers simplicity the highest refinement,
therefore the lowest parts count to do the job. Contrast with
professional engineering design which considers that refinement
requires designing the job perfectly, regardless of parts count, and
then cutting it back for reasons of cost.)

For instance, one wants to operate a 417A at 20mA or over to bring it
alive, and you want to input the maximum 2V straight out of a CD to get
the max swing so you can use only one voltage multiplication stage
rather than two, plus you need about 0.6V distance from the 0V negative
grid line because the tube doesn't wait until 0V to start drawing grid
current, plus you know from experience that you want to operate the
tube at say 180V (don't worry about this being max, just don't go over
it) and 20+mA and -2.6V but not much higher (study the curves falling
over to the right of the point defined by these numbers to discover why
not).

At this point you study the available batteries and what other people
have done. Lucas Cant in a famous, much copied potato amp, used a 2V D
cell which would permit an input of only about 1V--but he was dealing
with a far outie wing of the ultrafidelista who have huge, huge horns
(and no love-life). My own wing of the ultrafidelista, who love women
as much as music, have horns with more modest footprints which need the
extra swing, so we use two AA or AAA batteries to give 2.4V.

At this point you have two fixed numbers, the 2.4V out of the batteries
(not 2x the nominal 1.5V -- measure twice and build once!), and a
current draw of 20mA or more. Slide your ruler, still standing in for
-RL, some more along the 2.4V line you have interpolated on the curves,
find Ia 20mA and -Eg 2.4V and there you can drop a line to 175V Eb. I
could have gone a little higher in current and plate voltage for the
same -2,4V bias but in a circuit to be published for amateurs I didn't
want to hog the tube out to its whole official limit (even though I
know from experience that running it over max official dissipation
doesn't degrade its longevity noticeably--I've never had one break and
I have several with 10K+ hours).

Or does the value of Ra (in the absence of setting a specific
value for Rk) need to be calculated differently than a normal bypassed
grounded cathode circuit?

Rp, the particular plate resistance for any implementation, can be
calculated by drawing a line tangential to the -Eg curve through the
quiescent operating point. It won't make much difference if you accept
the published median number in the spec sheet as the gospel.

RL, the plate load, you decide. The higher RL is, the flatter the
transfer curve and the less the distortion in the stage, all other
things being equal. That is why I load my 300B power stages with a
power-choking 5K6, for the silence. That is why I load the 417A we've
been using as an example with 10K, which is 5.68 times the specific Rp
at the chosen operating point.

Once you've decided on your battery and operating point, you can check
the impedance of your design and discover that you have indeed chosen a
fixed Rk! 2.4V/0.020A equals a notional 120 ohm, which is a pretty
common choice for a resistor in an autobias 417A stage. See, a battery
for bias isn't such an extraordinary choice: it is just an ultra-silent
120 ohm replacement for a resistor and a cap!

Thanks, Wessel

HTH.

Andre Jute



2. Battery shunted in grid leak position between grid and ground,
parallel to leg of attenuator to ground.

3. Battery seriesed in line between wiper of attenuator and grid.

Your idea of a battery in series below the attenuator would be a fourth
version and has the disavantage that it would vary the bias with volume
setting. How the battery could be charged by the signal after the amp
is switched off, as you claim, beats me. Maybe you know some magic that
I don't, which doesn't quite seem likely.

My source in all these experiments was a Quad 67 CD player, which has a
capacitively coupled output. I don't have any sources without a cap on
the output. Very few audiophiles can buy such a thing, and no competent
DIYer would build such a thing.

It is true that I already know what the ultrafidelista think. Most of
them think what I think and change their when I change mine; it's like
a ballet. But I was wondering if there is something I overlooked about
battery bias that might be worth some margin of extra silence. Seems
not.

On the subject of noise, you're on a hiding to nothing. A battery is a
more silent component than anything else including a straight piece of
wire. What I was reaching for was a more, even a marginally more,
silent implementation of a battery. The truth is that I couldn't hear
that the battery in the grid circuit was superior to the battery in the
cathode circuit, so therefore the more convenient and conservatively
safer cathode circuit it is.

Of course, by the nature of tubes, where there is very little that is
not already known, most experiments do not lead to improvements. On the
other hand, it became to clear during the course of this discussion
that the vast majority of RATs don't know about battery bias and that
only a few have ever tried it. It was definitely a worthwhile
discussion for me, and I hope it has given those who have not tried
battery bias (in the cathode, gentlemen, in the cathode) something to
do this winter.

HTH.

Andre Jute
Visit Jute on Amps at http://members.lycos.co.uk/fiultra/
"an unbelievably comprehensive web site" -- Hi-Fi News & Record Review



Ian Iveson wrote:
I can't speak for other heads, but the inside of mine finds itself
no further forward. It would help if you were to say what you are
using as a source in your listening test.

Using the battery *instead of* the grid leak gives you a 1ohm shunt
across the lower leg of the attenuator, which is not a good test of
the principle. Perhaps you mean *in series with* the grid leak?

All methods of grid bias in the absence of DC blocking will result
in biasing the source, obviously. Here again it is not clear whether
that in itself may have degraded the sound, so again the test is
inconclusive.

A battery at the foot of the attenuator would not discharge itself
in the absence of a grid leak, and the circuit would be just as
simple as the one you have chosen. The battery could also be charged
through the signal input with the amp off, so recharging could be a
switched function of a dedicated preamp. I dare say there would be
objections to this option, and I wonder what they would be.

What I would really like to know is would it be better, leaving
aside the issue of DC blocking and recharging, to bias via the grid
or the cathode? And would a series or a shunt connection to the grid
be more appropriate?

Do batteries have the same impedance in both directions? Are they
more noisy when discharging, or charging, or in equilibrium? What
are the mechanisms of distortion and will it be less in the
high-impedance low current input or the low impedance high current
cathode circuit. Will it be less by a factor greater than the
relative gain?

I would be surprised if these issues had not already been resolved,
considering you say that the cathode connection is so common. It
seems unlikely that the obvious alternatives were not explored, or
that the merely practical charging and DC blocking issues won the
day with the "ultra fi" fetishists.

And don't forget all of your famous guru friends, who I am sure
would be happy to advise you. I would be interested to hear their
views.

cheers, Ian



"Andre Jute" wrote in message
oups.com...
Uh-uh, Bruce. What you're talking about is the battery in the
place of
the grid leak resistor, between grid and ground. That's a low
impedance
path to ground. I've tried that one with a DACT, a high precision
Swiss
medical switch with SMD resistors and ultrashort paths, with the
tube
in position to amplify the clicks if any, and the amp working into
ultra-sensitive horns, and I heard no clicks.

I republished the T39 KISS Ultrafi circuit with the batteries in
the
third possible position, in the cathode circuit, simply because I
could
hear no difference with the battery in the grid leak position, and
the
cathode circuit position has the very great advantage of being
more
failsafe than either of the other two positions, by virtue of the
trickle charge the tubes can enjoy there.

One final note on the *discussion* circuit I published (battery
between
wiper of attenuator and grid of signal tubes) : A "professional"
implementation would use at least a resistor and a bypass cap in
addition to the battery, as has been pointed out by you and
several
others. But I'm not interested in defending a circuit before
professional production engineers. I'm interested in simplifying
the
circuit to what sounds best, regardless of expense or
"professional
practice", and in this case, because I am publishing the circuit
for
amateurs, in making it reasonably longlasting (there is a
companion
circuit, called the T44 "Populaire", with autobias, recommended
for the
newest aspirants, precisely because it is a largely self-adjusting
circuit). Thus the return to the battery in the cathode...

Thanks to all who contributed valuable insights (even if we were
all
talking about different circuits!). I'm removing the discussion
circuit
because it really doesn't add much to our wisdom; it may be a
viable
alternative (in that it works) but it is a decidedly less
desirable one
than the battery in the cathode circuit. That's the point of
discussion
with one's peers, to choose the best course of action from among
possibilities.

Andre Jute

Maynard, later Lord, Keynes, was a director of the Ballet Lambert. One
day the company's manager came to him and said it was suggested that
the dancers decide the billing order themselves. Would he permit it?
Keynes looked the man straight in the eyes and said, "Only if you have
first called for an ambulance."

Stick around, Bruce. We can do with a quality engineer with patience
and a sense of humour. Learn to enjoy being horrified at much of what
we do; if you permit your blood pressure to rise you will not last
long.

Andre Jute.

bruce seifried wrote:
In article ,
Patrick Turner wrote:

snip



Aha... apparently we *are* talking about different cicuits.

Until it was changed several days ago, I had been referencing the T39
schematic on this page:

http://members.lycos.co.uk/fiultra/T...trafi-crct.jpg

I was never able to view the other schematic due to unknown browser
demons, but its description sounded like that of the T39 input biasing
arrangement, which I must say was bizarre, but easily fixable (...and I
happen to believe that a good quality capacitor makes a better capacitor
than a battery).

Carry on...

-bruce seifried

Both of Andre's circuits which use battery bias either in the grid or cathode
circuit
are NOT prone to switching transients when the gain is changed.

Patrick Turner.

snip



goddamn, Patrick, you are one stubborn bloke. Please pay attention.

There are at least *three* versions being discussed, not two:

1. batteries in the cathode circuit.
2. batteries *in series* with the grid.
3. batteries *in parallel* with the grid (grid leak position).

Circuit number three is what I have been talking about. The schematic
was locate at the link I referenced above, until a few days ago, when it
was redrawn to show the batteries in the cathode circuit. OK?

I will send you a copy of said schematic if you need to see it.

Can we move on now?

-bruce seifried

E&OE, as my phone company says when they overbill me by eight times.

Put a ruler across the curves at the angle of the plate load -RL =
-Eg/Ia. RL is traditionally between 2 and 4 times the plate load Rp,

should of course read

Put a ruler across the curves at the angle of the plate load -RL =
-Eg/Ia. RL is traditionally between 2 and 4 times the plate resistance Rp,

I imagine more similar slips will be discovered. Have pity on a harried
fellow trying to perform five tasks at once. Just as well this isn't my
day for mirror-writing...

Andre Jute
Visit Jute on Amps at http://members.lycos.co.uk/fiultra/
"an unbelievably comprehensive web site" -- Hi-Fi News & Record Review


Andre Jute wrote:
Wessel Dirksen wrote:
Andre Jute schreef:

Seems to me, my dear fellow, that your letter illustrates either that a
Marxist can make a meal out of very little or that you too can be an
"ultrafi fetishist". Those self-charging advantages of the battery in
the cathode circuit you are so keen to dismiss are huge benefits for no
cost. Your better ultrafidelista is not an audiophool. He is imbued
with the conviction that simplicity means better sound. The battery in
the cathode is the simplest safe implementation, requiring no other
components.

Once more, the three circuits considered in these discussions:

1. Battery in cathode circuit. Grid leak is taken care of by leg of
attenuator to ground. This is the common version. The entire circuit is
attenuator, tube, battery and a couple of pieces of wire and maybe a
grid stopper resistor.

A quick newbie question about this circuit #1 if I may:

You may, always. Whether anyone will know the answer is another
matter...

This looks interesting and I would like to try it but I don't know how
for sure how to calculate/implement this. Is the DC bias voltage of the
batteries all there is to it in determining the idle current through
the tube?

Yes. The battery fixes negative grid bias, in my example -2.4V.
Negative grid bias, together with your choice of plate voltage, in turn
fixes current drawn at quiescence (in a single-ended stage). Current
and negative grid bias together define the point through which the
transfer function must operate. The load you put on the plate then
defines the slope of the transfer function, which we commonly call the
loadline, through that point where quiescent plate voltage, current and
negative grid bias intersect.

The easiest way to understand these things is always to look at plate
voltage/current/negative grid bias Eb-Ia-Eg transfer curves.

Put a ruler across the curves at the angle of the plate load -RL =
-Eg/Ia. RL is traditionally between 2 and 4 times the plate load Rp,
but modern ultrafidelista design choses -RL somewhere between 4 and 8
times RP for the extra silence, eg. distance from the curves at the
bottom of the -Eg/Ia relationship.

Draw a vertical line from the Eb line at the desired plate voltage.
(Just for the sake of completeness, this line is angled very slightly
by the AC of the signal on top of the DC negative grid bias but in most
conservative designs you can forget this and draw it vertically.) Slide
your ruler, at the angle of -RL, up and down this plate voltage line
until you see an equidistant swing across the negative bias lines to
each side of the central point, or nearest. The idea is that you want
to cross the negatively grid bias lines where they are equidistant or
nearest to it. Low current traditional design down in the nasty curves
at the bottom of the transfer function is not for us. The point where
the vertical power supply line crosses the angled loadline and a
negative grid bias line (which you might have to interpolate) is the
quiescent operating point of an SE stage.

You can now choose a cathode resistor to drop the voltage over the grid
by the amount of negative bias you have determined graphically,
R=-Eg/Ia.

With battery bias you perform the same procedure and then juggle the
numbers until you find a compromise that suits available batteries. Or
you can of course add a multiturn minipot to adjust the voltage from
the batteries, though the most refined designs do without it.
(Ultrafidelista design considers simplicity the highest refinement,
therefore the lowest parts count to do the job. Contrast with
professional engineering design which considers that refinement
requires designing the job perfectly, regardless of parts count, and
then cutting it back for reasons of cost.)

For instance, one wants to operate a 417A at 20mA or over to bring it
alive, and you want to input the maximum 2V straight out of a CD to get
the max swing so you can use only one voltage multiplication stage
rather than two, plus you need about 0.6V distance from the 0V negative
grid line because the tube doesn't wait until 0V to start drawing grid
current, plus you know from experience that you want to operate the
tube at say 180V (don't worry about this being max, just don't go over
it) and 20+mA and -2.6V but not much higher (study the curves falling
over to the right of the point defined by these numbers to discover why
not).

At this point you study the available batteries and what other people
have done. Lucas Cant in a famous, much copied potato amp, used a 2V D
cell which would permit an input of only about 1V--but he was dealing
with a far outie wing of the ultrafidelista who have huge, huge horns
(and no love-life). My own wing of the ultrafidelista, who love women
as much as music, have horns with more modest footprints which need the
extra swing, so we use two AA or AAA batteries to give 2.4V.

At this point you have two fixed numbers, the 2.4V out of the batteries
(not 2x the nominal 1.5V -- measure twice and build once!), and a
current draw of 20mA or more. Slide your ruler, still standing in for
-RL, some more along the 2.4V line you have interpolated on the curves,
find Ia 20mA and -Eg 2.4V and there you can drop a line to 175V Eb. I
could have gone a little higher in current and plate voltage for the
same -2,4V bias but in a circuit to be published for amateurs I didn't
want to hog the tube out to its whole official limit (even though I
know from experience that running it over max official dissipation
doesn't degrade its longevity noticeably--I've never had one break and
I have several with 10K+ hours).

Or does the value of Ra (in the absence of setting a specific
value for Rk) need to be calculated differently than a normal bypassed
grounded cathode circuit?

Rp, the particular plate resistance for any implementation, can be
calculated by drawing a line tangential to the -Eg curve through the
quiescent operating point. It won't make much difference if you accept
the published median number in the spec sheet as the gospel.

RL, the plate load, you decide. The higher RL is, the flatter the
transfer curve and the less the distortion in the stage, all other
things being equal. That is why I load my 300B power stages with a
power-choking 5K6, for the silence. That is why I load the 417A we've
been using as an example with 10K, which is 5.68 times the specific Rp
at the chosen operating point.

Once you've decided on your battery and operating point, you can check
the impedance of your design and discover that you have indeed chosen a
fixed Rk! 2.4V/0.020A equals a notional 120 ohm, which is a pretty
common choice for a resistor in an autobias 417A stage. See, a battery
for bias isn't such an extraordinary choice: it is just an ultra-silent
120 ohm replacement for a resistor and a cap!

Thanks, Wessel

HTH.

Andre Jute



2. Battery shunted in grid leak position between grid and ground,
parallel to leg of attenuator to ground.

3. Battery seriesed in line between wiper of attenuator and grid.

Your idea of a battery in series below the attenuator would be a fourth
version and has the disavantage that it would vary the bias with volume
setting. How the battery could be charged by the signal after the amp
is switched off, as you claim, beats me. Maybe you know some magic that
I don't, which doesn't quite seem likely.

My source in all these experiments was a Quad 67 CD player, which has a
capacitively coupled output. I don't have any sources without a cap on
the output. Very few audiophiles can buy such a thing, and no competent
DIYer would build such a thing.

It is true that I already know what the ultrafidelista think. Most of
them think what I think and change their when I change mine; it's like
a ballet. But I was wondering if there is something I overlooked about
battery bias that might be worth some margin of extra silence. Seems
not.

On the subject of noise, you're on a hiding to nothing. A battery is a
more silent component than anything else including a straight piece of
wire. What I was reaching for was a more, even a marginally more,
silent implementation of a battery. The truth is that I couldn't hear
that the battery in the grid circuit was superior to the battery in the
cathode circuit, so therefore the more convenient and conservatively
safer cathode circuit it is.

Of course, by the nature of tubes, where there is very little that is
not already known, most experiments do not lead to improvements. On the
other hand, it became to clear during the course of this discussion
that the vast majority of RATs don't know about battery bias and that
only a few have ever tried it. It was definitely a worthwhile
discussion for me, and I hope it has given those who have not tried
battery bias (in the cathode, gentlemen, in the cathode) something to
do this winter.

HTH.

Andre Jute
Visit Jute on Amps at http://members.lycos.co.uk/fiultra/
"an unbelievably comprehensive web site" -- Hi-Fi News & Record Review



Ian Iveson wrote:
I can't speak for other heads, but the inside of mine finds itself
no further forward. It would help if you were to say what you are
using as a source in your listening test.

Using the battery *instead of* the grid leak gives you a 1ohm shunt
across the lower leg of the attenuator, which is not a good test of
the principle. Perhaps you mean *in series with* the grid leak?

All methods of grid bias in the absence of DC blocking will result
in biasing the source, obviously. Here again it is not clear whether
that in itself may have degraded the sound, so again the test is
inconclusive.

A battery at the foot of the attenuator would not discharge itself
in the absence of a grid leak, and the circuit would be just as
simple as the one you have chosen. The battery could also be charged
through the signal input with the amp off, so recharging could be a
switched function of a dedicated preamp. I dare say there would be
objections to this option, and I wonder what they would be.

What I would really like to know is would it be better, leaving
aside the issue of DC blocking and recharging, to bias via the grid
or the cathode? And would a series or a shunt connection to the grid
be more appropriate?

Do batteries have the same impedance in both directions? Are they
more noisy when discharging, or charging, or in equilibrium? What
are the mechanisms of distortion and will it be less in the
high-impedance low current input or the low impedance high current
cathode circuit. Will it be less by a factor greater than the
relative gain?

I would be surprised if these issues had not already been resolved,
considering you say that the cathode connection is so common. It
seems unlikely that the obvious alternatives were not explored, or
that the merely practical charging and DC blocking issues won the
day with the "ultra fi" fetishists.

And don't forget all of your famous guru friends, who I am sure
would be happy to advise you. I would be interested to hear their
views.

cheers, Ian



"Andre Jute" wrote in message
oups.com...
Uh-uh, Bruce. What you're talking about is the battery in the
place of
the grid leak resistor, between grid and ground. That's a low
impedance
path to ground. I've tried that one with a DACT, a high precision
Swiss
medical switch with SMD resistors and ultrashort paths, with the
tube
in position to amplify the clicks if any, and the amp working into
ultra-sensitive horns, and I heard no clicks.

I republished the T39 KISS Ultrafi circuit with the batteries in
the
third possible position, in the cathode circuit, simply because I
could
hear no difference with the battery in the grid leak position, and
the
cathode circuit position has the very great advantage of being
more
failsafe than either of the other two positions, by virtue of the
trickle charge the tubes can enjoy there.

One final note on the *discussion* circuit I published (battery
between
wiper of attenuator and grid of signal tubes) : A "professional"
implementation would use at least a resistor and a bypass cap in
addition to the battery, as has been pointed out by you and
several
others. But I'm not interested in defending a circuit before
professional production engineers. I'm interested in simplifying
the
circuit to what sounds best, regardless of expense or
"professional
practice", and in this case, because I am publishing the circuit
for
amateurs, in making it reasonably longlasting (there is a
companion
circuit, called the T44 "Populaire", with autobias, recommended
for the
newest aspirants, precisely because it is a largely self-adjusting
circuit). Thus the return to the battery in the cathode...

Thanks to all who contributed valuable insights (even if we were
all
talking about different circuits!). I'm removing the discussion
circuit
because it really doesn't add much to our wisdom; it may be a
viable
alternative (in that it works) but it is a decidedly less
desirable one
than the battery in the cathode circuit. That's the point of
discussion
with one's peers, to choose the best course of action from among
possibilities.

Andre Jute

bruce seifried wrote:

In article ,
Patrick Turner wrote:

snip


Aha... apparently we *are* talking about different cicuits.

Until it was changed several days ago, I had been referencing the T39
schematic on this page:

http://members.lycos.co.uk/fiultra/T...trafi-crct.jpg

I was never able to view the other schematic due to unknown browser
demons, but its description sounded like that of the T39 input biasing
arrangement, which I must say was bizarre, but easily fixable (...and I
happen to believe that a good quality capacitor makes a better capacitor
than a battery).

Carry on...

-bruce seifried

Both of Andre's circuits which use battery bias either in the grid or cathode
circuit
are NOT prone to switching transients when the gain is changed.

Patrick Turner.

snip

goddamn, Patrick, you are one stubborn bloke. Please pay attention.

There are at least *three* versions being discussed, not two:

1. batteries in the cathode circuit.
2. batteries *in series* with the grid.
3. batteries *in parallel* with the grid (grid leak position).

Circuit number three is what I have been talking about. The schematic
was locate at the link I referenced above, until a few days ago, when it
was redrawn to show the batteries in the cathode circuit. OK?

I will send you a copy of said schematic if you need to see it.

Can we move on now?

For 3 you'd need a cap somewhere.
But Andre sent me what he was discussing, and there was no way you'd get transient
or
biasing problems.

I am happily settled, ain't movin noplace :-

Although Nicole and Kylie have asked me over to the
tropical island they bought
but I said I'd need to bring a physio, chiropractor, therapist,
and there is limited room.....

Patrick Turner.



-bruce seifried

In article ,
Patrick Turner wrote:


big snip

For 3 you'd need a cap somewhere.

Thank you.


-bruce seifried

Current question about Option 1:

What if you were to use high current tubes such as 6H30 which often run
at 15 or so mA. Will a 1800 mAh be able to charge enough to be voltage
stable keeping within the hysteresis profile of a NiMH type battery?

Wessel Dirksen wrote:
Current question about Option 1:

Which is battery bias in the cathode.

What if you were to use high current tubes such as 6H30 which often run
at 15 or so mA. Will a 1800 mAh be able to charge enough to be voltage
stable keeping within the hysteresis profile of a NiMH type battery?

I think it was Grego Sanguinetti who told me, near my own start on the
valve odyssey: "The tube doesn't care where the current comes from."
That statement is worth serious thought. (The precise occasion was the
current delivered to the plate of a driver via the cathode of a power
tube in a direct coupling scheme, but the remark has very wide general
application.)

In the example of the 417A I worked for you, Wessel, the tube draws
20mA, and I and others have run them on battery bias (in the cathode!)
to 24mA. In my T68bis "Minus Zero" potato design I use two AAA
batteries per channel to save space. They were just batteries I had no
other use for that came in a bargain pack with a charger, rated
750mA/h. I see now (the amp is on the bench as I am using it to sort
WE417A that arrived from a lurker on RAT--thanks, John) that they
charge at 48mA but that is irrelevant. You can see the installation at
http://members.lycos.co.uk/fiultra/K...0T68MZ417A.jpg
where the two battery boxes are just above the two 9-pin valve bases in
the top photo. I use only ni-cads; there was a very high-level
discussion on the Joenet once about which batteries to use from which
it emerged that some types could be dangerous to the health of your
amplifier but that ni-cads are the best compromise.

Why does this 20mA scheme work? *Because the current is supplied by the
B+ to the plate.* Not by the batteries to the grid. The sort of cathode
battery bias we are talking about is for Class A single-ended stages.
We particularly don't want any current on the grid. That is why for 2V
input I bias at 2.4V. The battery thus is never called upon to deliver
current. In practice a truly minuscule amount of current flows, but it
is far, far less than the trickle charge available. Thus a battery in
the cathode circuit is said to have shelf life.

One more thing. It is precisely for the heavy current tubes that
battery bias is so very attractive, because it is current flowing that
generates the noise in the resistor you would otherwise use, and it is
the charge-recharge cycle of the capacitor you would otherwise use to
bypass the resistor which generates the noise. So the battery does away
with the heavy current reversal in the cap because the cap is not
required, and purely as a resistance the battery is very, very low
compared to the resistor you would otherwise use (in my example the
resistor would have to be 120 ohm to replace the battery).

HTH.

Andre Jute
Visit Jute on Amps at http://members.lycos.co.uk/fiultra/
"an unbelievably comprehensive web site" -- Hi-Fi News & Record Review

Andre Jute wrote:
Wessel Dirksen wrote:
Andre Jute schreef:

Seems to me, my dear fellow, that your letter illustrates either that a
Marxist can make a meal out of very little or that you too can be an
"ultrafi fetishist". Those self-charging advantages of the battery in
the cathode circuit you are so keen to dismiss are huge benefits for no
cost. Your better ultrafidelista is not an audiophool. He is imbued
with the conviction that simplicity means better sound. The battery in
the cathode is the simplest safe implementation, requiring no other
components.

Once more, the three circuits considered in these discussions:

1. Battery in cathode circuit. Grid leak is taken care of by leg of
attenuator to ground. This is the common version. The entire circuit is
attenuator, tube, battery and a couple of pieces of wire and maybe a
grid stopper resistor.

A quick newbie question about this circuit #1 if I may:

You may, always. Whether anyone will know the answer is another
matter...

This looks interesting and I would like to try it but I don't know how
for sure how to calculate/implement this. Is the DC bias voltage of the
batteries all there is to it in determining the idle current through
the tube?

Yes. The battery fixes negative grid bias, in my example -2.4V.
Negative grid bias, together with your choice of plate voltage, in turn
fixes current drawn at quiescence (in a single-ended stage). Current
and negative grid bias together define the point through which the
transfer function must operate. The load you put on the plate then
defines the slope of the transfer function, which we commonly call the
loadline, through that point where quiescent plate voltage, current and
negative grid bias intersect.

The easiest way to understand these things is always to look at plate
voltage/current/negative grid bias Eb-Ia-Eg transfer curves.

Put a ruler across the curves at the angle of the plate load -RL =
-Eg/Ia. RL is traditionally between 2 and 4 times the plate load Rp,
but modern ultrafidelista design choses -RL somewhere between 4 and 8
times RP for the extra silence, eg. distance from the curves at the
bottom of the -Eg/Ia relationship.

Draw a vertical line from the Eb line at the desired plate voltage.
(Just for the sake of completeness, this line is angled very slightly
by the AC of the signal on top of the DC negative grid bias but in most
conservative designs you can forget this and draw it vertically.) Slide
your ruler, at the angle of -RL, up and down this plate voltage line
until you see an equidistant swing across the negative bias lines to
each side of the central point, or nearest. The idea is that you want
to cross the negatively grid bias lines where they are equidistant or
nearest to it. Low current traditional design down in the nasty curves
at the bottom of the transfer function is not for us. The point where
the vertical power supply line crosses the angled loadline and a
negative grid bias line (which you might have to interpolate) is the
quiescent operating point of an SE stage.

You can now choose a cathode resistor to drop the voltage over the grid
by the amount of negative bias you have determined graphically,
R=-Eg/Ia.

With battery bias you perform the same procedure and then juggle the
numbers until you find a compromise that suits available batteries. Or
you can of course add a multiturn minipot to adjust the voltage from
the batteries, though the most refined designs do without it.
(Ultrafidelista design considers simplicity the highest refinement,
therefore the lowest parts count to do the job. Contrast with
professional engineering design which considers that refinement
requires designing the job perfectly, regardless of parts count, and
then cutting it back for reasons of cost.)

For instance, one wants to operate a 417A at 20mA or over to bring it
alive, and you want to input the maximum 2V straight out of a CD to get
the max swing so you can use only one voltage multiplication stage
rather than two, plus you need about 0.6V distance from the 0V negative
grid line because the tube doesn't wait until 0V to start drawing grid
current, plus you know from experience that you want to operate the
tube at say 180V (don't worry about this being max, just don't go over
it) and 20+mA and -2.6V but not much higher (study the curves falling
over to the right of the point defined by these numbers to discover why
not).

At this point you study the available batteries and what other people
have done. Lucas Cant in a famous, much copied potato amp, used a 2V D
cell which would permit an input of only about 1V--but he was dealing
with a far outie wing of the ultrafidelista who have huge, huge horns
(and no love-life). My own wing of the ultrafidelista, who love women
as much as music, have horns with more modest footprints which need the
extra swing, so we use two AA or AAA batteries to give 2.4V.

At this point you have two fixed numbers, the 2.4V out of the batteries
(not 2x the nominal 1.5V -- measure twice and build once!), and a
current draw of 20mA or more. Slide your ruler, still standing in for
-RL, some more along the 2.4V line you have interpolated on the curves,
find Ia 20mA and -Eg 2.4V and there you can drop a line to 175V Eb. I
could have gone a little higher in current and plate voltage for the
same -2,4V bias but in a circuit to be published for amateurs I didn't
want to hog the tube out to its whole official limit (even though I
know from experience that running it over max official dissipation
doesn't degrade its longevity noticeably--I've never had one break and
I have several with 10K+ hours).

Or does the value of Ra (in the absence of setting a specific
value for Rk) need to be calculated differently than a normal bypassed
grounded cathode circuit?

Rp, the particular plate resistance for any implementation, can be
calculated by drawing a line tangential to the -Eg curve through the
quiescent operating point. It won't make much difference if you accept
the published median number in the spec sheet as the gospel.

RL, the plate load, you decide. The higher RL is, the flatter the
transfer curve and the less the distortion in the stage, all other
things being equal. That is why I load my 300B power stages with a
power-choking 5K6, for the silence. That is why I load the 417A we've
been using as an example with 10K, which is 5.68 times the specific Rp
at the chosen operating point.

Once you've decided on your battery and operating point, you can check
the impedance of your design and discover that you have indeed chosen a
fixed Rk! 2.4V/0.020A equals a notional 120 ohm, which is a pretty
common choice for a resistor in an autobias 417A stage. See, a battery
for bias isn't such an extraordinary choice: it is just an ultra-silent
120 ohm replacement for a resistor and a cap!

Andre,
I just a bought a bunch of different AA battery holders (from 1 to 4
batteries) and will have a go at this.

Much appreciated,

Wessel