Re: NFB101 Part 3

Re: NFB101 Part 3



"Iain Churches" <IainNG@kolumbus.fi> wrote in message news:...
>
>
> "Patrick Turner" <info@turneraudio . com .au> wrote in message
> news:47FDDE34.F381F389@turneraudio . com .au...
>>
>>
>> Iain Churches wrote:
>>> But what does he do if he uses (as I do) a mu-follower
>>> for the input stage. There is no anode load.
>>
>> The Jolida 502 amp has a SRPP input stage using a 12AX7 with 470 ohms Rk
>> for the
>> bottom triode and 1k for the top triode.
>>

So, let's consider a simple mu-follower, with, say a 22k anode load
for the bottom triode, with 1k2 above it as the cathode resistor for the
top triode. Would I be right in thinking that one could start with a
2k2 and say 1nF (in series) across this 22k anode load.

Would this also have an effect upon the output from the low Z
output above it from the 1k2 cathode resistor?

Best regards
Iain

Re: NFB101 Part 3



Iain Churches wrote:
>
> "Iain Churches" <IainNG@kolumbus.fi> wrote in message news:...
> >
> >
> > "Patrick Turner" <info@turneraudio . com .au> wrote in message
> > news:47FDDE34.F381F389@turneraudio . com .au...
> >>
> >>
> >> Iain Churches wrote:
> >>> But what does he do if he uses (as I do) a mu-follower
> >>> for the input stage. There is no anode load.
> >>
> >> The Jolida 502 amp has a SRPP input stage using a 12AX7 with 470 ohms Rk
> >> for the
> >> bottom triode and 1k for the top triode.
> >>
>
> So, let's consider a simple mu-follower, with, say a 22k anode load
> for the bottom triode, with 1k2 above it as the cathode resistor for the
> top triode. Would I be right in thinking that one could start with a
> 2k2 and say 1nF (in series) across this 22k anode load.

No. All wrong this time.

You must specify the WHOLE schematic under consideration to avoid me
wasting time commenting.

But let's assume you have a true µ-follower with 6SN7 or 6CG7 hooked up
as follows :-

bottom Rk = 1k2, fully bypassed, but with 50 ohms unbypassed for
acceptance of NFB from an OPT secondary.

bottom anode has 22k RL in series with 1k5 then connecting to top triode
cathode.
top triode grid has 1M bias R to bottomk of 1k5.
top triode grid has cap of 0.47uF to bottom anode.

Let's say the load of this stage is connected to the top cathode, and is
220k at 1kHz.

See how you should have specified your µ-follower?

Please don't leave anything for us to assume, because everyone will get
muddled.

Now, the top triode has OLG about 18.
It drives a 220k load.
So if the output voltage is 18Vrms, grid signal is 19V.

1V appears across the 22k + 1.5k = 23.5k,
and the load current in the botom triode is 1V / 23.5k.
Neglect the tiny loading effect of the 1M biasing R.

The load on the bottom triode = 19V x 23.5 / 1V.

In other words, load on the bottom triode = (A + 1) x R between triodes
where this R > 5 x bias R.

So bottom tube has a load of roughly 400k, or well in excess of 20Ra,
so its gain max is approaching µ of about 20.
Expect gain to be approx 19.

Because there is only 50 ohms of Rk of the bottom triode
unbypassed, negelect its effects on Ra of the bottom triode.
At say 5mA, Ra would be about 10k.

The place to connect the Zobel is NOT across the 22k or 1k5.

This would not have much effect.

But the right place for the Zobel is from the lower triode anode to 0V.

When the Zobel of 2k2 and 0.001uF as you suggest is connected from anode
to 0V, the bottom tube gain will be down -3dB when Ra in parallel with
its active load
of 400k is equal to the reactance of the 0.001uF.

The 400k is so high it can be neglected.

So at what F does ZC = 10k?

F = 159,000 / ( 0.001 x 10,000 ) = 1.59kHz.

The presence of the 2k2 has almost zero effect.

Expect to see the OLG have a -3dB pole at around 1.6kHz.

But as F rises, Ra will be loaded ultimately at some very high F when
0.001uF has very low Z
with only 2k2 (in parallel with 400k which we can neglect,) so with only
2k2, gain = 3.5 approx.
The pole at some HF where gain is 3dB above 3.5, or 5, is where ZC =
2k2,
F = 159,000 / ( 2,200 x 0.001 ) = 75kHz approximately.
Using a calculator will give better results than my mental arithmatic.

So the response of the stage will be A = 20 up to about 300Hz, then
gain will turn down 3dB at 1.6kHz, and fall at nearly 6dB per octave
until 50kHz when the rate of gain decline
begins to level out to 5 at 75kHz, and finally by 200kHz, A = 3.5.
The shelving of gain has been achieved. The gain reduction is 20/3.5 =
approx 6 = approx -13.5dB

Such a drastic Zobel shelf would not be practical in 90% of cases, and C
should
rarely ever be more than 330pF, and R less than 3k3.

>
> Would this also have an effect upon the output from the low Z
> output above it from the 1k2 cathode resistor?

???

See above

Meanwhile, if you wanted to also place a LF gain stepping network
for the µ-follower input stage then an entirely different approach
is required, and a story for later.

But I never use a µ-follower or SRPP for the input or driver stage of
any power amp because of the
complexity, and because its entirely unecessary.
There isn't any need to have such a low Rout from the driver stages.

The drive of the top triode by RC coupling from the bottom
introduces yet another series tube and RC coupling time constant to deal
with.

Where you have a 6SN7 or 6CG7 as V1 in amy power amp, its better to
parallel the two triodes
then load them with a TRANSISTOR CCS, MJE350 is fine, and the whole mess
is SO MUCH
easier to deal with and to provide gain stepping.
The CCS dc supply will make the input tube have THD and IMD as low as
µ-follower
and sound will be fabulous. Speed and zest, dynamics will all be
present.
There does not need to be a follower in there at all.

The other alternative for a PP amp is to have the input stage as an LTP
with CCS common cathode tail, and have input to one side,
and FB to the other, and then have the usual LTP driver stage but with a
an R tail of say 10k to
say -150Vdc. The balance will be within 1% without balance adjust pots.
THD and IMD in the driver stage will be extremely low.

Mission accomplished.

Patrick Turner.


>
> Best regards
> Iain

Re: NFB101 Part 3


This post has my math errors conained in the original reply
corrected. Pleas ignore the previous post of mine about an hour ago...

I hav re-composed the reply......

Patrick Turner wrote:
>
> Iain Churches wrote:
> >
> > "Iain Churches" <IainNG@kolumbus.fi> wrote in message news:...
> > >
> > >
> > > "Patrick Turner" <info@turneraudio . com .au> wrote in message
> > > news:47FDDE34.F381F389@turneraudio . com .au...
> > >>
> > >>
> > >> Iain Churches wrote:
> > >>> But what does he do if he uses (as I do) a mu-follower
> > >>> for the input stage. There is no anode load.
> > >>
> > >> The Jolida 502 amp has a SRPP input stage using a 12AX7 with 470 ohms Rk
> > >> for the
> > >> bottom triode and 1k for the top triode.
> > >>
> >
> > So, let's consider a simple mu-follower, with, say a 22k anode load
> > for the bottom triode, with 1k2 above it as the cathode resistor for the
> > top triode. Would I be right in thinking that one could start with a
> > 2k2 and say 1nF (in series) across this 22k anode load.
>
> No. All wrong this time.
>
> You must specify the WHOLE schematic under consideration to avoid me
> wasting time commenting.
>
> But let's assume you have a true µ-follower with 6SN7 or 6CG7 hooked up
> as follows :-
>
> bottom Rk = 1k2, fully bypassed, but with 50 ohms unbypassed for
> acceptance of NFB from an OPT secondary.
>
> bottom anode has 22k RL in series with 1k5 then connecting to top triode
> cathode.
> top triode grid has 1M bias R to bottomk of 1k5.
> top triode grid has cap of 0.47uF to bottom anode.
>
> Let's say the load of this stage is connected to the top cathode, and is
> 220k at 1kHz.
>
> See how you should have specified your µ-follower?
>
> Please don't leave anything for us to assume, because everyone will get
> muddled.
>
> Now, the top triode has OLG about 18.
> It drives a 220k load.
> So if the output voltage is 18Vrms, grid signal is 19V.
>
> 1V appears across the 22k + 1.5k = 23.5k,
> and the load current in the botom triode is 1V / 23.5k.
> Neglect the tiny loading effect of the 1M biasing R.
>
> The load on the bottom triode = 19V x 23.5 / 1V.
>
> In other words, load on the bottom triode = (A + 1) x R between triodes
> where this R > 5 x bias R.
>
> So bottom tube has a load of roughly 400k, or well in excess of 20Ra,
> so its gain max is approaching µ of about 20.
> Expect gain to be approx 19.
>
> Because there is only 50 ohms of Rk of the bottom triode
> unbypassed, negelect its effects on Ra of the bottom triode.
> At say 5mA, Ra would be about 10k.
>
> The place to connect the Zobel is NOT across the 22k or 1k5.
>
> This would not have much effect.
>
> But the right place for the Zobel is from the lower triode anode to 0V.
>
> When the Zobel of 2k2 and 0.001uF as you suggest is connected from anode
> to 0V, the bottom tube gain will be down -3dB when Ra in parallel with
> its active load
> of 400k is equal to the reactance of the 0.001uF.
>
> The 400k is so high it can be neglected.
>
> So at what F does ZC = 10k?
>
F = 159,000 / ( 0.001 x 10,000 ) = 15.9kHz.

The presence of the 2k2 has almost zero effect on the first of TWO poles
to consider.

Expect to see the OLG have a -3dB pole at around 16kHz.

But as F rises, Ra will be loaded ultimately at some very high F when
0.001uF has very low Z
with only 2k2 (in parallel with 400k which we can neglect,) so with only
2k2, gain = 3.6 approx.
The pole at some HF where gain is 3dB above 3.6, or 5, is where ZC =
2k2,
F = 159,000 / ( 2,200 x 0.001 ) = 72kHz approximately.

>
> So the response of the stage will be A = 19.5 up to about 8kHz, then
> gain will turn down 3dB by 16kHz, and fall at nearly 6dB per octave
> until 50kHz when the rate of gain decline
> begins to level out to A = 5 at 72kHz, and finally by 200kHz, A = 3.6.
> The shelving of gain has been achieved. The gain reduction is 19.5/3.5 =
> approx 5.4x = approx -14.6dB

This will appear to be quite OK for some amps with pretty awful OPT.

A typical shelf used with a good OPT might only have a 6dB shelf and F1
pole
at 30kHz.
With a good OPT, and not much global NFB, the Zobel is often not needed.

But with Ra of the bottom tube much higher, say for 12AX7, the Zobel
would give a
shelf far too drastic and would not be practical in 90% of cases, and C
should rarely ever be more than 330pF, and R rarely less than 3k3.

The actual R&C vales MUST be trialled with a range of pure C loads,
and yet pure R loads should still give 65kHz of BW.
>
> >
> > Would this also have an effect upon the output from the low Z
> > output above it from the 1k2 cathode resistor?
>
> ???
>
> See above
>
> Meanwhile, if you wanted to also place a LF gain stepping network
> for the µ-follower input stage then an entirely different approach
> is required, and a story for later.
>
> But I never use a µ-follower or SRPP for the input or driver stage of
> any power amp because of the
> complexity, and because its entirely unecessary.
> There isn't any need to have such a low Rout from the driver stages.
>
> The drive of the top triode by RC coupling from the bottom
> introduces yet another series tube and RC coupling time constant to deal
> with.
>
> Where you have a 6SN7 or 6CG7 as V1 in amy power amp, its better to
> parallel the two triodes
> then load them with a TRANSISTOR CCS, MJE350 is fine, and the whole mess
> is SO MUCH
> easier to deal with and to provide gain stepping.
> The CCS dc supply will make the input tube have THD and IMD as low as
> µ-follower
> and sound will be fabulous. Speed and zest, dynamics will all be
> present.
> There does not need to be a follower in there at all.
>
> The other alternative for a PP amp is to have the input stage as an LTP
> with CCS common cathode tail, and have input to one side,
> and FB to the other, and then have the usual LTP driver stage but with a
> an R tail of say 10k to
> say -150Vdc. The balance will be within 1% without balance adjust pots.
> THD and IMD in the driver stage will be extremely low.
>
> Mission accomplished.
>
> Patrick Turner.
>
> >
> > Best regards
> > Iain

Re: NFB101 Part 3



"Patrick Turner" <info@turneraudio . com .au> wrote in message
news:47FF42F3.5B7B3601@turneraudio . com .au...
>
> This post has my math errors conained in the original reply
> corrected. Pleas ignore the previous post of mine about an hour ago...
>
>> > So, let's consider a simple mu-follower, with, say a 22k anode load
>> > for the bottom triode, with 1k2 above it as the cathode resistor for
>> > the
>> > top triode. Would I be right in thinking that one could start with a
>> > 2k2 and say 1nF (in series) across this 22k anode load.
>>
>> No. All wrong this time.
>>
>> You must specify the WHOLE schematic under consideration to avoid me
>> wasting time commenting.

'Umble apologies:-( Your interpretation of what I wrote was
100% correct. But to ease my conscience, I have sketched the
schematic at:

* w w w .kolumbus.fi/iain.churches/Pics/Schematics/MuFollower.png
>>

>> Now, the top triode has OLG about 18.
>> It drives a 220k load.
>> So if the output voltage is 18Vrms, grid signal is 19V.
>>
>> 1V appears across the 22k + 1.5k = 23.5k,
>> and the load current in the botom triode is 1V / 23.5k.
>> Neglect the tiny loading effect of the 1M biasing R.
>>
>> The load on the bottom triode = 19V x 23.5 / 1V.
>>
>> In other words, load on the bottom triode = (A + 1) x R between triodes
>> where this R > 5 x bias R.
>>
>> So bottom tube has a load of roughly 400k, or well in excess of 20Ra,
>> so its gain max is approaching µ of about 20.
>> Expect gain to be approx 19.

Yes. The gain is 24dB (x 17)
>>
>> Because there is only 50 ohms of Rk of the bottom triode
>> unbypassed, negelect its effects on Ra of the bottom triode.
>> At say 5mA, Ra would be about 10k.
>>
>> The place to connect the Zobel is NOT across the 22k or 1k5.
>>
>> This would not have much effect.
>>
>> But the right place for the Zobel is from the lower triode anode to 0V.


Ahaa!! (he says:-) Now I see!
>>
>> When the Zobel of 2k2 and 0.001uF as you suggest is connected from anode
>> to 0V, the bottom tube gain will be down -3dB when Ra in parallel with
>> its active load
>> of 400k is equal to the reactance of the 0.001uF.

My 2k2 and 1nF were clearly not good choices. I did not
realise that the Zobel had to go from anode 1 to ground.


>>
>> The 400k is so high it can be neglected.
>>
>> So at what F does ZC = 10k?

Sorry. Where does this 10k come from?


>>
> F = 159,000 / ( 0.001 x 10,000 ) = 15.9kHz.
>
> The presence of the 2k2 has almost zero effect on the first of TWO poles
> to consider.
>
> Expect to see the OLG have a -3dB pole at around 16kHz.
>
> But as F rises, Ra will be loaded ultimately at some very high F when
> 0.001uF has very low Z
> with only 2k2 (in parallel with 400k which we can neglect,) so with only
> 2k2, gain = 3.6 approx.
> The pole at some HF where gain is 3dB above 3.6, or 5, is where ZC =
> 2k2,
> F = 159,000 / ( 2,200 x 0.001 ) = 72kHz approximately.
>
>>
>> So the response of the stage will be A = 19.5 up to about 8kHz, then
>> gain will turn down 3dB by 16kHz, and fall at nearly 6dB per octave
>> until 50kHz when the rate of gain decline
>> begins to level out to A = 5 at 72kHz, and finally by 200kHz, A = 3.6.
>> The shelving of gain has been achieved. The gain reduction is 19.5/3.5 =
>> approx 5.4x = approx -14.6dB
>
> This will appear to be quite OK for some amps with pretty awful OPT.
>
> A typical shelf used with a good OPT might only have a 6dB shelf and F1
> pole
> at 30kHz.
> With a good OPT, and not much global NFB, the Zobel is often not needed.
>
> But with Ra of the bottom tube much higher, say for 12AX7, the Zobel
> would give a
> shelf far too drastic and would not be practical in 90% of cases, and C
> should rarely ever be more than 330pF, and R rarely less than 3k3.

OK. These are good "starting point" values.
>
>> Meanwhile, if you wanted to also place a LF gain stepping network
>> for the µ-follower input stage then an entirely different approach
>> is required, and a story for later.

Good. I am hoping that Ian's excellent NFB thread will lead on to
a Stability thread. From talking to other tube amp builders (and also
from personal experience:-)) I know this to be a *huge* stumbling
block. I have also learned that a little common sense, plus intelligent
use of a decade R and decade C box can save an awful of time
with a calculator.
>>
>> But I never use a µ-follower or SRPP for the input or driver stage of
>> any power amp because of the
>> complexity, and because its entirely unecessary.
>> There isn't any need to have such a low Rout from the driver stages.

I like the Mu follower because it is a little different to run-of-the-mill
input stages. It also performs very well indeed. You can use the
Hi Z output with a smaller coupling cap if you like.

>>
>> The drive of the top triode by RC coupling from the bottom
>> introduces yet another series tube and RC coupling time constant to deal
>> with.

Understood.
>>
>> Where you have a 6SN7 or 6CG7 as V1 in amy power amp, its better to
>> parallel the two triodes
>> then load them with a TRANSISTOR CCS, MJE350 is fine, and the whole mess
>> is SO MUCH
>> easier to deal with and to provide gain stepping.

That's deep and murky water for this country boy. I have no
experience with transistor CCS.



>> The CCS dc supply will make the input tube have THD and IMD as low as
>> µ-follower
>> and sound will be fabulous. Speed and zest, dynamics will all be
>> present.
>> There does not need to be a follower in there at all.


I am sure you have a sketch of such a circuit on your website.
Can you point me to it, please?

Thanks for a very informative post.
Best regards
Iain

Re: NFB101 Part 3



Iain Churches wrote:
>
> "Patrick Turner" <info@turneraudio . com .au> wrote in message
> news:47FF42F3.5B7B3601@turneraudio . com .au...
> >
> > This post has my math errors conained in the original reply
> > corrected. Pleas ignore the previous post of mine about an hour ago...
> >
> >> > So, let's consider a simple mu-follower, with, say a 22k anode load
> >> > for the bottom triode, with 1k2 above it as the cathode resistor for
> >> > the
> >> > top triode. Would I be right in thinking that one could start with a
> >> > 2k2 and say 1nF (in series) across this 22k anode load.
> >>
> >> No. All wrong this time.
> >>
> >> You must specify the WHOLE schematic under consideration to avoid me
> >> wasting time commenting.
>
> 'Umble apologies:-( Your interpretation of what I wrote was
> 100% correct. But to ease my conscience, I have sketched the
> schematic at:
>
> * w w w .kolumbus.fi/iain.churches/Pics/Schematics/MuFollower.png

This sketch is virtually the same as I postulate below, so all of what I
say below applies.

> >>
>
> >> Now, the top triode has OLG about 18.
> >> It drives a 220k load.
> >> So if the output voltage is 18Vrms, grid signal is 19V.
> >>
> >> 1V appears across the 22k + 1.5k = 23.5k,
> >> and the load current in the botom triode is 1V / 23.5k.
> >> Neglect the tiny loading effect of the 1M biasing R.
> >>
> >> The load on the bottom triode = 19V x 23.5 / 1V.
> >>
> >> In other words, load on the bottom triode = (A + 1) x R between triodes
> >> where this R > 5 x bias R.
> >>
> >> So bottom tube has a load of roughly 400k, or well in excess of 20Ra,
> >> so its gain max is approaching µ of about 20.
> >> Expect gain to be approx 19.
>
> Yes. The gain is 24dB (x 17)
> >>
> >> Because there is only 50 ohms of Rk of the bottom triode
> >> unbypassed, negelect its effects on Ra of the bottom triode.
> >> At say 5mA, Ra would be about 10k.
> >>
> >> The place to connect the Zobel is NOT across the 22k or 1k5.
> >>
> >> This would not have much effect.
> >>
> >> But the right place for the Zobel is from the lower triode anode to 0V.
>
> Ahaa!! (he says:-) Now I see!
> >>
> >> When the Zobel of 2k2 and 0.001uF as you suggest is connected from anode
> >> to 0V, the bottom tube gain will be down -3dB when Ra in parallel with
> >> its active load
> >> of 400k is equal to the reactance of the 0.001uF.
>
> My 2k2 and 1nF were clearly not good choices. I did not
> realise that the Zobel had to go from anode 1 to ground.

Read on, your guess wasn't too bad.....
>
> >>
> >> The 400k is so high it can be neglected.
> >>
> >> So at what F does ZC = 10k?
>
> Sorry. Where does this 10k come from?

10k is the Ra of the bottom triode.

Rememeber the bottom triode is equal to a voltage generator
putting out a voltage = µ x Vg but with 10k in series with its output
and the anode terminal.
>
> >>
> > F = 159,000 / ( 0.001 x 10,000 ) = 15.9kHz.
> >
> > The presence of the 2k2 has almost zero effect on the first of TWO poles
> > to consider.
> >
> > Expect to see the OLG have a -3dB pole at around 16kHz.
> >
> > But as F rises, Ra will be loaded ultimately at some very high F when
> > 0.001uF has very low Z
> > with only 2k2 (in parallel with 400k which we can neglect,) so with only
> > 2k2, gain = 3.6 approx.
> > The pole at some HF where gain is 3dB above 3.6, or 5, is where ZC =
> > 2k2,
> > F = 159,000 / ( 2,200 x 0.001 ) = 72kHz approximately.
> >
> >>
> >> So the response of the stage will be A = 19.5 up to about 8kHz, then
> >> gain will turn down 3dB by 16kHz, and fall at nearly 6dB per octave
> >> until 50kHz when the rate of gain decline
> >> begins to level out to A = 5 at 72kHz, and finally by 200kHz, A = 3.6.
> >> The shelving of gain has been achieved. The gain reduction is 19.5/3.5 =
> >> approx 5.4x = approx -14.6dB
> >
> > This will appear to be quite OK for some amps with pretty awful OPT.
> >
> > A typical shelf used with a good OPT might only have a 6dB shelf and F1
> > pole
> > at 30kHz.
> > With a good OPT, and not much global NFB, the Zobel is often not needed.
> >
> > But with Ra of the bottom tube much higher, say for 12AX7, the Zobel
> > would give a
> > shelf far too drastic and would not be practical in 90% of cases, and C
> > should rarely ever be more than 330pF, and R rarely less than 3k3.
>
> OK. These are good "starting point" values.

To do this quickly, use a double ganged tuning cap from an AM radio
in series with a pot of about 25k. A dual 50k linear pot paralleled is
OK.
The cap range will be from 40pF to 720pF with both sides paralleled.

Solder the pot onto the tuning gang and place 0.1uF dc blocking cap so
that if the
cap plates ever touch it won't allow dc flow that may upset what you are
doing
or ruin the pot.

Please think about everything while doing all things.

> >
> >> Meanwhile, if you wanted to also place a LF gain stepping network
> >> for the µ-follower input stage then an entirely different approach
> >> is required, and a story for later.
>
> Good. I am hoping that Ian's excellent NFB thread will lead on to
> a Stability thread. From talking to other tube amp builders (and also
> from personal experience:-)) I know this to be a *huge* stumbling
> block. I have also learned that a little common sense, plus intelligent
> use of a decade R and decade C box can save an awful of time
> with a calculator.

The LF gain shelving network is typically 0.047uF with 1M is parallel
placed in series after 0.47uF and before 220k to 0V which is the bias R
for the following
gain stage. See my website for examples.
The 0.47 + 0.047 + 1M + 220k works in 95% of amps, and can be used
without
any testing or calculations to vastly improve the LF stability and
overload
recovery behaviour.

Unlike the HF gain stepping network, the LF network is placed AFTER the
LOW Rout terminal
of the µ-follower, ie, at the cathode of the top triode. There are ways
to place the
LF network with 2 caps and two R between bottom anode and top tube grid
if
you have the 220k taken to a fixed bias point for the top triode to
operate from.
But at above 50Hz, the bottom triode is loaded by 220k and this robs
some gain.
So easier/better to place the LF shelver after the follower.


> >>
> >> But I never use a µ-follower or SRPP for the input or driver stage of
> >> any power amp because of the
> >> complexity, and because its entirely unecessary.
> >> There isn't any need to have such a low Rout from the driver stages.
>
> I like the Mu follower because it is a little different to run-of-the-mill
> input stages. It also performs very well indeed. You can use the
> Hi Z output with a smaller coupling cap if you like.
>
> >>
> >> The drive of the top triode by RC coupling from the bottom
> >> introduces yet another series tube and RC coupling time constant to deal
> >> with.
>
> Understood.
> >>
> >> Where you have a 6SN7 or 6CG7 as V1 in amy power amp, its better to
> >> parallel the two triodes
> >> then load them with a TRANSISTOR CCS, MJE350 is fine, and the whole mess
> >> is SO MUCH
> >> easier to deal with and to provide gain stepping.
>
> That's deep and murky water for this country boy. I have no
> experience with transistor CCS.

Try a CCS here and there in your tube circuits.
Your swamp wading boots will remain waterproof.

Your ears will be delighted by transistors used as slaves to the triodes
which really sound their best when as little current change as possible
occurs within them.

Triodes are really good for linear voltage changes.
>
> >> The CCS dc supply will make the input tube have THD and IMD as low as
> >> µ-follower
> >> and sound will be fabulous. Speed and zest, dynamics will all be
> >> present.
> >> There does not need to be a follower in there at all.
>
> I am sure you have a sketch of such a circuit on your website.
> Can you point me to it, please?

See the schematic for the power amp input stage, V2 on an SE amp at

* w w w .turneraudio . com .au/se35cfbmonobloc.html

In an SE amp, where there is a driver stage and following output stage,
the 2H of an input stage adds to the 2H of the output stage.
So if 2H at the input is minimised, there is less at the output.

The CCS dc supply allws the B+ of the input stage to be fairly low,
but supply current to the input tube can be high, and much higher than
if an R were used instead as most ppl do because they don't think much.
The total ac load on the paralleled triodes is the following bias R14,
220k.
So each 1/2 of the 6CG7 is loaded by 440k, a higher value load than if
it was in a µ-follower.
Linearity is thus better than the µ-follower, and you ain't got the
extra triode in series
with the signal where none is needed. Rout from V2 is about 5k, plenty
low enough.

In another sample from my site I have CCS for a preamp in anode and
cathode circuits....

* w w w .turneraudio . com .au/line-preamp-2003.html

Sounds like magic.

Patrick Turner.
>
> Thanks for a very informative post.
> Best regards
> Iain