NFB loop question

DVD-Software.info
Your one stop source for DVD Software

Profile

NFB loop question Goto page 1, 2, 3 Next

DVD-Software.info Forum Index -> Tubes

Author

Message

Tom Schlangen
Guest

Posted: Fri Dec 17, 2004 6:43 pm Post subject: NFB loop question

Gentlemen, consider a simple two staged SE amp, w/o gNFB. As we know, it might even be desirable to get a certain amount of 2H from the driver to cancel out some of the 2H produced by the output stage. Now consider a global NFB loop (voltage feedback) added around the stages. Do we still want the driver to produce a certain amount of 2H, or should the driver now act as clean as possible? Tom -- When in doubt, use brute force. - Ken Thompson

Patrick Turner
Guest

Posted: Fri Dec 17, 2004 9:19 pm Post subject: Re: NFB loop question

Tom Schlangen wrote:

Quote:

Gentlemen,

consider a simple two staged SE amp, w/o gNFB. As we know, it
might even be desirable to get a certain amount of 2H from the
driver to cancel out some of the 2H produced by the output
stage.

Now consider a global NFB loop (voltage feedback) added
around the stages.

Do we still want the driver to produce a certain amount of 2H,
or should the driver now act as clean as possible?

Firstly, the setting up of an input and driver tube to produce enough
thd to cancel the thd of the output tube is, imho, very bad practice.
It nearly always involves trying to make the driver tube distort more
than
it ever should have to, and at a lower voltage than it should.
So suppose we had an output tube which makes 5% at full po, and it takes

40 vrms to drive it, then to get good cancelation, 5% thd from the
driver
at 40 vrms is needed, and the 3H from both tubes will be additive, and
not insignificant.
Not only that, the rate of increase of 2H production for both tubes
won't be the same,
so there is usually a limited amount of 2H cancelation.

However, say the opv has 2% at 1/2 the full output voltage, or 1/4 full
po,
and then we chose a tube like a 12AU7 instead of a more linear 6SN7, and

just have the 12AU7 set up fairly optimally, and we manage to get say
1.5% 2H at 1/4 po, then we'll get a fairly useful amount of cancelation.

There will be harmonic distortion of the 2H fed into the opv, but
2% of 1.5% is only 0.03% of 4H.

Some 3H will be produced by the 2H fed into the output due to IMD, and
it all gets complicated very
easily, but at least we won't have too much 2H.

At the end of the day we might be doing rather well with "the amp that
doesn't try to cancel too much"
if we can get the 2H reduced from 5% to 2.5% at full po.
Then when we add say 12 dB of NFB we should see the 2H decline
to about 0.65%, and around 0.25 % at 1/4 full po, which may suit our
needs.
All the other products will be reduced although perhaps the 3H will not
be reduced by 12 dB.
When viewing each of the harmonics as the gain is cranked from 0.0 watts
and up,
the usual scene in SE amps is always predominantly 2H but with minor 3H,
sometimes
more measurable at low level if there is any input iron, and then mainly
2H,
and then as the gain is cranked more 5, 6, 7, 8, 9H becomes increasingly
measurable.

This "shark's fin profile of the thd products occurs in all amps and the
fin blade broadens to
include more products as the amp runs out of puff and clips, where a
huge increase begins, and
NFB has no useful effect on thd after clipping.

Now if you have perfectly linear driver tube, there would be more thd
for the same applied NFB,
but its spectra is likely to be slightly less complex.
I think we'd be splitting hairs to say cancelation + little NFB was
better than
no cancelation + moderate NFB.

The latter method would give a lower Ro of the amp.

Stability still has to be trimmed, and its surprising how little NFB
will require
considerable RC gain/phase shift tailoring to make the amp stable
at all types of loads.

IN a pair of my latest SE amps with 4 x 6CA7, I have CFB windings in the

output stage to give 12.5% CFB which is about 8 dB of applied NFB.
The thd is about 2% at 35 watts, and R about 1.5 ohms.
with the rated load = 5 ohms, and a linear driver tube, and no other
NFB.

I settled for a 12BH7 driver triode, cathode unbypassed, easily loaded,
to get a fair amount of cancelation at a few watts.
I tried 12AU7, 6CG7, and these gave good cancelation but away from the
range of load values I was aiming
for, namely between 3 and 6 ohms.
I settled for the 12BH7, and its other character appealed, low Ra,
very rugged, and easily able to drive the low input C of the 4 x 6CA7.
It only has to make 46 vrms of drive, and I got a 20 dB reduction in 2H
at 5 ohms
due to 2H cancelation, but declining cancelations away from 5 ohms, bit
overall
it was like a conventional SEUL with this big bite taken out of the thd
profiles of the thd
of various load values
With the 8 dB of global NFB added, the thd was down to 0.03%
when normally it'd be about 0.3% at 5 watts into 5 ohms.

Normal SEUL has the same high amount of thd as pentode but the odd order

products are much reduced and to a very similar spectral content of
triodes.
So the benefit of SEUL is to change the thd nature, yet keep the power
ceiling
of the pentode.
Then the global NFB is used to get the Ro down to where its wanted.

After taking many measurements, I figured the thd of the SE35 CFB amps
is probably lower than any
other SE amps of 35 watts for this amount of applied NFB and the thd
was no worse in spectral character than a good AB1 PP amp with the same
amount of NFB.
At elevated power levels ther PP amp of the same po begins to have an
edge,
but rarely does anyone have to make thse SE amps work hard.

I figured that if its OK to rely on the 2H **current cancelations** in a
PP amp then its OK
to rely on some moderate 2H **voltage cancelations** in an SE amp.

the warmth and creaminess of the tubes is still there but the dynamics
and accuracy
is tops. I didn't take long to sell them.

The odd thing is that at high value RLs, the 2H of the output stage
reverses its phase
in some pentodes and tetrodes, and certainly in the 6CA7, even with CFB
applied.
So at over 8 ohms, there is no cancelations, but the 2H becomes
additive.
Fortunately, since the amount of 2H at any power into high value RLs is
low in the output stage
since the amount of applied CFB increases with gain, the additive effect
only
had a slight effect.
Most of the error correction takes place in the output stage.
The global NFB and driver tube tends to then have a fairly gain constant
output stage to work with,
and that presumably is better to make less imd.

I'd say that what you get either way depends on what driver tubes you
have chosen
and what output tubes and op conditions you have.
And if the amount of total applied external NFB is about 12 dB with SE
triodes and you get
less than 0.5% thd, mainly 2H at clip then at 1/10 full po where the amp
may be used the sound should be OK.
It should be fabulous.

Basically, you want your tube amp to operate in its "sweet zone"
This is the power area where 90% of the programme level is heard,
yet when transients and crescendos occur you don't want a halo of fizz
to acompany
these extra efforts.

The guy with the SE35 has speakers of about 90 dB/W/M.
He's happy. He wouldn't be happy with a single 300B.
I have listened with him so I know the levels he uses.

One lone SE 6CA7 can make a very nice 8 watts, and its OK
with 96 dB/W/M speakers.
But you need 4 such tubes for the 90 dB speakers.

I am not one to think one MUST use NFB in an SE amp.
Certainly there is no "must" rule with a 300B amp.

But where tetrodes are employed, one has to use some NFB at least to
get the Ro down.

I don't believe it spoils the the music if its accurately applied.
Wide bandwidth OPTs are essential.

I did once try to use only 6 db of global NFB with a pair of Quad 2
where I had used EL34 in triode.
This seemed at least as good as an original Quad II.

But the sound was never quite as dynamic and engaging as a tiny 10 watt
amp using
6GW8 with plain UL that I'd built earlier for the same client.
I changed the EL34 to KT88, and went to 10 dB of global and the Quad II
revised then
seemed to have it all.

The use of triodes in Quad II reduces the amount of CFB from around 8 db
to about 3 dB,
so with 10 dB global, only 13 dB total is employed.

Ever since that episode, I have been loathe to abandon global NFB.

Then there have been the demos with the audio club, and it was there I
became
firm in my belief that NFB is OK, especially after hearing music from
another guys amps which
had very little NFB, and hearing what ppl had to say as the afternoon
went on.
Nobody else was impressed with the "only 6 dB of NFB" approach.

The SEUL amps with 13E1 didn't have any deliberate attempt at thd
cancellation.
There is an ECC32 driver tube, and it has to make 55 vrms, and its
fairly linear....
Despite measuring an average of 10 dB worse than the latest SE35,
they are well regarded.
The secret is proably that I have given ppl enough power ceiling.
The thd spectra is at its simplest and most benign levels in any class A
amp
with or without NFB when the amp is within its sweet zone of a few
watts.

Patrick Turner.

Quote:

Tom

--
When in doubt, use brute force.
- Ken Thompson

Ruud Broens
Guest

Posted: Fri Dec 17, 2004 10:01 pm Post subject: Re: NFB loop question

"Tom Schlangen" <t.schlangen@ndh.net> wrote in message news:pd4b92-m1i.ln1@julabi.snakes.netcologne.de... : Gentlemen, : : consider a simple two staged SE amp, w/o gNFB. As we know, it : might even be desirable to get a certain amount of 2H from the : driver to cancel out some of the 2H produced by the output : stage. : : Now consider a global NFB loop (voltage feedback) added : around the stages. : : Do we still want the driver to produce a certain amount of 2H, : or should the driver now act as clean as possible? : : Tom : : -- : When in doubt, use brute force. : - Ken Thompson You still want your driver transfer function to be as close as possible to the power stage transfer function as you can get :) Rudy

Tom Schlangen
Guest

Posted: Fri Dec 17, 2004 10:34 pm Post subject: Re: NFB loop question

Hello Patrick,

thank you for your extremely elaborate answer.

The key point for me is the following paragraph:

Quote:

Now if you have perfectly linear driver tube,
there would be more thd for the same applied NFB,
but its spectra is likely to be slightly less complex.
I think we'd be splitting hairs to say cancelation +
little NFB was better than no cancelation + moderate NFB.

The latter method would give a lower Ro of the amp.

Now a bit on the reason why I asked this question...

Quote:

IN a pair of my latest SE amps with 4 x 6CA7, I have CFB
windings in the output stage to give 12.5% CFB which is
about 8 dB of applied NFB.
The thd is about 2% at 35 watts, and R about 1.5 ohms.
with the rated load = 5 ohms, and a linear driver tube,
and no other NFB.

You know I have a similar project going, 2 x 6550 PSE tetrode
mode with 10% CFB. You saw some CRO screen shots, but these
were with ECL82/6BM8 (trioded) as gain-stage/driver. After
plotting the plate curves of trioded ECL82, I decided that
one could do better (with less distortion) with the driver,
and that was the reason I started plotting plate curves of
several other trioded pentodes ... ;-)

After quite some load line drawing, I decided on trioded
6V6 as driver. At 375V B+, Ra = 8k2, 197 volts at plate,
22mA plate current, bias -12V, bias resistor 470 ohms plus
82 ohms (aha - you probably know already where this is
leading to ...), this driver is capable of more than 170 Vpp
output swing at excellent linearity. According to the
graphical method shown at p549 of RDH, expected 2h distortion
would be in the 1% range at that swing. Given that only ca.
1/2 of that swing is needed to drive the 6550s, the expected
distortion of this driver stage would be maybe 0,5% or
something.

Since I was "trioding" other pentodes to measure them anyway,
I thought I might as well put a trioded EF86/6267 in V1 place,
but _outside_ the NFB loop, since this thing, at 300V B+,
Ra = 220k, bias -3V, Rk = 3k9, is so extremely linear (not
only at the output swing of ca. 25 Vpp to drive the following
trioded 6V6, but also at much higher plate swings) that I
doubt one could do _significantly_ better even with an active
CC plate loaded 6SN7 ...

So, at the moment, I have an extremely linear "isolated" gain
stage at the front, followed by a very linear & beefy driver,
followed by the (untouched) output stage, with a voltage NFB
loop from the OPT secondary to the drivers cathode resistor
devider network.

Preliminary schematic (w/o dimensioned R-NFB and C-NFB and
w/o stepped attenuators, I probably will do that this weekend)
at:

http://www.tubes.mynetcologne.de/roehren/6550pset/6550pset_amp_v31.pdf

I expect that this circuit will be quite easy to get it
unconditionally stable due to the NFB loop only being around
two stages. But that wasn't the primary intention behind
the change from ECL82 to EF86+6V6, of course ...

Tom

--
Okay, maybe i am paranoid. But that doesn't mean
they are not out to get me. - unknown

Tom Schlangen
Guest

Posted: Sat Dec 18, 2004 12:36 am Post subject: Re: NFB loop question

Hi Ruud,

Quote:

You still want your driver transfer function to be as
close as possible to the power stage transfer function
as you can get :)

Indeed :-)

Tom

--
fnord (now you see it, now you don't)
- R.A. Wilson / FZ

Patrick Turner
Guest

Posted: Sat Dec 18, 2004 5:37 am Post subject: Re: NFB loop question

Ruud Broens wrote:

Quote:

"Tom Schlangen" <t.schlangen@ndh.net> wrote in message
news:pd4b92-m1i.ln1@julabi.snakes.netcologne.de...
: Gentlemen,
:
: consider a simple two staged SE amp, w/o gNFB. As we know, it
: might even be desirable to get a certain amount of 2H from the
: driver to cancel out some of the 2H produced by the output
: stage.
:
: Now consider a global NFB loop (voltage feedback) added
: around the stages.
:
: Do we still want the driver to produce a certain amount of 2H,
: or should the driver now act as clean as possible?
:
: Tom
:
: --
: When in doubt, use brute force.
: - Ken Thompson

You still want your driver transfer function to be as close as possible
to the power stage transfer function as you can get :)
Rudy

This is easier said than done.

Patrick Turner.

Patrick Turner
Guest

Posted: Sat Dec 18, 2004 6:31 am Post subject: Re: NFB loop question

Tom Schlangen wrote:

Quote:

Hello Patrick,

thank you for your extremely elaborate answer.

The key point for me is the following paragraph:

Now if you have perfectly linear driver tube,
there would be more thd for the same applied NFB,
but its spectra is likely to be slightly less complex.
I think we'd be splitting hairs to say cancelation +
little NFB was better than no cancelation + moderate NFB.

The latter method would give a lower Ro of the amp.

Now a bit on the reason why I asked this question...

IN a pair of my latest SE amps with 4 x 6CA7, I have CFB
windings in the output stage to give 12.5% CFB which is
about 8 dB of applied NFB.
The thd is about 2% at 35 watts, and R about 1.5 ohms.
with the rated load = 5 ohms, and a linear driver tube,
and no other NFB.

You know I have a similar project going, 2 x 6550 PSE tetrode
mode with 10% CFB. You saw some CRO screen shots, but these
were with ECL82/6BM8 (trioded) as gain-stage/driver. After
plotting the plate curves of trioded ECL82, I decided that
one could do better (with less distortion) with the driver,
and that was the reason I started plotting plate curves of
several other trioded pentodes ... ;-)

After quite some load line drawing, I decided on trioded
6V6 as driver. At 375V B+, Ra = 8k2, 197 volts at plate,
22mA plate current, bias -12V, bias resistor 470 ohms plus
82 ohms (aha - you probably know already where this is
leading to ...), this driver is capable of more than 170 Vpp
output swing at excellent linearity. According to the
graphical method shown at p549 of RDH, expected 2h distortion
would be in the 1% range at that swing. Given that only ca.
1/2 of that swing is needed to drive the 6550s, the expected
distortion of this driver stage would be maybe 0,5% or
something.

Well you won't get much cancelation effect.

Quote:

Since I was "trioding" other pentodes to measure them anyway,
I thought I might as well put a trioded EF86/6267 in V1 place,
but _outside_ the NFB loop, since this thing, at 300V B+,
Ra = 220k, bias -3V, Rk = 3k9, is so extremely linear (not
only at the output swing of ca. 25 Vpp to drive the following
trioded 6V6, but also at much higher plate swings) that I
doubt one could do _significantly_ better even with an active
CC plate loaded 6SN7 ...

So you have 6V6 driving 6550 with CFB, and then an input
EF86 outside any FB loop around the former two.

Usually small signal triodes have insignificant thd below a couple of volts.
And you would be using only a tenth of a volt for most of your listening.

But the thd of the driver, the first of the 3 tube line up will be additive
to the
thd of the output.

I have 12AU7 as the inputs on the SE35, and with CCS loads to linearize the
12AU7 as much as possible which I think frees it to sing the best.
but is in the NFB loop.
If the 12AU7 wasn't in the loop it'd have to make nearly 3 times its output
voltage,
and that would mean more thd. Its still doable though.

Quote:

So, at the moment, I have an extremely linear "isolated" gain
stage at the front, followed by a very linear & beefy driver,
followed by the (untouched) output stage, with a voltage NFB
loop from the OPT secondary to the drivers cathode resistor
devider network.

That should be a nice set up, and free of the miller effect between the V1
and V2 you have.

Quote:

Preliminary schematic (w/o dimensioned R-NFB and C-NFB and
w/o stepped attenuators, I probably will do that this weekend)
at:

http://www.tubes.mynetcologne.de/roehren/6550pset/6550pset_amp_v31.pdf

I expect that this circuit will be quite easy to get it
unconditionally stable due to the NFB loop only being around
two stages. But that wasn't the primary intention behind
the change from ECL82 to EF86+6V6, of course ...

I have never been tempted to use 6BM8 or 6GW8 as input/driver tubes
when I could always have separate tubes to do the job more
puristically.
Well, I lie, I was tempted, but I resisted the temptation.

The EF86 has Ra somewhat lower than a 1/2 12AX7, and should drive the
6V6.
I say should.

The 6V6 has considerable C in when used as a triode because its anode is so
big.
The gain sets the miller C. But when NFB is applied from the OPT, the miller
effect
is reduced becaue the effective gain of the 6V6 is reduced.

Not many folks appreciate Rin is raised with normal series voltage NFB that
Tom is using.

Tom's 6V6 will have a gain of about 6 where no NFB is connected,
so for the 40 vrms of output he wants from it, about 7 vrms input is needed.
Suppose he uses 10 dB of NFB as his schematic may suggest,
then he'd need 21 vrms of input to the 6V6, with 14 vrms of NFB applied
at the 6V6 cathode.
Hey, that sounds like about 100% of the output voltage to the speaker.
There is no reason why all the speaker voltage could not be fed back
to the driver input cathode.

So for 40vrms out, he needs 21 vrms at the grid, and gain is only 2,
so the miller effect is very slight, Ca-g is multilpied only twice.

In PP amps where I use EL84 in triode the gain is about 18, and
the miller effect adds up to about 110 pF, so
if I had a triode input tube such as a 12AX7 with Ra = 35k for two halves
paralleled,
the HF pole would be 41 kHz, and that's still a bit low for me.

But In Tom's case the Ro of the EF86 will be quite low, especially since
its cathode is fully bypassed, and the Cin is also low, so he'll get maybe
100 kHz
easily.

But gee, he needs 21 vrms input to the 6V6.

That means the EF 86 has to earn its supper with some hard work,
and its thd would not be as low as I thought it would be.
It has 220k DC load, 220k Rg cap coupled load giving RL = 110k,
but since RL DC = RL cap coupled, the effective Ea for V1 is reduced
to about 1/2 way between 300v and 125v, or 213v.

For lowest thd the RL DC should be at least 1/3 of Rg.

I'd suggest RL DC be 100k, or better still, a CCS,
maybe the 6V6 bias R could be sneaked up to 390k.

I use global NFB around the 3 tube line up in my amps so far
because of the risk of making V1 work a bit too hard.
In my SE35, the 12BH7 has a gain of around 10 so for 45 vrms out
I need about 4.5 vrms input, and that's approaching all I'd ask from any
input tube.
The input tube gain is about 12, so with the 8 dB of global FB the
input is about 0.9v, and it allows easy driving from a CD player with a gain
pot.
There is no preamp needed, as it is already within the power amp, and
enclosed
by a harmless 8 dB NFB loop.

Maybe it don't sound as good as Tom's version.

Its a long trek to find out, so I'll have to leave that for another day.

Patrick Turner.

Quote:

Tom

--
Okay, maybe i am paranoid. But that doesn't mean
they are not out to get me. - unknown

Tom Schlangen
Guest

Posted: Sat Dec 18, 2004 8:24 am Post subject: Re: NFB loop question

Hello Patrick,

thank you for the analysis of my circuit.

Regarding trioded EF86:

Quote:

For lowest thd the RL DC should be at least 1/3 of Rg.

I forgot this rule :-( Maybe I should make a list of such
rules and repeat them like a Rosenkranz until my brains
is washed ... err ... conditioned accordingly ;-)

Quote:

I'd suggest RL DC be 100k, or better still, a CCS,

That's why I asked for high voltage BJTs for such purposes
recently here on RAT ...

But I would prefer to go as simple and reliable as possible
instead, no silicone, no SRPP or mu-follower, no additional
heater circuit for "upper" tubes ...

Maybe trioded EF184/6EJ7 could be an alternative as V1 (I
will re-check this on linearity later). Using it one surely
could forget about V1 Zo or RL / Rg ratio.

Quote:

maybe the 6V6 bias R could be sneaked up to 390k.

330k seems to do no harm.

Tom

--
fnord (now you see it, now you don't)
- R.A. Wilson / FZ

Tom Schlangen
Guest

Posted: Sat Dec 18, 2004 8:24 am Post subject: Re: NFB loop question

Quote:

Maybe trioded EF184/6EJ7 could be an alternative as V1 (I
will re-check this on linearity later). Using it one surely
could forget about V1 Zo or RL / Rg ratio.

Trioded EF184 at 350V B+, Ra = 56k (ca. 4,5 times the plate
resistance), 3,6mA and bias -2V looks like a good alternative
to me regarding linearity. Gain is about 44. Using 330k bias
res at V2, this gives a RL / Rg ratio of about 5,9.

Tom

--
The first rule of magick is simple: Don't waste your time
waving your hands and hoping, when a rock or club will do.

Patrick Turner
Guest

Posted: Sat Dec 18, 2004 9:30 pm Post subject: Re: NFB loop question

Tom Schlangen wrote:

Quote:

Hello Patrick,

thank you for the analysis of my circuit.

Regarding trioded EF86:

For lowest thd the RL DC should be at least 1/3 of Rg.

I forgot this rule :-( Maybe I should make a list of such
rules and repeat them like a Rosenkranz until my brains
is washed ... err ... conditioned accordingly ;-)

It wouldn't matter if you only wanted 2v from the EF86,
but I think you want 21 v, and that's quite a lot, and perhaps the
thd will be somewhat high, and additive with the output stage.
OK at low levels though.

Quote:

I'd suggest RL DC be 100k, or better still, a CCS,

That's why I asked for high voltage BJTs for such purposes
recently here on RAT ...

But I would prefer to go as simple and reliable as possible
instead, no silicone, no SRPP or mu-follower, no additional
heater circuit for "upper" tubes ...

Only one rational alternative then.
Have DC RL to the EF86 of about 47k,
and then a choke of 50H.

At 1 kHz, the choke will beover 300 kOhms, and a negigible load,
and the 47k will isolate the V1 from ever seeing the Lp and Cshunt of
the choke.
Gainchange at F extremes will be minimal, and the choke Dn is also
isolated
because of the low Ra of the V1 triode.
Several thousand turns of 0.15 wire would be fine.
place a low current fuse in series though.

Quote:

Maybe trioded EF184/6EJ7 could be an alternative as V1 (I
will re-check this on linearity later). Using it one surely
could forget about V1 Zo or RL / Rg ratio.

You'd still need the ideal RL/Rg ratio.
But indeed the 6EJ7 would be good. Maybe 6BX6.
But it'll have a gain of about 50.
So the input voltage will only be about 0.42 volts.

Quote:

maybe the 6V6 bias R could be sneaked up to 390k.

330k seems to do no harm.

6EJ7 can have a far lower value of DC RL; maybe 47k is OK.
Get some current into V1; EF86 is a bit of a pidler...
Then Rg = 330k, and you have an uncle called bob.

Have you considered the use of EL84 as the driver triode?
Its gain would be twice that of the 6V6,
and then you'd need much less input voltage even with the cathode
FB in the output stage.

14 mA of idle current in an '84 is plenty.

Just a few thoughts.

Patrick Turner.

Quote:

Tom

--
fnord (now you see it, now you don't)
- R.A. Wilson / FZ

Tom Schlangen
Guest

Posted: Sun Dec 19, 2004 1:12 am Post subject: Re: NFB loop question

Hi Patrick,

Quote:

Maybe trioded EF184/6EJ7 could be an alternative as V1

You'd still need the ideal RL/Rg ratio.
But indeed the 6EJ7 would be good. Maybe 6BX6.
But it'll have a gain of about 50.
So the input voltage will only be about 0.42 volts.

I think I can live with this :-)

Quote:

6EJ7 can have a far lower value of DC RL; maybe 47k is OK.
Get some current into V1; EF86 is a bit of a pidler...

Please have a look at:

http://www.tubes.mynetcologne.de/roehren/6550pset/6550pset_amp_v32.pdf

which shows actually measured values.

With trioded EF184, V1 stage gain is ca. 45, at 350V B+, 56k Ra,
130V plate, 3,6 mA plate current and -2,2V bias. Linearity is
excellent at that op point. V2 grid res is set to 180k according
to the 1/3 Ra/Rg rule.

Quote:

Have you considered the use of EL84 as the driver triode?

Sure.

Quote:

Its gain would be twice that of the 6V6,
and then you'd need much less input voltage even with the
cathode FB in the output stage.
14 mA of idle current in an '84 is plenty.

Trioded EL84 w/o doubt would make an excellent driver, indeed,
I agree. I don't have any _rational_ reason against it.

On the other hand, the trioded EF184 + trioded 6V6 combo as
shown in the schematic above, will do the job blamelessly also,
I am sure.

An admittedly completely non-rational reason for me to follow
the EF184/6V6 path is that I invested quite some hours of
taking their trioded plate curves by hand and now there is
a chance to bring a decent harvest in :-)

Quote:

Just a few thoughts.

Much appreciated!

Tom

--
The opposite of a correct statement is a false statement. But the
opposite of a profound truth may well be another profound truth.
- Niels Bohr

Ian Iveson
Guest

Posted: Sun Dec 19, 2004 8:23 am Post subject: Re: NFB loop question

"Ruud Broens" <broensr@wanadoo.nl> wrote

Quote:

You still want your driver transfer function to be as close as
possible
to the power stage transfer function as you can get :)

Quite what does "close" mean here, Ruud? And why?

For a pp stage, if the two are equal then the even harmonics are
cancelled completely. If the characteristics of the valves are
different, then some even H may remain. Odd H are summed, but let's
assume that we're not concerned about them at the moment.

Can the part of the transfer function that gives rise to the even
harmonics be isolated, for the purpose of comparing odd candidates
for pp operation? Does "close" only concern that part of the
function?

For cascaded stages, the situation is more complicated, because we
are multiplying rather than summing. Here, if the functions are
equal wrt even H, then is it cancelled? I don't think so. Is it
possible to get closer to cancellation if the functions are
different wrt even H?

What you really need is the inverse function I suppose...except you
would have no gain. So part of the function needs inverting, part
not. Tricky.

Proper wondering, don't know.

cheers, Ian

Chris Hornbeck
Guest

Posted: Sun Dec 19, 2004 8:23 am Post subject: Re: NFB loop question

On Fri, 17 Dec 2004 18:01:22 +0100, "Ruud Broens" <broensr@wanadoo.nl>
wrote:

Quote:

You still want your driver transfer function to be as close as possible
to the power stage transfer function as you can get :)

How will you simulate and scale the (undefined) loudspeaker
in the load line? The unstated assumption of a resistive
load line on the final stage is faulty.

Chris Hornbeck

Patrick Turner
Guest

Posted: Sun Dec 19, 2004 5:57 pm Post subject: Re: NFB loop question

Ian Iveson wrote:

Quote:

"Ruud Broens" <broensr@wanadoo.nl> wrote

You still want your driver transfer function to be as close as
possible
to the power stage transfer function as you can get :)

Quite what does "close" mean here, Ruud? And why?

For a pp stage, if the two are equal then the even harmonics are
cancelled completely. If the characteristics of the valves are
different, then some even H may remain. Odd H are summed, but let's
assume that we're not concerned about them at the moment.

Can the part of the transfer function that gives rise to the even
harmonics be isolated, for the purpose of comparing odd candidates
for pp operation? Does "close" only concern that part of the
function?

For cascaded stages, the situation is more complicated, because we
are multiplying rather than summing. Here, if the functions are
equal wrt even H, then is it cancelled? I don't think so. Is it
possible to get closer to cancellation if the functions are
different wrt even H?

What you really need is the inverse function I suppose...except you
would have no gain. So part of the function needs inverting, part
not. Tricky.

Proper wondering, don't know.

cheers, Ian

Mr Broens was bringing the issue to ppls attention in the context of an
SE amp, and not a PP amp.

Usually the phase of the 3H of a PP driver amp is the same as the phase
of the
3H in the output stage, so the 3H sums. 2H is only present due to tube
mismatches
or if a CPI phase inverter is used.
Sometimes the least 2H can be gained in a CPI type amp where the
output tubes are swapped, and cancellation of the out of balance 2H is
possible.
Certainly this is possible in almost all PP amps as output tubes or the
pair of driver tubes
do have some 2H, but you need careful measuring of thd to get the
benefit of the slightly
lower thd.

Some PP amp output stages generate 3H that has the opposite 3H phase of
the driver, and some 3H
cancellation occurs, but usually we are never lucky enough to have the
class
A driver stage 3H match the phase and magnitude of 3H of the output
stage between 0 watts and clipping.
Any attempt to force the match is usually unsuccessful, and gives
cancellation
at a small range of power, and not with all RLs.

One can have more luck with the 2H in SE amps.

Patrick Turner.

Patrick Turner
Guest

Posted: Sun Dec 19, 2004 6:04 pm Post subject: Re: NFB loop question

Chris Hornbeck wrote:

Quote:

On Fri, 17 Dec 2004 18:01:22 +0100, "Ruud Broens" <broensr@wanadoo.nl
wrote:

You still want your driver transfer function to be as close as possible
to the power stage transfer function as you can get :)

How will you simulate and scale the (undefined) loudspeaker
in the load line? The unstated assumption of a resistive
load line on the final stage is faulty.

This is a very good point.
Speakers produce mainly 2H and 3H thd to start with, and more H as
volume increases.

The phase of 3H of a speaker usually involves a flattening of both +ve and
-ve
peaks of the waves, like slow gradual limiting.
The same type of 3H occurs in amplifiers, so 3H in amps and speakers usually
sum.

But the level of 2H may well vary depending which way the speakers
are phased.

The lower the Ro of the amp, the less able it is for a distortion voltage to
be present
across the speaker terminals.
But the distortion in the acoustic waves still happens.

Patrick Turner.

Quote:

Chris Hornbeck

	DVD-Software.info Forum Index -> Tubes	All times are GMT Goto page 1, 2, 3 Next
Page 1 of 3

You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum