Reply
 
Thread Tools Display Modes
  #1   Report Post  
Posted to rec.audio.tubes
Patrick Turner Patrick Turner is offline
external usenet poster
 
Posts: 3,964
Default Ultralinear equations for µ, Ra, gain.

I recently cobbled together an integrated stereo amp with ultralinear
class A PP output stages using 66% UL tappings.
I had a nice pair of OPT and a PT from amoung stock I have advertised
for sale at my website and I had an old defunct LUX amp which was made
in Japan for Japanese 100V mains and for 50CA10 output tubes which are a
PITA when you need to replace them. I had a customer wanting to buy a
bit of good music. So, I have converted the Lux to to a much better
Turner amp and I'll be able to pay some hefty bills with the rewards.

The input stage is a paralleled SET 12AT7 with 15dB global NFB applied
to its cathode, then the phase splitting and extra required gain is via
an LTP with single 6CG7 with MJE340 CCS for cathode sink.

For the output stage I have EL34, 8.5k a-a, turn ratio = 41.8:1, ZR =
1,750 approx, winding R at primary = 440 ohms, and Ia = 50mA approx and
Ea = +375V. Cathode bias is used with 560 ohm for each Rk to each output
tube and each Ck = 1,000 uF.
So, the nominal loading with such an OPT is 8,500 ohms : 4.86 ohms at
the sec. Such loading makes this amp able to cope with any load between
3 and 23 ohms, although loads above 12 ohms could optimally be higher
sensitivity than lower value loads. Usually 16 ohms speakers are old
ones and do have usefully higher sensitivity above 93dB/W/W.

All loads above 8 ohms result in pure class A power for pentode, UL or
triode connection.

I did try triode, but PO was a bit low and so I experimented with UL and
found that with the loading I have the best UL% was at the 66% tapping
which my OPTs could use because a tap was available, along with taps for
33% UL.

The 66% tapping gives near triode like performance ( and truly wondrous
sound ) but still gives up to 80% of the pentode PO at low load values.

During this whole process I gave the UL connection some renewed
consideration and I share some findings with you all. I have some
observations and formulas which ppl may find useful.

Now all of us were brought up to believe the data sheets for EL34 and
for all the other pentode and beam tetrodes. For an EL34, ppl have been
expected to believe the pentode Ra = 12k, Gm = 11mA/V, and hence the
pentode µ = 132. And these parameters are **only** correct at a certain
Ea and Ia, which in fact are seldom ever used, unless its an SE
application. In PP applications, invariably, Ea is higher and Ia lower,
and like the conditions I stated above, Ia = 50mA, Ea = +375Vdc.

So I measured the parameters, and found that at the higher Ea of 375Vdc
and lower Ia for usual common PP operation at 50mA, Ra = 20k and µ =
130, and gm = 6.5mA/V only, way lower than the 11mA/V at much higher Ia.
Eg2 was a fixed 375V.

During normal operation of a mainly class A PP amp I have just sewn
together, the amp never moves out of class A and in fact the load each
OP tube sees is about 4.5k ohms for the full V swing, and since the
power likely to be used from this amp is only ever going to be an
average of 0.5W for average speakers, the load value doesn't change and
THD is negligible, and all our measurements need only be concerned about
what happens for the first 5 Watts while we know that sure there is a
25W AB ceiling for some low value load, but we won't be using 25 watts,
ever, unless teenagers have installed themselves at the volume control
without parental supervision.

How did I get the figures for the operating conditions?
OK, I set up the tube without any load, and used a 1kHz test sine wave
and the OPT shunt primary inductance and shunt capacitance at 1kHz is
such a high impedance it loads the OP tubes negligibly, and the dc
supply device, ie, the unloaded OPT primary can be considered to be a
CCS.
The gain with zero RL connected across the anode is equal to µ for the
tubes. I measured gain, A = µ = 130.However, to be sure, I used two
different loads without changing the input voltage at each EL34 grid,
and confirmed the µ value, and found the Ra with two gain equations each
for different RL values. Low anode voltages below 20Vrms were used to
stay well away from distortion effects.

So, Ra = RL1 x RL2 x ( A2 - A1 ) / ( [A1 x RL2] - [A2 x RL1] )

where.... RL1 is the first load you try,
A1 is the voltage gain for RL1,
RL2 is the second load you try,
A2 is the measured gain for RL2.

Note, for class A PP circuts the load for one tube = 0.5 x RL a-a.

Now this gives you only Ra.

The Gm could be measured by applying a common same phase signal of
1.0Vrms to both EL34 grids and measuring the current between the OPT CT
and the B+ supply cap which in my case is a 470uF cap with 0.34 ohms
reactance at 1kHz. So the current may be measured by measuring the
voltage across a series 10 ohm resistance, and the current halved for
*one* OP EL34.
I found Gm to be around 6.5mA/V.

But if you measure the µ being the gain without a load at 1kHz, then
apply this formula, Gm = µ / Ra, and in this case Gm was also found to
be the same either measured or calculated. So far, so good.

Now all this is nice and simple for normal pentode operation of the
EL34.

The equivalent generator model would have an imaginery voltage generator
producing µ x Vg internal imaginery output, and this output feeds a 20k
of equivalent Ra dynamic anode resistance and the anode output is at the
far end of this 20k anode resistance.

But what about the screen? In triode mode, you have the anode signal
connected to the screen, and the anode voltage when fed back into the
tube to the screen reduces the Vo to being the much lower "triode
connection" gain figure.

Now the screen should be considered for what it really is, ie, an
additional control grid, although one operating at the same potential as
Ea. What does it really do? Well, the data pages in old books are almost
dead useless in describing what happens with the screen.

Consider the EL34 set up and idling away with cathode bias and about
50mA for Ia, and Ea = 375V, and a fixed Eg2 screen voltage of +375V.
Let us load the anode with the high primary inductance only which at
1kHz is a load so high its high enough to be considered a CCS. Let us
connect the g1 control grid to 0V. Now consider we apply an ac 1kHz
signal voltage to the g2 via a transformer fed by low impedance,
and thus modulate the anode current similarly to how g1 does it, by
means of changing the control voltage.

With no load, I found the screen produced voltage gain = 15 approx. So
screen µ = 15. I also found the Ra to be marginally higher than measured
when using g1 as the control grid with a fixed Eg2. These parameters
are no-where to be found in data books. But anyway, the same loads as
used for finding gains for true pentode connections can be used to
determine the Ra for screen driven triode operation which is what I am
considering in effect. So we have screen µ, and screen Ra, and we can
confirm what we measure when applying the gain formula of A = µ x RL / (
RL + Ra ).

In this case when RL = 4.5k, screen gain = 2.5 approx.

This has important results. The effect of the active screen reduces
pentode distortion and pentode Ra.

Say we have the tube set up in normal pentode operation, and we have
1Vrms at g1.

AP ( pentode gain ) = µ x RL / ( RL + Ra ) = 130 x 4.5 / ( 20 + 4.5 ) =
23.8.

So for 1V at the g1, we have -23.8V at the anode with RL = 4.5k.

Let us suppose we have a fraction of the anode signal voltage fed to the
screen, and suppose it is = -3V.

If there was -3V at g2, the screen gain effect produces this to + ( -3 x
-2.5 ) = +7.5V, and this subtracts from the pure pentode anode signal of
23.8 to give a resultant voltage = -16.3V. Now -3V is 18% of the
resultant anode signal and so if we had a UL transformer with a tap at
18%, you'd get the UL gain with RL = 4.5k of 16.7.

Now you can go on repeating this experiment and gradually increase the
applied signal to g2 until you will find that lo and behold, the g2
signal equals the anode signal. It is then that you have the same gain
condition as triode connection.

You can run a whole series of calcuations for plotting a graph for the
UL gain results for a given RL and hence find gain for UL fractions or
percentages you didn't calculate.

But in fact its somewhat easier to use the following useful simple
formula I have devised which takes into account the pentode µ, Ra, and
screen µ and the UL fraction.

So, we get, for any wanted RL and UL fraction....

Voltage Gain, A = µP x RL / ( Ra + [ RL x { 1 + [N x µG2] }] )

Where...
A = voltage gain grid to anode, µP = pentode µ, RL is the class A load
considered, Ra is the pentode Ra, N = fraction of UL with 0.0 = pure
pentode and 1.0 = triode connection, and µG2 is the screen µ we
measured.

For example, for the above EL34 case and for 66% UL, ie, N = 0.67.

A = 130 x 4.5k / ( 20k + [ 4.5k x { 1 + [0.67 x 15] } ) = 8.42.

Now say N = 1, and you use triode connection.
You will find A = 6.35.

And so gain for any UL % and RL can be worked out. Not only that, one
can do some short cutting and you will find that the screen Gm is
approximately proportional to the g1 Gm as Ia to Ig2, and where you have
say Ia = 50mA and Ig2 = 5mA, then g2 Gm = 5/50 times g1 gm,
and in this case g1 Gm = 6.5mA/V, and I found G2 Gm = approx 0.65mA/V,
and if you assume screen drive triode Ra = pentode Ra then g2 µ = Gm x
Ra. In this case 0.00065 amps per volt x 20,000 ohms = 13, close to what
we observe.

By similar reasoning and a few pages of derivations worked out that....

The µ of the UL connected tube = µP / ( 1 + [ N x µG2 ] ).

In this case where N = 0.0 for pentode connected, µP = µP / 1 = µP, =
130.

For triode, and for N = 1, µT = 130 / ( 1 + [ 1 x 15 ] ) = 8.13.

For 66% UL, N = 0.67,

µUL = 130 / ( 1 + [ 0.67 x 15 ] ) = 11.8.

Similarly, it rurns out the Ra for the UL connection is also similarly
calculated at

ULRa = RaP / ( 1 + [N x ug2] ) .

And in this case for triode connection where N = 1.0,

RaT = 20k / ( 1 + [1 x 15] ) = 1.25k, which is what we might measure for
triode Ra.

For 66% UL,
RaUL = 20k / ( 1 + [ 0.67 x 15 ] ) = 1.82k.

For 43% UL,
RaUL = 2.68k.

This seems about right because a typical recommended loading for EL34
and UL is 6k a-a.

And UL connection always gives Ra-a somewhat similar to the value of
RLa-a, and in this case for the "standard 43% taps", Ra-a = 5.4k.

Once you know µ for UL and Ra for UL, you can work out the UL tube gain
from the standard univerally aplicable A = µ x RL / ( RL + Ra ).


When you have cathode FB in the output stage as in Quad-II amps, the
same rules for gain calculation can be considered.

Take Quad-II for an example. There is 10% of the anode to cathode signal
voltage appearing at the cathode. Say +90V at anode and -10V at cathode.
Therefore because g2 is at a fixed voltage, the signal between cathode
and g2 is in fact like a +10V signal in a UL amp with the same electrode
signal voltage ratios, so the KT66 gain can be worked right out from my
above formulas if we say N = 0.1.
Of course the gain in the Quad-II case is the *open loop* gain and =
anode to cathode signal voltage divided by the grid to cathode signal
voltage. Suppose we found KT66 gain = -10 for some load. Then -10V is
needed between g1 and k to generate +100V between anode and k.
And because there is already -10V at the cathode, then the grid to 0V
will be -20V. There is 6dB of cathode FB.

Now I have not gone as far as I could to give formulas for Ra with CFB
of a given % and for fixed or variable values of screen signal voltages.

But I have never needed to use more than 20% of the OPT primary as a
cathode winding. This 20% is ideal and where used with a fixed g2 supply
voltage because you have a drive voltage that is not excessively large
so that its high value increases drive amp distortions.
In McIntosh for example, the drive voltages need to be applied to the OP
tubes with half the a-k signal appearing at the k are over 100Vrms at
clipping, and the while you reduce THD in the OP stage you increase it
in the drive amp so you have a case of 2 steps forward, two steps
backward, or worse, and you begin to waste glasware trying to get
required gain and get a class B amp to behave.

Now why EL34 with 66% UL taps. Well, if I had pure triode, the PO is
limited severly by the extent to which the Va can swing negative before
the grid 1 begins to draw grid current, thus losing control of the tube
and causing early clipping. As we move the screen tap away from the
anode connection to some other winding point between B+ and anode, and
if we look to the pentode Ra curves, the UL gain rises and the anode is
allowed to swing nearly down to the Eg1 = 0V line on the Ra curve
sheets. You'll get enough swing with 66% UL, especially where the load
is a high value one, and all class A and nominally 8.5ka-a in this case.
The amp is going to power 16 ohms speakers, so in fact RLa-a will be
27.2k approx, and there will be only 10W of power available. And how
**well** this amp will sing!!! I've tried it on my own much harder to
drive lower Z speakers and it sings. But with the higher load and
because the speakers are more sensitive, THD will be under 0.02% or
less, and music will survive grandly.

To get 10W into 27k, you need 520Vrms a-a, and that is what I found to
be just possible, ie, a peak swing of 366V where Ea = 375V.
With pure triode connection you couldn't get this.

I suspect 66% UL has lower THD at 5W than triode at 5W does.

Now that America seems to have found a New Voice, and that it may embark
on a Better Course, then you may all focus on audio without the
anxieties brought on by the uncertainties of what may otherwise have
been.

I wish Mr Obama well, and hope that he learns to soon control the forces
of darkness arrayed around him, and even to win them over to a better
way of getting along.

Lance Armstrong, that brillaint athlete from Texas graces our country
right now, and I hope one day that it might be possible to have a race
Tour de Iraq and Afghanistan without all these people getting angry and
taking shots at participants.

It would be an enormous task to ensure the very disgruntled peoples of
the middle east get justice and aknowledgement of their land rights so
they lay down their weapons, and come together for peaceful construct.

Somebody pinch me, I might be dreamin.....


Patrick Turner.
  #2   Report Post  
Posted to rec.audio.tubes
tubegarden tubegarden is offline
external usenet poster
 
Posts: 343
Default Ultralinear equations for µ, Ra, gain.

Hi RATs!

So, Mr. Numbercruncher, are you going to try Mr. Blumlein's idle
current balancing "garter" circuits?

Or are you so sure we are crazy you can't waste a few hours hearing
some good news?

He did document the circuit before WW2, you know, just not everyone
thought better sound was worth the huge investment, in your case four
resistors and two caps per channel

Happy Ears!
Al

PS Yes, I think my time and the solder are free

  #3   Report Post  
Posted to rec.audio.tubes
Patrick Turner Patrick Turner is offline
external usenet poster
 
Posts: 3,964
Default Ultralinear equations for µ, Ra, gain.



tubegarden wrote:

Hi RATs!

So, Mr. Numbercruncher, are you going to try Mr. Blumlein's idle
current balancing "garter" circuits?

Or are you so sure we are crazy you can't waste a few hours hearing
some good news?

He did document the circuit before WW2, you know, just not everyone
thought better sound was worth the huge investment, in your case four
resistors and two caps per channel

Happy Ears!
Al

PS Yes, I think my time and the solder are free


The Garter circuit has Rk for each tube twice the normal value and cross
couples the bias R to try to balane Ia of each OP tube. Its a brave
attempt to equalise Ia dc for each tube and it is **slightly** better at
doing this job than the conventional method of using one Rk and Ck for
each tube. The better sound you hear is due to more equal Idc in each
1/2 of the OPT primary. Where you are supposed to have 50mA per tube and
have 45mA in one and 55mA in the other, you have 10mA of dc diference,
and I leave you to work out the dc field strength in Tesla and maybe it
makes the OPT saturate with ac at a higher F than it was designed to
cope with. Ac Saturation F is proportional to the applied ac voltage
amplitude as well as inversely proportional to frequency, and core Afe
and primary turn numbers. More turns and iron and less applied acV =
lower Fsat. With dc, the saturation is worse with more turns unless the
iron is gapped, or low µ iron. Pardon, but I am adressing the issue far
too briefly, but many OPT manufacturers only pay lip service to the
rules so clearly spelled out in RDH4. Most OPT used in many commercial
amps have been designed by accountant bean counters to reduce the copper
and iron to the barest minimums, Quad-II and many Leak models are
examples. RDH4 gives a formula for core caused distortions in a dc
balanced OPT but hasn't got much on OPT with specific dc offsets.
Tremain's Audio Encyclopedia has more to appall the mind about THD/IMD
with increasing amounts of dc offset. I've often heard Quad-II with
hopelessly unbalanced Idc, 90mA in one KT66, 40mA in the other, and real
crap sound. Placing a Garter circuit in there would improve things, but
you'd loose 30V of B+ headroom, and the Garter won't help class AB op
much. So there is better than the Garter. CCS for each cathode plus
something to limit Ek rise during class AB, or my system of Dynamic Bias
Stabilisation at my website, etc.

In one pair of Quad-II amps I re-engineered I have cathode bias with
each cathode with Rk = 470 ohms plus 1,000uF bypass. There is a
transistor LTP to drive a pair of red LED to indicate dc balance and an
adjustable trimpot to balance the applied slightly negative grid bias to
each OP tube. There is a small circuit board to do this monitoring job
plus active shut down protection if Ia goes too high for longer than 4
seconds. I can say to you this works better than thre Garter ever will
to ensure the best sound and to give amp owners peace of mind about the
condition of their amps and output tubes. As long as each LED shines
with equal brightness, Idc balance is very close. And if OP tubes drift
or a fault develops, there is immediate indication that something is
wrong with the biasing, and blooming bad biasing is the most common
problem in tube amps.


Patrick Turner.
  #4   Report Post  
Posted to rec.audio.tubes
Alex Alex is offline
external usenet poster
 
Posts: 111
Default Ultralinear equations for µ, Ra, gain.


"Patrick Turner" wrote in message
****************
Now the screen should be considered for what it really is, ie, an
additional control grid, although one operating at the same potential as
Ea. What does it really do? Well, the data pages in old books are almost
dead useless in describing what happens with the screen.

Consider the EL34 set up and idling away with cathode bias and about
50mA for Ia, and Ea = 375V, and a fixed Eg2 screen voltage of +375V.
Let us load the anode with the high primary inductance only which at
1kHz is a load so high its high enough to be considered a CCS. Let us
connect the g1 control grid to 0V. Now consider we apply an ac 1kHz
signal voltage to the g2 via a transformer fed by low impedance,
and thus modulate the anode current similarly to how g1 does it, by
means of changing the control voltage.


That is interesting, Patrick -- to use a pentode as a triode with the g1
connected to the cathode and with g2 as a control grid. Such mode will
always involve grid current and will require quite a large drive to get a
decent peak anode current at low anode voltage.

However, line deflection pentodes can be chosen. Those can produce enormous
plate currents from a relatively small g2 voltages, about 50..100V (Vg1=0).
Therefore, to drive such a "triode" by g2 one will need a cathode follower
on the basis of a smaller tube, capable of delivering peak screen current
(when anode is going low), I guess up to 30mA. For that driver a 6AQ5 can be
used (in a triode connection) or even 6N6P triode.

I am wondering, has anyone tried to use pentodes as screen grid controlled
triodes with g1 permanently connected to cathode?

Regards,
Alex


  #5   Report Post  
Posted to rec.audio.tubes
John Byrns John Byrns is offline
external usenet poster
 
Posts: 1,441
Default Ultralinear equations for µ, Ra, gain.

In article ,
"Alex" wrote:

"Patrick Turner" wrote in message
****************
Now the screen should be considered for what it really is, ie, an
additional control grid, although one operating at the same potential as
Ea. What does it really do? Well, the data pages in old books are almost
dead useless in describing what happens with the screen.

Consider the EL34 set up and idling away with cathode bias and about
50mA for Ia, and Ea = 375V, and a fixed Eg2 screen voltage of +375V.
Let us load the anode with the high primary inductance only which at
1kHz is a load so high its high enough to be considered a CCS. Let us
connect the g1 control grid to 0V. Now consider we apply an ac 1kHz
signal voltage to the g2 via a transformer fed by low impedance,
and thus modulate the anode current similarly to how g1 does it, by
means of changing the control voltage.


That is interesting, Patrick -- to use a pentode as a triode with the g1
connected to the cathode and with g2 as a control grid. Such mode will
always involve grid current and will require quite a large drive to get a
decent peak anode current at low anode voltage.

However, line deflection pentodes can be chosen. Those can produce enormous
plate currents from a relatively small g2 voltages, about 50..100V (Vg1=0).
Therefore, to drive such a "triode" by g2 one will need a cathode follower
on the basis of a smaller tube, capable of delivering peak screen current
(when anode is going low), I guess up to 30mA. For that driver a 6AQ5 can be
used (in a triode connection) or even 6N6P triode.

I am wondering, has anyone tried to use pentodes as screen grid controlled
triodes with g1 permanently connected to cathode?


Why would one want to do this vs. simply driving the control grid as
intended? The only reason I can see is higher potential power output
due to increased peak plate current, assuming the tube can tolerate
abnormally high peak screen voltages while the plate voltage is low.

--
Regards,

John Byrns

Surf my web pages at, http://fmamradios.com/


  #6   Report Post  
Posted to rec.audio.tubes
JC JC is offline
external usenet poster
 
Posts: 7
Default Ultralinear equations for µ, Ra, gain.


"John Byrns" wrote in message news:byrnsj-
Why would one want to do this vs. simply driving the control grid as
intended? The only reason I can see is higher potential power output
due to increased peak plate current, assuming the tube can tolerate
abnormally high peak screen voltages while the plate voltage is low.
Regards,
John Byrns



Reply by Jim, WD5JKO,

Many years ago the Gonset company did exactly this with amateur radio AM
transmitters. One example is where a pair of 6DQ6's plate modulated a single
class C 6DQ5. The 6DQ6's ran with G1 and the K grounded. The audio drive
came from a 6CM6 (similar to 6AQ5/6V6 but 9 pin), and drove either the
speaker (receive) or the 6DQ6 G2's (transmit).

Why would they do this? This was a transceiver designed for base or mobile
operation. Therefore size, weight, and power consumption were all big
considerations. Running the 6DQ6's this way with zero bias eliminated the
need for a bias supply and a screen supply. Since several watts of audio was
already available, why not use that to drive the 6DQ6's into G2 and have
them act like a hi - mu power triode (kind of like a baby 811a). Running the
modulators in class B also boosts the efficiency.

This modulator was capable of an easy 50 watts RMS of audio with a plate
voltage of about 600v.

This circuit in the Gonset with a few simple modifications is pretty clean,
and when compared to other designs of that AM ham era, this was sometimes
cleaner. Running screen grid tubes class AB2 and G1 driven with high
impedance drive, and high impedance G1 and G2 supplies can result in far
more distortion.

My experience with the G76 is that if you maintain the 6DQ6 G2 waveform
clean, then the plate waveform is almost an exact replica of the input. This
takes more NFB, and beefing up the 6CM6 audio driver (lower Rk, and higher
Ck). I raise the B+ to the 6CM6 on transmit to about 300v, and reduce it to
about 225 during receive. This keeps that tube cooler during extended
receive periods.

For such a simple circuit the modified G76 sounds quite good, and the Gonset
topology of driving the modulators as they do is very effective and
efficient.

Jim


see schematic 3 at the following link:
http://bama.edebris.com/manuals/gonset/g76/

  #7   Report Post  
Posted to rec.audio.tubes
Alex Alex is offline
external usenet poster
 
Posts: 111
Default Ultralinear equations for µ, Ra, gain.


"John Byrns" wrote in message news:byrnsj- Why would
one want to do this vs. simply driving the control grid as
intended? The only reason I can see is higher potential power output
due to increased peak plate current, assuming the tube can tolerate
abnormally high peak screen voltages while the plate voltage is low.


Exactly! To achieve the same high plate current at a low plate voltage in a
conventional triode would require higher (control) grid current, though the
positive grid voltage might be smaller compared to driving a beam tetrode
via g2.

And you know, a screen grid does tolerate high peak currents at a constantly
high voltage (in a conventional pentode operation). Therefore it can
tolerate the same instantaneous operating point in a
quasi-triode-screen-driven configuration.

Regards,
Alex

Regards,

John Byrns

Surf my web pages at, http://fmamradios.com/



  #8   Report Post  
Posted to rec.audio.tubes
Alex Alex is offline
external usenet poster
 
Posts: 111
Default Ultralinear equations for µ, Ra, gain.


"JC" wrote in message
...

"John Byrns" wrote in message news:byrnsj-
Why would one want to do this vs. simply driving the control grid as
intended? The only reason I can see is higher potential power output
due to increased peak plate current, assuming the tube can tolerate
abnormally high peak screen voltages while the plate voltage is low.
Regards,
John Byrns



Reply by Jim, WD5JKO,

Many years ago the Gonset company did exactly this with amateur radio AM
transmitters. One example is where a pair of 6DQ6's plate modulated a

single
class C 6DQ5. The 6DQ6's ran with G1 and the K grounded. The audio drive
came from a 6CM6 (similar to 6AQ5/6V6 but 9 pin), and drove either the
speaker (receive) or the 6DQ6 G2's (transmit).

Why would they do this? This was a transceiver designed for base or mobile
operation. Therefore size, weight, and power consumption were all big
considerations. Running the 6DQ6's this way with zero bias eliminated the
need for a bias supply and a screen supply. Since several watts of audio

was
already available, why not use that to drive the 6DQ6's into G2 and have
them act like a hi - mu power triode (kind of like a baby 811a). Running

the
modulators in class B also boosts the efficiency.

This modulator was capable of an easy 50 watts RMS of audio with a plate
voltage of about 600v.

This circuit in the Gonset with a few simple modifications is pretty

clean,
and when compared to other designs of that AM ham era, this was sometimes
cleaner. Running screen grid tubes class AB2 and G1 driven with high
impedance drive, and high impedance G1 and G2 supplies can result in far
more distortion.

My experience with the G76 is that if you maintain the 6DQ6 G2 waveform
clean, then the plate waveform is almost an exact replica of the input.

This
takes more NFB, and beefing up the 6CM6 audio driver (lower Rk, and higher
Ck). I raise the B+ to the 6CM6 on transmit to about 300v, and reduce it

to
about 225 during receive. This keeps that tube cooler during extended
receive periods.

For such a simple circuit the modified G76 sounds quite good, and the

Gonset
topology of driving the modulators as they do is very effective and
efficient.

Jim


Thanks, Jim.

It is interesting to know that g2 driven beam terrodes is a proven
technique. Maybe someone will revive that topology?

1. Low mu conventional triode in an audio power stage requires a large
negative bias, which complicates the circuit. Attempts to go into AB2 for
higher output requires a cathode follow driver with bipolar supply, which
complicates the circuit more.

2. Hi-mu conventional triode can work totaly in positige grid voltage
domain, but would require quite a large grid current (30...50% of plate) on
peaks (though low drive voltage). Getting this g1 drive current from the
main B+ supply would be a ridiculous waste of efficiency. Instead an emitter
follower powered from a local 15...20V supply would be a good solution, but
then it will not be a "valve" amplifier. (This should not concern Patrick
who uses to stick lots of transistors in his circuits still calling them
"valve".)

3. It looks like the screen-grid-driven-quasi-triode is the most convenient
solution:
- it has a low-to-medium mu;
- it works totally in the positive g2 voltages domain;
- it does not require negative bias supply;
- g2 currents are small (10...20% of plate current).

This opportune combination is I believe because:
a) screen grid is quite rare -- hence low g2 current;
b) cathode emission is subdued by the control grid umbrella connected to the
cathode -- hence all the curves are shifted into positive domain.

I am not sure of the dynatron effect (secondary emission from the plate),
but hopefully the beamforming plates can keep it at bay.

Regards,
Alex


  #9   Report Post  
Posted to rec.audio.tubes
Patrick Turner Patrick Turner is offline
external usenet poster
 
Posts: 3,964
Default Ultralinear equations for µ, Ra, gain.



Alex wrote:

"Patrick Turner" wrote in message
****************
Now the screen should be considered for what it really is, ie, an
additional control grid, although one operating at the same potential as
Ea. What does it really do? Well, the data pages in old books are almost
dead useless in describing what happens with the screen.

Consider the EL34 set up and idling away with cathode bias and about
50mA for Ia, and Ea = 375V, and a fixed Eg2 screen voltage of +375V.
Let us load the anode with the high primary inductance only which at
1kHz is a load so high its high enough to be considered a CCS. Let us
connect the g1 control grid to 0V. Now consider we apply an ac 1kHz
signal voltage to the g2 via a transformer fed by low impedance,
and thus modulate the anode current similarly to how g1 does it, by
means of changing the control voltage.


That is interesting, Patrick -- to use a pentode as a triode with the g1
connected to the cathode and with g2 as a control grid. Such mode will
always involve grid current and will require quite a large drive to get a
decent peak anode current at low anode voltage.


You don't connect the g1 to k, but to 0V, ansd you retain the use of Rk
and Ck so that the tube has an idle condition as my example stipulated
and similar to conventional ordinary pentode op.

There would be zero grid current in g1, but plenty of screen current so
CF drive to the screen is essential.



However, line deflection pentodes can be chosen. Those can produce enormous
plate currents from a relatively small g2 voltages, about 50..100V (Vg1=0).



Indeed you are very correct. 6CD6 and 6DQ6 are examples where the screen
gm is much higher than for tubes like 6L6 or EL34.


Therefore, to drive such a "triode" by g2 one will need a cathode follower
on the basis of a smaller tube, capable of delivering peak screen current
(when anode is going low), I guess up to 30mA. For that driver a 6AQ5 can be
used (in a triode connection) or even 6N6P triode.


Indeed.

I am wondering, has anyone tried to use pentodes as screen grid controlled
triodes with g1 permanently connected to cathode?


Some may have but you may find the linearity isn't too good, and maybe
its better to have the g1 at 0V and the k at a cathode bias voltage Ek
so that to get the wanted idle current Ia you'd have the Eg2 idling at
the normal Eg2 for normal pentode operation.
Looking at the curves given in the RCA tube book for some tubes (but not
all), the linearity for such g2 drive is far better than for normal op
but you will have Ra about the same as for normal pentode/beam tet.

Usually the screen driven triode operation of beam tets or pentodes
gives much lower gain and much higher Ra than triode op using g1 drive
and g2 connected to the anode.

The use of pentodes and tetrodes with both screen g2 and control grid g1
connected together gives interesting results but not for audio. Such use
has been mainly confined to RF PP amps.

Patrick Turner.



Regards,
Alex

  #10   Report Post  
Posted to rec.audio.tubes
Patrick Turner Patrick Turner is offline
external usenet poster
 
Posts: 3,964
Default Ultralinear equations for µ, Ra, gain.



John Byrns wrote:

In article ,
"Alex" wrote:

"Patrick Turner" wrote in message
****************
Now the screen should be considered for what it really is, ie, an
additional control grid, although one operating at the same potential as
Ea. What does it really do? Well, the data pages in old books are almost
dead useless in describing what happens with the screen.

Consider the EL34 set up and idling away with cathode bias and about
50mA for Ia, and Ea = 375V, and a fixed Eg2 screen voltage of +375V.
Let us load the anode with the high primary inductance only which at
1kHz is a load so high its high enough to be considered a CCS. Let us
connect the g1 control grid to 0V. Now consider we apply an ac 1kHz
signal voltage to the g2 via a transformer fed by low impedance,
and thus modulate the anode current similarly to how g1 does it, by
means of changing the control voltage.


That is interesting, Patrick -- to use a pentode as a triode with the g1
connected to the cathode and with g2 as a control grid. Such mode will
always involve grid current and will require quite a large drive to get a
decent peak anode current at low anode voltage.

However, line deflection pentodes can be chosen. Those can produce enormous
plate currents from a relatively small g2 voltages, about 50..100V (Vg1=0).
Therefore, to drive such a "triode" by g2 one will need a cathode follower
on the basis of a smaller tube, capable of delivering peak screen current
(when anode is going low), I guess up to 30mA. For that driver a 6AQ5 can be
used (in a triode connection) or even 6N6P triode.

I am wondering, has anyone tried to use pentodes as screen grid controlled
triodes with g1 permanently connected to cathode?


Why would one want to do this vs. simply driving the control grid as
intended? The only reason I can see is higher potential power output
due to increased peak plate current, assuming the tube can tolerate
abnormally high peak screen voltages while the plate voltage is low.


Indeed John, why?.

But I raised the issue in my OP to give some idea about how the screen
g2 works.

Patrick Turner.

--
Regards,

John Byrns

Surf my web pages at, http://fmamradios.com/



  #11   Report Post  
Posted to rec.audio.tubes
Patrick Turner Patrick Turner is offline
external usenet poster
 
Posts: 3,964
Default Ultralinear equations for µ, Ra, gain.



Alex wrote:

"JC" wrote in message
...

"John Byrns" wrote in message news:byrnsj-
Why would one want to do this vs. simply driving the control grid as
intended? The only reason I can see is higher potential power output
due to increased peak plate current, assuming the tube can tolerate
abnormally high peak screen voltages while the plate voltage is low.
Regards,
John Byrns



Reply by Jim, WD5JKO,

Many years ago the Gonset company did exactly this with amateur radio AM
transmitters. One example is where a pair of 6DQ6's plate modulated a

single
class C 6DQ5. The 6DQ6's ran with G1 and the K grounded. The audio drive
came from a 6CM6 (similar to 6AQ5/6V6 but 9 pin), and drove either the
speaker (receive) or the 6DQ6 G2's (transmit).

Why would they do this? This was a transceiver designed for base or mobile
operation. Therefore size, weight, and power consumption were all big
considerations. Running the 6DQ6's this way with zero bias eliminated the
need for a bias supply and a screen supply. Since several watts of audio

was
already available, why not use that to drive the 6DQ6's into G2 and have
them act like a hi - mu power triode (kind of like a baby 811a). Running

the
modulators in class B also boosts the efficiency.

This modulator was capable of an easy 50 watts RMS of audio with a plate
voltage of about 600v.

This circuit in the Gonset with a few simple modifications is pretty

clean,
and when compared to other designs of that AM ham era, this was sometimes
cleaner. Running screen grid tubes class AB2 and G1 driven with high
impedance drive, and high impedance G1 and G2 supplies can result in far
more distortion.

My experience with the G76 is that if you maintain the 6DQ6 G2 waveform
clean, then the plate waveform is almost an exact replica of the input.

This
takes more NFB, and beefing up the 6CM6 audio driver (lower Rk, and higher
Ck). I raise the B+ to the 6CM6 on transmit to about 300v, and reduce it

to
about 225 during receive. This keeps that tube cooler during extended
receive periods.

For such a simple circuit the modified G76 sounds quite good, and the

Gonset
topology of driving the modulators as they do is very effective and
efficient.

Jim


Thanks, Jim.

It is interesting to know that g2 driven beam terrodes is a proven
technique. Maybe someone will revive that topology?


I doubt it because to drive screens you need a much higher g2-g2 drive
voltage than for normal g1-g1 voltage and then the g2 drive must be low
impedance and by CF.

1. Low mu conventional triode in an audio power stage requires a large
negative bias, which complicates the circuit. Attempts to go into AB2 for
higher output requires a cathode follow driver with bipolar supply, which
complicates the circuit more.


A negative supply is very easy to arrange for bias.

And for mainly class AB amps cathode bias is fine.



2. Hi-mu conventional triode can work totaly in positige grid voltage
domain, but would require quite a large grid current (30...50% of plate) on
peaks (though low drive voltage). Getting this g1 drive current from the
main B+ supply would be a ridiculous waste of efficiency. Instead an emitter
follower powered from a local 15...20V supply would be a good solution, but
then it will not be a "valve" amplifier. (This should not concern Patrick
who uses to stick lots of transistors in his circuits still calling them
"valve".)


I don't fully understand the first part of your paragraph.

Yes, I do have bjt's strewn around in my tube circuits. But they act as
passive constant current sources or current sinks and as such are merely
willing slaves to the demands by the tubes for their appetite for
blameless dc supply without dc carry resistors which waste ac power and
increase circuit distortions. Apart from using a j-fet in the early low
signal level gain path of a tubed phono amp not one single bjt is used
anywhere to perform buffering or voltage gain applications.



3. It looks like the screen-grid-driven-quasi-triode is the most convenient
solution:
- it has a low-to-medium mu;
- it works totally in the positive g2 voltages domain;
- it does not require negative bias supply;
- g2 currents are small (10...20% of plate current).


All this is correct.

Linearity look good if well set up but I have not done it.

My original concern was the effect of screen drive in terms of the Ra
and screen µ and screen gm and enabling ppl to work out gains for UL
output stages and while the amp was used well below its maximum PO.


This opportune combination is I believe because:
a) screen grid is quite rare -- hence low g2 current;
b) cathode emission is subdued by the control grid umbrella connected to the
cathode -- hence all the curves are shifted into positive domain.


With g1 tied to k the g2 voltage would have to be low for a low Ia at
idle.

Maybe you'd find the linearity will be poor.

My OP merely investigated the screen's effect and at Eg2 = normal
operation levels for UL output stages, and while the cathodes had
cathode bias and with g1 tied to 0V, not to the k.




I am not sure of the dynatron effect (secondary emission from the plate),
but hopefully the beamforming plates can keep it at bay.


I'm not sure either.

Patrick Turner.

Regards,
Alex

Reply
Thread Tools
Display Modes

Posting Rules

Smilies are On
[IMG] code is On
HTML code is Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Simple negative feedback equations for Ra'. Patrick Turner Vacuum Tubes 0 October 13th 06 12:38 PM
Gain equations : ref frequency response Pooh Bear Vacuum Tubes 37 December 22nd 05 12:10 PM
"Debating trade technique" was Gain equations : ref frequency response Andre Jute Audio Opinions 20 December 21st 05 03:56 AM
line level input gain control vs console pre-amp gain? saturation question. perry Pro Audio 2 June 21st 04 10:30 PM
PA CONUNDRUM: AMP GAIN KNOBS vs MIXER GAIN SETTINGS Richard Kuschel Pro Audio 7 September 22nd 03 08:30 PM


All times are GMT +1. The time now is 09:38 PM.

Powered by: vBulletin
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
Copyright ©2004-2024 AudioBanter.com.
The comments are property of their posters.
 

About Us

"It's about Audio and hi-fi"