Reply
 
Thread Tools Display Modes
  #1   Report Post  
John L Stewart John L Stewart is offline
Senior Member
 
Location: Toronto
Posts: 301
Smile Pentode Screen Resistance (rs) Estimation Example

John L Stewart Jan 2016

In the Heathkit UA-1 12W Amplifier we see a 0.1 microF capacitor bypassing a one Meg resistor supplying the screen current. We might think that we can simply take that RC time constant & easily estimate the low frequency rolloff. But we would be dead wrong & get a result with a gross error.

The actual R the supply resistor is driving into is the screen resistance, analogous to a plate resistance of a triode. So the R in the RC time constant then becomes the parallel resistance of the screen & supply resistances. The problem is complicated by the lack of any published data with regards to screen resistance.

Fred Terman (And Wm. Hewlett) tell us that screen resistance for a pentode can be estimated as follows-

rs = ( ( Ib + Ic2 ) / Ic2 ) * Rp

From this we need to know 6AN8 triode Rp, another spec not published for the 6AN8. For the attached 6AN8 pentode plate family draw a line from the origin to the intersection of the Ec1 zero bias curve where Ec2 & the plate are at equal voltage. In this example that would be at 150 plate volts.

Triode plate Rp for the pentode section at zero bias is approximately as follows-

Rp ~ Change of Eb / Change of Ib

From the 6AN8 Average Transfer page, Change of Eb = 150 Volts
& Change of Ib ~ 28mA (plate) + 9mA (screen)
So triode Rp ~ ( 150 / 37 ) K or 4.05K

Aside from the scale factors of voltage & current all triode plate families look very much the same. As an example refer to the 6BQ7 plate family. When G1 bias is increased plate resistance increases. As an educated guess we can assume Rp for a triode connected 6AN8 pentode section to be something like 10K at the operating point.

Using the typical operating currents split on the 6AN8 pentode we get-

rs ~ ( ( 9 + 2.8 ) / 2.8 ) * 12K
or rs ~ 42.1K

That is the number we need to compute the rolloff frequency of the amplifier front end.

The R in the RC calculation now becomes rs in parallel with one M.

R ~ 40.4 K, a factor of about 20 times different than where we started.

RC = 0.0040 sec F = 0.159 / RC Hz 39.4 Hz

Looks like it is a step built in as part of the NFB stabilization circuits.

Note 1- Wm Hewlett & David Packard of HP were Fred Terman’s Graduate Students at Stanford U.

More to come
Attached Images
   
  #2   Report Post  
John L Stewart John L Stewart is offline
Senior Member
 
Location: Toronto
Posts: 301
Smile

Measurement of Screen Resistance Example John L Stewart Jan 2016

When designing or building an ordinary vacuum tube audio amplifier with a pentode front end we can for the most part ignore the screen supply resistance. Simply set it to be in the range of 3-4 times the plate resister & the circuit will perform satisfactorily. As long as the screen resister is adequately bypassed there are no significant problems.

That all changes when the intention is to use NFB. Most folks would assume that the RC time constant is simply the screen resister times the screen bypass capacitor. That can lead to problems since the screen supply resister is actually in parallel with the resistance of the screen grid itself as seen looking into the tube. Some calculations using the available published tube data indicates the screen resistance of common audio voltage amplifier pentodes to be in the range of 40K.

This simple setup makes measurements leading to the incremental screen resistance of the pentode section of a 6U8 vacuum tube while in operation. The tube is connected to a plate supply of 300 volts. The screen is fed from this supply thru a total of 730K resistance. But interposed on that is a means of applying an interfering One KHz test signal. Any audio transformer of high impedance primary & secondary can be used to couple the audio signal generator to the high voltage on the screen. I used an old Hammond 447 Interstage Transformer.

Just two measurements are required. Using a differential probe the AC voltage drop across the 730K is measured, then the AC voltage from common to screen.

The results are as follows-

First Pass- Drop across 730K was 0.9V

So Ig2 is 0.9 V / 0.73 M, 1.23 microA And Eg2 measured 0.043V

So rs is delta E / delta I rs = (0.043V / 1.23 microA)K or 35.0K

2nd Pass- Drop across 730K was 2.83V

So Ig2 is 2.83 V / 0.73 M, 3.88 microA And Eg2 measured 0.133V

So rs is delta E / delta I rs = (0.133V / 12.83 microA)K or 35.0K
Attached Images
   
  #3   Report Post  
John L Stewart John L Stewart is offline
Senior Member
 
Location: Toronto
Posts: 301
Smile

Measure Step Size caused by the Screen Bypassing Network
John L Stewart Jan 2016

This paper covers measurements made to determine the frequency response step size caused by the screen bypass circuit in a voltage amplifier pentode.

Two similar circuits were used in this trial. Neither is optimized. In one circuit the screen grid is supplied DC in the ordinary way from the plate supply. In the alternative circuit a screen DC connexion thru a suitable resistance back from the following cathode of a DC coupled split load phase inverter is used. Each has certain advantages & problems. Refer to the schematics.

The signal generator is direct connected to grid one of the 6U8 pentode section. There is no capacitor in order that there be no other RC time constant in circuit that might cause errors during the tests. Notice the screen bypass capacitor is rather low capacity. The test equipment would take a very long time while operating at low frequency in order to acquire the data. Shifting the frequency by using a smaller RC time constant eliminates that problem. The resulting step remains the same size.

The Pico Technology Scope/ Spec A has a max input of +/- 20 volts. With the X10 Differential Probe that becomes +/- 200 volts. So the gain measurements are all taken off the cathode of the triode section of the 6U8. If taken off the plate connexion of the 6U8 pentode the probe would need to be set to X100 which would result in less measurement resolution. The experimenter needs to know as much about what his equipment will not do as what it will.

For the ordinary circuit- in out Gain Gain db


Bypass cap connected 100 mV 12.54 V 125 42
Bypass cap disconnected 100 3.59 35.9 31
Step is ~ 11 db


Bypass cap disconnected 182 mV 6.14 V 33.7 30.6
Bypass cap connected 182 18.8 102.2 40.2
Step is ~ 10 db almost clipping

For the CF to G2 circuit- in out Gain Gain db

Bypass cap disconnected 99 mV 2.11 V 21.2 26.5
Bypass cap connected 103 7.89 76.6 37.7
Step is ~ 11 db overloaded


Bypass cap connected 50 mV 4.87 V 97.4 39.8
Bypass cap disconnected 48 1.04 21.7 26.7
Step is ~ 13.1 db almost clipping

Terman reports phase shift for screen steps in this range have maximums of 25-40 degrees.
Attached Images
   
  #4   Report Post  
John L Stewart John L Stewart is offline
Senior Member
 
Location: Toronto
Posts: 301
Smile

Here are the measurement results with the formatting still intact.

Cheers to all, John L Stewart
Attached Images
  
  #5   Report Post  
Posted to rec.audio.tubes
Big Bad Bob Big Bad Bob is offline
external usenet poster
 
Posts: 366
Default Pentode Screen Resistance (rs) Estimation Example

On 01/31/16 13:56, John L Stewart so wittily quipped:
Measure Step Size caused by the Screen Bypassing Network
John L Stewart Jan 2016

This paper covers measurements made to determine the frequency response
step size caused by the screen bypass circuit in a voltage amplifier
pentode.

Two similar circuits were used in this trial. Neither is optimized. In
one circuit the screen grid is supplied DC in the ordinary way from the
plate supply. In the alternative circuit a screen DC connexion thru a
suitable resistance back from the following cathode of a DC coupled
split load phase inverter is used. Each has certain advantages &
problems. Refer to the schematics.

The signal generator is direct connected to grid one of the 6U8 pentode
section. There is no capacitor in order that there be no other RC time
constant in circuit that might cause errors during the tests. Notice the
screen bypass capacitor is rather low capacity. The test equipment would
take a very long time while operating at low frequency in order to
acquire the data. Shifting the frequency by using a smaller RC time
constant eliminates that problem. The resulting step remains the same
size.

The Pico Technology Scope/ Spec A has a max input of +/- 20 volts. With
the X10 Differential Probe that becomes +/- 200 volts. So the gain
measurements are all taken off the cathode of the triode section of the
6U8. If taken off the plate connexion of the 6U8 pentode the probe would
need to be set to X100 which would result in less measurement
resolution. The experimenter needs to know as much about what his
equipment will not do as what it will.

For the ordinary circuit- in out Gain Gain db


Bypass cap connected 100 mV 12.54 V 125 42
Bypass cap disconnected 100 3.59 35.9 31
Step is ~ 11 db


Bypass cap disconnected 182 mV 6.14 V 33.7 30.6
Bypass cap connected 182 18.8 102.2 40.2
Step is ~ 10 db almost clipping

For the CF to G2 circuit- in out Gain Gain db

Bypass cap disconnected 99 mV 2.11 V 21.2 26.5
Bypass cap connected 103 7.89 76.6 37.7
Step is ~ 11 db overloaded


Bypass cap connected 50 mV 4.87 V 97.4 39.8
Bypass cap disconnected 48 1.04 21.7 26.7
Step is ~ 13.1 db almost clipping

Terman reports phase shift for screen steps in this range have maximums
of 25-40 degrees.


+-------------------------------------------------------------------+
|Filename: Screen Decoupling in DC NFB Version 6W.jpg |
|Download: http://www.audiobanter.com/attachment.php?attachmentid=427|
|Filename: Screen Decoupling in Ordinary Version 6W.jpg |
|Download: http://www.audiobanter.com/attachment.php?attachmentid=428|
|Filename: Keep-On-Truckin-T-Shirt-(8395).jpg |
|Download: http://www.audiobanter.com/attachment.php?attachmentid=429|
+-------------------------------------------------------------------+




I would expect screen resistance to change based on G-K volts, according
to some kind of general curve, and it's probably NON-linear at that,
based on the actual screen volts, plate volts, plate current, blah blah
blah.

But yeah, get it wrong and you howl and screech like a poorly configured
regenerative receiver.




  #6   Report Post  
John L Stewart John L Stewart is offline
Senior Member
 
Location: Toronto
Posts: 301
Smile

I would expect screen resistance to change based on G-K volts, according
to some kind of general curve, and it's probably NON-linear at that,
based on the actual screen volts, plate volts, plate current, blah blah
blah.

But yeah, get it wrong and you howl and screech like a poorly configured
regenerative receiver.

------------------------------------------------------------------

OK Bob, you bring up a good point. But things are not as ominous as that. See below-

Terman’s Explanation of Change of rs with Change of Screen Current

Translated it looks like this-

If Ig2 the screen current increases by 25% we would take the 3rd root of 1.25

(1.25)^(1/3) = 1.077

Then the reciprocal 1 / 1.077 = 0.928

If the screen resistance rs was 40K then at the new current it is 0.928* 40K

rs = 37.1 K at the new current

Fortunately the test setup is relatively simple. It is easy to make measurements under any set of conditions. If one has an ordinary scope there is no need for a differential probe as I have used. Just need two channels set for AC in, x10 probes & A-B to measure the difference resulting.

Some measurements are a long way off ground with large DC offsets so be sure to set AC inputs. With x10 probes the circuit loading will be 20M so in most cases that will not have to be taken into account.

I gave my analogue scope away several years ago. The Pico Tech provides a lot more information. And that can go right to a file for further use. But I forget to check the diff probe input Z, it is 4M. So I would need to go back & take that into account for the measurements across the 730K resister in the example. That is easy on an HP Calculator running RPN. So the 730K screen resister in the example becomes 617.3K & on from there. But the basic procedure is the otherwise the same.

The step in frequency response provided by the partially bypassed screen is a convenient adjustment one can make to get LF stability in a NFB amplifier. If it were a triode front end then a partially bypassed cathode resister could perform the same function. For the HF something else needs doing.

Cheers to all the Faithful still on here, John L Stewart
Attached Images
 
  #7   Report Post  
Posted to rec.audio.tubes
Big Bad Bob Big Bad Bob is offline
external usenet poster
 
Posts: 366
Default Pentode Screen Resistance (rs) Estimation Example

On 02/01/16 10:10, John L Stewart so wittily quipped:
I would expect screen resistance to change based on G-K volts,
according
to some kind of general curve, and it's probably NON-linear at that,
based on the actual screen volts, plate volts, plate current, blah blah
blah.

But yeah, get it wrong and you howl and screech like a poorly
configured
regenerative receiver.

------------------------------------------------------------------

OK Bob, you bring up a good point. But things are not as ominous as
that. See below-

Terman’s Explanation of Change of rs with Change of Screen Current


(etc.)

this probably plays a big role in determining the series resistance
needed for a proper ultra-linear config. As for me, I think I prefer
using a fixed and well regulated (and short-circuit protected) voltage
for G2 in the power stages. It also tends to give you a bit more power
out of the same tubes, with somewhat higher distortion depending on the
amplifier configuration [if you're doing AB2, it's probably less
significant, and G1 current probably becomes the primary source of
distortion and whatnot]. My focus is normally on 'final stage' beam
power and power pentodes.

In low power amplification stages with pentodes, this is a completely
different thing.

Biggest problems with pentodes overall is the nonlinearity. It's great
for mixer stages, great for frequency synth in radios, great for IF with
AGC, not so good for audio amplification. Basically, a pentode can act
as nonlinearly as a bipolar transistor. Correcting for this in audio
circuits requires lots of negative feedback, and so you point out the
problems with feedback and G2 series resistors and bypass capacitors [etc.].

Most pentode tube usage I've seen is in RF, not AF, and RF is where they
do the best, particularly the 'sharp cutoff' variety [a side effect of
their nonlinearity] for AGC and similiar circuits.

I'm not a fan of using them in AF circuitry. A dual triode typically
gives you as much open loop amplification using 2 stages, with
significantly lower distortion.

Anyway, this whole thing about G2's biasing and filtering could become
the topic of an entire series of articles or even a book.


  #8   Report Post  
John L Stewart John L Stewart is offline
Senior Member
 
Location: Toronto
Posts: 301
Smile

Biggest problems with pentodes overall is the nonlinearity. .



I'm not a fan of using them in AF circuitry. A dual triode typically
gives you as much open loop amplification using 2 stages, with
significantly lower distortion.

``````````````````````````````````````````````````

Are your conclusions based on hearsay or did you make some measurements to back up your opinion? If you have some comparison data perhaps you could post it here.

Thanx, John
  #9   Report Post  
John L Stewart John L Stewart is offline
Senior Member
 
Location: Toronto
Posts: 301
Smile

Biggest problems with pentodes overall is the nonlinearity. It's great
for mixer stages, great for frequency synth in radios, great for IF with
AGC, not so good for audio amplification. Basically, a pentode can act
as nonlinearly as a bipolar transistor. Correcting for this in audio
circuits requires lots of negative feedback, and so you point out the
problems with feedback and G2 series resistors and bypass capacitors [etc.].

Most pentode tube usage I've seen is in RF, not AF, and RF is where they
do the best, particularly the 'sharp cutoff' variety [a side effect of
their nonlinearity] for AGC and similiar circuits.

I'm not a fan of using them in AF circuitry. A dual triode typically
gives you as much open loop amplification using 2 stages, with
significantly lower distortion.

--------------------------------------------------------

For sure a pair of triodes will get more gain than a pentode. But the 2nd triode of the pair amplifies all the distortion of the first as well as the fundamental resulting in several even higher order harmonics, not exactly what we need.

As it turns out a pentode does have lower D than it does as a triode over a useful section of its output voltage range. Refer to the attached work done many years ago.

That range corresponds to the region where most hi G power pentodes such as EL34, KT66, 6550 & son on can be driven to full power. And running thru a split load phase invertor can easily drive a PP pair to full power, something Dyna & others took advantage of. Fewer stages translates to fewer high order harmonics before NFB is applied. And much better stability margin & reliability.

In the normal listening range the pentode looks much better than it does hooked up as a triode.

Cheers to all, John L Stewart
Attached Images
 
  #10   Report Post  
Posted to rec.audio.tubes
Big Bad Bob Big Bad Bob is offline
external usenet poster
 
Posts: 366
Default Pentode Screen Resistance (rs) Estimation Example

(I said)
Most pentode tube usage I've seen is in RF, not AF, and RF is where
they
do the best, particularly the 'sharp cutoff' variety [a side effect of
their nonlinearity] for AGC and similiar circuits.

I'm not a fan of using them in AF circuitry. A dual triode typically
gives you as much open loop amplification using 2 stages, with
significantly lower distortion.


then, On 02/04/16 06:43, John L Stewart so wittily quipped:

For sure a pair of triodes will get more gain than a pentode. But the
2nd triode of the pair amplifies all the distortion of the first as well
as the fundamental resulting in several even higher order harmonics, not
exactly what we need.


yeah, that's somewhat unavoidable. so you suggest that single stage has
lower overall distortion? Hard to tell, but I suppose it would depend
on the overall circuit design, operating range, etc.. and whether you're
using 'mu factor' gain or an unbypassed cathode resistor to provide a
small amount of negative feedback in the stage.

[my favorite includes partial bypass to give you proper cathode bias but
still control the gain, maybe 1/2 or 1/3 of the mu factor per stage]

As it turns out a pentode does have lower D than it does as a triode
over a useful section of its output voltage range. Refer to the attached
work done many years ago.


saw that, measured IM distortion but for the same tube [not an actual
triode, just triode-configured pentode]. But it doesn't address the
higher noise typically found in pentode amplifiers due to the additional
grids, etc..

So maybe it makes the case for a pentode at the mid-point, then? Low
signal input levels sort of demand a triode to improve S:N (and high
quality low noise grid resistors with values below ~200k).

That range corresponds to the region where most hi G power pentodes such
as EL34, KT66, 6550 & son on can be driven to full power.


OK - that would make sense. dual pentodes driving power output tubes,
maybe, or single-ended pentode driving a phase-split transformer. Since
off-the-shelf pentodes typically run at higher currents than triodes,
AB2 amplifiers would possibly work better.

In the normal listening range the pentode looks much better than it does
hooked up as a triode.


I'd still like to see the noise measurements, comparing actual triodes
to pentode, maybe 12AX7 [which would be the typical triode to use for
low level signals]. I suppose I'd run my own o-scope test for that one,
if I had any pentodes laying about to test. There are a number of
things to look at, from grid resistor values to capacitor materials to
power supply rejection. Those would have to be factored out somehow for
a fair comparison.




  #11   Report Post  
Posted to rec.audio.tubes
Patrick Turner Patrick Turner is offline
external usenet poster
 
Posts: 3,964
Default Pentode Screen Resistance (rs) Estimation Example



On Monday, 1 February 2016 05:37:06 UTC+11, John L Stewart wrote:
Measurement of Screen Resistance Example John L Stewart Jan 2016


The large amount of John's text re screen resistance is below my reply :-

A tube amp might have a 6U8 triode-pentode used for a driver tube but I don't recall too many, although the RCA inspired Dynaco type circuit comes to mind where pentode is input tube and triode is concertina phase inverter.

I have not found it important to know what the dynamic screen resistance actually is, and which could be calculated as signal Vg2 / Ig2 ac after measuring current and Vac across a screen feed resistor of low value. The Rs is different from the DC input resistance which is just Eg2 / Ig2 dc.

If the B+ rail = +300V, probable Ea would be say +150V, and Eg2 can be between +120V to +150V with little change of gain performance. Cathode Rk will determine the idle Iadc, and the gm, and gain, and Rg2 supply resistance may be 3 times the RLdc for anode, but Rg2 needs to be adjusted by trial and error to get Ea 150V, Eg2 120V and with whatever Rk is needed to get wanted Ia dc which is best measured by Vdc across anode RLdc / RLa dc. When these interactive voltages are correct, then the pentode is set up right. Usually, Rk will be two series R, say 220r + 47r, with 220r bypassed with say 470uF, and NFB brought to unbypassed 47r. So how to bypass screen? Most circuits show screen bypassed to cathode, and for LF stability with NFB the the bypass C value should be larger than the accountant allowed low value of say 0.1uF; better to use 1uF, and it won't be the screen circuit phase shift causing LF oscillations if there are any, it'll be the high open loop gain extending to too low a LF, and having insufficient primary inductance on OPT. This problem has always been common in many tube amps which prompted incompetent manufacturers to stipulate that their amp never ever be turned on without a speaker connected - they'd oscillate badly without load, because open loop gain goes high in output stages with no load. But even Williamson's amp was dodgy in this regard.
Just read my schematics at my site with critical damping networks for stability.

But some makers bypassed input pentode g2 to 0V, and when cathode FB is used, there's largish Vac between cathode and g2, so the effective gain of pentode changes and the WHOLE operation of the schematic in terms of OLG and amount of applied NFB changes and one needs to fully analyse a specific schematic before relying fully on any simplistic discussions here. There are just too many variables involved. But I do recommend that ppl interested go to their workshop, solder up a circuit, document it fully, post all results with schematic at a website, and then everyone will know what are the facts, well, pending presentation without errors.
Patrick Turner.


When designing or building an ordinary vacuum tube audio amplifier with
a pentode front end we can for the most part ignore the screen supply
resistance. Simply set it to be in the range of 3-4 times the plate
resister & the circuit will perform satisfactorily. As long as the
screen resister is adequately bypassed there are no significant
problems.

That all changes when the intention is to use NFB. Most folks would
assume that the RC time constant is simply the screen resister times the
screen bypass capacitor. That can lead to problems since the screen
supply resister is actually in parallel with the resistance of the
screen grid itself as seen looking into the tube. Some calculations
using the available published tube data indicates the screen resistance
of common audio voltage amplifier pentodes to be in the range of 40K.

This simple setup makes measurements leading to the incremental screen
resistance of the pentode section of a 6U8 vacuum tube while in
operation. The tube is connected to a plate supply of 300 volts. The
screen is fed from this supply thru a total of 730K resistance. But
interposed on that is a means of applying an interfering One KHz test
signal. Any audio transformer of high impedance primary & secondary can
be used to couple the audio signal generator to the high voltage on the
screen. I used an old Hammond 447 Interstage Transformer.

Just two measurements are required. Using a differential probe the AC
voltage drop across the 730K is measured, then the AC voltage from
common to screen.

The results are as follows-

First Pass- Drop across 730K was 0.9V

So Ig2 is 0.9 V / 0.73 M, 1.23 microA And Eg2 measured 0.043V

So rs is delta E / delta I rs = (0.043V / 1.23 microA)K or 35.0K

2nd Pass- Drop across 730K was 2.83V

So Ig2 is 2.83 V / 0.73 M, 3.88 microA And Eg2 measured 0.133V

So rs is delta E / delta I rs = (0.133V / 12.83 microA)K or 35.0K


+-------------------------------------------------------------------+
|Filename: Incremental Screen Resistance Plot w Captions 5W.jpg |
|Download: http://www.audiobanter.com/attachment.php?attachmentid=424|
|Filename: Screen Resistance rs Determination 5W.jpg |
|Download: http://www.audiobanter.com/attachment.php?attachmentid=425|
|Filename: G2 Resistance Test Setup 5W E.jpg |
|Download: http://www.audiobanter.com/attachment.php?attachmentid=426|
+-------------------------------------------------------------------+



--
John L Stewart


  #12   Report Post  
Posted to rec.audio.tubes
[email protected] johnnhelen4@gmail.com is offline
external usenet poster
 
Posts: 13
Default Pentode Screen Resistance (rs) Estimation Example

On Wednesday, February 17, 2016 at 7:31:55 PM UTC-5, Patrick Turner wrote:
On Monday, 1 February 2016 05:37:06 UTC+11, John L Stewart wrote:
Measurement of Screen Resistance Example John L Stewart Jan 2016


The large amount of John's text re screen resistance is below my reply :-

A tube amp might have a 6U8 triode-pentode used for a driver tube but I don't recall too many, although the RCA inspired Dynaco type circuit comes to mind where pentode is input tube and triode is concertina phase inverter.

I have not found it important to know what the dynamic screen resistance actually is, and which could be calculated as signal Vg2 / Ig2 ac after measuring current and Vac across a screen feed resistor of low value. The Rs is different from the DC input resistance which is just Eg2 / Ig2 dc.

If the B+ rail = +300V, probable Ea would be say +150V, and Eg2 can be between +120V to +150V with little change of gain performance. Cathode Rk will determine the idle Iadc, and the gm, and gain, and Rg2 supply resistance may be 3 times the RLdc for anode, but Rg2 needs to be adjusted by trial and error to get Ea 150V, Eg2 120V and with whatever Rk is needed to get wanted Ia dc which is best measured by Vdc across anode RLdc / RLa dc. When these interactive voltages are correct, then the pentode is set up right. Usually, Rk will be two series R, say 220r + 47r, with 220r bypassed with say 470uF, and NFB brought to unbypassed 47r. So how to bypass screen? Most circuits show screen bypassed to cathode, and for LF stability with NFB the the bypass C value should be larger than the accountant allowed low value of say 0.1uF; better to use 1uF, and it won't be the screen circuit phase shift causing LF oscillations if there are any, it'll be the high open loop gain extending to too low a LF, and having insufficient primary inductance on OPT. This problem has always been common in many tube amps which prompted incompetent manufacturers to stipulate that their amp never ever be turned on without a speaker connected - they'd oscillate badly without load, because open loop gain goes high in output stages with no load. But even Williamson's amp was dodgy in this regard.
Just read my schematics at my site with critical damping networks for stability.

But some makers bypassed input pentode g2 to 0V, and when cathode FB is used, there's largish Vac between cathode and g2, so the effective gain of pentode changes and the WHOLE operation of the schematic in terms of OLG and amount of applied NFB changes and one needs to fully analyse a specific schematic before relying fully on any simplistic discussions here. There are just too many variables involved. But I do recommend that ppl interested go to their workshop, solder up a circuit, document it fully, post all results with schematic at a website, and then everyone will know what are the facts, well, pending presentation without errors.
Patrick Turner.


When designing or building an ordinary vacuum tube audio amplifier with
a pentode front end we can for the most part ignore the screen supply
resistance. Simply set it to be in the range of 3-4 times the plate
resister & the circuit will perform satisfactorily. As long as the
screen resister is adequately bypassed there are no significant
problems.

That all changes when the intention is to use NFB. Most folks would
assume that the RC time constant is simply the screen resister times the
screen bypass capacitor. That can lead to problems since the screen
supply resister is actually in parallel with the resistance of the
screen grid itself as seen looking into the tube. Some calculations
using the available published tube data indicates the screen resistance
of common audio voltage amplifier pentodes to be in the range of 40K.

This simple setup makes measurements leading to the incremental screen
resistance of the pentode section of a 6U8 vacuum tube while in
operation. The tube is connected to a plate supply of 300 volts. The
screen is fed from this supply thru a total of 730K resistance. But
interposed on that is a means of applying an interfering One KHz test
signal. Any audio transformer of high impedance primary & secondary can
be used to couple the audio signal generator to the high voltage on the
screen. I used an old Hammond 447 Interstage Transformer.

Just two measurements are required. Using a differential probe the AC
voltage drop across the 730K is measured, then the AC voltage from
common to screen.

The results are as follows-

First Pass- Drop across 730K was 0.9V

So Ig2 is 0.9 V / 0.73 M, 1.23 microA And Eg2 measured 0.043V

So rs is delta E / delta I rs = (0.043V / 1.23 microA)K or 35.0K

2nd Pass- Drop across 730K was 2.83V

So Ig2 is 2.83 V / 0.73 M, 3.88 microA And Eg2 measured 0.133V

So rs is delta E / delta I rs = (0.133V / 12.83 microA)K or 35.0K


+-------------------------------------------------------------------+
|Filename: Incremental Screen Resistance Plot w Captions 5W.jpg |
|Download: http://www.audiobanter.com/attachment.php?attachmentid=424|
|Filename: Screen Resistance rs Determination 5W.jpg |
|Download: http://www.audiobanter.com/attachment.php?attachmentid=425|
|Filename: G2 Resistance Test Setup 5W E.jpg |
|Download: http://www.audiobanter.com/attachment.php?attachmentid=426|
+-------------------------------------------------------------------+



--
John L Stewart


Thus spake the Oracle!!

I would have thought Phil A would have made some comment by now. Hey Phil, lets hear your opinion!

Cheers to all, John L Stewart
  #13   Report Post  
Posted to rec.audio.tubes
Big Bad Bob Big Bad Bob is offline
external usenet poster
 
Posts: 366
Default Pentode Screen Resistance (rs) Estimation Example

On 02/17/16 16:31, Patrick Turner so wittily quipped:
There are just too many variables involved. But I do recommend that ppl interested go to their workshop, solder up a circuit, document it fully, post all results with schematic at a website, and then everyone will know what are the facts, well, pending presentation without errors.


ack, there are plenty of curves and equations described online in
various sources describing the behavior of pentode amplifiers, and they
can have VERY high gain if you set them up right, but maybe you're
simply explaining why I prefer to use triodes.

12AX7 good enough for Leo Fender, should be ok for me, heh.

A bipolar transistor single-stage amplifier can have a gain of 1000. I
wouldn't want to hear the audio quality of that, not at all. use an
emitter resistor (like unbypassed K resistor on pentode) for NFB and
keep the actual stage gain a bit lower, and all that gain might keep the
signal:noise and distortion down.

hard to say which is better. build and test, measure with good
equipment, use low noise metal film resistors and high quality
capacitors, shield the tubes, and additional low capacitance filtering
in the power supply in parallel with electrolytics...


  #14   Report Post  
John L Stewart John L Stewart is offline
Senior Member
 
Location: Toronto
Posts: 301
Smile

Patrick is correct, we don’t see many audio amplifiers making use of the 6U8. However, I did not have to look far to find one. It is one of Patrick’s very own early efforts. It has two 6U8s. The schematic is attached so that everyone may marvel at it.

NTL an interesting topology as anyone who has tried it will agree. Patrick manages to get about 95% of the signal at the plates of the output tubes all the way back to their grids in the form of NFB. If one looks carefully we can see an anode follower.

Without the MU followers as plate loads one is lucky to get ½ the signal all the way round the loop. In Patrick’s amp the result is the output tubes look very much like very stiff triodes driving the OPT. That will help the LF response very much. But not so much the HF since the signal still has to get past the OPT leakage inductance & winding capacities.

Each of the mu followers present what looks like a constant current load (Hi R) to the pentodes. Under ordinary circumstances that often creates a problem wherein the operating point (Q) is said to be ‘undefined’. Very small changes in either the pentode or mu follower result in large shifting of the Q point. The circuit could be unstable. In Patrick’s circuit the plate 470K NFB resister will keep things under control. Refer to the attached.

The circuit utilizes many parts to accomplish this. Some would say too many. Unfortunately the Law of Diminishing Returns sets in rather quickly. A simpler circuit using one less tube & fewer parts can accomplish the same thing.

I used the 6U8 for the screen resistance tests simply because I had one. But any ordinary small signal pentode would have done just as well to illustrate how the measurement can be made. The data is useful when designing an amplifier that might use the screen of an amplifier input tube to establish an LF step in the frequency response.

And I used a soldering iron as Patrick often suggests. But I also frequently use modern tools such as simulation software to get some idea if a particular circuit is worth pursuing. Before I plug in the iron! And it is better not to build the entire amp to measure only a part of a circuit. Best to look at each section in isolation to better understand what its function will be in the whole.

And I don’t use my slide rule a lot anymore either. The HP calculator running RPN is a hell of a lot faster. For anyone interested in RPN there is a less than one MB download at this link-

https://sourceforge.net/projects/fwcalc/

RPN does the calcs from the inside out, just as we do with pencil & paper. It stores the intermediate results so you won’t need the pencil until the final result. No Drucker!!

Cheers to all, John L Stewart
Attached Images
  
  #15   Report Post  
Posted to rec.audio.tubes
Big Bad Bob Big Bad Bob is offline
external usenet poster
 
Posts: 366
Default Pentode Screen Resistance (rs) Estimation Example

On 02/25/16 17:21, John L Stewart so wittily quipped:
NTL an interesting topology as anyone who has tried it will agree.
Patrick manages to get about 95% of the signal at the plates of the
output tubes all the way back to their grids in the form of NFB. If one
looks carefully we can see an anode follower.

Without the MU followers as plate loads one is lucky to get ½ the signal
all the way round the loop. In Patrick’s amp the result is the output
tubes look very much like very stiff triodes driving the OPT. That will
help the LF response very much. But not so much the HF since the signal
still has to get past the OPT leakage inductance & winding capacities.


biggest problem with high NFB from downstream of the output transformer
is the phase shifting at the edges of frequency response. Careful
choice of component values (and possible filtering/peaking reactive
components) might be needed to avoid the inevitable oscillation that
would take place without it.

but yeah, if you can get away with it, it would be like a typical op amp
with very high open loop gain, and relatively low 'stage gain'. The end
result is very low distortion.




  #16   Report Post  
Posted to rec.audio.tubes
Patrick Turner Patrick Turner is offline
external usenet poster
 
Posts: 3,964
Default Pentode Screen Resistance (rs) Estimation Example

I reply to John L Stewart...

Patrick is correct, we don't see many audio amplifiers making use of the
6U8. However, I did not have to look far to find one. It is one of
Patrick's very own early efforts. It has two 6U8s. The schematic is
attached so that everyone may marvel at it.

I could not see the attachment address, so here is is, see the schematic at top of page....
http://www.turneraudio.com.au/miscel...chematics2.htm

I also found many 6U8 around, they have useful sharp cut-off pentode, a bit 6AU6, and triode with lowish Ra medium µ. Probably best might be 2 x 6BX6 with above triode "CCS" being 1/2 12AT7.


NTL an interesting topology as anyone who has tried it will agree.
Patrick manages to get about 95% of the signal at the plates of the
output tubes all the way back to their grids in the form of NFB. If one
looks carefully we can see an anode follower.

**I don't get 95% local FB to output grids. The circuit in above link has rather low ß, ie, the fraction of N&D at anode appearing at grids is probably not above 0.5; its years since I build the test circuit in about 1997, in response to what Allan Wright was doing with a secret module for power amps ( Forced Symmetry ) and involving shunt FB with a secret j-fet. He'd pinched the idea from RDH4.

**But where output pentodes / beam tetrodes are used, they will have enough open loop gain to render a ß = 0.5 as being quite enough to do more than 15% CFB or any UL connection and with more linearazing effect because of 2 reasons I can think of.


Without the MU followers as plate loads one is lucky to get 1/2 the signal
all the way round the loop. In Patrick's amp the result is the output
tubes look very much like very stiff triodes driving the OPT. That will
help the LF response very much. But not so much the HF since the signal
still has to get past the OPT leakage inductance & winding capacities.

** I've used my balanced shunt FB idea ( stolen from RDH4 ) in this SS amp.....
http://www.turneraudio.com.au/solids...ono-mosfet.htm

Each of the mu followers present what looks like a constant current load
(Hi R) to the pentodes. Under ordinary circumstances that often creates
a problem wherein the operating point (Q) is said to be 'undefined'.
Very small changes in either the pentode or mu follower result in large
shifting of the Q point. The circuit could be unstable. In Patrick's
circuit the plate 470K NFB resister will keep things under control.
Refer to the attached.

** It all worked OK with a very poor quality OPT.

The circuit utilizes many parts to accomplish this. Some would say too
many. Unfortunately the Law of Diminishing Returns sets in rather
quickly. A simpler circuit using one less tube & fewer parts can
accomplish the same thing.

** Maybe, depends, I got 4 tubes in input amps; I think its better than the RCA inspired circuit with just one triode-pentode. Williamson with 4 x 1/2 6SN7 was a kind of gold standard and much better than RCA, Dynaco et all suggest, I like 2 cascaded LTP, as in http://www.turneraudio.com.au/RCA-reformed-30W.html
and then this one, with radical re-engineering, lotsa input tubes, but a most magnificent amp, which benefitted with its Chinese OPTs which were close to bein accidently good....http://www.turneraudio.com.au/Ming-D...-reformed.html


I used the 6U8 for the screen resistance tests simply because I had one.
But any ordinary small signal pentode would have done just as well to
illustrate how the measurement can be made. The data is useful when
designing an amplifier that might use the screen of an amplifier input
tube to establish an LF step in the frequency response.

** Possibly g2 of input pentode can be used for OLG shaping for stability, so at mid F the tube has high gain while at VLF and VHF (audio) the pentode operates as a lower gain triode.
A number of commercially made amps in 1950s used g2 on say 6SJ7 as a NFB port for GNFB. If the screen signal was equal to the anode signal driving the next stage, then the pentode is working as a triode; and because the anode signal at V1 is low and a small % of possible dynamic range then Vg2 FB signal could be easily twice Va, putting 6SJ7 into realm of supercharged triode.

And I used a soldering iron as Patrick often suggests.

**Currently I use 80W rated Chinese made, driven with switched windings from mains PT in box because with full 240Vac it runs way too hot for Pb-Sn solder, it is hotter for the lead free solder. With about 180Vac it runs just right and at about 40W for PbSn without oxide forming quickly. I can switch up heat for chassis and big leads. The iron clad tips are 8mm dia, and didn't tin easily at first, but did after some use. So these irons now last many years and are now better than old crap from old days....

I also frequently use modern tools such as simulation software to get some idea
if a particular circuit is worth pursuing. Before I plug in the iron!
And it is better not to build the entire amp to measure only a part of a
circuit. Best to look at each section in isolation to better understand
what its function will be in the whole.

** I never ever found time to learn to drive anywhere in a simulated vehicle; it IS GOOD, but my brain still works OK, rather like the guy who can think up a good chess move without needing to spend all day and night using 30 online programs to find a better move. But I was a hopeless chess player, all the club guys could easily beat me, but that's clubs for ya, fulla ppl wanting to beat you up on a table.
They were hopeless with electronics and use of any tools, and often extremely obsessive-compulsive and some were quite insane.
Horses for courses, eh?

Keep well, and sure that's a hard one, but do try to not go insane,
Patrick Turner.
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
Estimation of inter-sample peaks Mark Pro Audio 2 November 11th 10 06:48 PM
White Paper: Automated Sound Quality Estimation [email protected] Pro Audio 0 February 21st 09 02:28 PM
What's your winding resistance? Bob H. Vacuum Tubes 2 February 20th 07 05:28 AM
Pre-out to sub cable resistance [email protected] Tech 11 February 7th 06 11:07 AM
ohm (resistance) KLA Audio Opinions 10 July 11th 05 08:30 PM


All times are GMT +1. The time now is 12:29 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"