[Patrick J. Jankowiak]

Hello List,

On the 6L6/G/GA/GB, Please help me to better understand these class
AB1 6L6 operating conditions, and compare them to some observed
conditions. I assume the published conditions, having been published
for years by RCA, Sylvania, et al, were once derived from actual test
data.

The official data is shown first, with the tube types according to the
manuals. Last, some measured data is shown. All measurements are done
by driving the amplifier barely to clipping, measuring the peak
voltage on a scope, and figuring the RMS voltage to be 0.707 of this,
and from there using the load resistance to calculate current, then
power, so it is the usual method. To vary the plate to plate load, a
70V/25 watt line matching transformer was used with an 8 ohm load.
70.7V on this amp's 250 ohm output, as called for in the manual, is 20
watts.

At issue is this seemingly incongruous set of data: For the published
conditions of a 9000 ohm load, the output is supposed to be 24.5
watts. The actual amplifier puts out 8 watts. Ok, so I figure I might
have flat tubes, (even though they measure 80% of what a NOS JAN
6L6WGB did on the tube tester). Maybe they cannot pass enough current.
So I tried loading it more lightly so that the load was 18000 ohms and
the power did not increase. At 9000 ohms, there is no extra room for
more voltage excursion under lighter loads. So, why not load more
heavily? Ok. I loaded to 4500 ohms, and Behold! 22 watts. So, no
problem with the current-carrying capacity of the tubes.

But why do I not get the correct power output at 9Kct as stated in the
venerable tube manuals?

Note, the 9K CT specs only seem to show up for early 6L6G material. In
later material it is omitted or relegated to the "additional
data/obsolete" sections.

Could either of these possibilities be true:

1. The data was in error with regard to output?

2. The amplifier in question, a military unit made by David Bogen,
came with two speakers having 250/500/1000/2500 ohm inputs. The output
terminals are stated to be 250 ohms. The mess was intended to be
connected all together, in parallel, with instructions to assure a 250
ohm match regardless of using 1, 2, 4, or more speakers. Could the
amplifier have a wiring mistake? I could not find anything!

Although it is not unusual to get more power when overloading an
amplifier, especially one with a high impedance load to begin with,
this is just too strange.

The bottom line is that I have tried varying the load resistance using
a decade box, and with the 'correct' match, 250 ohms, I just don't get
the power.



tube manual specs:
[[[[[[[[[[[[[
6L6G / 6L6GB

plate 360V, 88/132mA
screen 270v, 5/15mA
bias -22.5
load CT 6600
PoutW 26.5W

plate 360V, 88/132mA
screen 270v, 5/18mA
bias 248 OHMS CATHODE RESISTOR
load CT 9000
PoutW 24.5W

-------------
6L6G / 6L6GB / 6L6 / 6L6GC

plate 360V, 88/132mA
screen 270v, 5/15mA
bias -22.5
load CT 6600
PoutW 26.5W

-------------
6L6 / 6L6GC

plate 360V, 88/132mA
screen 270v, 5/11mA
bias -22.5
load CT 3800
PoutW 18W

[[[[[[[[[[[[[
6L6G in the AM-20 amplifier, has 250 ohm (70V) output terminals.
Tubes check at 80% of a NOS JAN 6L6WGB, 1 KHz sine wave.

plate 380V
screen 265V
bias 250 OHMS CATHODE RESISTOR @23.5V
load CT 9000
PoutW 8W (when load = rated value [250 ohms])
PoutW 22W (when load = 1/2 rated value [125 ohms])




Any comments are welcome.
PJ

[Phil Allison]

"Patrick J. Jankowiak"


plate 380V
screen 265V
bias 250 OHMS CATHODE RESISTOR @23.5V
load CT 9000
PoutW 8W (when load = rated value [250 ohms])
PoutW 22W (when load = 1/2 rated value [125 ohms])



** Errr - power output CANNOT increase by a factor more than 2 when the
load impedance is halved.

YOU have stuffed up somewhere.

The old, published 9000 ohms and 24 watts figure looks like a misprint to
me - as it requires the plate of the 6L6 to swing down to only 31 volts.

24 watts into 9000 ohms means there is 465 volts rms p-p.

465 rms = 657 volts peak.

So, the down swing is 329 volts with a 360 volt supply.




........ Phil

[Mark Aitchison]

Patrick J. Jankowiak wrote:
Hello List,

On the 6L6/G/GA/GB, Please help me to better understand these class AB1
6L6 operating conditions, and compare them to some observed conditions.

Let's see: you have two 6L6 tubes in push-pull with a 9 kilohm
center-tapped o/p transformer, 360V plate supply, but 8 Watts output
instead of the manual's 24.5W. If the output voltage swing is about as
much as you can get (you said "there is no extra room for more voltage
excursion...") then you should be getting over 300v peak into each 4500
ohm side of the transformer, so over 20W into the primary. Maybe the
load impedance isn't what you think it is... check it by putting a 100
ohm resistor between one of the plates and the transformer and carefully
measure the AC voltage across it... remember the high voltage is beyond
what cheap meters say they safely handle. It should give about 4.4 V
(RMS) when there is 200V (RMS) from plate to positive supply (ie across
half the output transformer). If you cannot get 4.4 V (44mA AC) the
load is too high, or the drive too little or the transformer has a fault
or circuitry around it has a problem or the screen grid isn't connected
properly (although the last one is unlikely).

Mark

[Mark Aitchison]

I said...
... you should be getting over 300v peak into each 4500
ohm side of the transformer, so over 20W into the primary. ...

Thinking about it some more, you wouldn't get much more than 300V peak
since the best a 6L6 might do is drop 25-30v plate-to-cathode at 70mA,
so (with the grid bias) we wouldn't have much more than 300V left for
the transformer. Still, you should be getting closer to 20W than 8W. I
still suspect the load isn't really 4500 ohms each side of center.

Mark

[Phil Allison]

"Mark Aitchison"


** Kiwi alert !!!



... you should be getting over 300v peak into each 4500
ohm side of the transformer, so over 20W into the primary. ...

Thinking about it some more, you wouldn't get much more than 300V peak
since the best a 6L6 might do is drop 25-30v plate-to-cathode at 70mA,


** The *peak* plate current is approx 150 mA per tube.

Supply idle is specified as 44mA and full power is 100mA.

http://www.mif.pg.gda.pl/homepages/f...155/6/6L6G.pdf


so (with the grid bias) we wouldn't have much more than 300V left for the
transformer. Still, you should be getting closer to 20W than 8W. I still
suspect the load isn't really 4500 ohms each side of center.


** A 9000 ohm CT tranny is ** 2250 ohms ** per side.

Full output current peaks are supplied by one tube at a time.





........ Phil

[Ned Carlson]

Patrick J. Jankowiak wrote:

But why do I not get the correct power output at 9Kct as stated in the
venerable tube manuals?

First thing, are you using cathode bias resistor as the data
sheet shows? Note that this load is ONLY (I've never seen different)
shown as example in tube manuals, used with cathode bias.

Second, what kind of 6L6 are you using? Not everything that's
labelled as a 6L6 is going to have the same characteristics
as a bogey RCA 6L6 from the 40's. I'd bet a lot of them don't.
If you dig up a couple of real old metal 6L6's that test 100%,
they'll probably come close to operating as tube manual example
indicates.

--
Ned Carlson
SW side of Chicago, USA
www.tubezone.net

[Mark Aitchison]

Phil Allison wrote:
"Mark Aitchison"
... you should be getting [over] 300v peak into each 4500 ohm...
Still, you should be getting closer to 20W than 8W. I still
suspect the load isn't really 4500 ohms each side of center.
** The *peak* plate current is approx 150 mA per tube.
A 9000 ohm CT tranny is ** 2250 ohms ** per side

OOps and oops! I forgot the idle current and the centre-tapped trannies
always get me. I'll have to think about that 4500 ohm/2250 ohm issue
some more. Meanwhile, I do still think the circuit should be capable of
closer to 20W than 8W, if the load is right. As far as I can see from
the graphs it isn't a huge distance away from 30 V plate-cathode and the
difference between 600 V p-p and (say) 550V, even when squared, isn't
tremendous. If it was 600V p-p into 9k that is still 20W (so I must
have made 2 cancelling mistakes before). I think measuring the AC
current from the plate is still a good test (but I should have said to
block DC).

[Phil Allison]

"Mark Aitchison"

OOps and oops! I forgot the idle current and the centre-tapped trannies
always get me. I'll have to think about that 4500 ohm/2250 ohm issue some
more.


** Half turns ratio = 1/4 impedance.

Thinking done.


Meanwhile, I do still think the circuit should be capable of closer to 20W
than 8W,


** The OP's figure of 8 watts is ********.




........ Phil

[Mark Aitchison]

Phil Allison wrote:
"Mark Aitchison"


... I'll have to think about that 4500 ohm/2250 ohm issue some
more.

** Half turns ratio = 1/4 impedance.

Thinking done.

Hmmm and more Hmmm. That half-turns 1/4 impedance business applies when
you drive ONLY half the coil, so half the voltage but twice the current
to get the same power (so 1/4 times impedance). But we are driving both
halves (well, certainly if we're still in PP class A) so same current
and half the voltage (compared with peak-peak voltage into 9k) because
half the power per half of the transformer. And 9k CT means the
impedance of the total is 9k, so each half has to be 4.5k.

[Phil Allison]

"Mark Aitchison" = another Kiwi menace



... I'll have to think about that 4500 ohm/2250 ohm issue some more.

** Half turns ratio = 1/4 impedance.

Thinking done.

Hmmm and more Hmmm.


** That is a bad hummm you have.



That half-turns 1/4 impedance business applies when you drive ONLY half
the coil, so half the voltage but twice the current to get the same power
(so 1/4 times impedance). But we are driving both halves (well, certainly
if we're still in PP class A)


** The specified conditions are NOT class A !!!!!!!!!!!!

The plate supply current varies widely as does the screen supply current.

It is CLEARLY stated to be class * AB1* in the data sheets.

BTW:

Damn well READ what I posted to YOU and DO NOT ****ing SNIP the bits you
can't comprehend or don't like.

Like this one:

" Full output current peaks are supplied by one tube at a time."

That is how * class AB 1 * operates.




........ Phil

[Mark Aitchison]

Phil Allison wrote:
"Mark Aitchison" = another Kiwi menace

More menacing to come :-) (a.k.a. hmmming)

Damn well READ what I posted to YOU and DO NOT [snipped :-)] SNIP the bits you
can't comprehend or don't like.

Sorry to annoy you. I didn't put much time into thinking before typing,
and it shows. The "if we are still in class A" bit was meant to show my
thought process - that I could justify the half-power-per-side logic
(and so arithmentic) so long as the it was class A push pull. More
thinking:

1. each valve has a quiescent current of 44mA.

2. Up until the point one valve cuts off (ignoring nonlinearlities it
would be at about 44mA into 9k, which is about 17 Watts peak) we are
still in class A and each side of the center-tap is supplying equal
power. So I think that, up to that point, we can say each half of the
transformer is 4500 ohms (44mA into 4.5k is about 8.7W pk).

3. After one tube is turned off its negative current swing isn't
contributing to the output, so the other valve, at this point, has to
increase in current at twice the rate. We are about 40% to the top in
terms of power output (about 40 degrees into a sinewave?). So, in
effect, the load the "on" tube now "sees" changes from 4500 ohms to 2250
ohms part way through the cycle, so it still has to put out the 300V pk
(as it would in Class A) to get 40W pk, but the current has to be 133mA
(plus its 44mA quiescent but minus the other tube's 44mA?), which is
roughly in the ballpark of what you said (150mA peak) and near the
average with-signal current consumption (don't want to integrate it to
work it out properly) so maybe I'm on the right track.

4. That means the effective impedance of each each side depends on the
amplitude, with an average somewhere between 2250 and 4500 ohms.

5. Back to the original question posed: why 8W RMS? Either the load is
wrong or the measurement of 8W is wrong (which you might have been
saying), or (from the above) it isn't being driven hard enough because
8WRMS is the point where is shifts out of class A (and so grid input has
to increase lots else it would distort). The things to test, then,
would be:

(a) current going through plate (on a scope or whatever) to check load
(b) grid voltage (with respect to cathode) to make sure it is being
driven close to 0 volts on peaks.

Hope you don't mind me thinking out loud.

Mark

[patrick jankowiak]

Ned Carlson wrote:

Patrick J. Jankowiak wrote:

But why do I not get the correct power output at 9Kct as stated in the
venerable tube manuals?


First thing, are you using cathode bias resistor as the data
sheet shows? Note that this load is ONLY (I've never seen different)
shown as example in tube manuals, used with cathode bias.

Second, what kind of 6L6 are you using? Not everything that's
labelled as a 6L6 is going to have the same characteristics
as a bogey RCA 6L6 from the 40's. I'd bet a lot of them don't.
If you dig up a couple of real old metal 6L6's that test 100%,
they'll probably come close to operating as tube manual example
indicates.


I am certain it is cathode bias as shown in the schematic, measured
the resistance to ground, and measured the voltage at the cathode as
in my measurements of the voltages in the original post.

After looking at the tubes again, I am using 6L6GC for both. One is a
GE and one is a Sylvania. They are reasonably close on the tube
tester, all I have is a B&K 747 which is not so great but it does show
up differences between tubes pretty well.

Since the amp is old, someone may have fooled with it, I will go and
re-verify the connections on the OPT secondary since there are several
taps there as in terminals on the bottom, and they are solderable. I
did RTFM which is posted here and it implies the 250 ohm output is
supposedly a 70.7 volt output and tells the repairman to measure 70.7
volts AC there. The circuit might answer some questions.

http://www.bunkerofdoom.com/mil/AN_TIQ2/ANTIQ20.HTML

So tomorrow I will go back and measure some of the things suggested.
Will it be sufficient to measure the voltage across the cathode
resistor to obtain the current, or must I go to the plates? it would
be easier to insert a small resistor in the cathode lead from a tube.
Not sure how to get the peak plate current with a meter. I could try
a scope in differential mode. I can also look at the plate waveforms
and see where the voltage bottoms out. I will also re-do the power
measurements in case I made a mistake. I did them at 1KHz.