[AMRadio] Testing tubes

Tue Oct 14 07:45:41 EDT 2003

	John,

	Wow! I have been inundated with stuff from higher minds than mine! Besides
your excellent replies, I got a somewhat similar, yet different perspective
from Patrick Turner from the R.A.T. group, (newsgroup: rec.audio.tubes) that
I paste below. The guy wants me to calculate Mu, and plate resistance as
well. After all, Gm = U / Ra.
Patrick's reply:

> What you are doing will broaden your ideas about tubes,
> but why concentrate on the Gm?
>
> If I want to find out how to set up a tube I find out what
> all three important parameters are, the Ra, U, and Gm.
>
> You can set up a tube with a choke load, and strap various values
> of RL across the choke.
>
> Then after the bias is set apply a fixed signal input voltage to get about
> half the maximum anode voltage swing before clipping, so thd is
negligible.
> Then measure the voltage gain, and do so for at least two different RLs,
> say 5k, and say 10k.
>
> Voltage Gain, A, = U x RL / ( RL + Ra ), for all tubes.
>
> So since we know what A is, and RL is, we have two equations for the
> two different loads, and from these we can work out
> Ra, U and thus Gm, because for all tubes,
> Gm = U / Ra.
>
> The most constant parameter for all tubes is the U,
> because its determined mainly by electrode distances,
> but Ra and Gm vary widely.
>
> The other way to measure Gm is to simply have no choke load,
> or large value RL, and just have a 10 ohm R to the B+,
> and measure the undistorted current change with grid voltage changes.
> This don't tell you about Ra or U, which change dramatically
> when G2 is connected to the anode.
>
> The first method of mine involves algebra.
> But from one of with two equations, U can be expressed as a factor x Ra,
> and this can be substituted where U appears in the second equation so that
> this then has only numbers and Ra present, so Ra can be found, and its
value
> inserted into either equation, and the value of U found,
> and thus Gm.
>
> The idle conditions for the tubes can be changed, and the 3
> parameters will change.
>
> The tests and measurements can be done using a class AB PP pair,
> which will indicate what the 3 design values are when combined.
> In AB circuits, the Gm usually drops a bit after one tube cuts off,
> resulting in a flattening of the amplified sign wave,
> which is compression, or lot of 3H present.
> But when biased to near class B, the crossover distortion indicates
> that Gm is initially low, then becomes greater as more tube current flows,
> and the phase of the resulting 3H is opposite that of compression.
> Some tubes, like the 6DQ6, display increasing Gm
> even in class AB circuits at the extreme of the amplitude,
> so that the wave peaks up before clipping, and again the 3H
> is opposite phase to a compressed wave, so Gm is increasing lots,
> even up to just before clipping.
>
> In triode signal tubes, such as 12AT7, 6DJ8,
> quite large variations in Gm and Ra occur for different
> idle currents, and some have deliberately engineered
> variable parameters by means of variably spaced grid wires.
> Many pentodes have also thus been made to have variable U and Gm,
> to make the gain vary a heck of a lot, which is handy for
> gain control in radio receivers, or to provide a more
> constant sound level from microphones, regardless of the distance away
from
> the person using it becomes.
>
> Once you have established that a given load is ideal for
> a pair of say 6L6GC tubes, then quite a few others might be substituted.
> Speaker loads vary between say 4 and 40 ohms, so absolutely strict
> load matches are impossible.
> It pays to err on the conservative side, IMHO,
> ie, allow for a high value RL than a low one,
> as the best fidelity is then possible, and tube substitutions
> are more successful.
>
> Patrick Turner.

-----Original Message-----
From: amradio-admin at mailman.qth.net
[mailto:amradio-admin at mailman.qth.net]On Behalf Of John Coleman, ARS
WA5BXO
Sent: Monday, October 13, 2003 9:37 PM
To: amradio at mailman.qth.net
Subject: RE: [AMRadio] Testing tubes

Actually Jim,
	You could probably use a 100 ohm resistor in series with the
plate so as to have 1 volt measured change = 10 ma change.  The 10 volt
drop in plate voltage will not alter the plate current enough to be
perceptible so long as the screen is constant.  But it is easier to
measure and you wouldn't need quite the large filter on the power
supply.  It need only be large enough capacitance to bypass the testing
frequency.  And if the testing frequency was at about 1000 CPS then it
would be more accurate than at 60 CPS.  I have drawn up what I think
would be a good testing circuit for Gm. It might even show distortion
factors if the scope was to be connected vertical to input and
horizontal to output.
http://wa5bxo.shacknet.nu/HAMPICTURES/transconductance_check.gif

What do you think?

John, WA5BXO

-----Original Message-----
From: amradio-admin at mailman.qth.net
[mailto:amradio-admin at mailman.qth.net]On Behalf Of John Coleman
Sent: Monday, October 13, 2003 10:04 AM
To: amradio at mailman.qth.net
Subject: RE: [AMRadio] Testing tubes

Hi Jim
	The idea that you have a triode connection is not exactly valid for
this type of test because there is no series load resistor.  The plate and
screen voltage are held constant regardless of the plate current.  It is
however important that you measure the plate current separately from the
screen current, if you want to measure the GM of a tetrode connection.
Changes in screen current, is not what the test is about.  The plate and
screen voltage may come from the same source.  The screen voltage is most
critical and must be maintained constant as close as possible.  Since plate
and screen come from the same source then the source voltage should be
monitored with a DVM and adjusted to be the constant of 250 Volts.  This
will require a little juggling each time you make a change in grid voltage.
The actual plate current that you start with (90-110 ma) is not so
important.  But the change (delta) Ip with grid voltage change is what we
are looking for.  The only other thing that might be a problem is that the
plate dissipation will be about 25 watts.  Is this tube OK for that during
the test?  I would not use a delta Eg of more than 5 volts to make the test
this would yield a change in plate current of more than 46.5 ma which is
about 50 percent of the quiescent current of 100 ma.  It is important that
the increase in current when the Eg is made more negative be equal to the
decrease in current when the Eg is made less negative by the same amount.
In other words it should be linear. Keep in mind that when the plate current
goes to 150 ma or more that the plate dissipation will be 50 percent more so
make you measurement fast and don't for get to hold the EP constant with
that DVM.

You could also place a 1 ohm resistor in the plate lead and capacitively
couple to a sensitive AC voltmeter (scope) and measure the AC current
produced as a change in Eg AC. A 100 ma change would be represented as
0.1VPTP on the scope.  You would need a very large capacitor to Filter the
source with.

73
John, WA5BXO

-----Original Message-----
From: amradio-admin at mailman.qth.net [mailto:amradio-admin at mailman.qth.net]
On Behalf Of Jim candela
Sent: Sunday, October 12, 2003 11:08 PM
To: amradio at mailman.qth.net
Subject: RE: [AMRadio] Testing tubes

    Hi Group,

    I got a wild hair, and made a setup to compare a set of used Genalex
KT77's to a new MP (matched pair) of Sovetek EL34's. I was following a note
in the KT77 data sheet that stated when a KT77 is considered bad (< 9.3
ma/volt transconductance when EP, Eg2 = 250 volts held constant, and IP =
100 ma). Following this criteria, I made a test setup. I am not too sure of
my technique as it is a work in progress. I am curious that the Gm can vary
considerably when changing Vg1 +/- 1 volt from the value used to obtain 100
ma plate current. Maybe this is all wrong, but I averaged the two numbers
assuming that is the final answer. Since Genalex stated to run the screen
and plate at 250 volts, I connected them together (Fluke 8012a looks at
plate current only), so this test is in triode mode (or is it not?). For a
high voltage supply I used my Sherwood S8000 stereo amp / FM with a variac
(get ~ 250 vdc at 65 volts AC input). Hey it was there (awaiting a re-cap
job), convenient, and able to do the job!

    Is the data I've taken valid? If not, can someone enlighten me? Maybe I
need to vary Vg1 less than +/- 1 volt, for better data?

Regards,
Jim Candela

Test results:
http://pages.prodigy.net/jcandela/KT77/genalex_kt77.htm

KT77 Data Sheet:
http://pages.prodigy.net/jcandela/KT77/kt77.pdf

Test Setup:
http://pages.prodigy.net/jcandela/KT77/PA110015.JPG

_______________________________________________
AMRadio mailing list
AMRadio at mailman.qth.net
http://mailman.qth.net/mailman/listinfo/amradio

_______________________________________________
AMRadio mailing list
AMRadio at mailman.qth.net
http://mailman.qth.net/mailman/listinfo/amradio