I am looking at how to set up the Hickok 580A to test a tube using the
tube specs in a tube data manual.

In the example given in the Hickok manual, a tube is set up for plate,
screen, and bias voltages. The instructions then say that the Gm should
be referenced to that given in the tube manual.

Here's what has me scratching my head. Gm is a function of the input
signal voltage. The higher the signal voltage, the higher the Gm, the
lower the input signal voltage, the lower the Gm.

The 580A has a signal voltage of around .28 vrms. However, the tube
specs in the tube manual do not provide a signal voltage used to obtain
the resultant Gm listed in the tube manual.

Is .28 vrms a universal signal voltage used by tube manufacturers to
obtain these arbitrary Gm values? If it is not, there is really no way
to obtain accurate comparison between the 580A(or any tester that can
use these tube manual specs instead of the roll chart) and the tube
data manuals.

Thanks

Here's what has me scratching my head. Gm is a function of the input
signal voltage.


** No it's not.

The higher the signal voltage, the higher the Gm, the
lower the input signal voltage, the lower the Gm.


** No way.

Gm is a tube's RATIO between an incremental ( big word for small ) change
in the grid voltage and the resulting change in plate current.




......... Phil

Phil Allison wrote:


Here's what has me scratching my head. Gm is a function of the input
signal voltage.


** No it's not.

The higher the signal voltage, the higher the Gm, the
lower the input signal voltage, the lower the Gm.


** No way.

Gm is a tube's RATIO between an incremental ( big word for small ) change
in the grid voltage and the resulting change in plate current.

What I stated is absolutely true. In fact, the Hickok 752 varies the
input signal voltage to change it's Gm ranges.

Delta Vin can be expressed in rms voltage. Delta I(plate) can be
expressed as rms plate current.

Transconductance in umhos is equal to Ip(in ma) divided by the signal
voltage x 1000.

So if an input signal of 2 vrms causes an Iplate(rms) of 10 ma, the
transconductance of that tube under those conditions is 4000 umhos

Sorry... 4000 umhos should be 5000 umhos

...

Phil Allison wrote:

Here's what has me scratching my head. Gm is a function of the input
signal voltage.


** No it's not.

The higher the signal voltage, the higher the Gm, the
lower the input signal voltage, the lower the Gm.


** No way.

Gm is a tube's RATIO between an incremental ( big word for small )
change
in the grid voltage and the resulting change in plate current.

What I stated is absolutely true.


** Bull**** - you ****ing dumb ass - it was absolute crap.



In fact, the Hickok 752 varies the signal voltage to change it's Gm
ranges.


** Sure - that agrees with me.

You are so ****ing thick you dunno what a ratio is.


Delta Vin can be expressed in rms voltage. Delta I(plate) can be
expressed as rms plate current.

Transconductance in umhos is equal to Ip(in ma) divided by the signal
voltage x 1000.

So if an input signal of 2 vrms causes an Iplate(rms) of 10 ma, the
transconductance of that tube under those conditions is 4000 umhos


** So what ????????????????????

Double the input- you get double the output.

Gm stays the same.


**** OFFFFFF !!!!!




........... Phil

Phil Allison wrote:

** So what ????????????????????

Double the input- you get double the output.

Gm stays the same.

Okay, I don't know what I was thinking yesterday. My brain was stuck in
calibration mode :-)

But, since the transfer curves are not exactly linear, varying the
input signal by half, will not necessarily vary the output current by
half. In fact, I blueprint calibrated my 752 and even though the input
signal is halved, the Gm varies sometimes as much as 100-150 umhos on
the various ranges using a 12AX7, where it was initially reading 1100.
That's a pretty good difference. This "error" varies from tube to tube,
depending on the linearity of it's transfer curve at those particular
operating conditions.



**** OFFFFFF !!!!!

Don't take it personally ;-)





.......... Phil

wrote

Is .28 vrms a universal signal voltage used by tube manufacturers
to
obtain these arbitrary Gm values? If it is not, there is really no
way
to obtain accurate comparison between the 580A(or any tester that
can
use these tube manual specs instead of the roll chart) and the
tube
data manuals.

Ideally and literally you are right but the difference won't be
great. I sense a deeper misunderstanding.

You're lucky. With my AVO you have to jiggle the grid voltage by
hand using a telephone dial with a non-linear scale, and no meter,
with rectified AC on the anode. Still works though.

Essentially you are trying to find the gradient of the
characteristic curve at a particular point. Ideally you would need
to construct a tangent. Your method approximates to a tangent by
using a *small* incremental change. gm varies only slightly with
small changes in voltage.

You get gm at the particular DC operating point you are using. I
assume that a quoted figure for gm in a data sheet is at the given
typical operating point.

Does anyone know if there is a standard way of measuring it? Close
as you can get I guess. Perhaps use a curve tracer and draw a
tangent? Perhaps they all used Hickok 580s? Don't worry about it.

cheers, Ian

On 7 Aug 2005 13:01:18 -0700, wrote:

Phil Allison wrote:

** So what ????????????????????

Double the input- you get double the output.

Gm stays the same.

Okay, I don't know what I was thinking yesterday. My brain was stuck in
calibration mode :-)

But, since the transfer curves are not exactly linear, varying the
input signal by half, will not necessarily vary the output current by
half. In fact, I blueprint calibrated my 752 and even though the input
signal is halved, the Gm varies sometimes as much as 100-150 umhos on
the various ranges using a 12AX7, where it was initially reading 1100.
That's a pretty good difference. This "error" varies from tube to tube,
depending on the linearity of it's transfer curve at those particular
operating conditions.



Obviously the gain will change a bit as you change the operating point of the
tube and its signal level, but that's not the point to testing tubes, is it?

Normally one just sets up a test with a 'working' bias and a signal that stays
in the linear region, and then uses it to compare tubes. Or run a signal from
zero up to some set point, it doesn't matter if all tubes are tested the same.

You could also set up an automatic tester to vary the op point until you reach
maximum gain, but then that tube would only perform in one circuit...

It's a bit like transistors, but not as bad... I've seen transistor beta range
from 200 to 900 on the same part number!

"Robbie"

It's a bit like transistors, but not as bad... I've seen transistor beta
range
from 200 to 900 on the same part number!


** You have simply not lived.


Transistor beta values ranges from near zero to several hundreds for the
same device, depending on the actual collector current chosen for the test.
For any power transistor - beta is a curve on a graph NOT a number.

Most power transistors are not even speced for beta below 100mA, while at
near maximum rated current, beta values fall to single digit numbers.

Check out the 2N3055 for example - see figure 2 in this pdf.

http://perso.wanadoo.es/chyryes/comp...55-MJ15016.pdf





........... Phil

wrote:

I am looking at how to set up the Hickok 580A to test a tube using the
tube specs in a tube data manual.

In the example given in the Hickok manual, a tube is set up for plate,
screen, and bias voltages. The instructions then say that the Gm should
be referenced to that given in the tube manual.

Here's what has me scratching my head. Gm is a function of the input
signal voltage. The higher the signal voltage, the higher the Gm, the
lower the input signal voltage, the lower the Gm.

The 580A has a signal voltage of around .28 vrms. However, the tube
specs in the tube manual do not provide a signal voltage used to obtain
the resultant Gm listed in the tube manual.

Is .28 vrms a universal signal voltage used by tube manufacturers to
obtain these arbitrary Gm values? If it is not, there is really no way
to obtain accurate comparison between the 580A(or any tester that can
use these tube manual specs instead of the roll chart) and the tube
data manuals.

Thanks

Tube testers test everyone's patience it seems.

I am not aware of the 0.28v being a universal signal input
for a grid signal in a tube tester, but it would be OK for most tubes
since even with a triode with a gain of 100, the output voltage would only
be 28vrms, and with most test conditions, such an output voltage shouldn't
be in the
high distortion region of the tube under test.

But then I never use a tube tester.
They never tell me enough about the tube I am testing,
and if I test a tube i wanna know if its microphonic or noisy
or gassy or plum wore out etc and to know all I wanna know
I have a test jig where I can change the RL to work the gain for two values
of RL,
and work out what Ra, Gm and µ are for the tube at the operating point I
have selected, and which I can vary.

For all tubes, gain, A, = µ x RL / ( RL + Ra ).
Therefore µ = A x ( RL + Ra ) / RL.

And for all tubes, Gm = µ / Ra
µ is the amplification factor which is the most constant parameter,
Gm is the transconductance in A/V,
Ra is the dynamic plate resistance in ohms.

After accurately measuring the gain of a tube well below clipping
with say 50K = RL1 and 100K = RL2, you will get two gains, A1 and A2.
And you will get two equations to work out µ with RL1, and µ with RL2.
Since µ has the same value for both equations,
A1 x ( RL1 + Ra ) / RL1 = A2 x ( RL2 + Ra ) / RL2

Notice that you will have measured A1 and A2, and you know what RL1 and RL2
are,
and the only unknown is Ra, which will be the same for both RL1 and RL2
**providing we have done the test with quiescent Ia for both RLs being the
same**

Therefore Ra can be easily worked out, and from that the value of µ by
substitution,
and after that Gm.
It was simple high school maths when I went to school 40 years ago.

Then you can compare what you have measured to the data sheets in the old
books.
If you test a 12AX7 and get µ = 95 to 105, and Ra about 65kohms,
all is well with the AX7, but that still leaves the tests for noise and
grid current at idle after the tube has been on for 1/2 an hour.

The grid voltage should not be much above the bias supply value;
If Rg = 470k, and Eg is 2V above the bias supply, the tube is probably
stuffed; most are noisy
if you read this +ve grid voltage.
The noise is best measured by using a preamp with a gain of 1,000 and
bandwidth of 20kHz, and
not any more than 20 kHz, unless you are measuring RF tubes for RF noise.
..
Ground the grid with a short wire to the bias supply so no noise
from external sources can enter the grid.
Connect the preamp to the tube anode via a .47 uF cap.
Say the tube creates a grid input noise of 2uV.
( And if you are lost about noise and all this, you have two choices,
read up and learn from the old books, or remain ignorant ).
Say you have measured the gain of the tube with 100k RL at say 60
for your 12AX7.
Then its safe to say you will have approximately 60 x 2uV of noise at the
anode,
or 0.120mV, which is too small to measure accurtaely, but it is *much*
larger than
the noise of your preamp, even if its an opamp chip.

With a gain of 1,000, the prreamp will amplify the noise to 120mV which is
easily
measured.

Obviously, you won't know what the noise is when you start, but knowing the
preamp and tube gain will allow you
to work out what the noise must be at the input grid of the tube under
test.
If you get input noise = 10uV, your 12AX7 is stuffed.
2 uV is a reasonable figure.

Hum from the heaters may dominate the noise measurements, so have a DC
supply
ready for the heaters. If when AC is used, hum is bad despite what nulling
you have,
then into the bin with the tube.
The hum will show itself amoungst the other noise on the CRO, and a speaker
will tell you what it sounds like.
a loud thumpetythumpety noise means there is flicker noise, and the tube is
stuffed.
you should aim for tubes to give a continuous uniform
"ssssssssssssssssssssssssssssssssssssssssssssssss" ,
without any sputtering noises.
When you test a few tubes microphonic tubes become obvious; the quiet ones
can be tapped with a finger and not make much noise, the stuffed tubes
clang
like the bells at St Mary's.

After testing 20 tubes from your junk box you'll get pretty good at lobbing
tubes into the bin,
and treasuring the ones you find are quiet, unmicrophonic, and which have
full expected gains
They will sound ok.

Patrick Turner.

Phil Allison wrote:



Here's what has me scratching my head. Gm is a function of the input
signal voltage.

** No it's not.

The higher the signal voltage, the higher the Gm, the
lower the input signal voltage, the lower the Gm.

** No way.

Gm is a tube's RATIO between an incremental ( big word for small ) change
in the grid voltage and the resulting change in plate current.

True. But the Gm, ( the plate current change / grid voltage change )
does actually rise *slightly* as the Vg is raised above the quiescent Egq
and it decreases when Vg falls below Egq for most triodes etc, but the reason
for using a low grid test signal
is to keep the Gm variation to a very small quantity which does not
alter the average Gm for that Egq and Eaq and Iaq test condition.
The variations in Gm during a wave cycle are what vary Ra, and hence what cause
the distortion in a tube.

There's a heck of a lot more info in the old books.

Patrick Turner.




........ Phil

wrote:

Phil Allison wrote:


Here's what has me scratching my head. Gm is a function of the input
signal voltage.


** No it's not.

The higher the signal voltage, the higher the Gm, the
lower the input signal voltage, the lower the Gm.


** No way.

Gm is a tube's RATIO between an incremental ( big word for small ) change
in the grid voltage and the resulting change in plate current.

What I stated is absolutely true. In fact, the Hickok 752 varies the
input signal voltage to change it's Gm ranges.

Delta Vin can be expressed in rms voltage. Delta I(plate) can be
expressed as rms plate current.

Transconductance in umhos is equal to Ip(in ma) divided by the signal
voltage x 1000.

So if an input signal of 2 vrms causes an Iplate(rms) of 10 ma, the
transconductance of that tube under those conditions is 4000 umhos

This would be right with no gain, ie, the anode taken to a B+
voltage supply with no RL.
The 2V is still small enough not to cause much distortion current,
and the Gm for the Ea & Iaq could be established.

You see if you had a CCS load at the anode, the 2V grid input
could not possibly cause any anode current change.


Patrick Turner.

Phil Allison wrote:

"Robbie"

It's a bit like transistors, but not as bad... I've seen transistor beta
range
from 200 to 900 on the same part number!


** You have simply not lived.

Transistor beta values ranges from near zero to several hundreds for the
same device, depending on the actual collector current chosen for the test.
For any power transistor - beta is a curve on a graph NOT a number.

Most power transistors are not even speced for beta below 100mA, while at
near maximum rated current, beta values fall to single digit numbers.

Check out the 2N3055 for example - see figure 2 in this pdf.

http://perso.wanadoo.es/chyryes/comp...55-MJ15016.pdf

.......... Phil

Transistors become more manageable with current FB and voltage FB,
without which they'd be useless.

But with tubes the parameters seem easier to get one's head around,
since less extreme variations occur between samples.

If "living" must include knowing bjt BS then its a hard life for
people at r.a.t :-]

Patrick Turner.

Ian Iveson wrote:

Ideally and literally you are right but the difference won't be
great.


I sense a deeper misunderstanding.

Na... just too much thinking on too little sleep. The next morning I
awoke and it all made sense to me. I've been caught up in the
calibration thing and keep thinking that a real tube behaves like the
calibration fixture. One of these days, I'll quit asking dumb
questions, or at least sleep on them before I ask :-)



cheers, Ian

On Mon, 8 Aug 2005 11:44:58 +1000, "Phil Allison"
wrote:


"Robbie"

It's a bit like transistors, but not as bad... I've seen transistor beta
range
from 200 to 900 on the same part number!


** You have simply not lived.


Transistor beta values ranges from near zero to several hundreds for the
same device, depending on the actual collector current chosen for the test.
For any power transistor - beta is a curve on a graph NOT a number.

Most power transistors are not even speced for beta below 100mA, while at
near maximum rated current, beta values fall to single digit numbers.

Check out the 2N3055 for example - see figure 2 in this pdf.

http://perso.wanadoo.es/chyryes/comp...55-MJ15016.pdf



Very interesting, Phil... no wonder SS amplifiers need tons of feedback!

I used to have an amp using those 2N3055 transistors, I didn't know they were
the same as the MJ15015... I may have some of those in stock too!

I never saw any MJ2955's, only the MJE2955, which is plastic... and a funny
pinout if I recall...

Thanks for the memories!!


.......... Phil

Phil Allison wrote:

Transistor beta values ranges from near zero to several hundreds for the
same device, depending on the actual collector current chosen for the test.


For any power transistor - beta is a curve on a graph NOT a number.

Really? It's spec'ed 3 seperate times(for different operating
characteristics) in the pdf you listed below.



http://perso.wanadoo.es/chyryes/comp...55-MJ15016.pdf





.......... Phil

Phil Allison wrote:

Transistor beta values ranges from near zero to several hundreds for the
same device, depending on the actual collector current chosen for the
test.


For any power transistor - beta is a curve on a graph NOT a number.

Really?


** Yes really - you dumb ****.


It's spec'ed 3 seperate times(for different operating
characteristics) in the pdf you listed below.


** So three seperate points on the curve.

Gotta see the whole curve to know all the points.





............ Phil

Phil Allison wrote:

Phil Allison wrote:

Transistor beta values ranges from near zero to several hundreds for the
same device, depending on the actual collector current chosen for the
test.


For any power transistor - beta is a curve on a graph NOT a number.

Really?


** Yes really - you dumb ****.


It's spec'ed 3 seperate times(for different operating
characteristics) in the pdf you listed below.


** So three seperate points on the curve.

You said they never list it numerically. In fact, typical and maximum
hfe values are always given. Just wanted to give you a heads up.

Sit back, take a deep breath and think happy thoughts... It'll be okay
:-)

Phil Allison wrote:

Transistor beta values ranges from near zero to several hundreds for
the
same device, depending on the actual collector current chosen for the
test.


For any power transistor - beta is a curve on a graph NOT a number.

Really?


** Yes really - you dumb ****.


It's spec'ed 3 seperate times(for different operating
characteristics) in the pdf you listed below.


** So three seperate points on the curve.

You said they never list it numerically.


** Where ??????

You must be hallucinating.



In fact, typical and maximum
hfe values are always given.


** That is to allows for device to device variations, plus is quoted at a
particular Ic.

But for a given power device, "hfe" is not a single number - it is a curve
on a graph.



Just wanted to give you a heads up.


** Go to hell - ****head.




.............. Phil

"flipper"
Phil Allison

" For any power transistor - beta is a curve on a graph NOT a number. "


You said they never list it numerically.

No, he didn't. He said "beta is a curve" and not "a number."

That is a characteristic of "beta" and no where did he make a
statement of things 'never listed'.



** The utterly mad idea that transistor beta is a simple number,
characteristic of a particular sample of a device, has been around for
decades. One of the idiotic conclusions that follows from this error is
using a basic "transistor tester", intended for small signal devices, to
check power devices.

Results are all over the place, with some samples showing credible numbers
but most others showing very low or else no beta at all !!!!

I know of a case where hundreds of new Motorola power devices were returned
as faulty because beta readings on a $15 meter were only 5 to 10 !!

A power transistor beta tester must operate at a known Ic of at least 200 mA
to get sensible numbers - this means adjusting the base current to a device
under test get an Ic of 200 mA or more - then noting the actual base
current value and computing beta.





......... Phil

"flipper"
"Phil Allison"

** The utterly mad idea that transistor beta is a simple number,
characteristic of a particular sample of a device, has been around for
decades. One of the idiotic conclusions that follows from this error is
using a basic "transistor tester", intended for small signal devices, to
check power devices.

You'd think the min-max range would be clue enough to realize it isn't
a uniform constant..



** Try switching your brain on.


" The utterly mad idea that transistor beta is a simple number,
characteristic of a particular sample of a device, "

The meter is assumed to reveal the beta of the device under test.

The false assumption is that although there is a known wide variation from
sample to sample - a single, low current, test is enough for any one
device.




....... Phil

"flipper"
Phil Allison

** Try switching your brain on.

My fault for attempting a cordial conversation.



** Won't happen with your brain switched off.

Note - I gave you credit for having one available.




.......... Phil

Phil Allison wrote:

** Go to hell - ****head.

I bet you think you're pretty slick, don't 'cha?





............. Phil