I've always been frustrated when trying to test large tubes (813, 805,
826, etc) because none of my tube testers can handle them. I suppose
its possible to make an adaptor to use a regular tube tester, but using an
external filament supply, but that would be a crude test at best. After
thinking
about it for a while, I've decided to build a curve tracer that will handle
large tubes (and smaller ones as well). Actually, its not that complex.
I'll use a Variac driving a 20 volt @10 amp filament transformer for the
filament supply, and another Variac to drive a 1600 volt transformer for
the plate. By stepping the grid between +100 v and -100 v (DC, and variable
ranges), a set of I-V plate curves should be possible.

At this point, I'm just thinkng about it. Has anyone else built something
like this?

Steve

Steve,

Yep, I have. An example is in eBay auction # 5725492214 (no one's bought it,
alas). I used something similar to what you say, although I will add that I
think it would be best if the filament tranny is center tapped (which I did not
do). I used a 1 ohm resistor between the cathode and ground to get the current
measurement -- which means the filament supply cannot be grounded -- and a
voltage divider to get the plate voltage, a 100 : 1 ratio seems to work pretty
well. It doesn't work very well for the small tubes, however. I think the 1 ohm
current resistor is the main culprit (I need 10 ohm and 100 ohm option). It
might eventually be necessary to use a different voltage ratio as well, but I
don't think so. Note that a resistor in the cathode lead introduces an error by
changing the grid-cathode voltage as the plate current rises (unless you can
tie the grid supply to the cathode), but it's not much of an error in most
cases (0.1 V at 100 mA for a 211 tube, for example). I have a Fluke voltage
supply with switches that make it pretty easy to get a set of curves in the 15
seconds that my digital camera will allow.

Lessons learned: Most scopes have a "z" input which will blank the traces while
switching from one grid voltage to another. You could press a button while
switching and eliminate the "extra" traces like you see on my photos.

You really need some kind of power limit. Withough some type of power limit, it
is almost impossible to get a decent set of curves, because not only is it
difficult to hand adjust the plate voltage for each curve, from memory of the
proceeding curve, to the correct point, but in addition, my scope (a Tek
2215A), and probably others, shift the curves as a function of the peak x-y
position! A good approximation for power is a resistor in series with the
primary of the transformer, although this makes a series of curves that descend
in a straight line, actually like you see in many of the old sets of plate
curves. ;-) I am in the process of updating my tracer, and I am going to try
something much fancier, but I have an old 1000 W 50 ohm rheostat that I am
repairing (bought off eBay). It works as is (now that I cleaned up the
contacts), but the ends were cracked, and they don't fit properly into the
metal pieces at each end, so I'm going to try to fix that before putting it
back up on eBay (or selling it to you, if you wish). That and a few big power
resistors, which I also have, would make it possible to set the plate voltage
once, and then simply switch the grid voltage, and wind up with a nice looking
set of curves like you can see in auction # 5719015311 (this happened to be the
unusually high resistance of my variac and output transformers).

I used the natural diode ability of a tube to take care of the negative voltage
swing, but I suspect it would be much better to use a real diode on the input
of the HV tranny.

That's all I can think of for now, but we can obviously discuss this much more!

Phil

Steve wrote:

I've always been frustrated when trying to test large tubes (813, 805,
826, etc) because none of my tube testers can handle them. I suppose
its possible to make an adaptor to use a regular tube tester, but using an
external filament supply, but that would be a crude test at best. After
thinking
about it for a while, I've decided to build a curve tracer that will handle
large tubes (and smaller ones as well). Actually, its not that complex.
I'll use a Variac driving a 20 volt @10 amp filament transformer for the
filament supply, and another Variac to drive a 1600 volt transformer for
the plate. By stepping the grid between +100 v and -100 v (DC, and variable
ranges), a set of I-V plate curves should be possible.

At this point, I'm just thinkng about it. Has anyone else built something
like this?

Steve

As I thought about this, the thing to consider is using a differential or
instrumentation amplifier on the current sense resistor -- you can get an
off-the-shelf device with gain of 1000 set by a single resistor -- plenty of
devices from Analog Devices or Burr-Brown (TI). Some of the diff amps
already have clamp diodes tied to the V+/V- rails on the input pins. You
can get away with a resistor small enough not to affect the voltage in tube
applications.

The problem with any of these devices is that tubes tend to be noisy
(compared to other things you might want to measure) so the measurements
have to be integrated...

You might want to consider using an oscillator and a power amp to drive the
variac at 1kHz -- the 60Hz signal is pretty darn slow for viewing meaningful
information with a scope. If you have a signal generator with RAMP
function, you don't have to worry about the retrace. Of course, if you are
doing this you can get rid of the variac too, just control the oscillator
output. The tube tracer article in one of the first issues of GLASS AUDIO
used the 60 Hz to sweep the plate voltage, and ganged C- supplies which were
controlled by a couple CMOS chips. I think that the author pointed out that
you needed to mentally integrate the traces !

Jack

"Phil" wrote in message ...
Steve,

Yep, I have. An example is in eBay auction # 5725492214 (no one's bought
it,
alas). I used something similar to what you say, although I will add that
I
think it would be best if the filament tranny is center tapped (which I
did not
do). I used a 1 ohm resistor between the cathode and ground to get the
current
measurement -- which means the filament supply cannot be grounded -- and a
voltage divider to get the plate voltage, a 100 : 1 ratio seems to work
pretty
well. It doesn't work very well for the small tubes, however. I think the
1 ohm
current resistor is the main culprit (I need 10 ohm and 100 ohm option).
It
might eventually be necessary to use a different voltage ratio as well,
but I
don't think so. Note that a resistor in the cathode lead introduces an
error by
changing the grid-cathode voltage as the plate current rises (unless you
can
tie the grid supply to the cathode), but it's not much of an error in most
cases (0.1 V at 100 mA for a 211 tube, for example). I have a Fluke
voltage
supply with switches that make it pretty easy to get a set of curves in
the 15
seconds that my digital camera will allow.

Lessons learned: Most scopes have a "z" input which will blank the traces
while
switching from one grid voltage to another. You could press a button while
switching and eliminate the "extra" traces like you see on my photos.

You really need some kind of power limit. Withough some type of power
limit, it
is almost impossible to get a decent set of curves, because not only is it
difficult to hand adjust the plate voltage for each curve, from memory of
the
proceeding curve, to the correct point, but in addition, my scope (a Tek
2215A), and probably others, shift the curves as a function of the peak
x-y
position! A good approximation for power is a resistor in series with the
primary of the transformer, although this makes a series of curves that
descend
in a straight line, actually like you see in many of the old sets of plate
curves. ;-) I am in the process of updating my tracer, and I am going to
try
something much fancier, but I have an old 1000 W 50 ohm rheostat that I am
repairing (bought off eBay). It works as is (now that I cleaned up the
contacts), but the ends were cracked, and they don't fit properly into the
metal pieces at each end, so I'm going to try to fix that before putting
it
back up on eBay (or selling it to you, if you wish). That and a few big
power
resistors, which I also have, would make it possible to set the plate
voltage
once, and then simply switch the grid voltage, and wind up with a nice
looking
set of curves like you can see in auction # 5719015311 (this happened to
be the
unusually high resistance of my variac and output transformers).

I used the natural diode ability of a tube to take care of the negative
voltage
swing, but I suspect it would be much better to use a real diode on the
input
of the HV tranny.

That's all I can think of for now, but we can obviously discuss this much
more!

Phil

Steve wrote:

I've always been frustrated when trying to test large tubes (813, 805,
826, etc) because none of my tube testers can handle them. I suppose
its possible to make an adaptor to use a regular tube tester, but using
an
external filament supply, but that would be a crude test at best. After
thinking
about it for a while, I've decided to build a curve tracer that will
handle
large tubes (and smaller ones as well). Actually, its not that complex.
I'll use a Variac driving a 20 volt @10 amp filament transformer for the
filament supply, and another Variac to drive a 1600 volt transformer for
the plate. By stepping the grid between +100 v and -100 v (DC, and
variable
ranges), a set of I-V plate curves should be possible.

At this point, I'm just thinkng about it. Has anyone else built
something
like this?

Steve

John Walton wrote:

As I thought about this, the thing to consider is using a differential or
instrumentation amplifier on the current sense resistor -- you can get an
off-the-shelf device with gain of 1000 set by a single resistor -- plenty of
devices from Analog Devices or Burr-Brown (TI). Some of the diff amps
already have clamp diodes tied to the V+/V- rails on the input pins. You
can get away with a resistor small enough not to affect the voltage in tube
applications.

The problem with any of these devices is that tubes tend to be noisy
(compared to other things you might want to measure) so the measurements
have to be integrated...

You might want to consider using an oscillator and a power amp to drive the
variac at 1kHz -- the 60Hz signal is pretty darn slow for viewing meaningful
information with a scope.

The early TEK curve tracers used full wave rectified AC for the main sweep.
After getting that info from the TEK salesman of the time (1960) I built a
curve tracer which used a Polaroid camera on a TEK 541 to do some
curve tracing. The G1 voltage was switched manually. In the end I got
a complete set of curves on the photo for a number of tubes.

Some of those photos are still here, somewhere.

Cheers, John Stewart

If you have a signal generator with RAMP
function, you don't have to worry about the retrace. Of course, if you are
doing this you can get rid of the variac too, just control the oscillator
output. The tube tracer article in one of the first issues of GLASS AUDIO
used the 60 Hz to sweep the plate voltage, and ganged C- supplies which were
controlled by a couple CMOS chips. I think that the author pointed out that
you needed to mentally integrate the traces !

Jack

"Phil" wrote in message ...
Steve,

Yep, I have. An example is in eBay auction # 5725492214 (no one's bought
it,
alas). I used something similar to what you say, although I will add that
I
think it would be best if the filament tranny is center tapped (which I
did not
do). I used a 1 ohm resistor between the cathode and ground to get the
current
measurement -- which means the filament supply cannot be grounded -- and a
voltage divider to get the plate voltage, a 100 : 1 ratio seems to work
pretty
well. It doesn't work very well for the small tubes, however. I think the
1 ohm
current resistor is the main culprit (I need 10 ohm and 100 ohm option).
It
might eventually be necessary to use a different voltage ratio as well,
but I
don't think so. Note that a resistor in the cathode lead introduces an
error by
changing the grid-cathode voltage as the plate current rises (unless you
can
tie the grid supply to the cathode), but it's not much of an error in most
cases (0.1 V at 100 mA for a 211 tube, for example). I have a Fluke
voltage
supply with switches that make it pretty easy to get a set of curves in
the 15
seconds that my digital camera will allow.

Lessons learned: Most scopes have a "z" input which will blank the traces
while
switching from one grid voltage to another. You could press a button while
switching and eliminate the "extra" traces like you see on my photos.

You really need some kind of power limit. Withough some type of power
limit, it
is almost impossible to get a decent set of curves, because not only is it
difficult to hand adjust the plate voltage for each curve, from memory of
the
proceeding curve, to the correct point, but in addition, my scope (a Tek
2215A), and probably others, shift the curves as a function of the peak
x-y
position! A good approximation for power is a resistor in series with the
primary of the transformer, although this makes a series of curves that
descend
in a straight line, actually like you see in many of the old sets of plate
curves. ;-) I am in the process of updating my tracer, and I am going to
try
something much fancier, but I have an old 1000 W 50 ohm rheostat that I am
repairing (bought off eBay). It works as is (now that I cleaned up the
contacts), but the ends were cracked, and they don't fit properly into the
metal pieces at each end, so I'm going to try to fix that before putting
it
back up on eBay (or selling it to you, if you wish). That and a few big
power
resistors, which I also have, would make it possible to set the plate
voltage
once, and then simply switch the grid voltage, and wind up with a nice
looking
set of curves like you can see in auction # 5719015311 (this happened to
be the
unusually high resistance of my variac and output transformers).

I used the natural diode ability of a tube to take care of the negative
voltage
swing, but I suspect it would be much better to use a real diode on the
input
of the HV tranny.

That's all I can think of for now, but we can obviously discuss this much
more!

Phil

Steve wrote:

I've always been frustrated when trying to test large tubes (813, 805,
826, etc) because none of my tube testers can handle them. I suppose
its possible to make an adaptor to use a regular tube tester, but using
an
external filament supply, but that would be a crude test at best. After
thinking
about it for a while, I've decided to build a curve tracer that will
handle
large tubes (and smaller ones as well). Actually, its not that complex.
I'll use a Variac driving a 20 volt @10 amp filament transformer for the
filament supply, and another Variac to drive a 1600 volt transformer for
the plate. By stepping the grid between +100 v and -100 v (DC, and
variable
ranges), a set of I-V plate curves should be possible.

At this point, I'm just thinkng about it. Has anyone else built
something
like this?

Steve

!doctype html public "-//w3c//dtd html 4.0 transitional//en"
html
Yeah, 60 Hz will leave a lot of flicker, no doubt -- especially half-wave
60 Hz -- but I also found out, probably like you, that although you can
get a rough feel for a tube on the scope, you need a picture to really
do anything, like match tubes, or answer those questions like, "For this
design, I wonder how tube X would work?" You will always want to go straight
to a hardcopy of the plate curves. At present, therefore, I'm willing to
make a tracer with flicker, the tradeoff being a relatively inexpensive
3000 V, 1000 W, 1 A tracer, since I'm going to want a digital pic anyway.
Looking at some Tek plate curve photos, I noticed that they switched from
one curve to the next at the top of the curves, so they really got 240
curves per second with their full wave power supply!
pPhil
pJohn Stewart wrote:
blockquote TYPE=CITEJohn Walton wrote:
p As I thought about this, the thing to consider is using a differential
or
br instrumentation amplifier on the current sense resistor -- you can
get an
br off-the-shelf device with gain of 1000 set by a single resistor --
plenty of
br devices from Analog Devices or Burr-Brown (TI).  Some of the
diff amps
br already have clamp diodes tied to the V+/V- rails on the input pins. 
You
br can get away with a resistor small enough not to affect the voltage
in tube
br applications.
br
br The problem with any of these devices is that tubes tend to be noisy
br (compared to other things you might want to measure) so the measurements
br have to be integrated...
br
br You might want to consider using an oscillator and a power amp to
drive the
br variac at 1kHz -- the 60Hz signal is pretty darn slow for viewing
meaningful
br information with a scope.
pThe early TEK curve tracers used full wave rectified AC for the main
sweep.
brAfter getting that info from the TEK salesman of the time (1960) I
built a
brcurve tracer which used a Polaroid camera on a TEK 541 to do some
brcurve tracing. The G1 voltage was switched manually. In the end I got
bra complete set of curves on the photo for a number of tubes.
pSome of those photos are still here, somewhere.
pCheers, John Stewart
p  If you have a signal generator with RAMP
br function, you don't have to worry about the retrace.  Of course,
if you are
br doing this you can get rid of the variac too, just control the oscillator
br output.  The tube tracer article in one of the first issues
of GLASS AUDIO
br used the 60 Hz to sweep the plate voltage, and ganged C- supplies
which were
br controlled by a couple CMOS chips.  I think that the author
pointed out that
br you needed to mentally integrate the traces !
br
br Jack
br
br "Phil" wrote in message a "news:416F36 /a...
br Steve,
br
br Yep, I have. An example is in eBay auction # 5725492214 (no one's
bought
br it,
br alas). I used something similar to what you say, although I will
add that
br I
br think it would be best if the filament tranny is center tapped
(which I
br did not
br do). I used a 1 ohm resistor between the cathode and ground to
get the
br current
br measurement -- which means the filament supply cannot be grounded
-- and a
br voltage divider to get the plate voltage, a 100 : 1 ratio seems
to work
br pretty
br well. It doesn't work very well for the small tubes, however. I
think the
br 1 ohm
br current resistor is the main culprit (I need 10 ohm and 100 ohm
option).
br It
br might eventually be necessary to use a different voltage ratio
as well,
br but I
br don't think so. Note that a resistor in the cathode lead introduces
an
br error by
br changing the grid-cathode voltage as the plate current rises (unless
you
br can
br tie the grid supply to the cathode), but it's not much of an error
in most
br cases (0.1 V at 100 mA for a 211 tube, for example). I have a Fluke
br voltage
br supply with switches that make it pretty easy to get a set of curves
in
br the 15
br seconds that my digital camera will allow.
br
br Lessons learned: Most scopes have a "z" input which will blank
the traces
br while
br switching from one grid voltage to another. You could press a button
while
br switching and eliminate the "extra" traces like you see on my photos.
br
br You really need some kind of power limit. Withough some type of
power
br limit, it
br is almost impossible to get a decent set of curves, because not
only is it
br difficult to hand adjust the plate voltage for each curve, from
memory of
br the
br proceeding curve, to the correct point, but in addition, my scope
(a Tek
br 2215A), and probably others, shift the curves as a function of
the peak
br x-y
br position! A good approximation for power is a resistor in series
with the
br primary of the transformer, although this makes a series of curves
that
br descend
br in a straight line, actually like you see in many of the old sets
of plate
br curves. ;-) I am in the process of updating my tracer, and I am
going to
br try
br something much fancier, but I have an old 1000 W 50 ohm rheostat
that I am
br repairing (bought off eBay). It works as is (now that I cleaned
up the
br contacts), but the ends were cracked, and they don't fit properly
into the
br metal pieces at each end, so I'm going to try to fix that before
putting
br it
br back up on eBay (or selling it to you, if you wish). That and a
few big
br power
br resistors, which I also have, would make it possible to set the
plate
br voltage
br once, and then simply switch the grid voltage, and wind up with
a nice
br looking
br set of curves like you can see in auction # 5719015311 (this happened
to
br be the
br unusually high resistance of my variac and output transformers).
br
br I used the natural diode ability of a tube to take care of the
negative
br voltage
br swing, but I suspect it would be much better to use a real diode
on the
br input
br of the HV tranny.
br
br That's all I can think of for now, but we can obviously discuss
this much
br more!
br
br Phil
br
br Steve wrote:
br
br I've always been frustrated when trying to test large tubes (813,
805,
br 826, etc) because none of my tube testers can handle them. I
suppose
br its possible to make an adaptor to use a regular tube tester,
but using
br an
br external filament supply, but that would be a crude test at best.
After
br thinking
br about it for a while, I've decided to build a curve tracer that
will
br handle
br large tubes (and smaller ones as well). Actually, its not that
complex.
br I'll use a Variac driving a 20 volt @10 amp filament transformer
for the
br filament supply, and another Variac to drive a 1600 volt transformer
for
br the plate. By stepping the grid between +100 v and -100 v (DC,
and
br variable
br ranges),  a set of I-V plate curves should be possible.
br
br At this point, I'm just thinkng about it. Has anyone else built
br something
br like this?
br
br         & nbsp;      
Steve
br /blockquote
/html

Maybe you will want to borrow the power supply for my HT45 "Loudenboomer" -- will run full legal limit.

One of the guys on the DIYAUDIO site worked at Eimac in the 1960's.

Jack
"Phil" wrote in message ...
Yeah, 60 Hz will leave a lot of flicker, no doubt -- especially half-wave 60 Hz -- but I also found out, probably like you, that although you can get a rough feel for a tube on the scope, you need a picture to really do anything, like match tubes, or answer those questions like, "For this design, I wonder how tube X would work?" You will always want to go straight to a hardcopy of the plate curves. At present, therefore, I'm willing to make a tracer with flicker, the tradeoff being a relatively inexpensive 3000 V, 1000 W, 1 A tracer, since I'm going to want a digital pic anyway. Looking at some Tek plate curve photos, I noticed that they switched from one curve to the next at the top of the curves, so they really got 240 curves per second with their full wave power supply!
Phil

John Stewart wrote:

John Walton wrote:
As I thought about this, the thing to consider is using a differential or
instrumentation amplifier on the current sense resistor -- you can get an
off-the-shelf device with gain of 1000 set by a single resistor -- plenty of
devices from Analog Devices or Burr-Brown (TI). Some of the diff amps
already have clamp diodes tied to the V+/V- rails on the input pins. You
can get away with a resistor small enough not to affect the voltage in tube
applications.

The problem with any of these devices is that tubes tend to be noisy
(compared to other things you might want to measure) so the measurements
have to be integrated...

You might want to consider using an oscillator and a power amp to drive the
variac at 1kHz -- the 60Hz signal is pretty darn slow for viewing meaningful
information with a scope.

The early TEK curve tracers used full wave rectified AC for the main sweep.
After getting that info from the TEK salesman of the time (1960) I built a
curve tracer which used a Polaroid camera on a TEK 541 to do some
curve tracing. The G1 voltage was switched manually. In the end I got
a complete set of curves on the photo for a number of tubes.

Some of those photos are still here, somewhere.

Cheers, John Stewart

If you have a signal generator with RAMP
function, you don't have to worry about the retrace. Of course, if you are
doing this you can get rid of the variac too, just control the oscillator
output. The tube tracer article in one of the first issues of GLASS AUDIO
used the 60 Hz to sweep the plate voltage, and ganged C- supplies which were
controlled by a couple CMOS chips. I think that the author pointed out that
you needed to mentally integrate the traces !

Jack

"Phil" wrote in message ...
Steve,

Yep, I have. An example is in eBay auction # 5725492214 (no one's bought
it,
alas). I used something similar to what you say, although I will add that
I
think it would be best if the filament tranny is center tapped (which I
did not
do). I used a 1 ohm resistor between the cathode and ground to get the
current
measurement -- which means the filament supply cannot be grounded -- and a
voltage divider to get the plate voltage, a 100 : 1 ratio seems to work
pretty
well. It doesn't work very well for the small tubes, however. I think the
1 ohm
current resistor is the main culprit (I need 10 ohm and 100 ohm option).
It
might eventually be necessary to use a different voltage ratio as well,
but I
don't think so. Note that a resistor in the cathode lead introduces an
error by
changing the grid-cathode voltage as the plate current rises (unless you
can
tie the grid supply to the cathode), but it's not much of an error in most
cases (0.1 V at 100 mA for a 211 tube, for example). I have a Fluke
voltage
supply with switches that make it pretty easy to get a set of curves in
the 15
seconds that my digital camera will allow.

Lessons learned: Most scopes have a "z" input which will blank the traces
while
switching from one grid voltage to another. You could press a button while
switching and eliminate the "extra" traces like you see on my photos.

You really need some kind of power limit. Withough some type of power
limit, it
is almost impossible to get a decent set of curves, because not only is it
difficult to hand adjust the plate voltage for each curve, from memory of
the
proceeding curve, to the correct point, but in addition, my scope (a Tek
2215A), and probably others, shift the curves as a function of the peak
x-y
position! A good approximation for power is a resistor in series with the
primary of the transformer, although this makes a series of curves that
descend
in a straight line, actually like you see in many of the old sets of plate
curves. ;-) I am in the process of updating my tracer, and I am going to
try
something much fancier, but I have an old 1000 W 50 ohm rheostat that I am
repairing (bought off eBay). It works as is (now that I cleaned up the
contacts), but the ends were cracked, and they don't fit properly into the
metal pieces at each end, so I'm going to try to fix that before putting
it
back up on eBay (or selling it to you, if you wish). That and a few big
power
resistors, which I also have, would make it possible to set the plate
voltage
once, and then simply switch the grid voltage, and wind up with a nice
looking
set of curves like you can see in auction # 5719015311 (this happened to
be the
unusually high resistance of my variac and output transformers).

I used the natural diode ability of a tube to take care of the negative
voltage
swing, but I suspect it would be much better to use a real diode on the
input
of the HV tranny.

That's all I can think of for now, but we can obviously discuss this much
more!

Phil

Steve wrote:

I've always been frustrated when trying to test large tubes (813, 805,
826, etc) because none of my tube testers can handle them. I suppose
its possible to make an adaptor to use a regular tube tester, but using
an
external filament supply, but that would be a crude test at best. After
thinking
about it for a while, I've decided to build a curve tracer that will
handle
large tubes (and smaller ones as well). Actually, its not that complex.
I'll use a Variac driving a 20 volt @10 amp filament transformer for the
filament supply, and another Variac to drive a 1600 volt transformer for
the plate. By stepping the grid between +100 v and -100 v (DC, and
variable
ranges), a set of I-V plate curves should be possible.

At this point, I'm just thinkng about it. Has anyone else built
something
like this?

Steve