[Richard[_12_]]

http://www.r-type.org/pdfs/kt88.pdf

I'm wanting to get 35W plate dissipation running unvarying DC current
through a KT88 and run it for several hours on a the tube that has not
been run for two decades. I'm hoping that this might activate the getter
and reduce some glassiness.

I'm trying to make the circuit as simple as possible.

I have access to 350V dc (not variable as it is) at 120mA max.

To get 35W plate dissipation we can have 350V x 100mA. Or lowest
voltage, 152V x 230mA (tube can pass 230mA cathode current). But max
current I have is 120mA, so that means lowest voltage is 291V x 120mA).

Is the simplest circuit to?:

Ground the grid - have static plate voltage of 350V, and adjust Vg2
voltage by placing a resistance in Vg2 circuit?

I'm looking at page 6 in the above document, top graph, and thinking if
I got Vg2 to about 100V, then Ia would be about 125mA. That's too hight,
so Vg2 would have to be lower than 100V.

This is the simplest circuit arrangement I can think of, because I don't
have arrange for grid bias, or an adjustable PSU voltage, or putting a
high-wattage resistor in series with cathode.

I don't know how to calculate Vg2 series resistance to get about 100V.

Am I on the right track - or not. :c). Thanks.

[Big Bad Bob]

On 10/06/10 16:41, Richard so witilly quipped:
http://www.r-type.org/pdfs/kt88.pdf

I'm wanting to get 35W plate dissipation running unvarying DC current
through a KT88 and run it for several hours on a the tube that has not
been run for two decades. I'm hoping that this might activate the getter
and reduce some glassiness.

I'm trying to make the circuit as simple as possible.

I have access to 350V dc (not variable as it is) at 120mA max.

To get 35W plate dissipation we can have 350V x 100mA. Or lowest
voltage, 152V x 230mA (tube can pass 230mA cathode current). But max
current I have is 120mA, so that means lowest voltage is 291V x 120mA).

Is the simplest circuit to?:

Ground the grid - have static plate voltage of 350V, and adjust Vg2
voltage by placing a resistance in Vg2 circuit?

Grounding the grid might be unstable unless you want to control plate
current via screen current/voltage. If you don't want to use a grid
bias (cathode resistor or -V bias supply) then you might try varying
screen current using a series triode and a single cathode resistor.

If you look on the performance curve for the triode, you should see an
approximate voltage for a desired plate current. Then pick a resistor
that will drop that voltage, place it in series with the cathode and
wire the triode's grid and screen on the other end of the cathode
resistor (yes the triode's cathode potential will be at 100V, more on
this later). Then wire the triode's plate to 350V DC. A typical triode
will dissipate 5W or so so you might have to put 2 in parallel but they
often come in pairs anyway. For more fine adjustment you could put a
potentiometer on the triode's grid and drop a bit more voltage than
necessary with the cathode resistor on the triode. Then you could adjust
the triode's grid voltage to vary the screen current (and plate current)
on the KT88.

FYI you might need a separate heater supply for the triode. +200V on a
cathode might not do well, depending on the tube.

(hopefully this isn't too complex, just 2 or maybe 3 additional components)

[Richard[_12_]]

On 10/10/2010 07:01, Big Bad Bob wrote:
On 10/06/10 16:41, Richard so witilly quipped:
http://www.r-type.org/pdfs/kt88.pdf

I'm wanting to get 35W plate dissipation running unvarying DC current
through a KT88 and run it for several hours on a the tube that has not
been run for two decades. I'm hoping that this might activate the getter
and reduce some glassiness.

I'm trying to make the circuit as simple as possible.

I have access to 350V dc (not variable as it is) at 120mA max.

To get 35W plate dissipation we can have 350V x 100mA. Or lowest
voltage, 152V x 230mA (tube can pass 230mA cathode current). But max
current I have is 120mA, so that means lowest voltage is 291V x 120mA).

Is the simplest circuit to?:

Ground the grid - have static plate voltage of 350V, and adjust Vg2
voltage by placing a resistance in Vg2 circuit?

Grounding the grid might be unstable unless you want to control plate
current via screen current/voltage. If you don't want to use a grid bias
(cathode resistor or -V bias supply) then you might try varying screen
current using a series triode and a single cathode resistor.

If you look on the performance curve for the triode, you should see an
approximate voltage for a desired plate current. Then pick a resistor
that will drop that voltage, place it in series with the cathode and
wire the triode's grid and screen on the other end of the cathode
resistor (yes the triode's cathode potential will be at 100V, more on
this later). Then wire the triode's plate to 350V DC. A typical triode
will dissipate 5W or so so you might have to put 2 in parallel but they
often come in pairs anyway. For more fine adjustment you could put a
potentiometer on the triode's grid and drop a bit more voltage than
necessary with the cathode resistor on the triode. Then you could adjust
the triode's grid voltage to vary the screen current (and plate current)
on the KT88.

FYI you might need a separate heater supply for the triode. +200V on a
cathode might not do well, depending on the tube.

(hopefully this isn't too complex, just 2 or maybe 3 additional components)


Well, let's start from my current situation and see if I'm doing
something not quite right.

http://homepage.ntlworld.com/richard.../kt88test.html

Here you see a picture of my current setup.

I'm using a transformer with secondary of 350V 0 350V at 120mA, an 8
henry choke and a 5Z4G rectifier.

It's currently wired as choke input, then a 4 uF capacitor after the choke.

When there is a suitable load to enable the choke input circuit to
operate correctly, the output voltage is about 310V.

I'v wired the KT88 triode connection. Anode and screen are at same
potential because they are connected.

I had trouble with parasitic oscillation, but putting ferrite beads on
anode, screen and cathode wires appears to have killed it. The grid has
no bead, but a there is a 1K0 resistor close to the the g1 pin on the
valve socket.

There is a 270R cathode bias resistor, which serves to effectively place
a bias on g1, because g1 connects to "earthy" end of the bias resistor.

Okay, with this setup (which may be deficient, but I would not know why)
anode current goes way up to 240mA, the voltage output from PSU drops to
about 250V. Under these conditions bias resistor develops 64.8V.

Of course I cannot maintain the level of current but for several seconds.

Remember, I'd be trying to employ the graph for triode connection he

http://www.r-type.org/pdfs/kt88.pdf

I would say, does this heavy current mean the tube is terribly gassy?

Of course, I cannot use the circuit to maintain current to the tube that
would really get the tube heated over a period.

Is my current circuit arrangement not reliable to ascertain the
condition of the tube?

Should I even try another circuit to bring the current to 125mA and
leave it running for several hours to see if gassyness reduces? Given
the results achieved so far with the current circuit.

[Richard[_12_]]

errata:

http://homepage.ntlworld.com/richard.../kt88test.html

Here you see a picture of my current setup.

[Richard[_12_]]

I said:

Remember, I'd be trying to employ the graph for triode connection he

http://www.r-type.org/pdfs/kt88.pdf

If the HT is 250V and the current is 240mA, should not the bias be
around -7V according the the triode connection graph?

But it's around -65V.

Is this the "proof" of gassyness?

[glenbadd]

On Oct 13, 7:37*am, Richard wrote:
I had trouble with parasitic oscillation, but putting ferrite beads on
anode, screen and cathode wires appears to have killed it. The grid has
no bead, but a there is a 1K0 resistor close to the the g1 pin on the
valve socket.
http://homepage.ntlworld.com/richard.../kt88test.html

A nice clean layout, but I suspect the loops of wire all around the
place
form inductors which have Q and couple to each other at very high
frequencies,
just enough to oscillate. The ferrite beads kill the Q. Rearranging
the wires
may have a similar effect.

G.

[Patrick Turner]

On Oct 13, 11:14*am, glenbadd wrote:
On Oct 13, 7:37*am, Richard wrote:

I had trouble with parasitic oscillation, but putting ferrite beads on
anode, screen and cathode wires appears to have killed it. *The grid has
no bead, but a there is a 1K0 resistor close to the the g1 pin on the
valve socket.
http://homepage.ntlworld.com/richard.../kt88test.html

A nice clean layout, but I suspect the loops of wire all around the
place
form inductors which have Q and couple to each other at very high
frequencies,
just enough to oscillate. The ferrite beads kill the Q. *Rearranging
the wires
may have a similar effect.

G.

I agree that the layout seems neat enough but of course leads are
rather long. I use shorter leads. I also never would use screwed
connections for a test circuit and only use all soldered connections.
Such connections can be made or un-made to 16mm long brass plated
4guauge wood screws screwed into the breadboard, so the screwed
connectors don't need to be used at all.
But usually with such tests of a tube there is NO TENDENCY for any
oscillation even with overlong leads everywhere.
I've done all this sort of thing many times and the never noticed much
oscillation at RF or any other frequency.

However, when a tube becomes faulty and conducts way too much Ia for a
given grid bias voltage then it can become unstable and oscillate.
When left to run with excessive Ia the anode dissipation is excessive
and the anode will turn bright red.
One should be able to quickly measure the Ea and Ia and multiply them
to calculate the Pda, ie, say 300Vdc from anode to cathode x 100mA of
current measured in the cathode resistor gives Pda = 300Vdc x 0.1Adc =
30 watts of heat power being liberated; Pda means Power Dissipated at
Anode. 30W would be OK for KT88, but 55W will produce red glowing
anodes.

After fusing and wrecking a few new or old tubes or PSU parts then the
message about all this stuff sinks in unless you cannot learn from
experience or ask the right questions about what is in front of you.

Probably the tube the OP is testing is gassy, and very second hand.
But he should try to test 4 tubes in successsion and if they all show
the same behaviour then either there is something wrong with the test
circuit or there is something similarly wrong with all 4 tubes which
would be unlikely.
Many people have kept gassy tubes in the hope that some day a good
cure can be found. But none has ever been found and nobody is
researching for a fix, so gassy tubes belong only in the rubbish bin;
they won't cure and they won't ever get better.

He could have easily stopped any oscillations with 0.1uF x 630V rated
from grid to cathode and from anode to cathode. But if the tube had
been healthy he could have left these caps in place and done signal F
measurements at say 60Hz using a portion of the mains F as a test
signal via pot of say 20k placed across the heater supply where one
side is at 0V and the other is hot at 6.3Vrms.
The capacitance won't affect measurements at low F.

However, he should have a much bigger C value for his choke input C if
he expects to make accurate ac measurements with LF test signals. I
can only guess the guy is a beginner and strapped for cash and he
doesn't know which bins in which to find old cheap junk he can addapt.
But 470uF caps rated for +450V are only about $12 and surely that is
MUCH better than the ancient low value C in a can he presently is
using.

Having 10 ohm resistors in series with anode, cathode, grid and screen
allow fast measurement of the Vdc across each and later calculation of
Idc which is Vdc / 10.

Patrick Turner.

[Richard[_12_]]

On 12/10/2010 21:37, Richard wrote:

I said

I'v wired the KT88 triode connection. Anode and screen are at same
potential because they are connected.

There is a 270R cathode bias resistor, which serves to effectively place
a bias on g1, because g1 connects to "earthy" end of the bias resistor.

Okay, with this setup (which may be deficient, but I would not know why)
anode current goes way up to 240mA, the voltage output from PSU drops to
about 250V. Under these conditions bias resistor develops 64.8V.


Okay, I made a change which altered the readings significantly.

I have a triode connection, ferrite beads on anode, screen and grid
wires, but I had only a 1K0 grid stopper next to g1 pin on the tube
socket. A 270R cathode bias resistor.

Ia was 240mA
Voltage across bias resistor was 64.8V
PSU voltage 250V (dropped 60V due to overdrawing PSU design current)

I took out the grid stopper, and simply placed ferrite beads on the grid
wire.

Now the figures a

Ia 90mA
Voltage across bias resistor is 24.3V
PSU Voltage is 310V

Removing that 1K0 grid stopper altered the whole thing. Things look much
better.

Why did removing the 1K0 resistor in g1 circuit change things so
significantly?

[Richard[_12_]]

On 13/10/2010 12:37, flipper wrote:
On Wed, 13 Oct 2010 11:45:37 +0100, wrote:

On 12/10/2010 21:37, Richard wrote:

I said

I'v wired the KT88 triode connection. Anode and screen are at same
potential because they are connected.

There is a 270R cathode bias resistor, which serves to effectively place
a bias on g1, because g1 connects to "earthy" end of the bias resistor.

Okay, with this setup (which may be deficient, but I would not know why)
anode current goes way up to 240mA, the voltage output from PSU drops to
about 250V. Under these conditions bias resistor develops 64.8V.


Okay, I made a change which altered the readings significantly.

I have a triode connection, ferrite beads on anode, screen and grid
wires, but I had only a 1K0 grid stopper next to g1 pin on the tube
socket. A 270R cathode bias resistor.

Ia was 240mA
Voltage across bias resistor was 64.8V
PSU voltage 250V (dropped 60V due to overdrawing PSU design current)

I took out the grid stopper, and simply placed ferrite beads on the grid
wire.

Now the figures a

Ia 90mA
Voltage across bias resistor is 24.3V
PSU Voltage is 310V

Removing that 1K0 grid stopper altered the whole thing. Things look much
better.

Why did removing the 1K0 resistor in g1 circuit change things so
significantly?

It shouldn't have.. unless it wasn't connected to ground or the tube
has an incredible amount of grid leakage.

You may possibly have been right on the first point or there was an
oscillation I never knew about because there were no ferrites on wire to
g1. Not sure which was right.

I reconnected the 1K0 resistor, but this time I put two beads near g1
pin. So, that's a slightly different setup, because before the 1K0 was
wired in without the beads.

Anyway I also now wired in a switch to short out that 1K0 resistor or
not as required, and either way the plate current is (as you say it
should be)precisely the same.

So, perhaps the tubes are still good.

I think to check for gassiness I need to get a reading of grid current.

I have a 200uA meter. Should I place this in series in the g1 circuit
and get a reading? Or read voltage across the 1k0? If so what voltage
would I be looking for? How should I get an indication of grid current?

[Richard[_12_]]

I said:

I think to check for gassiness I need to get a reading of grid current.

I have a 200uA meter. Should I place this in series in the g1 circuit
and get a reading? Or read voltage across the 1k0? If so what voltage
would I be looking for? How should I get an indication of grid current?

Well, I put the 200uA meter in series with the 1K0 resistor, or with the
wire that goes to g1.

If this correctly reads grid current, then it looks like I've no grid
current on one two and about 4uA on the other.

What is acceptable grid current for a GEC KT88?

[Big Bad Bob]

glenbadd wrote:
On Oct 13, 7:37 am, Richard wrote:
I had trouble with parasitic oscillation, but putting ferrite beads on
anode, screen and cathode wires appears to have killed it. The grid has
no bead, but a there is a 1K0 resistor close to the the g1 pin on the
valve socket.
http://homepage.ntlworld.com/richard.../kt88test.html

A nice clean layout, but I suspect the loops of wire all around the
place
form inductors which have Q and couple to each other at very high
frequencies,
just enough to oscillate. The ferrite beads kill the Q. Rearranging
the wires
may have a similar effect.

alternately electrolytic capacitors are NOTORIOUS for misbehaving at high
frequencies. HF oscillation can be killed by placing a 0.1 microfarad
capacitor in parallel with the electrolytic filter capacitor, and (as you
pointed out) not having wires curving about. Straight point-to-point wiring
helps with stray oscillation when there's no negative feedback in place to
limit the overall gain.

I'm guessing the tube at the top of the photo is a full wave rectifier and
there's a filter choke and potted filter capacitor to the right of it (in
that order). In that case the KT88 would be the tube at the bottom of the
picture, with the ammeter in series with the plate, correct?

[Patrick Turner]

On Oct 13, 9:45*pm, Richard wrote:
On 12/10/2010 21:37, Richard wrote:

I said

I'v wired the KT88 triode connection. Anode and screen are at same
potential because they are connected.
There is a 270R cathode bias resistor, which serves to effectively place
a bias on g1, because g1 connects to "earthy" end of the bias resistor.

Okay, with this setup (which may be deficient, but I would not know why)
anode current goes way up to 240mA, the voltage output from PSU drops to
about 250V. Under these conditions bias resistor develops 64.8V.

Okay, I made a change which altered the readings significantly.

I have a triode connection, ferrite beads on anode, screen and grid
wires, but I had only a 1K0 grid stopper next to g1 pin on the tube
socket. A 270R cathode bias resistor.

Ia was 240mA
Voltage across bias resistor was 64.8V
PSU voltage 250V (dropped 60V due to overdrawing PSU design current)

I took out the grid stopper, and simply placed ferrite beads on the grid
wire.

Now the figures a

Ia 90mA
Voltage across bias resistor is 24.3V
PSU Voltage is 310V

Removing that 1K0 grid stopper altered the whole thing. Things look much
better.

Why did removing the 1K0 resistor in g1 circuit change things so
significantly?

I think you may be a long way from knowing enough to answer your
questions by yourself and part of the problem may seem to be that you
have no oscilloscope to monitor whether or not there are RF
oscillations or what frequency they may be, or whether your ferrite
beads have any real effect at all. Bypassing ALL electrodes to a
common ground point using say 0.47uF 630V rated plastic caps should
eliminate any possibility of RF oscillations.

Your obsevations so far lead me to think you have an almost completely
****ed tube.

What happens when you test new tubes, or other tubes of the same age?

There are many questions you should be asking, IMHO.

Patrick Turner.r

i.0uF plaPlaicin

[Patrick Turner]

On Oct 14, 1:14*am, Richard wrote:
On 13/10/2010 12:37, flipper wrote:





On Wed, 13 Oct 2010 11:45:37 +0100, *wrote:

On 12/10/2010 21:37, Richard wrote:

I said

I'v wired the KT88 triode connection. Anode and screen are at same
potential because they are connected.

There is a 270R cathode bias resistor, which serves to effectively place
a bias on g1, because g1 connects to "earthy" end of the bias resistor.

[Patrick Turner]

On Oct 14, 7:21*am, Big Bad Bob BigBadBob-at-mrp3-
wrote:
glenbadd wrote:
On Oct 13, 7:37 am, Richard wrote:
I had trouble with parasitic oscillation, but putting ferrite beads on
anode, screen and cathode wires appears to have killed it. *The grid has
no bead, but a there is a 1K0 resistor close to the the g1 pin on the
valve socket.
http://homepage.ntlworld.com/richard.../kt88test.html

A nice clean layout, but I suspect the loops of wire all around the
place
form inductors which have Q and couple to each other at very high
frequencies,
just enough to oscillate. The ferrite beads kill the Q. *Rearranging
the wires
may have a similar effect.

alternately electrolytic capacitors are NOTORIOUS for misbehaving at high
frequencies. *HF oscillation can be killed by placing a 0.1 microfarad
capacitor in parallel with the electrolytic filter capacitor, and (as you
pointed out) not having wires curving about. *Straight point-to-point wiring
helps with stray oscillation when there's no negative feedback in place to
limit the overall gain.

I'm guessing the tube at the top of the photo is a full wave rectifier and
there's a filter choke and potted filter capacitor to the right of it (in
that order). *In that case the KT88 would be the tube at the bottom of the
picture, with the ammeter in series with the plate, correct?

Indeed SOME but not all electrolytics become poor capacitors at HF and
it is possible the test circuit at HF then resembles a HF oscillator
and hence becomes unstable. Yes, 0.1uF polyester or polypropylene
metal film caps are usually OK but 0.47 even moreso.

The caps should be connected to lugs of the tube socket from anode,
grid, screen and cathode and all grounded leads taken to a common
ground point within 30mm of the tube socket. Cap leads all need to be
less than 50mm. This way all the long test leads are bypassed and
their inductances and high impedances at RF frequencies, HF, are
bypassed and their effects negated.

Patrick Turner.

[Big Bad Bob]

Patrick Turner wrote:
Indeed SOME but not all electrolytics become poor capacitors at HF and
it is possible the test circuit at HF then resembles a HF oscillator
and hence becomes unstable. Yes, 0.1uF polyester or polypropylene
metal film caps are usually OK but 0.47 even moreso.

right - some electrolytics are better than others. In every well-designed
power supply I've ever seen there is a non-electrolytic capacitor in
parallel with at least one of the electrolytics for this very purpose (as
well as for surge and HF noise suppression for anything that might get
through upper stage filtering).

There are also input/output impedance mismatches which (from what I recall)
_could_ also be the source of oscillations.

Or maybe it's just 'bad gas' like you suggested already.

A new tube is probably ~$20 on Amazon.com - except for the 'cool experiment
with glass and metal and glowy filaments' factor, I'd also consider buying a
new one.

Just thinking about it, even though we have microchips with zillions of
gates less than the size of a dime, there's still a 'coolness factor'
associated with glass and metal and glowy filaments. Tubes look like
they're actually "doing something". Oh yeah, and there's that blue glow you
get from beam power tetrodes. It's sorta like functional art.

[Patrick Turner]

On Oct 17, 7:46*pm, Big Bad Bob BigBadBob-at-mrp3-
wrote:
Patrick Turner wrote:
Indeed SOME but not all electrolytics become poor capacitors at HF and
it is possible the test circuit at HF then resembles a HF oscillator
and hence becomes unstable. Yes, 0.1uF polyester or polypropylene
metal film caps are usually OK *but 0.47 even moreso.

right - some electrolytics are better than others. *In every well-designed
power supply I've ever seen there is a non-electrolytic capacitor in
parallel with at least one of the electrolytics for this very purpose (as
well as for surge and HF noise suppression for anything that might get
through upper stage filtering).

Today, many electro caps are far better caps than existed 40 years ago
and with much broader bandwidth because they are designed for use in
myriad applications where their HF performance is very important, eg,
switch mode power supplies and for rail bypassing on boards full of
digital devices with lots of switching pulsed signals.
So in many apps it is not necessary to bypass modern electros in audio
circuits because their impedance increase as F rises happens at well
over 1MHz where stray shunt capacitances elsewhere will prevent
oscillations at HF, as in the case of a 12AX7 used in a phono amp.

But say you try to use a high gn j-fet sch a a 2SK369 to drive a high
gm frame grid pentode strapped as a triode in a phono stage for moving
coil and in cascode mode, then you WILL GET HF oscillations and maybe
at 200MHz unless you are very careful about bypassing and the use of
ceramic and plastic caps and RF chokes for heater feeds etc. In other
words one must treat some critical audio amps as HF amps but be
careful to prevent them becoming oscillators, and lt's face it, an
oscillator is an amplifier with positive FB.

There are also input/output impedance mismatches which (from what I recall)
_could_ also be the source of oscillations.

Or maybe it's just 'bad gas' like you suggested already.

A new tube is probably ~$20 on Amazon.com - except for the 'cool experiment
with glass and metal and glowy filaments' factor, I'd also consider buying a
new one.

Just thinking about it, even though we have microchips with zillions of
gates less than the size of a dime, there's still a 'coolness factor'
associated with glass and metal and glowy filaments. *Tubes look like
they're actually "doing something". *Oh yeah, and there's that blue glow you
get from beam power tetrodes. *It's sorta like functional art.

Indeed tubes have lots of aesthetic appeal. But they really do well
with music when well set up.

Someone in the Hong Kong audio club was once asked in about 1994 to
comment about the Halcro class AB mosfet amps and about the sound.
"Ah, Hacro, it like 300B, but go louder" ..... inscrutable words from
an asian audiophile, and rather deflating for the makers of the Halcro
brand which costs about 50 grand for 2 x 200Watt channels where THD at
200W and at 20kHz is 0.0001%, and quite unmeasurable at 1Watt and
1kHz.

That blue glow should be a royal dark blue, and rather a subtle
effect, and not the bright sky blue of a tube in its death throes.

Patrick Turner.

[Richard[_11_]]

On 17/10/2010 04:34, Patrick Turner wrote:
On Oct 13, 9:45 pm, wrote:
On 12/10/2010 21:37, Richard wrote:

I said

I'v wired the KT88 triode connection. Anode and screen are at same
potential because they are connected.
There is a 270R cathode bias resistor, which serves to effectively place
a bias on g1, because g1 connects to "earthy" end of the bias resistor.

Okay, with this setup (which may be deficient, but I would not know why)
anode current goes way up to 240mA, the voltage output from PSU drops to
about 250V. Under these conditions bias resistor develops 64.8V.

Okay, I made a change which altered the readings significantly.

I have a triode connection, ferrite beads on anode, screen and grid
wires, but I had only a 1K0 grid stopper next to g1 pin on the tube
socket. A 270R cathode bias resistor.

Ia was 240mA
Voltage across bias resistor was 64.8V
PSU voltage 250V (dropped 60V due to overdrawing PSU design current)

I took out the grid stopper, and simply placed ferrite beads on the grid
wire.

Now the figures a

Ia 90mA
Voltage across bias resistor is 24.3V
PSU Voltage is 310V

Removing that 1K0 grid stopper altered the whole thing. Things look much
better.

Why did removing the 1K0 resistor in g1 circuit change things so
significantly?

I think you may be a long way from knowing enough to answer your
questions by yourself and part of the problem may seem to be that you
have no oscilloscope to monitor whether or not there are RF
oscillations or what frequency they may be, or whether your ferrite
beads have any real effect at all. Bypassing ALL electrodes to a
common ground point using say 0.47uF 630V rated plastic caps should
eliminate any possibility of RF oscillations.

Your obsevations so far lead me to think you have an almost completely
****ed tube.

What happens when you test new tubes, or other tubes of the same age?

There are many questions you should be asking, IMHO.

Patrick Turner.r

i.0uF plaPlaicin


There was strong oscillation without the ferrite beads. Even oscillation
with the 1KO grid stopper. That made the tube conduct heavy current.

When beads were put on every pin, stopped oscillations dead. Now the
tube can be tested.

There is in fact 4uA ionic current.

[Patrick Turner]

On Nov 10, 6:48*am, Richard wrote:
On 17/10/2010 04:34, Patrick Turner wrote:





On Oct 13, 9:45 pm, *wrote:
On 12/10/2010 21:37, Richard wrote:

I said

I'v wired the KT88 triode connection. Anode and screen are at same
potential because they are connected.
There is a 270R cathode bias resistor, which serves to effectively place
a bias on g1, because g1 connects to "earthy" end of the bias resistor.

[Big Bad Bob]

On 11/09/10 15:33, Patrick Turner so witilly quipped:
Either leads are kept very short, or you use ferrite beads or you
bypass anode to cathode, anode to grid using sufficient C values which
will allow signal testing at LF. I have always found ferrite beds to
be of limited value, and one does not know exactly what value of
inductance results with their use. I've built a number of SS amps
using both bjts and mosfets, and the latter can often give oscillation
troubles in a circuit even when leads have been kept short.

maybe it's acting like a 'Dynatron Oscillator' circuit, through the
power supply, due to a non-linear region of the tube's performance
curve. Proving that would be more difficult than adding a series
resistor or capacitor into the circuit to try to make it NOT happen. It
would have to happen in the plate/cathode circuit with the screen grid
causing the extra gain. Or something like that. Normally the load
would have enough internal resistance to prevent it (as well as
impedence reflected from the secondary). There's a formula that has to
do with transconductance and matching the load impedence with the input
and overall gain. I forget what it is exactly. 'Servo Theory' stuff.
If it's mismatched, you get spurious (and sometimes continuous) oscillation.

[Patrick Turner]

On Nov 11, 8:08*pm, Big Bad Bob BigBadBob-at-mrp3-
wrote:
On 11/09/10 15:33, Patrick Turner so witilly quipped:

Either leads are kept very short, or you use ferrite beads or you
bypass anode to cathode, anode to grid using sufficient C values which
will allow signal testing at LF. *I have always found ferrite beds to
be of limited value, and one does not know exactly what value of
inductance results with their use. I've built a number of SS amps
using both bjts and mosfets, and the latter can often give oscillation
troubles in a circuit even when leads have been kept short.

maybe it's acting like a 'Dynatron Oscillator' circuit, through the
power supply, due to a non-linear region of the tube's performance
curve. *Proving that would be more difficult than adding a series
resistor or capacitor into the circuit to try to make it NOT happen. *It
would have to happen in the plate/cathode circuit with the screen grid
causing the extra gain. *Or something like that. *Normally the load
would have enough internal resistance to prevent it (as well as
impedence reflected from the secondary). *There's a formula that has to
do with transconductance and matching the load impedence with the input
and overall gain. *I forget what it is exactly. *'Servo Theory' stuff..
If it's mismatched, you get spurious (and sometimes continuous) oscillation.

I agree entirely.

Amplifiers are bandwidth limited devices. But any tube could be made
to oscillate, ie, amplify in the presence of positive FB over a wider
frequency range than the amp in which they might be used.

At radio frequencies the actual full schematic of a tube under test
might resemble something very different to what someone knows is
present at say 1kHz.

Patrick Turner.

[Richard[_11_]]

On 09/11/2010 23:33, Patrick Turner wrote:
On Nov 10, 6:48 am, wrote:
On 17/10/2010 04:34, Patrick Turner wrote:





On Oct 13, 9:45 pm, wrote:
On 12/10/2010 21:37, Richard wrote:

I said

I'v wired the KT88 triode connection. Anode and screen are at same
potential because they are connected.
There is a 270R cathode bias resistor, which serves to effectively place
a bias on g1, because g1 connects to "earthy" end of the bias resistor.

Okay, with this setup (which may be deficient, but I would not know why)
anode current goes way up to 240mA, the voltage output from PSU drops to
about 250V. Under these conditions bias resistor develops 64.8V.

Okay, I made a change which altered the readings significantly.

I have a triode connection, ferrite beads on anode, screen and grid
wires, but I had only a 1K0 grid stopper next to g1 pin on the tube
socket. A 270R cathode bias resistor.

Ia was 240mA
Voltage across bias resistor was 64.8V
PSU voltage 250V (dropped 60V due to overdrawing PSU design current)

I took out the grid stopper, and simply placed ferrite beads on the grid
wire.

Now the figures a

Ia 90mA
Voltage across bias resistor is 24.3V
PSU Voltage is 310V

Removing that 1K0 grid stopper altered the whole thing. Things look much
better.

Why did removing the 1K0 resistor in g1 circuit change things so
significantly?

I think you may be a long way from knowing enough to answer your
questions by yourself and part of the problem may seem to be that you
have no oscilloscope to monitor whether or not there are RF
oscillations or what frequency they may be, or whether your ferrite
beads have any real effect at all. Bypassing ALL electrodes to a
common ground point using say 0.47uF 630V rated plastic caps should
eliminate any possibility of RF oscillations.

Your obsevations so far lead me to think you have an almost completely
****ed tube.

What happens when you test new tubes, or other tubes of the same age?

There are many questions you should be asking, IMHO.

Patrick Turner



There was strong oscillation without the ferrite beads. Even oscillation
with the 1KO grid stopper. That made the tube conduct heavy current.

When beads were put on every pin, stopped oscillations dead. Now the
tube can be tested.

There is in fact 4uA ionic current

OK, then use the ferrite beads. I have not found any need during my
many tests of OPT tubes over the last 15 years. I just kept leads
short. The only time I've noticed RF oscillations have been in an OPT
tubes in an amp circuit that had begun to overheat and glow red.
Usually this had been caused by incorrect bias setting or having too
low an anode load so that the AC signal Pda exceeded the rating. It
can happen in amps with an OPT which has shorted turns. OP Tubes can
then oscillate at some RF which saturates the tube and hasens their
melt down unless the amp is turned off or a fuse blows.

Have you measured the oscillation frequency? Probably it is at some
some RF. The beads act to raise the inductance of the lead wires to a
higher value. Beads change the effective circuit to one where RF
oscillation becomes impossible due to the arrangement of L and C
elements.

Either leads are kept very short, or you use ferrite beads or you
bypass anode to cathode, anode to grid using sufficient C values which
will allow signal testing at LF. I have always found ferrite beds to
be of limited value, and one does not know exactly what value of
inductance results with their use. I've built a number of SS amps
using both bjts and mosfets, and the latter can often give oscillation
troubles in a circuit even when leads have been kept short.

Patrick Turner.

Well, this is the layout:

http://homepage.ntlworld.com/richard.../kt88test.html

But anyway, I've stopped oscillations.

I'm now working out an arrangement to activate the getter.

You will like it. (Maybe if it works: :c)