[Alex]

A glow gas-discharve voltage regulator (VR150/30 or OD3, 150V) I am using in
a preamp power supply is behaving strangely.

It has quite large internal differential resistance -- about 800 ohm. When
current through the tube changes from 4 to 30mA, voltage drop would change
from 149.5 to 177V. It is too bad. (A good gas discharge regulator tube
would change voltage drop by about 1 or 2 Volts at worst.) But this one
regulates not bad in 3...8mA range, then voltage drop starts increase
greatly with current.

I looked inside the tube and noticed that the glow area on the cathode
(cylinder) refuses to grow more than to about 25% of the cathode surface
area. In a good tube the glow area would progressively enlarge with the
current rising until it covers the whole cathode. But in this one would not
grow above approximately 25% of the cathode. When current rises the
brightness of the glow increases instead of the glow area growing..

I suspect this tube is simply faulty, but it looks odd: what can stop the
discharge to creep further and wider? There is nothing visible on the
cathode to limit the area.

Who could please throw some light on the physical cause of such behaviour?

Regards,
Alex

[Patrick Turner]

Alex wrote:

A glow gas-discharve voltage regulator (VR150/30 or OD3, 150V) I am using in
a preamp power supply is behaving strangely.

It has quite large internal differential resistance -- about 800 ohm. When
current through the tube changes from 4 to 30mA, voltage drop would change
from 149.5 to 177V. It is too bad. (A good gas discharge regulator tube
would change voltage drop by about 1 or 2 Volts at worst.) But this one
regulates not bad in 3...8mA range, then voltage drop starts increase
greatly with current.

I looked inside the tube and noticed that the glow area on the cathode
(cylinder) refuses to grow more than to about 25% of the cathode surface
area. In a good tube the glow area would progressively enlarge with the
current rising until it covers the whole cathode. But in this one would not
grow above approximately 25% of the cathode. When current rises the
brightness of the glow increases instead of the glow area growing..

I suspect this tube is simply faulty, but it looks odd: what can stop the
discharge to creep further and wider? There is nothing visible on the
cathode to limit the area.

Who could please throw some light on the physical cause of such behaviour?

Light is extremely difficult to grab and hold and throw, but maybe the
tube is a bit on the stuffed side.
If a replacement works, then that is what is needed, because you can't
go inside the tube to repair it.

You have a voltage change of 27.5V with a current change of 26mA, so
effectively the Rout is just over 1k.

It should not matter with a preamp running only in class A and with very
small signal changes in signals.

You could have a normal quasi shunt reg using say EL84 in the CRCRC
filter so that the R2 is divided into 2 R with the join to the anode and
the and the voltage at C3 fed to the grid via a cap so that slight
signal at C3 causes oppositely phased signal so appear at the anode,
thus opposing the V change at C3. It won't have any effect at DC, unless
you are clever with time constants used in your circuit.

I like CRCRCRC, with all the C being 470uF and R being about 150 ohms.
Each RC stage has attenuation factor of 100Hz hum = 0.0225,
and so after 3 such filters its 0.0000126, so that the hum with 30mAdc
at C1 = 0.14V is reduced to 1.8uV at C4, not a lot. Even less if hum is
at 120Hz.
You will also find very low F also fairly well attenuated and the very
slow drift up and down of the rail Vdc due to mains variations just
won't matter. The pole between 150 ohms and 470uF is 2.25Hz. Chokes on
the other hand will be resonant with C to make the B+ rail appear to
bounce up and down on the resonant F. 10H plus 470uF has Fo = 2.3Hz, and
the peak will be about 12dB if the series R of the choke isn't high,
or about equal to ZC at Fo, or 150 ohms. But you can use a choke plus
some added R to make up enough R to damp the resonance and get a
"maximally flat filter response" as the theory books say. But the
RCRCRCRC is much cheaper to make than a CRCLC filter.
Using some silly low value C of 47uF will make Fo = 7.4Hz, which is
worse.

You don't get resonances with RC filtering.

They say motor start caps are the best, but you are limited to about
40uF. Then ZC at 100Hz = 40 ohms, so R needs to be at least 390 ohms,
and after 4 RC filters you have nearly 48Vdc drop. Ripple would be 1.65V
at C1 at 30mA, and at C5 its 0.165mV, much higher than when using 470uF
caps, so 5 RC sections may need to be used but still the LF wobble due
to mains LF noise will worse than with 470uF caps.

Patrick Turner.





Regards,
Alex

[Alex]

On Jun 19, 2:06*am, Patrick Turner wrote:
Alex wrote:

A glow gas-discharve voltage regulator (VR150/30 or OD3, 150V) I am using in
a preamp power supply is behaving strangely.

It has quite large internal differential resistance -- about 800 ohm. When
current through the tube changes from 4 to 30mA, voltage drop would change
from 149.5 to 177V. It is too bad. (A good gas discharge regulator tube
would change voltage drop by about 1 or 2 Volts at worst.) But this one
regulates not bad in 3...8mA range, then voltage drop starts increase
greatly with current.

I looked inside the tube and noticed that the glow area on the cathode
(cylinder) refuses to grow more than to about 25% of the cathode surface
area. In a good tube the glow area would progressively enlarge with the
current rising until it covers the whole cathode. But in this one would not
grow above approximately 25% of the cathode. When current rises the
brightness of the glow increases instead of the glow area growing..

I suspect this tube is simply faulty, but it looks odd: what can stop the
discharge to creep further and wider? There is nothing visible on the
cathode to limit the area.

Who could please throw some light on the physical cause of such behaviour?

Light is extremely difficult to grab and hold and throw, but maybe the
tube is a bit on the stuffed side.
If a replacement works, then that is what is needed, because you can't
go inside the tube to repair it.

You have a voltage change of 27.5V with a current change of 26mA, so
effectively the Rout is just over 1k.

It should not matter with a preamp running only in class A and with very
small signal changes in signals.

You could have a normal quasi shunt reg using say EL84 in the CRCRC
filter so that the R2 is divided into 2 R with the join to the anode and
the and the voltage at C3 fed to the grid via a cap so that slight
signal at C3 causes oppositely phased signal so appear at the anode,
thus opposing the V change at C3. It won't have any effect at DC, unless
you are clever with time constants used in your circuit.

I like CRCRCRC, with all the C being 470uF and R being about 150 ohms.
Each RC stage has attenuation factor of 100Hz hum = 0.0225,
and so after 3 such filters its 0.0000126, so that the hum with 30mAdc
at C1 = 0.14V is reduced to 1.8uV at C4, not a lot. Even less if hum is
at 120Hz.
You will also find very low F also fairly well attenuated and the very
slow drift up and down of the rail Vdc due to mains variations just
won't matter. The pole between 150 ohms and 470uF is 2.25Hz. Chokes on
the other hand will be resonant with C to make the B+ rail appear to
bounce up and down on the resonant F. 10H plus 470uF has Fo = 2.3Hz, and
the peak will be about 12dB if the series R of the choke isn't high,
or about equal to ZC at Fo, or 150 ohms. But you can use a choke plus
some added R to make up enough R to damp the resonance and get a
"maximally flat filter response" as the theory books say. But the
RCRCRCRC is much cheaper to make than a CRCLC filter. * *
Using some silly low value C of 47uF will make Fo = 7.4Hz, which is
worse.

You don't get resonances with RC filtering.

They say motor start caps are the best, but you are limited to about
40uF. Then ZC at 100Hz = 40 ohms, so R needs to be at least 390 ohms,
and after 4 RC filters you have nearly 48Vdc drop. Ripple would be 1.65V
at C1 at 30mA, and at C5 its 0.165mV, much higher than when using 470uF
caps, so 5 RC sections may need to be used but still the LF wobble due
to mains LF noise will worse than with 470uF caps.

Patrick Turner.

I appreciate your discourse on RC filtering in power supplies.
Yes, that voltreg tube is stuffed, because a replacemet works just
fine. And of course I can not get inside the tube to repair it (even
if I knew what repairs to do.) My question was rather about what
physical mechanism which could possibly prevent the glow discharge to
spread. It is like part of the cathode covered by some invisible
insulatin film. Is it cathode poisoning? Foreighn gases ingress? Even
extra 27V above the normal voltage drop can not make the dischage
strike outside of some area!

Of course a 150V Zener is better than a VR150, a pair of 1N4007 is
better than 5Z4G, an op-amp is better than a valve (sorry to be
blasphemous on this newsgroup), etc., but in some cases you want to
have your equipment to be "valve" equipment, unlike you "valve"
amplifiers in which MJE340/350 greatly outnumber tubes.

In this project (test instrument) a preamp is a precision preamp with
a stable voltage gain, and supply voltage shall be regulated. Besides,
having too much capacitance after a vacuum rectifier (6X4) will harm
it. On each power-up, while its cathode is luke-warm, and current
demand to charge hundreds of uFs is large, tiny hot spots will be
developing on the cathode oxide coating, for a fraction of a second,
evaporating it little by little.

This mechanism is as follows:
On power-up electrolytics are not charged, the cathode is cold,
emission is low, field is strong, and (important!) internal resistance
of the barium oxide is high. Imagine a certain spot of the cathode got
slightly better emission than the rest. More current is flowing from
the nickel core of the cathode to this spot. Because of high
resistance of the coating (cathode is not hot yet) lots of power is
dissipated in the oxide in this spot. This power dissipation in the
cathode(!) causes local temperature of this spot to rise, emission
increases, current density in the spot rises, more power is
dissipated, etc., in an avalanche maner, until either the whole
cathode pickes up temperature from the heater or the electrolytics
fully charge.

But a damage is done -- some oxide coating has evaporated from the
temporary overheated hot spot. Next time a hot spot will form
(randomly) somewhere else...

Regards,
Alex