Ran across this little curiosity while going through an old GE neon "glow" tube manual to help ID some very nice NOS NE-17 bulbs I bought mainly as indicators and "bypass" devices for avoiding high grid start-up voltages in some kinds of tube circuits.
I'd seen this application before in schematics for old digital circuits and a few other odd places. Didn't seem at all suitable for audio since neon tubes' freq response rolls off around 1000Hz. But read all the way to the bottom. More interesting than I thought.
Even if flawed, it might be an interesting application in a "third channel bass" tube amp. Anyone ever try this?
I'd seen this application before in schematics for old digital circuits and a few other odd places. Didn't seem at all suitable for audio since neon tubes' freq response rolls off around 1000Hz. But read all the way to the bottom. More interesting than I thought.
Even if flawed, it might be an interesting application in a "third channel bass" tube amp. Anyone ever try this?
It looks interesting. But it is too small for me to read the text. Oops, I was able to zoom in.
However, I can see that using the neons as shown in the schematics might cause you to draw grid current. Careful design of the unbypassed cathodes is required. Do you want to draw grid current? Try and draw 1mA on a 12AX7 grid, and you might flash burn the super fine grid wires. I remember a 12AX7 that had a constant current source to the cathode, and the grid was tied to a low resistance to ground. When B+ went away, the grid smoked. How about checking the tube curves of a 6L6GC, with grid current of a typical neon to the grid?
Real Power RF tubes do have control grid current ratings, and even some small signal tubes used for oscillator service (a 6J6 for example). Now we might find use for a neon there.
I have considered using a neon to ground with the other end tied to the R in an RC coupled circuit, to prevent smoke (in a cathode circuit that has an unbypassed current source). It will limit the grid voltage to about 90V during tube warm up, where the plate of the previous stage rises to B+ during warm-up. (the current source did not have hundreds of volts rating).
I found that a protection diode, diode anode to the grid stopper, and diode cathode to the tube cathode is much simpler, as I do not want to draw grid current during normal and linear operation.
The advantage of the neon is the voltage drop. The disadvantage may be the noise. Another disadvantage is the driver stage has to drive the extra resistor, and can reduce gain and increase distortion. There are other ways to solve DC coupling design.
However, I can see that using the neons as shown in the schematics might cause you to draw grid current. Careful design of the unbypassed cathodes is required. Do you want to draw grid current? Try and draw 1mA on a 12AX7 grid, and you might flash burn the super fine grid wires. I remember a 12AX7 that had a constant current source to the cathode, and the grid was tied to a low resistance to ground. When B+ went away, the grid smoked. How about checking the tube curves of a 6L6GC, with grid current of a typical neon to the grid?
Real Power RF tubes do have control grid current ratings, and even some small signal tubes used for oscillator service (a 6J6 for example). Now we might find use for a neon there.
I have considered using a neon to ground with the other end tied to the R in an RC coupled circuit, to prevent smoke (in a cathode circuit that has an unbypassed current source). It will limit the grid voltage to about 90V during tube warm up, where the plate of the previous stage rises to B+ during warm-up. (the current source did not have hundreds of volts rating).
I found that a protection diode, diode anode to the grid stopper, and diode cathode to the tube cathode is much simpler, as I do not want to draw grid current during normal and linear operation.
The advantage of the neon is the voltage drop. The disadvantage may be the noise. Another disadvantage is the driver stage has to drive the extra resistor, and can reduce gain and increase distortion. There are other ways to solve DC coupling design.
Last edited:
Interesting article - but the bypassing of the neons with a capacitor to improve the frequency response (figs 4.9 and 4.10) also has the potential to create a simple relaxation oscillator right in the signal path due to the negative resistance characteristic shown by neons when they strike. Perhaps it can be made to work, but I can't help feeling there are better ways to do things.
Dunno... this is (to me) why the good Doctor Zenner invented the Zenner diode. No glow, but also no potential for a relaxation oscillator either. The problem of course is Zenner noise.
I do not think that using a neon bulb will work properly for this purpose. Also, as you already mentioned yourself, the frequency response is impaired and drops @1Khz.
Please be cautious when DC coupling small-signal tubes. You will need to think about protecting the grid otherwise you can easily smoke up the grid during warm-up conditions if you're not careful.
If you really want to DC couple then I would advise you to look for other ways to couple between stages. Have you already looked at screen-drive with pentodes (e.g. EL509/PL519)?
Personally, I would start by exploring screen drive. I have built simple DC coupled push pull amplifiers in the past that only consisted out of:
- A simple CLC power supply
- 12 resistors and 2 diodes per channel
- Two small tubes for the VAS/phase splitter stage (I prefer a long tail cascaded ECC88 stage with a CCS in the tail)
- One tube to buffer the VAS from the output stage (ECC99 will work, otherwise use small power pentodes like a EL84)
- Two power tubes (JJ EL509)
Negative grid voltage can be applied for the power tubes, or you can use a cathode resistor to bias the power tubes.
Of course, there are other ways to DC couple between stages, but these mostly result in complicated circuitry that I would avoid if you do not have allot of experience.
To get a basic idea for screen drive you can research the Synola Amp. It has a couple of flaws, but you will get a basic understanding of how it works.
The main flaw in this circuit is the driver stage for the output tube. A ECC82 cannot sink enough current to keep the screen grid satisfied. This can easily be overcome by changing the ECC82 for a ECC99 (single triode will work) or a EL84. If you replace the driver tube then you can also leave out R5 and R6 (the screen grid will source enough current to keep the driver tube happy, R5 is not needed). Also, there is no protection diode for the driver tube which means that its grid will see full (380V !!) B+ voltage during startup.
Please be cautious when DC coupling small-signal tubes. You will need to think about protecting the grid otherwise you can easily smoke up the grid during warm-up conditions if you're not careful.
If you really want to DC couple then I would advise you to look for other ways to couple between stages. Have you already looked at screen-drive with pentodes (e.g. EL509/PL519)?
Personally, I would start by exploring screen drive. I have built simple DC coupled push pull amplifiers in the past that only consisted out of:
- A simple CLC power supply
- 12 resistors and 2 diodes per channel
- Two small tubes for the VAS/phase splitter stage (I prefer a long tail cascaded ECC88 stage with a CCS in the tail)
- One tube to buffer the VAS from the output stage (ECC99 will work, otherwise use small power pentodes like a EL84)
- Two power tubes (JJ EL509)
Negative grid voltage can be applied for the power tubes, or you can use a cathode resistor to bias the power tubes.
Of course, there are other ways to DC couple between stages, but these mostly result in complicated circuitry that I would avoid if you do not have allot of experience.
To get a basic idea for screen drive you can research the Synola Amp. It has a couple of flaws, but you will get a basic understanding of how it works.
The main flaw in this circuit is the driver stage for the output tube. A ECC82 cannot sink enough current to keep the screen grid satisfied. This can easily be overcome by changing the ECC82 for a ECC99 (single triode will work) or a EL84. If you replace the driver tube then you can also leave out R5 and R6 (the screen grid will source enough current to keep the driver tube happy, R5 is not needed). Also, there is no protection diode for the driver tube which means that its grid will see full (380V !!) B+ voltage during startup.
The K2-W tube based operational amplifier (1952) had neon lamps as level shifters between stages.
philbrickarchive.org
philbrickarchive.org
Last edited:
I tried it once, but never got it to work reliably. I built a simple two-stage DC-coupled negative-feedback amplifier with shunt feedback at the input and series feedback at the output for a college demonstration. The neon lamps were bypassed with capacitors, with resistors in between the lamps and the capacitors to prevent relaxation oscillations. The amplifier worked, but when overdriven it started oscillating and never stopped oscillating. It only started working reliably after I replaced the neon lamps with simple bypassed resistors. I still don't know what I did wrong.
By the way, when you use neon lamps as voltage limiting devices, make sure there is some light shining on them. They respond in tens of microseconds when illuminated, but in tens or even hundreds of milliseconds when they are used in darkness.
By the way, when you use neon lamps as voltage limiting devices, make sure there is some light shining on them. They respond in tens of microseconds when illuminated, but in tens or even hundreds of milliseconds when they are used in darkness.
Last edited:
It works fine if well designed. As the K2-W shows.
The following grid won't go positive if the amplifier is happy. If it gets unhappy, the neon is likely to die before the following grid melts; most neons have low current ratings. In the K2-W there is only <1mA available from the first stage plate resistor, and a mA in the grid of the next tube does no harm (though tends to go uselessly non-linear, so you will fix it quick).
The frequency response is a red herring. The neon is stable with some pFd across it. Worst-case, a few K of series resistor swamps negative resistance with negligible effect on gain.
Agree that *DC* coupling too much of an audio amp is just wrong. Even if DC NFB stabilizes operating points, there is still the cold-start problem.
Neons go bad with age (they don't start well without a pinch of radioactive, and the stuff used has half-life of a decade or so), and are a dying product. The ones you buy today to replace too-tired vintage neons won't work as well, and by 2028 you might have a hard-hard time finding any new ones.
The following grid won't go positive if the amplifier is happy. If it gets unhappy, the neon is likely to die before the following grid melts; most neons have low current ratings. In the K2-W there is only <1mA available from the first stage plate resistor, and a mA in the grid of the next tube does no harm (though tends to go uselessly non-linear, so you will fix it quick).
The frequency response is a red herring. The neon is stable with some pFd across it. Worst-case, a few K of series resistor swamps negative resistance with negligible effect on gain.
Agree that *DC* coupling too much of an audio amp is just wrong. Even if DC NFB stabilizes operating points, there is still the cold-start problem.
Neons go bad with age (they don't start well without a pinch of radioactive, and the stuff used has half-life of a decade or so), and are a dying product. The ones you buy today to replace too-tired vintage neons won't work as well, and by 2028 you might have a hard-hard time finding any new ones.
Heck, the ones I'm buying now were made when I was 5 years old, and I'm on the far side of 50 years from that date. But the comments on reliability are spot on. I've tried building neon clocks, etc. at various times and found I needed to burn in and hand-match all the (old) tubes. Many were way out of spec. And yes, illumination is needed for them to fire reliably.
(side note - I did get a neon ring-counter clock working for awhile, but it was impossible to take a picture with a flash camera. Even shining a bright flashlight on it would cause it to stop working, as well as no illumination at all. There are miniature cold-cathiode thyratron tubes that are more reliable, but they are expensive curiosities now. I own a few...)
Really this was more a point of interest, I'm probably not going to try it.
On second thoughts - there don't appear to be any suitable radioactive sources available to diy some extra performance from those vintage NE-2's (if needed). The smoke detector material mainly emits alpha particles, which won't penetrate the lamp glass.
I'm aware of a couple of systems that were used in high-performance cold-cathode tubes.
One was to put a tiny bit of radioactive material in the tube, a beta emitter I imagine. The vendor's typically spec'ed these as "small amount of radioactive material". This wasn't done for ordinary indicator bulbs (night lights, etc.), but for tubes used in switching and regulation. I assume that the radioactivity is essentially dead in any of these tubes and they must be illuminated from outside.
The other system I saw when was was at Lawrence Berkeley Labs decades ago, and used some tritium gas in some scary huge thyratron tubes in an accelerator. These puppies never left the experimental site (I hope!).
There is a third system that uses a "pilot discharge" of a few microamps using an auxiliary electrode to keep a few ions floating around in the tube. The Z70U was one such device, and they can still be purchased. Very small, rugged, reliable, and used as computer components for a few years during the transition from vacuum tube to transistor computers.
Last edited:
Neon voltage reference tubes like the 85A2 usually have either tritium or uranium dioxide as their radioactive primer. Tritium has a half-life of about 12.32 years, uranium dioxide has a very long halflife.
I don't know what primer, if any, is used for normal neon indicator lamps, but the one I measured a couple of years ago could be primed very effectively with just a bit of light. I measured an average ignition delay of 21.6 ms in darkness and about 10 us with a bit of light, if I remember well.
I don't know what primer, if any, is used for normal neon indicator lamps, but the one I measured a couple of years ago could be primed very effectively with just a bit of light. I measured an average ignition delay of 21.6 ms in darkness and about 10 us with a bit of light, if I remember well.
By the way, the British National Museum of Computing has a working computer from the early 1950's that uses dekatron neon valves, see
First generation - WITCH & EDSAC | The National Museum of Computing
First generation - WITCH & EDSAC | The National Museum of Computing
I used to do the same thing, until I realized you can now buy ultra high efficiency LEDs that will light up brightly on a single milliamp of current. Some will be plenty bright with ten times less current, say a hundred microamps.
Now you can have neon-bulb current draw, and neon bulb brightness levels, without the neon-bulb problems!
The starting point is a 3 mm or 5 mm LED with a brightness rating of 3000 mcd or more (you can routinely get up to 20,000 mcd now.) Then try 1k/volt of current-limiting resistance in series to start with, i.e. about 390k in series for a typical B+ of 360V - 400 V. This well set LED current at roughly 1 mA.
Increase the 390k resistor if too bright, decrease if too dim. Depending on the LED, you might find yourself using several megohms (!) in series if you have a really high efficiency LED.
As a bonus, the LED and series resistor act as a safety discharge cap for the B+ filter capacitors, making the amp safe to touch a few minutes after it has been unplugged (depending on filter cap values, of course.)
Not so long ago, it used to be a given that you needed 10 mA to light up an indicator LED. It's amazing how much the new high-efficiency LEDs have changed that.
-Gnobuddy
I don't have those type LEDs, unfortunely. My blue ones are too bright without 100k@12V! I want a dim light, and the neon does that for me with 100k dropper for an off the shelf 230V indicator neon running on 560V.
For some reason, blue LEDs are always glaringly bright. I don't think human eyes were designed to view pure monochromatic blue of such short wavelength.
However: 100k, 12V, assume 2V dropped in the LED, that means you like a current of about 0.1 mA (or 100 uA) through your blue LEDs.
You can get that same current with, say, a 400V B+ by using about 4000 kilo ohms - the nearest convenient standard value would be 3.9 megohms. That should give you the same brightness you get with 100k and 12V.
-Gnobuddy
- Status
- Not open for further replies.