Solid state equivalent for an old valve lamp?

[Artlav]

I've got an old UV induction lamp, which is basically a bulb of mercury vapour inside a coil, with driving circuitry below it.


It used to have a crappy AC power supply that i have since then replaced with a DC-DC converter.
And really, i wouldn't mind replacing the rest of it.


As far as i can tell, it's some sort of a valve oscillator.
It runs at 40.68MHz, producing about 5W of light out of the bulb.


Question is - how can i reproduce this circuit using modern components?
Hopefully with an ability to regulate the output power level.

[Seppy]

I have no real experience with valve electronics, however I may be able to come up with a circuit that emulates what it does if you give me a scope capture of the lamp input.

What is the voltage output of the DC-DC converter section?

If you like them nifty analog circuits, the following may be of relevance:
http://www.electronics-tutorials.ws/opamp/opamp93.gif

If you grab that circuits V_c and plug it into a comparator (Its level should be defined by a potentiometer) you have a PWM waveform generator. If you then use the output of the PWM waveform generator to control the gate of a N-channel MOSFET you can control your lamp .

Alright so I've made a fun simple circuit that may or may not go the job depending on the voltages you have available, you might also need a virtual ground.

The circuit simulator falstad allows the export of its saved files to a weird and wonderful link
http://www.falstad.com/circuit/#%24+3+5.0E-6+1.3241202019156522+66+5.0+50%0Aa+256+288+360+288+0+15.0+-15.0+1000000.0%0Ar+256+216+360+216+0+100.0%0Aw+360+216+360+288+0%0Aw+256+272+256+216+0%0Aw+256+304+256+352+0%0Aw+256+352+360+352+0%0Ar+360+352+360+416+0+100.0%0Ag+200+416+200+440+0%0Aw+200+272+256+272+0%0Ag+360+416+360+440+0%0Ar+360+352+360+288+0+100.0%0Ac+200+416+200+272+0+1.0E-6+-4.628696703640642%0Aa+256+144+360+144+0+15.0+-15.0+1000000.0%0Aw+200+272+200+160+0%0Aw+200+160+256+160+0%0A174+160+168+216+88+0+1000.0+0.48020000000000007+Duty+Cycle+Pot%0Aw+216+128+256+128+0%0Aw+360+144+448+144+0%0Af+448+144+504+144+0+1.5%0Ag+504+160+504+208+0%0AR+160+88+160+48+0+0+40.0+7.0+0.0+0.0+0.5%0AR+504+64+504+24+0+0+40.0+12.0+0.0+0.0+0.5%0A162+504+64+504+128+1+2.1024259+1.0+0.0+0.0%0Ax+349+112+453+118+0+24+By+Seppy%0AR+160+168+160+192+0+0+40.0+-7.0+0.0+0.0+0.5%0Ao+22+2+0+33+1.1692013098647223+2.993155353253689+0+-1%0A

NOTE: The values are a little silly in the simulation, I couldn't be bothered calculating them for your frequency. 40.68MHz is quite high for an OPAMP multivibrator so it may not work at all.
Ah well if your interested a link to calculate the values:
http://hyperphysics.phy-astr.gsu.edu/hbase/electronic/square.html#c4

A crystal could be used, in a similar way to the circuit at the following link, this wasn't by me:
https://www.circuitlab.com/editor/#?id=zr448b

[Pjotr]

Funny and intriguing thing. The basic elements here are the coil around the bulb and C20/C21. They form a resonance circuit at, as you say, 40 MHz. There is a hell lot of HF current circulating there. The tube is an oscillator and spitting over 10W I guess, at 40 Mhz into that resonance circuit. Although it can be done, it will not be simple to generate that amount of power at 40 MHz with solid state devices. That needs a kind of a substantial SW transmitter. Vacuum tubes are still the devices to keep it simple for that.

[Seppy]

Funny and intriguing thing. The basic elements here are the coil around the bulb and C20/C21. They form a resonance circuit at, as you say, 40 MHz. There is a hell lot of HF current circulating there. The tube is an oscillator and spitting over 10W I guess, at 40 Mhz into that resonance circuit. Although it can be done, it will not be simple to generate that amount of power at 40 MHz with solid state devices. That needs a kind of a substantial SW transmitter. Vacuum tubes are still the devices to keep it simple for that.

Yeah it wouldn't be easy at such a frequency, its inductive so just a coil right? Perhaps driving an N-MOSFET with the above oscillator circuit (the final link i sent), there are some very fast fets.

There are also options like this (1MHz though):

from rmcybernetics

[bktemp]

This simple royer type oscillator does not work very well above a few MHz: The main problems are the lowpass due to the gate capacity and stray inductance in the drain path.
Class E or similar type of circuit is the way to go: Any parasitic components of the transistor must be included in the tuned circuit.
Depending on the load driving a tuned circuit using a fixed frequency may not work, because the load de-tunes the resonant coil as soon as the gas ignites.

[Seppy]

I believe the circuit (Posted above):
https://www.circuitlab.com/editor/#?id=zr448b

Driving(via ZXGD3001E6
) an N-MOSFET lowside might meet the requirements. What do you recon bktemp?

[amyk]

Interesting operating principle - reminds me of the rubidium lamp used in frequency standards.

[bktemp]

Maybe a plasma tweeter driving circuit might work:
http://www.plasmatweeter.de/images/magnat/Mpp02_s.gif
Description of the system (in German):
http://www.plasmatweeter.de/magnat.htm
I think it runs at 27MHz and produces >20W.

You probably need special rf mosfets for 40MHz operation, but this type of circuit should work:
http://www.richieburnett.co.uk/hfcirc.gif
from
http://www.richieburnett.co.uk/hfsstc.html
I have seen similiar circuits used at least up to 20MHz at much higher power levels.
The gate must be tuned, otherwise you need more power for driving the gate than the circuit ouputs into the lamp.

[richard.cs]

Take a look at ham radio transmitter circuits for inspiration. I haven't checked - is this in an ISM band? It probably should be and the frquency probably isn't all that critical to the lamp so you could re-tune if you need to.

If you want an all solid state design then you're probably looking at a switching supply to convert down to around 10V and a fast RF mosfet as the output PA inductivly coupled in to the output coil. It's not a trivial piece of design work and remember that the output is into what's probably a fairly high-q coil so the resonant capacitor will see a few hundred Watts of circulating power. Could be self oscillating or driven from it's own oscillator.

I'd actually be inclined to make it a self-resonant valve oscillator not too dissimilar to the current design. Something like a TV line output valve run off of unsmoothed mains (or doubled mains in the USA) perhaps with a capacitive dropper for the heater. It'll give you a much simpler design with less to go wrong. A self oscillating design also means you always drive your output coil on tune.

If you wanted to and felt brave you could shrink the heater circuit to a resistive dropper that initially heats it before being switched out (probably semiconductors here or just a ptc thermistor) and let a second coil on the tank circuit provide RF heating current. That would be quite cool and your resistive dropper can be small as it's not on for long. That's probably a bit tricky to get right but personally I think it'd be really cool. :-)

[Zero999]

I think you should leave it alone. A valve seems like a simple and elegant solution. When the valve fails then maybe you could consider replacing it with something solid state but until then if it's not broken, don't fix it.

[Artlav]

Interesting.
So, this is a case when the valve is actually superior, or at least simpler and cheaper than anything solid state?
If so, i might keep it as is. I wanted to get rid of HV supply and a few seconds of startup delay.

Somehow i thought it would be something simple, since i made a 500MHz RF generator that can ignite various neon lamps out of a single RF FET (rd02mus1) and a few passives.
It did, however, turned most of 1W of power consumed into heat.


So i thought this would be just as dead simple as that.

What is the voltage output of the DC-DC converter section?

It takes 12V input, and produces 2.3V for the filament and 250V for the anode.
The old supply was resistor-limited rectified mains for the anode, and a capacitative divider for the filament.

so I've made a fun simple circuit that may or may not go the job

Getting 40MHz is the easy part.
Switching 5W of it into the coil is the hard part.

however I may be able to come up with a circuit that emulates what it does if you give me a scope capture of the lamp input.

Bad idea - it's a resonant tank circuit, could have a lot of volts on it.
Basically, a sine wave.

I have seen similiar circuits used at least up to 20MHz at much higher power levels.

Hm, tesla coil tuff seem to be in the right domain.
Never saw a coil running at 40MHz, however.
Not even VTTC.

It probably should be and the frquency probably isn't all that critical to the lamp so you could re-tune if you need to.

AFAIK, the frequency should be fairly precise - it won't ignite or will burn much dimmer unless precisely tuned.
Might be due to mismatch between the tank and the oscillator, however.
I only had to tune it once.

I think you should leave it alone. A valve seems like a simple and elegant solution.

Apparently it is.
I didn't expect that, and thought there would be a dead simple modern parts solution, but that does not appear to be so.

[Zero999]

I think you should leave it alone. A valve seems like a simple and elegant solution.

Apparently it is.
I didn't expect that, and thought there would be a dead simple modern parts solution, but that does not appear to be so.

[/quote]
It still often is the most efficient way of producing a cheap RF oscillator. Look inside a microwave and what do you find? A cavity magnetron!

[T3sl4co1l]

Interesting.
So, this is a case when the valve is actually superior, or at least simpler and cheaper than anything solid state?
If so, i might keep it as is. I wanted to get rid of HV supply and a few seconds of startup delay.

Superior in various ways, perhaps not; all that old Soviet surplus might perhaps still be cheaper though.

The 1?24? is quite a capable little tube,
http://tec.org.ru/_bd/17/1755_124-.pdf
The nearest American equivalent I might suggest would be something the size of 6DK6, but with a much weaker grid (about 1/5th the transconductance), and much more peak current capacity (the curves only go up to -12Vg1).  5686 is a miniature transmitter type with similar transconductance, but is quite a bit more powerful.

Getting 5W of real RF power from a 2.5W tube is quite excellent, suggesting a properly matched class C oscillator.

Somehow i thought it would be something simple, since i made a 500MHz RF generator that can ignite various neon lamps out of a single RF FET (rd02mus1) and a few passives.
It did, however, turned most of 1W of power consumed into heat.

This is basically all you need, you just need it matched by impedance.

The vacuum tube is expecting a load on the order of 200V and 20mA average, but as a class C oscillator-amplifier, perhaps 60 or 100mA peak.  This suggests a load resistance on the order of 2kohms.

If you can find a FET with voltage and current parameters like that (offhand, I know of 400V and 600V devices that would probably reach the frequencies required, but they're in the several ampere range, too large to be comparable), you could build a direct replacement.  Using more traditional RF transistors (designed for ~24VDC supply), you'll need a different matching circuit.

Probably this is the same limitation of your 500MHz oscillator; very poor matching to the plasma, so the SWR / mismatch is very bad.

Frequencies of 13.56MHz (ISM), 27, 40, 60 and so on have been used for such purposes as the present lamp, and for diathermy (heating biological tissue) and technical applications (plasmas for chemical analysis (e.g., ICP-MS), semiconductor fabrication, etc.).

Such high frequencies are necessary to achieve voltages sufficient to ignite the plasma; although plasma induction can be done at lower frequencies, it has to be ignited somehow.

Ed: mega forum fail, it seems I cannot write "1P24B" in Cyrillic!  Perhaps our new admin can offer some services addressing the forum's international support,

Tim

[SArepairman]

How stable is this oscillation frequency?

What frequency range of output at 1-10watt power levels is possible using the same tube topology shown here?

[T3sl4co1l]

I would expect pretty stable; phase noise will be limited by random motion in the plasma.  Probably still tight enough to fit within the ISM band as such (40.68 MHz +/-20kHz)?

Frequency range is adjustable by the coils and caps, of course.  The tube's maximum power is probably dropping off in the 200MHz range due to transit time effects and electrode inductances, maybe lower if the electrodes are connected carelessly inside.  You might not use the same circuit at those frequencies (neutralization may be necessary for an amplifier, having consequences for the oscillator), and grid input resistance will be lower (i.e., the "work" done by the grid squishing the space charge -- the equivalent resistance goes as 1/f^2), so it may need a different turns ratio at the grid-cathode circuit.

A similar circuit will work down as low as you like (use iron cored coils to build an audio frequency oscillator!), but something like a conventional Hartley may give better efficiency.  Of course, frequencies  below 10s of MHz probably won't have enough electric field to excite the plasma without electrodes.  But you could still make, like, a CCFL driver... which is hilarious, because the bulb will be cold cathode, but its driver won't be..

Tim

[calexanian]

I bet ya its not running at as high of a frequency as you think. Also those submineature tubes are not good for too much power. 2 watts or so even in C class is about all they are good for. Cute design though.