Pigtail Neons :)

[Art]

Hi Guys,
I did a circuit a while back... a schematic flogged off the net to drive a neon with low voltage.
It's 555 timer controlled, to pulse power to the transformer.
The MJE355 transistor gets very hot very quickly, and I never sorted that out.
THe transformer itself, stays cool.

http://img.photobucket.com/albums/v186/ArtArt/Neons_zps7f873823.jpg

This time I can check the impedance of the transformer primary isn't real low,
and also I want to replace the 555 timer with a pic so I can vary the pulse width duty cycle and frequency.
Maybe some combination will cool it down.
Also two channels, and the potential quick fix of dimming the neon on & off in steps rather
than leave it on constantly.
There will be some compromise on the dimming speed to ensure it can be cool.

Our local electronics supplier do blue, green , and original red neons now.

[Ioannis]

MAybe the switching of the MJE355 has slow slope and operates in linear region for long time. Have you checked with a scope?

Ioannis

[pedja089]

I can't find datasheet for MJE355. I suppose that you meant MJE3055. I would try with logic level mosfet.
One reason for getting to hot is slope, as Ioannis mention, but probably isn't that.
Other:
1) What is type of transformer core, maybe frequency is to high.
2) Is that flyback topology? Maybe core is saturated.
3) Maybe emitter collector breakdown occur, I don't see any clamp on collector.
4) Neon have very low impedance when glowing, so try to put resistor in series to limit current.
etc...

[pedja089]

Forget to mention, if is flyback then you need diode on output, if is single switch forward, then you need to "reset" core, and you must have an inductor on output.

[NickMu]

I’m not an expert in this field but few months ago I’ve been playing with non-contact power transfer.

I agree with the points made by Ionnis and pedja089 and add few notes of my own:

Your LC circuit and its Q factor will determine your frequency needed and your efficiency.
For the same setup your duty cycle will change the power input / output ratio drastically.
In your case you should draw no more than few tens of milliamps. If your outputs are getting hot you are drawing too much and inefficiently.
I’m not sure what kind of diagnostic tools you have on your bench but an oscilloscope and a variable frequency generator will be more than helpful. Monitor the output signal from your Mosfet with the scope and change the frequency until you get maximum output level, cleanest signal shape and minimum current draw from your power source. If you don’t have an oscilloscope just by monitoring the current draw will get you close to optimum. You want to stimulate your circuit with its resonant frequency for best results. Once you got over this step start changing the duty cycle of the driving signal with the same goals in mind. As we discuss in one of the treads I started in about two months ago you can use a low caliber PIC to create your output. Dealing with low frequencies will also give you access to a wide duty cycle range.
The shape of the signal driving your output device (try to make it as square as possible) is extremely important. The amplitude of this signal is also important. Try using TTL level Mosfets or use a Mosfet driver. The Mosfet driver made all the difference for my application and it is TTL compatible. Most of the Mosfets available have the needed diode built in.
If you are using audio type transformers your optimum frequency should be really low.

HTH

Regards,

Nick

[NickMu]

One clarification to my above reply:
If I recall it correctly older Neons used to draw less than 1mA from about 100 - 150 V supply which makes it about 100 – 150 mW per bulb in the worst case. I think that about 100K series limiting resistors were commonly used (the value of the resistor varies with the voltage). Depending on your input voltage to the converter you should draw 25 – 35 mA from 5V and 10 – 20 mA from 12V input. The estimated current draw includes generous losses in your system.

Regards,

Nick

[Art]

I just got my first scope which is why I'm doing it now.

I know it isn't right, so that's why on breadboard.
Also, I have a slightly different pair of transformers additional to what is in the picture.

Now don't laugh... it's still dual trace 25 MHz...
http://img.photobucket.com/albums/v186/ArtArt/Scope_zpsc7265cda.jpg

[Art]

Ok, I got it :) No heat from anything, but I'm not sure the 1N004 protection
diodes are working, I get a zap with my fingers across the backs of both of them.

It's hard to see with the video, but they are two channel software PWM,
the light is moving back and forth between the two neons.

I think the transformer primary in the original circuit was very low impedance.
It's a short on my continuity meter, it's so low. That would look bad for the MJE3055 output.

Still have to sort out this stray high voltage problem.

https://www.youtube.com/watch?v=1F8kiKHg0xo

[Art]

pic:

http://img.photobucket.com/albums/v186/ArtArt/Neon_Driver_zps6871fe8c.png

:)

[pedja089]

1N diode won't do much at higher frequency.
Try to find some UF diode, or some schottky 60V diode, 1A...
Edit:
Add a lot of decoupling capacitors(10nF,100nF)
I think that you didn't put diode right way.
It should be like this
http://i.stack.imgur.com/Oarz3.gif
You can try to put 1N4148, but it probably won't last long...

[Art]

I have used the diode like the second example on this page:
http://electronics.stackexchange.com/questions/31014/where-should-i-put-the-kickback-diode-in-a-transistor-switch
The first way actually prevented the neons lighting, no matter what polarity the diode across the primary.

I don't have anything other than rectifiers at the moment, but will see how I go tomorrow.

[pedja089]

Second example won't do anything to spikes, it just prevent transistor from negative voltage.
I think that mosfet would be best solution. Because it have zener diode between source and drain.
You can also make RCD clamp, like this
http://m.eet.com/media/1055546/C0541-Figure4.gif

[Art]

Second example won't do anything to spikes, it just prevent transistor from negative voltage.
I think that mosfet would be best solution. Because it have zener diode between source and drain.
You can also make RCD clamp, like this
http://m.eet.com/media/1055546/C0541-Figure4.gif


Will a P-Channel MOSFET directly replace the MJE3055?
http://www.jaycar.com.au/products_uploaded/ZT-2467.pdf

Note this schematic has no protection diodes at all.

http://img.photobucket.com/albums/v186/ArtArt/Neon_Schema_zps376e8aa3.jpg?t=1385515523

[pedja089]

N channel mosfet will replace MJE3055... But you need one with logic level gate, to be able to drive them directly from pic output.
EDIT:
This might be helpful
http://www.dos4ever.com/flyback/flyback.html

[Art]

It looks like IRF540N then. It has 1-4Volt gate.. I could use a resistor on the pic output because it is 5 Volts.

[Art]

Well IRF540N works but it's only back to square one.
I'm still getting zapped if I put my fingers across the MOSFETs.

[pedja089]

That is ok, you can leave everything as is.
Zener in that mosfet have 100V breakdown voltage, so when you touch 100V...
That diode will protect transistor from both positive and negative spikes.

[Art]

Ok, thanks.
It's a software relay switch and indicators for an amplifier,
and later should be able to monitor some voltages as well.

http://img.photobucket.com/albums/v186/ArtArt/Soft_Power_zps71c21be4.png

[pedja089]

Nice project:)
I never got chance to play with valves...

[Art]

Thanks :)
Some vintage amps don't even have a clunk switch, and you're supposed to just turn it on at the power point,
and this "was" one of those.

Also, Just about every serious problem will make the transformer heat up, which could be detected for auto switch off,
but that might be too late for something, so it would be better to directly monitor some voltages,
if that doesn't present clock noise to it once shielded.

[pedja089]

Put some ferrite beads on pcb near connector on every pin except ground, also put toroid on wires. That should remove lot switching noise...

[Art]

Thanks for your help, it's not introducing noise, but will follow our advice anyway.

[Art]

My first PCB that isn't just a tagstrip or something.
Perhaps something only I can love ;)

http://img.photobucket.com/albums/v186/ArtArt/Neon_Driver_Art_zps46dc2cab.jpg

I'm glad it works, and since it's the 60's, it's fine to hand draw it,
and use DIL ICs and components like they did in the 60's :D

[Ioannis]

Wow! What you reminded me... OK, I was not building anything in the 60's but later started with similar pcb's.

With today's way of thinking, this seems too much effort to light two neon bulbs, when you can do it with just a LED and a resistor.

Congrats, Art for your persistence and make it working!

Where did you find those retro transformers?

Ioannis

[Art]

Hi, the transformers are new stock from my local electronics shop,
they have three types like this, and call them "transformers for decoupling",
and quote primary impedance, and nothing else. I was sure to grab spares.

One, I think was dodgy from the start, and should be replaced,
I can hear the PWM (even though very quiet) through one of the transformers.
I can even hear SOUND and DTMFOUT through that transformer.

It's not just for the neons.. I got the software relay switch working too so far:

https://www.youtube.com/watch?v=UfkuBugvrAk

It is possible to drive either side of a neon.
With either two transformers per channel,
or, I think, one transformer, and more MOSFETS,
you could move the light from side to side within one neon.