Where did all the kits go?

OK, here's what I wrote in 2003...

Observe that a glow plug has a trivial resistance when cold and soon goes up as it starts to glow; in other words, the element is non-linear just like a light globe. I made this observation in my note book "My power supply can support a glow plug at 1.42V drawing 2.67A, this works out to

The output stage of my circuit consists of a power transistor, ballast resistor and the glow plug itself, all in series. I selected the ballast resistor so that the maximum collector current of the trany during the on-cycle would not be exceeded when the glow plug (or whatever the connector happened to touch) was at zero resistance. Then I calculated the on-cycle current and power based on 0.5 Ohm from above; and I finished the paper design by predicting a duty cycle which would create about 4W equivalent in the glow plug - 11% at approx 12V. YMMV depending on the voltage drop across your chosen trany and ballast resistor. The pulse train was generated using the ubiquitous '555. The power transistor had the on-current forced into its base through the collector of a driver trany.

For testing, I started with an automotive light globe. This strategy enabled me to tweak the duty cycle a bit. When I replaced the globe with a glow plug, my CRO showed that the switching transients were huge! (More so than with the light globe who's non-linear resistance is actually higher at the particular duty cycle). The EMI issue was fixed after considerable trial and error using a capacitor to earth on the node consisting of the driver trany collector and the output trany base. This worked far better than attempts to quench ringing along the high current output chain.

For what its worth... ballast resistance is 0.7 Ohm 10W tied to +ve rail, my output transistor is 2N3055 between ballast and plug, plug returns to earth, behind this chain... driver transistor is BC558, emitter of driver trany is on the +ve rail, its collector feeds base of output trany shared with 0.22uF EMI cap its base is fed from 555 pin 3 via 1K Ohm and... the +ve rail is protected with a 6Amp diode in case of reverse connection to 12V SLA

Had I been brave I might have reduced or eliminated the ballast resistor and adjusted the duty cycle accordingly.

Despite the fixed duty cycle there is little or no problem with wet plug. Its resistance goes down, it draws more current, it gets hot, its resistance goes up. I understand that some folks use a D size NiCad cell to drive the plug, no regulation, no problem.

Thanks for that Branko - the switching transients issue is interesting - what frequency did you use? Perhaps a higher frequency would reduce this as the plug wouldn't cool as much between pulses?

I'll definitely use a ballast resistor, not only to limit the maximum current, but I can also use it as the shunt for my ammeter which will be a

It worked out to about 700Hz. I didn't think it would make much difference and concentrated only on the duty cycle. If you experiment with it, please post your findings.

I wouldn't attribute plug cooling to much because thermal inertia is large by comparison to the pulse energy.

When it was on the bread board, I tried larger and larger capacitors on the output stage with little or no effect. It was quite a surprise that a modest capacitor on the base of the final drive transistor would reduce switching transients through a trivial reduction in rise time.

Of course you will need a sample and hold to do this but as a level shift and gain stage will also be necessary both objectives could be met with an op amp. On the other hand, once you hear the output transistor scream under load, you may think about using audio feedback.

By the way, my choice of ballast was based on Ic-max which was published by the OEM based on continuous duty. The component is really quite wastefull and it could be a worthy exercise to reduce its value in combination with an even shorter duty cycle.

That's true, although I wonder if the longer rise time might lead to more power dissapation in the transistor? Then again, as you suggest, it might well be trivial.

The 3914 idea was a hang over from my DC-DC converter idea where you're dealing with stable DC. As you say, with a pulsed circuit it'd be flickering like mad and would require a bit more circuitry to settle it down. The audio idea might be more realistic (and simpler!) and more useful at the field. I'll have to get on and do a prototype...

I guess with a 0R7 resistor it'd be dissapating almost as much power as the plug, but I am concerned about the very real risk of a short-circtuit and the ensuing smoke from the switching device. I'll have to check what mosfets I've got laying about and see what they can handle.

Many thanks for your wisdom :)

Technically, they do already...in the sense of foreign equity invested in the USA...

I think so..

It depends on what you class as investment...but in terms of actual wall street funds in bonds stocks and shares..do some research,. and start to worry...