xenon flasher experts needed

[ronald_jansen]

Hello,

I am currently designing a xenon-flasher for lighting a machine vision camera.
The flash frequency should be able to run up to 100Hz

I have already looked at a large number of schematics.
There seem to be two common ways to do it.
The one fires the trigger transformer with a Thyristor. The other uses an IGBT.

The interesting part is, that all the schematics that fire the trigger transformer with an IGBT, also switch the tube with the same IGBT. This wonders me. Why would you want the switch the tube? Such a switch is going to eat some of my precious energy. The tube is not going to conduct any current until it is fired. Afterwards it will automatically switch of when the buffer is empty.
I use a switching power supply to generate the pulse voltage. During the pulse the supply is set to standby, so there is no chance that the tube will remain on on the current delivered by the power supply.

Is is not possible to use the IGBT to switch the trigger transformer only? Or maybe even a FET?

 

[Nigel Goodwin]

I would imagine it's switched in order to turn it OFF before the capacitor is fully discharged, this allows shorter flashes, and faster flash rates.

 

[ronald_jansen]

I know for a fact that this is one of the reasons to do it like this. But the interesting part is, that all schematics using IGBTs do it like this. Also the ones from which I cannot imagine that they would want to interrupt the flash.

 

[dougy83]

The visible light output of the tube is greater with increased current (more infrared with loweer current). If you interrupt the tube current before the bulk cap is fully discharged, you don't waste as much of your energy making IR.

As for triggering the transformer, it's ok to use scr, mosfet, bjt, igbt, triac, switch - pretty much anything that can control passage of current. Obviously the blocking voltage of the device will have to be sufficiently rated for the voltage applied to the transformer.

 

[ronald_jansen]

I am always a bit wary of doing things nobody else has done before. They might have a reason to do it that way.

It might be that all SCRs just have high breakdown voltages. A MOSFET might be too sensitive and therefore not reliable. An IGBT should work, I guess. If they even switch the tube with it.

Do you know which factors should determine the trigger capacitors capacity?
A manufacturer recommended me a tube and a trigger transformer that goes with it. Then I found that the trigger transformer datasheet recommends a 47nF cap and the tube datasheet recommends a 220nf cap. Who is right? Or is it not that critical? I would prefer the 47nF. This would considerably reduce the charging resistors dissipation.
But I do not know if it will fire reliably under all circumstances if I do not use the cap recommended for this tube.

 

[gary350]

A friend gave me a used 5000 watt xenon flash lamp from the airport. When they are 95% used up they replace them before they burn out.

I have a cap bank, 3 caps in parallel. Caps say 410 MFD, 5000 VDC, 18,000. amp high energy caster oil filled. 18K x 3 = 54K 5K x 54K = 270 Meg Watts.

I discharged my cap bank into the 5000 watt flash lamp from about 100 feet away. It flashed as bright as the Sun. WOW. The flash lamp totally vaporized. The only thing I ever found was a tiny piece of melted glass about 1/8" diameter.

The cap bank will vaporize a 1/8" steel wire 5 feet long. It makes a BOOM like a high powered rifle. Very loud and lots of fun on the 4th of July.

I ran a piece of #24 copper wire around the yard on electric fence poles and insulators. 120 ft across the front and back yard and 210 ft along each side. Discharged the cap bank into that wire it went BOOM like det cord and make a flash like lightning all the way around the yard.

 

[BrownOut]

I wondered what that glow from the direction of Murfreesboro was

 

[ronald_jansen]

Wow 5000V! Be *very* careful not to vaporize yourself!

I think that comes pretty near to real lightning.

 

[schmitt trigger]

Apparently the IGBTs are very rugged with respect to high energy pulse handling capabilities.
They can also switch on reasonably fast, and can be turned off. A thyristor, which as you correctly mentioned, will only turn off until the capacitor's charge is completely depleted.

 

[ronald_jansen]

I did some more reading on the subject.

Apparently MOSFETS run into real problems because of the miller capacitance. The turn on forces the parasitic miller capacitance to discharge into the gate, causing th FET to close again. This considerably slows down the turn on process. Especially at higher voltages.

Bipolar should work theoretically. I am only not sure about which currents to count with. The tube manufacturers do not give a whole lot of data. The caps are foil caps with very low resistance. So I guess the tube will be the main current limiting factor. Which current will be there? I don't know. Theoretically it could run up to 1000s of amps.

Another thing that comes to mind is the testing.

Do you have some tips to increase safety?

The power supply is a flyback regulator. If the feedback loop is malfunctioning (you never know with those proto's) the voltage can get really high.
I was thinking about using a gas discharge tube to discharge the caps in an emergency situation.

Anything else I could do (except for wearing safety glasses)?

 

[schmitt trigger]

If you are concerned about the high voltage running out of control, you can always add a crowbar circuit. Goggle crowbar circuits.
For personal protection I would use the standard fare used in machine shops...you know, the transparent full face mask, the leather chest protector and gloves.... probably an overkill, but if you can borrow those from a shop, it could be worthwhile.

 

[Oznog]

I tried to make a xenon flasher once.

The current IS crazy high. So high it's actually generally impractical to switch. Hundreds of amps is not uncommon. If nothing else, check your capacitance and look at the dv/dt when fired and calculate current from that. Note that it's not just a matter of the transistor's surge load at say 0.4v Vsat * 300A or 300A^2*rds-on. How much gate or base drive can the transistor actually take to keep operating as a switch? You can't exceed Vgs-max of course. A bipolar, you could probably surge the base current far above Ibe-max for a flash, but how much? If your gain's 40, you'd need a 7.5A surge.

The xenon multi-flashers I know of used one really big storage cap, a smaller flash cap, and a resistor between them that allowed amps to flow but still allows the flash cap to drain so fast that it's guaranteed that the tube will quench and require a retriggering. You can't control the pulse width doing this unfortunately, but you can control the period.

I've not seen an IGBT, or any transistor, controlling the main current between the flash cap and the tube on the devices I took apart, or in any schematics. Again, the current's just crazy high. The turn-ON time may not be an issue because the transistor gate can be "on" before the tube is triggered. The problem is in being able to support a drive of hundreds of amps and switch "off" in those conditions.

gary350, good lord, where did you get those caps?
Did you build a "quarter shrinker" or what? Isn't that enough to shrink quarters?

 

[Mr RB]

I did one about 20 years ago that was very simple and worked at about 100Hz, it was a strobe for moving machinery.

I just used a automotive ignition coil with the secondary attached to the xenon tube and the 12v primary driven by a big old TO3 transistor, I think it was a BUX80 but it might have been a TV transistor. The transistor was driven by a 555 timer. It worked fine with coil current about 12v 5 amps and a lowish duty cycle.