Xenon arc lamp igniter problem

I found the igniter circuit from a short-arc xenon lamp power supply is
not reliable. most of the time, ionized plasma can be seen in lamps ,
but the lamp can't establish stable conducting current. The open
circuit voltage is ok (~120v), and current regulation is working fine
too. So I think the problem is the igniter pulse.

According to the article "Cermax lamp engineering guide" from
perkinElmer.com, peak voltage, rise time, and pulse width are all
related with triggerability. Since the power supply was working well
before, the following are the possible reasons that I'm now
considering:

1. The capacitance of the discharging (through primary winding)
capacitor decreased. (decreased trigger pulse width? or decreased peak
voltage?)
2. The dv/dt of the discharging capacitor deteriorated. (larger rise
time?)
3. the aging of the spark gap tube. the spark gap looks good, and I can
see spark inside and the sound is also a normal ping. but I don't konw
if an aging spark gap can decrease its sparkover voltage or increase
its on resistance.

any advice?

Another thing that bothers me is the estimation of the peak voltage.
The trigger transformer has 2 turns on primary and 25 turns on
secondary, and wound on a rod core (L~1.4", D~0.4", material unknown).
If I want 35KV peak voltage on secondary, how large the discharging
capacitor on the primary and how large the sparkover voltage I shall
choose? It seems to me that the larger the capacitance, the larger the
peak voltage and the pulse width on the secondary, is this true? The
power supply uses 2 6.8nF series connected ceramic disk(rated 12kV) as
the discharging capaictors. how large the sparkover voltage of the
spark gap I shall choose(assuming the capacitors can be charged up to
12kV)? Is 7.5kV too large?

Thanks!

 

[Dan Mills]

shrsmas@yahoo.com wrote:

I found the igniter circuit from a short-arc xenon lamp power supply is
not reliable. most of the time, ionized plasma can be seen in lamps ,
but the lamp can't establish stable conducting current. The open
circuit voltage is ok (~120v), and current regulation is working fine
too. So I think the problem is the igniter pulse.

How much 'boost' capacitance do you have (and what series resistor), normal
practice (at least on the 1.6 Kw + stuff I am familiar with), you have a
bank of 20,000uF or so in series with a few ohms across the power supply
outputs just before the igniter. The idea is that this cap bank supplies
the energy to heat the surface of the electrodes before the voltage
collapses from 120V to the normal operating region of 20V or so.

Without this getting the lamp to strike reliably is hard, as the plasma will
form then immediately collapse due to the cold electrodes having too high a
work function.

According to the article "Cermax lamp engineering guide" from
perkinElmer.com, peak voltage, rise time, and pulse width are all
related with triggerability. Since the power supply was working well
before, the following are the possible reasons that I'm now
considering:

1. The capacitance of the discharging (through primary winding)
capacitor decreased. (decreased trigger pulse width? or decreased peak
voltage?)
2. The dv/dt of the discharging capacitor deteriorated. (larger rise
time?)
3. the aging of the spark gap tube. the spark gap looks good, and I can
see spark inside and the sound is also a normal ping. but I don't konw
if an aging spark gap can decrease its sparkover voltage or increase
its on resistance.

any advice?

If the silly thing is flashing over initially but the plasma is failing to
stabilise, then I would suspect the boost supply rather then the igniter.
That said short arc Xe igniters are IMHO black magic and I wouldn't rule
anything out.

Another thing that bothers me is the estimation of the peak voltage.
The trigger transformer has 2 turns on primary and 25 turns on
secondary, and wound on a rod core (L~1.4", D~0.4", material unknown).
If I want 35KV peak voltage on secondary, how large the discharging
capacitor on the primary and how large the sparkover voltage I shall
choose? It seems to me that the larger the capacitance, the larger the
peak voltage and the pulse width on the secondary, is this true? The
power supply uses 2 6.8nF series connected ceramic disk(rated 12kV) as
the discharging capaictors.

Ahh, but the whole igniter circuit is designed to be resonant, this is not a
classical transformer! Think something more like a tesla coil (at least in
the ones I have seen).
Basically it is a high Q resonant circuit until the lamp ionises at which
point it is damped by the low impedance of the lamp plasma.

how large the sparkover voltage of the
spark gap I shall choose(assuming the capacitors can be charged up to
12kV)? Is 7.5kV too large?

I would think half that or less, but without actually seeing the circuit it
is kind of difficult to know.

Regards, Dan (Sometime Cinema tech).

 

[shrsmas@yahoo.com]

Dan Mills wrote:

How much 'boost' capacitance do you have (and what series resistor),
normal
practice (at least on the 1.6 Kw + stuff I am familiar with), you
have a
bank of 20,000uF or so in series with a few ohms across the power
supply
outputs just before the igniter. The idea is that this cap bank
supplies
the energy to heat the surface of the electrodes before the voltage
collapses from 120V to the normal operating region of 20V or so.
The boost capacitor is 10000uF, and it is also the power supply's main

filter capacitor. The power supply parallels a triple voltage
multiplier (boost) with a normal bridge (main) rectifier. The circuit
is current regulated by a pair of IRF250, so the initial series
resistor the lamp sees is nearly nothing. The boost supply seems fine
to me. BTW, the xenon lamp is merely 200W.

Ahh, but the whole igniter circuit is designed to be resonant, this
is not a
classical transformer! Think something more like a tesla coil (at
least in
the ones I have seen).
Basically it is a high Q resonant circuit until the lamp ionises at
which
point it is damped by the low impedance of the lamp plasma.
I think it's more like a pulse transformer without load.

how large the sparkover voltage of the
spark gap I shall choose(assuming the capacitors can be charged up
to
12kV)? Is 7.5kV too large?

I would think half that or less, but without actually seeing the
circuit it
is kind of difficult to know.
It seems that the calculation/simulation of such circuits is very

difficult for me. maybe the only way to repair this power supply is to
do some experiment. I really don't want to buy lot's of high-priced HV
capacitors and spark gaps.

Steve.

 

Dan Mills wrote:

shrsmas@yahoo.com wrote:

I found the igniter circuit from a short-arc xenon lamp power
supply is
not reliable. most of the time, ionized plasma can be seen in lamps
,
but the lamp can't establish stable conducting current. The open
circuit voltage is ok (~120v), and current regulation is working
fine
too. So I think the problem is the igniter pulse.

How much 'boost' capacitance do you have (and what series resistor),
normal
practice (at least on the 1.6 Kw + stuff I am familiar with), you
have a
bank of 20,000uF or so in series with a few ohms across the power
supply
outputs just before the igniter. The idea is that this cap bank
supplies
the energy to heat the surface of the electrodes before the voltage
collapses from 120V to the normal operating region of 20V or so.

Without this getting the lamp to strike reliably is hard, as the
plasma will
form then immediately collapse due to the cold electrodes having too
high a
work function.

I put together a trigger circuit for a xenon arc lamp back in 1972.

I used a mains-driven voltage multiplier to charge a stack of 1uF film
capacitors up to about 2kV, where they discharged through a spark gap
into a single-turn primary on a 24:1 step-up transformer, in which the
secondary winding was the power lead to the lamp (12 turns of each
lead).

I used a pair of big ferrite U-cores with a roughly 0.1mm air gap as my
core.

The first prototype used a home-made spark gap between two steel
spheres (1cm balls from a big ball-bearing) centred in a chunk of
silica-glass tube.

I let the UV flash from this spark-gap illuminate the electrodes of the
xenon arc lamp, and this arrangement triggered the arc lamp better than
90% of the time.

Later copies used a commercial spark-gap in a borosilicate glass
envelope, and they triggered the arc closer to 10% of the time.

My theory was that the initial 40kV voltage spike had to coincide with
the presence of a free electron between the electrodes of the arc lamp
if it was going to initiate a discharge. With the prototype, the hard
UV from the triggering spark-gap produced a photo-electron at the right
time, and there was enough energy stored in the transformer inductance
to sustain the discharge through the glow-discharge period until the
electrode surfaces in the arc lamp had heated up enough to sustain an
arc discharge - I found figures suggesting that only took about a
microsecond.

Once the we had an arc, the regular power supply - a linear constant
current source with a peak voltage output of about 40V DC - could
sustain the discharge.

In the later copies, the transformer-capacitor circuit had to ring
repeatedly until a cosmic ray or the like delivered an electron between
the arc lamp electrodes to start the discharge, and by then, enough of
the stored energy had been dissipated in the ringing that there was
enough left over to carry the discharge through into the arc regime.

Sorting out what might be going on took quite a while - even with the
0.1mm gap, the ferrite core in the pulse trnasformer spent most of its
time saturated, so the energy transfer from the primary to the
secondary side must have taken a number of cycles before there was
enough energy stored in the secondary side to generate the necessary
break-down voltage across the arc lamp electrodes.

Once the gap had broken down, the core would not have been saturated
and the primary and secondary sides would have been tightly coupled.

Back then, nobody made UV leds.

-------
Bill Sloman, Nijmegen