Problems with a power supply for an arc lamp

[Mathias]

Dear ng,

I need to drive a high-pressure mercury short arc lamp (HBO100) with a
lowcost power supply as a microscope light source (I've got a proper
secure lamp house, so don't worry about exploding lamps). The lamp needs
a current of exactly 5A at 20V, and so far I've used two computer ATX
power supplies in series (24V, max 30A), and a ~0.8R resistor (with
additional manual adjustment for the lamp's warm-up phase). The lamp
needs a curent pulse of ~4kV to ignite, which I use a piezo element for.
I hope you can read my ASCII-art, but so far it's really simple

P+24V:--0.8R-+----+
S L:5A Ig:4kV
U-:----------+----+

PSU: 2xATX, L: HBO100 lamp, Ig: Piezo ignition.

Now this works somehow (pressing the piezo button ~40 times is tiring,
but ok...) but only for a few starts.
I guess that the high voltage pulses gradually destroy the capacitors in
the ATX PSUs (do they?).

So my question is: do you have suggestions how to seperate the high
voltage pulse from the PSUs? If I use a 10kV rated capacitor at the PSUs
output, it prevents the lamp from igniting. If I applied the ignition
pulse with reverse polarity, I might use diodes to prevent the pulse
from reaching the PSU, but are there 5A rated diodes which are quick
enough for the ~1ms (?) pulse? Would it help?

Alternatively, I could of course use a commercial PSU but that is beyond
150..200$ even at ebay, and I need something an order of magnitude
cheaper


Thanks a lot for any suggestions!
Mathias



PS: does anybody know a way to generate 4kV pulses easily without manual
button-pressing?

 

[John O'Flaherty]

Dear ng,

I need to drive a high-pressure mercury short arc lamp (HBO100) with a
lowcost power supply as a microscope light source (I've got a proper
secure lamp house, so don't worry about exploding lamps). The lamp needs
a current of exactly 5A at 20V, and so far I've used two computer ATX
power supplies in series (24V, max 30A), and a ~0.8R resistor (with
additional manual adjustment for the lamp's warm-up phase). The lamp
needs a curent pulse of ~4kV to ignite, which I use a piezo element for.
I hope you can read my ASCII-art, but so far it's really simple

P+24V:--0.8R-+----+
S L:5A Ig:4kV
U-:----------+----+

PSU: 2xATX, L: HBO100 lamp, Ig: Piezo ignition.

Now this works somehow (pressing the piezo button ~40 times is tiring,
but ok...) but only for a few starts.
I guess that the high voltage pulses gradually destroy the capacitors in
the ATX PSUs (do they?).

So my question is: do you have suggestions how to seperate the high
voltage pulse from the PSUs? If I use a 10kV rated capacitor at the PSUs
output, it prevents the lamp from igniting. If I applied the ignition
pulse with reverse polarity, I might use diodes to prevent the pulse
from reaching the PSU, but are there 5A rated diodes which are quick
enough for the ~1ms (?) pulse? Would it help?

Alternatively, I could of course use a commercial PSU but that is beyond
150..200$ even at ebay, and I need something an order of magnitude
cheaper


Thanks a lot for any suggestions!
Mathias



PS: does anybody know a way to generate 4kV pulses easily without manual
button-pressing?

You might use an inductor along with your output capacitor to keep the
pulse away from the supply. Another thing, some discharge lamps, like
in camera flash units, use a third electrode wrapped around the
exterior of the lamp to create enough ions to break the gas down. Maybe
something like that would work, and wouldn't inject anything into your
power supply.

 

[Joerg]

Hello Mathias,

I need to drive a high-pressure mercury short arc lamp (HBO100) with a
lowcost power supply as a microscope light source (I've got a proper
secure lamp house, so don't worry about exploding lamps). The lamp needs
a current of exactly 5A at 20V, and so far I've used two computer ATX
power supplies in series (24V, max 30A), and a ~0.8R resistor (with
additional manual adjustment for the lamp's warm-up phase). The lamp
needs a curent pulse of ~4kV to ignite, which I use a piezo element for.
I hope you can read my ASCII-art, but so far it's really simple

P+24V:--0.8R-+----+
S L:5A Ig:4kV
U-:----------+----+

PSU: 2xATX, L: HBO100 lamp, Ig: Piezo ignition.

Now this works somehow (pressing the piezo button ~40 times is tiring,
but ok...) but only for a few starts.
I guess that the high voltage pulses gradually destroy the capacitors in
the ATX PSUs (do they?).

So my question is: do you have suggestions how to seperate the high
voltage pulse from the PSUs? If I use a 10kV rated capacitor at the PSUs
output, it prevents the lamp from igniting. If I applied the ignition
pulse with reverse polarity, I might use diodes to prevent the pulse
from reaching the PSU, but are there 5A rated diodes which are quick
enough for the ~1ms (?) pulse? Would it help?

As John wrote, an inductor might work. You could even get away with a
version that has a high enough inductance for the duration of the piezo
spark but let it saturate after the lamp has ignited. The inductance
would then drop to almost zilch but it's no longer needed anyway. Just
make the wire strong enough so it won't smoke.

A word of caution here: Switchers such as PC power supplies often do not
like being wired in series with their outputs. They might "regulate"
each other to death. Also, you might want to place a big zener on the
supply side to suppress what's leaking through the inductor since there
will always be some stray capacitance.

Alternatively, I could of course use a commercial PSU but that is beyond
150..200$ even at ebay, and I need something an order of magnitude
cheaper

When I lived in Germany it was pretty easy to obtain large 24V switcher
units from industrial electronics scrap yards. Reason is everybody
wanted 12V and consequently those were more expensive, nobody was
interested in 24V. Sometimes I got them for less than one Deutschmark
per Kilogram.

PS: does anybody know a way to generate 4kV pulses easily without manual
button-pressing?

A flyback transformer scrapped out of an old TV plus a pulser on the
primary side. Of course it has to be a TV that wasn't discarded because
of a blown flyback transformer

The oil furnaces you guys have also contain a high voltage transformer,
to generate the spark that starts the burner. These can often spark for
half a minute or more without getting hot. Talk to a furnace installer
(Heizungsinstallateur in German), they usually throw this stuff away
upon a new install so you might get one for free. I believe they are
called "Zuendtransformator".

 

[John O'Flaherty]

You might use an inductor along with your output capacitor to keep the
pulse away from the supply. Another thing, some discharge lamps, like
in camera flash units, use a third electrode wrapped around the
exterior of the lamp to create enough ions to break the gas down. Maybe
something like that would work, and wouldn't inject anything into your
power supply.

I probably wasn't clear about that third electrode thing. It's pretty
close to the ground end of the lamp, and you apply the high voltage
pulse to it w/r to ground. It would probably be really easy to try by
wrapping a small piece of aluminum foil close to the ground end of the
lamp, and hooking your high voltage source to ground and to it.

 

[[email protected]]

Dear ng,

I need to drive a high-pressure mercury short arc lamp (HBO100) with a
lowcost power supply as a microscope light source (I've got a proper
secure lamp house, so don't worry about exploding lamps). The lamp needs
a current of exactly 5A at 20V, and so far I've used two computer ATX
power supplies in series (24V, max 30A), and a ~0.8R resistor (with
additional manual adjustment for the lamp's warm-up phase). The lamp
needs a curent pulse of ~4kV to ignite, which I use a piezo element for.
I hope you can read my ASCII-art, but so far it's really simple

P+24V:--0.8R-+----+
S L:5A Ig:4kV
U-:----------+----+

PSU: 2xATX, L: HBO100 lamp, Ig: Piezo ignition.

Now this works somehow (pressing the piezo button ~40 times is tiring,
but ok...) but only for a few starts.
I guess that the high voltage pulses gradually destroy the capacitors in
the ATX PSUs (do they?).

So my question is: do you have suggestions how to seperate the high
voltage pulse from the PSUs? If I use a 10kV rated capacitor at the PSUs
output, it prevents the lamp from igniting. If I applied the ignition
pulse with reverse polarity, I might use diodes to prevent the pulse
from reaching the PSU, but are there 5A rated diodes which are quick
enough for the ~1ms (?) pulse? Would it help?

Alternatively, I could of course use a commercial PSU but that is beyond
150..200$ even at ebay, and I need something an order of magnitude
cheaper

Two points. Your short arc mercury lamp probably presents a slightly
negative impedance, and any power supply you use to drive it had better
be happy running as a current regulator.

If you want to put a couple of kV across the lamp to start itm one way
is to use a symmetrical step-up transformer. The interesting feature of
this transformer is that the 5A lamp current runs through the output
windings, so you need to use fairly heavy wire with enough insulation
to survive the voltage you apply to the single-turn primary.

For a 450W xenon arc lamp, drawing 24A, I used a big ferrite core,
gapped by about 0.1mm, with twelve turns of heavy wire on each side to
make a split 24-turn secondary.

I then used a Cockroft-Walton voltage multiplier to build up about 2kV
across a spark gap, and applied the 2kV pulse to a 1-turn widing on the
transformer.

The core saturated, but the circuit rang for long enough to eventually
transfer half the energy stored in the multiplier into the secondary,
where it duly fired the arc lamp.

The currents circulating in the seconary during ignition weren't all
that high - I put a high current reverse biassed diode across the
terminals of the power supply to cope with the negative-going current
excursions - and the system worked pretty well.

The elctronics shop tried to improve it after I left. and managed to
wreck a couple of the power supplies in the process.

 

[Frithiof Andreas Jensen]

Great to see some initiative ;-)

So my question is: do you have suggestions how to seperate the high
voltage pulse from the PSUs? If I use a 10kV rated capacitor at the PSUs
output, it prevents the lamp from igniting. If I applied the ignition
pulse with reverse polarity, I might use diodes to prevent the pulse
from reaching the PSU, but are there 5A rated diodes which are quick
enough for the ~1ms (?) pulse? Would it help?

A simple inductor in series with the supply should be enough - I would
try a few turns on a ring core out of the junk box first; try and see
if it reduces the number of ingnition attempts. If it does, it would
be a good experiment to add turns until it capacitance takes over and
the ingnition starts to fail again.

I think the supplies will not be damaged, the high voltage is probably
dissipated by the output filters.

For absolute stability, you should parallel each supply with a
resistor (57 - 120 Ohms, 10 W f.ex.) and a diode so that each supply
always "see" a minimum load and each supply cannot push current
through the other one. I have occasionally seen supplies latch "On"
when current was pushed through them and they hit the current limit.

Alternatively, I could of course use a commercial PSU but that is beyond
150..200$ even at ebay, and I need something an order of magnitude
cheaper

You showed "them" already - way to go.

PS: does anybody know a way to generate 4kV pulses easily without manual
button-pressing?

Ignition transformers/units for gas cookers - probably cheaper than
anything one can make. Run it off a one-shot timer so that a "long
enough" chain of pulses is generated when the supply comes on. The
local appliance spare parts pusher is a good place to get small bits
of functionality for one-off projects!

PS: .... and if is a prototype, the part is often generic and can be
sourced in thousands once one finds the real manufacturer. Automotive
spares is another source for odd bits that are expensive to make but
cheap to buy.

 

[jasen]

Two points. Your short arc mercury lamp probably presents a slightly
negative impedance, and any power supply you use to drive it had better
be happy running as a current regulator.

he has .8R in series with it, is that not enough?

Bye.
Jasen

 

[Mathias]

he has .8R in series with it, is that not enough?

Bye.
Jasen

During warm-up of the lamp the resistance drops, so I'd exceed the 5A,
thus at the moment I manually regulate the resistance up (and added a
large halogen light in series whose resistance rises with heat (i.e.
current) and counteracts to some degree).

Thx,
Mathias