Maker Pro
Maker Pro

Q-switch driver

C

Chris Carlen

Jan 1, 1970
0
Hi:

Commercial q-switch drivers are upwards of $4000 to put 3-4kV into a few
pF in 1.5ns or so.

The drivers in the lasers I work with are custom made Marx banks done by
the laser manufacturer, it seems, and the schematics are not complicated
at all. Looks like something I can do if I just copied their circuit.
I don't think it can do 1.5ns though, more like in the 5-15ns range. I
am not sure of the exact relationship between q-switch switching time
and Nd:YAG laser pulse duration, but I'd like to get in the 10ns range
for a custom built laser.

I can barely afford the q-switch, so I'd have to make my own driver.

Does 3-4kV into 6pF in 5-10ns seem all that hard? It has to be done
only at 1-30Hz or so, and held for only a little longer than the laser
pulse, let's say 100ns.

I wonder how I'd measure the performance? It seems it would only be
possible indirectly by measuring perhaps the rise/fall times of the
modulation of a CW laser beam aimed through the q-switch.


Comments appreciated.


Good day!


--
____________________________________
Christopher R. Carlen
Principal Laser/Optical Technologist
Sandia National Laboratories CA USA
[email protected]
 
R

Rick

Jan 1, 1970
0
Chris Carlen said:
Hi:

Commercial q-switch drivers are upwards of $4000 to put 3-4kV into a few
pF in 1.5ns or so.

The drivers in the lasers I work with are custom made Marx banks done by
the laser manufacturer, it seems, and the schematics are not complicated
at all. Looks like something I can do if I just copied their circuit.
I don't think it can do 1.5ns though, more like in the 5-15ns range. I
am not sure of the exact relationship between q-switch switching time
and Nd:YAG laser pulse duration, but I'd like to get in the 10ns range
for a custom built laser.

I can barely afford the q-switch, so I'd have to make my own driver.

Does 3-4kV into 6pF in 5-10ns seem all that hard? It has to be done
only at 1-30Hz or so, and held for only a little longer than the laser
pulse, let's say 100ns.

I wonder how I'd measure the performance? It seems it would only be
possible indirectly by measuring perhaps the rise/fall times of the
modulation of a CW laser beam aimed through the q-switch.


Comments appreciated.

Sounds like an EO q-switch :)

There is an example of a 1KV 10000pF piezo driver in Art of Electronics on page
169

4KV into 6pF in 5ns...that means you need about 4.8 amps...hmmm, that sounds
like quite a bit
 
R

Rene Tschaggelar

Jan 1, 1970
0
Chris said:
Hi:

Commercial q-switch drivers are upwards of $4000 to put 3-4kV into a few
pF in 1.5ns or so.

The drivers in the lasers I work with are custom made Marx banks done by
the laser manufacturer, it seems, and the schematics are not complicated
at all. Looks like something I can do if I just copied their circuit. I
don't think it can do 1.5ns though, more like in the 5-15ns range. I am
not sure of the exact relationship between q-switch switching time and
Nd:YAG laser pulse duration, but I'd like to get in the 10ns range for a
custom built laser.

I can barely afford the q-switch, so I'd have to make my own driver.

Does 3-4kV into 6pF in 5-10ns seem all that hard? It has to be done
only at 1-30Hz or so, and held for only a little longer than the laser
pulse, let's say 100ns.

I wonder how I'd measure the performance? It seems it would only be
possible indirectly by measuring perhaps the rise/fall times of the
modulation of a CW laser beam aimed through the q-switch.


The requirements are quite something.
I was thinking about that subject too. First you spend some time
for the security measures to prevent getting yourself killed.
Then get bagfull of semiconductors. Each shot that was somehow
unperfect costs you a handfull of of them, solder them out -
replacement in.

The highvoltage probeheads are not suitable.

You can measure the performance by cutting a modelocked laser and cut
during the pulse. Delaying the cut by picoseconds should let the laser
sample the slope of the Q switch. Measure with photodiode.

You can get a ready switch from Behlke. It works and has a pricetag.
Whatever you do may be cheaper from the parts, but may take you weeks.

Just in case you succeed, let us know.

Rene
 
J

John Larkin

Jan 1, 1970
0
Hi:

Commercial q-switch drivers are upwards of $4000 to put 3-4kV into a few
pF in 1.5ns or so.

The drivers in the lasers I work with are custom made Marx banks done by
the laser manufacturer, it seems, and the schematics are not complicated
at all. Looks like something I can do if I just copied their circuit.
I don't think it can do 1.5ns though, more like in the 5-15ns range. I
am not sure of the exact relationship between q-switch switching time
and Nd:YAG laser pulse duration, but I'd like to get in the 10ns range
for a custom built laser.

I can barely afford the q-switch, so I'd have to make my own driver.

Does 3-4kV into 6pF in 5-10ns seem all that hard? It has to be done
only at 1-30Hz or so, and held for only a little longer than the laser
pulse, let's say 100ns.

I wonder how I'd measure the performance? It seems it would only be
possible indirectly by measuring perhaps the rise/fall times of the
modulation of a CW laser beam aimed through the q-switch.


Comments appreciated.


Good day!


Sounds like a string of Zetex SOT-23 avalanche transistors might work
here. Say 350 volts each, 25 amps peak or so, sub-nanosecond turnon.
You only have to trigger one of them and the whole string breaks down.

How does the laser manufacturer do it? Fets?

The optical measurement would obviously be the ultimate test, but it's
easy to make a blindingly fast HV scope divider with one of those
Caddock non-inductive axial resistors.

John
 
C

Chris Carlen

Jan 1, 1970
0
John said:
Sounds like a string of Zetex SOT-23 avalanche transistors might work
here. Say 350 volts each, 25 amps peak or so, sub-nanosecond turnon.
You only have to trigger one of them and the whole string breaks down.

How does the laser manufacturer do it? Fets?

The last I looked, a Marx bank of 7 BJTs. Looked quite simple.
The optical measurement would obviously be the ultimate test, but it's
easy to make a blindingly fast HV scope divider with one of those
Caddock non-inductive axial resistors.

John


Thanks for the reply!


--
____________________________________
Christopher R. Carlen
Principal Laser/Optical Technologist
Sandia National Laboratories CA USA
[email protected]
 
B

Boris Mohar

Jan 1, 1970
0
Hi:

Commercial q-switch drivers are upwards of $4000 to put 3-4kV into a few
pF in 1.5ns or so.

The drivers in the lasers I work with are custom made Marx banks done by
the laser manufacturer, it seems, and the schematics are not complicated
at all. Looks like something I can do if I just copied their circuit.
I don't think it can do 1.5ns though, more like in the 5-15ns range. I
am not sure of the exact relationship between q-switch switching time
and Nd:YAG laser pulse duration, but I'd like to get in the 10ns range
for a custom built laser.

I can barely afford the q-switch, so I'd have to make my own driver.

Does 3-4kV into 6pF in 5-10ns seem all that hard? It has to be done
only at 1-30Hz or so, and held for only a little longer than the laser
pulse, let's say 100ns.

I wonder how I'd measure the performance? It seems it would only be
possible indirectly by measuring perhaps the rise/fall times of the
modulation of a CW laser beam aimed through the q-switch.

May be you use some info from here

http://www.avtechpulse.com/papers/
 
R

Rene Tschaggelar

Jan 1, 1970
0
John said:
On Thu, 22 Jan 2004 14:30:07 -0800, Chris Carlen


[snip]


Sounds like a string of Zetex SOT-23 avalanche transistors might work
here. Say 350 volts each, 25 amps peak or so, sub-nanosecond turnon.
You only have to trigger one of them and the whole string breaks down.

How does the laser manufacturer do it? Fets?

The optical measurement would obviously be the ultimate test, but it's
easy to make a blindingly fast HV scope divider with one of those
Caddock non-inductive axial resistors.

John,
Thanks for the notion of the Zetex Transistors.
The scope probe doesn't look that easy to me.
Assume the load capacity of the marx bank, the pockels cell, is
in the order of 5 to 10pF. So the probe cannot have 1pF, is has
to be somewhat less. The fast inputs of the fast scopes are not
1MOhm, but 50 Ohms. I cannot dream up a 1000:1, 50 Ohm output
divider that does 1ns. One usually has the expensive FET probes
in this frequency region.
How would that be done with a resistive divider ?

Rene
 
F

Fred Bartoli

Jan 1, 1970
0
Rene Tschaggelar said:
John said:
On Thu, 22 Jan 2004 14:30:07 -0800, Chris Carlen


[snip]


Sounds like a string of Zetex SOT-23 avalanche transistors might work
here. Say 350 volts each, 25 amps peak or so, sub-nanosecond turnon.
You only have to trigger one of them and the whole string breaks down.

How does the laser manufacturer do it? Fets?

The optical measurement would obviously be the ultimate test, but it's
easy to make a blindingly fast HV scope divider with one of those
Caddock non-inductive axial resistors.

John,
Thanks for the notion of the Zetex Transistors.
The scope probe doesn't look that easy to me.
Assume the load capacity of the marx bank, the pockels cell, is
in the order of 5 to 10pF. So the probe cannot have 1pF, is has
to be somewhat less. The fast inputs of the fast scopes are not
1MOhm, but 50 Ohms. I cannot dream up a 1000:1, 50 Ohm output
divider that does 1ns. One usually has the expensive FET probes
in this frequency region.
How would that be done with a resistive divider ?


Why a resistive divider to measure a pulse ?

Use a capacitive one with a 350MHz 10:1 probe.

Say you want 3Kv for 50V at the probe tip. This 60:1 ratio will give you an
input cap of 10p/60 = 0.17p
The input time constant is 10M*10p = 100us is more than adequat.

0.17p is a 2cm^2 @ 1cm if neglecting fringing effects that increase the
effective capacitance.
That sounds like it could work.

The 100uS time constant lives you plenty of time to calibrate this with an
ordinary LF generator.

One could also use a Fet probe with a 3KV/5V ratio, giving a low 0.017p
load.
Of course be sure to not let it spark.


Fred.
 
R

Rene Tschaggelar

Jan 1, 1970
0
Fred said:
John said:
On Thu, 22 Jan 2004 14:30:07 -0800, Chris Carlen


[snip]


Sounds like a string of Zetex SOT-23 avalanche transistors might work
here. Say 350 volts each, 25 amps peak or so, sub-nanosecond turnon.
You only have to trigger one of them and the whole string breaks down.

How does the laser manufacturer do it? Fets?

The optical measurement would obviously be the ultimate test, but it's
easy to make a blindingly fast HV scope divider with one of those
Caddock non-inductive axial resistors.

John,
Thanks for the notion of the Zetex Transistors.
The scope probe doesn't look that easy to me.
Assume the load capacity of the marx bank, the pockels cell, is
in the order of 5 to 10pF. So the probe cannot have 1pF, is has
to be somewhat less. The fast inputs of the fast scopes are not
1MOhm, but 50 Ohms. I cannot dream up a 1000:1, 50 Ohm output
divider that does 1ns. One usually has the expensive FET probes
in this frequency region.
How would that be done with a resistive divider ?

Why a resistive divider to measure a pulse ?

Use a capacitive one with a 350MHz 10:1 probe.

Say you want 3Kv for 50V at the probe tip. This 60:1 ratio will give you an
input cap of 10p/60 = 0.17p
The input time constant is 10M*10p = 100us is more than adequat.

0.17p is a 2cm^2 @ 1cm if neglecting fringing effects that increase the
effective capacitance.
That sounds like it could work.

The 100uS time constant lives you plenty of time to calibrate this with an
ordinary LF generator.

One could also use a Fet probe with a 3KV/5V ratio, giving a low 0.017p
load.
Of course be sure to not let it spark.


The capacitive divider is great. Do your own and calibrate.
I'll have a look into it.

Thanks

Rene
 
J

John Larkin

Jan 1, 1970
0
John said:
On Thu, 22 Jan 2004 14:30:07 -0800, Chris Carlen


[snip]


Sounds like a string of Zetex SOT-23 avalanche transistors might work
here. Say 350 volts each, 25 amps peak or so, sub-nanosecond turnon.
You only have to trigger one of them and the whole string breaks down.

How does the laser manufacturer do it? Fets?

The optical measurement would obviously be the ultimate test, but it's
easy to make a blindingly fast HV scope divider with one of those
Caddock non-inductive axial resistors.

John,
Thanks for the notion of the Zetex Transistors.
The scope probe doesn't look that easy to me.
Assume the load capacity of the marx bank, the pockels cell, is
in the order of 5 to 10pF. So the probe cannot have 1pF, is has
to be somewhat less. The fast inputs of the fast scopes are not
1MOhm, but 50 Ohms. I cannot dream up a 1000:1, 50 Ohm output
divider that does 1ns. One usually has the expensive FET probes
in this frequency region.
How would that be done with a resistive divider ?

Rene

A Caddock MD1248 resistor is about a tenth of an inch in diameter and
an inch long. If you cut the leads short, you can poke one end into an
SMA female (on a 50 ohm cable, with attenuators as needed) and the
other end becomes the scope probe tip. A 950 ohm resistor makes a 20:1
divider probe with 250 fF capacitance and about 6 GHz bandwidth.
Higher value resistors sacrifice bandwidth, but a 5K would still be
above a GHz response, plenty good for what Chris is doing, with a load
of 'under' an amp peak at 4KV. Avalanche transistors would hardly
notice an extra amp, and that's well below what a 1 pF capacitive load
would pull at this slew rate.

That's basicly all an HP 54006A probe consists of, although they do
have a nice mechanical setup and grounding pin.

Actually, at a mere GHz or so, a string of surface-mount resistors
works pretty well, too.

The capacitive thing is nice, too, but you probably have to use a
homemade capacitor, something in the 0.1 pF range is nice, and you
have to calibrate it. How about using a very short piece of 50 ohm
hardline coax as the cap?



in---======== ==========================

------------------------------------------- out

======== ========================== gnd

gap


This will differentiate, which an RC terminator will fix at the end.


John
 
S

Scott Stephens

Jan 1, 1970
0
Chris said:
Does 3-4kV into 6pF in 5-10ns seem all that hard? It has to be done
only at 1-30Hz or so, and held for only a little longer than the laser
pulse, let's say 100ns.

That has been discussed here before, so Google on avalanche, nanosecond,
et. Find some patent references that use common mosfets (irf740) to dump
a pulse through common HV rectifiers (snap or trapped-plasma mode er
something). Kilovolt levels in nanoseconds, from your local Radioshack?

--
Scott

**********************************

DIY Piezo-Gyro, PCB Drill Bot & More Soon!

http://home.comcast.net/~scottxs/

**********************************
 
J

John Larkin

Jan 1, 1970
0
That has been discussed here before, so Google on avalanche, nanosecond,
et. Find some patent references that use common mosfets (irf740) to dump
a pulse through common HV rectifiers (snap or trapped-plasma mode er
something). Kilovolt levels in nanoseconds, from your local Radioshack?


Search for "drift step recovery diode" or "Grekhov diode".

John
 
Top