[Gaddaffyduck]

Hello,

First, I'm from Norway, quite young, so I apologize for bad grammar/spelling in advance.
Also unsure where to ask, so please move this thread if necessary.

I have an idea that I need some help with.
I can nothing about chemistry, and I refers me into this forums to hear what your opinion is about this idea, and if this is possible to realize.

Briefly about the idea:
You are driving on a dark winter evening, it snows quite close, and you have almost no visibility.
High beam makes it almost impossible to see anything because the glare from the car's light hitting the snow is high.
Especially vulnerable vehicles will be cars with off Road Light with High Lumen (?)
then the reflection so high that driving lights are preferred instead.
My idea is to create a transparent film, which you can place inside the window of the vehicle, which will react to the bright light that is reflected when we have the high beam.
below you see illustrated how I think.



It is not intended that the film will be a "photochromism" that they use in transitions glasses, it would be downright dangerous in traffic since the entire window will be dimmed I guess.
but the same principle as photochromism. High beams hit the snow, the snow is dimmed down like illustrated above.

Is this possible and create at all?
Thoughts?

[Borek]

Doesn't sound realistic to me. Note that the piece of glass through which you see the snow flake depends on the head position, so you would need a system that adapts to the driver head movements. That on top on many other problems.

[DrCMS]

The photochromic dyes used in reactive glasses needs quite a bit of light (usually UV) to work. 

The situation you are trying to fix involves much less light and significantly less UV and if you use more dye it would go very dark during daylight hours.

A polarizing filter might be a possibility?

[Enthalpy]

Hi Gaddaffyduck, welcome here!
No worry with your English, it isn't my mothertongue neither, and I'm happy to share a language with you.

Seeing remote objects while the car lights illuminate the snowflakes falling near the driver is a difficult task. Many people have tried hard at it, and also against raindrops, but no widespread solution has been found. I don't believe a polarizer screen would help.

I saw the report of an experiment (last year?) where the car had cameras to detect the positions of individual raindrops as they fell within sight. Then, a computer deduced the future positions of the drops, and (if I remember properly) controlled many thin agile beams of light to illuminate the path only where the drops were not. Less than simple... but reportedly it worked. Adapt it to snowflakes?

An imaging radar to replace lights and eyes? At 50mm wavelength to be less sensitive to snowflakes, an antenna spread over 1.5m width and height would distinguish only 1m separation at 50m distance.

I'd favour no lights at all at the car. Amplify natural light if some is available. This illuminates remote objects, not just the nearby snowflakes. For the car to be seen, alternate quickly between lights on and the display of amplified natural light.

Several cameras can rebuild a 3D scene, where the distance of the individual snowflakes is identified. This is a difficult but classic task. With enough cameras, software can erase the near snowflakes and put instead the part of the distant scene taken by an other camera that has a clear sight in this direction. Or, replace the near snowflake with the part of the distant scene that was filmed just before; this accepts fewer cameras but may miss some new objects.

----------

I long wanted to make the following setup, but I suppose affordable technology doesn't have the components ready. Imagine a shutter, say a camera sensor, or a film on the car's screen that can be switched transparent for just 100ns (or better) in our case, cycling say every 100µs. Then you have lights synchronized with the shutter: strong at the beginning of a cycle, weak at the end. Light emitted first has more time to go forth and back before the shutter's opening hence illuminates farther, and is stronger to compensate the distance, so distant parts of the scene appear as bright as the near ones, which receive weaker light emitted later.

Alternately, the light pulse can be uniform, and the detector's sensitivity increase over the cycle duration, for instance as a shutter becomes progressively transparent. Several superimposed shutter films could also vary between less and more transparent, maybe just switch each between two transparencies, and cumulate to the needed dynamic range. Yes, radars do similar things.

As camera sensors have a big dynamic range and can be fast, successive pictures can be taken within a cycle of uniform light pulse, and the later pictures are more amplified, as they correspond to a bigger distance; for instance, their exposure time can be longer. This needs a fast information (=charges) transfer only at the pixel scale: each pixel sends the light-created charges to its own charge accumulator or to a dump, in a proportion resulting from relative timings or from attenuation, that is designed to compensate the attenuation resulting from distance. The accumulator can be read just once per lighting cycle, or even, it can sum over several lighting cycles and be read with a period convenient to the use, say 100 time per second for human sight. So the global data transfer from the sensor can be reasonably slow.

Instead of varying the light intensity or the receiver's sensitivity within one illumination cycle, it can vary from cycle to cycle, together with the delay between illumination and reception.

A lidar, which detects light only for one or few pixels at a time, can realize the distance compensation more easily, but is difficult to make fast enough for a car. It could be nice on a boat.

Maybe I add a few sketches later. After all, members of a chemistry forum are not supposed to be fluent on radar, lidar and camera sensors, so I realize the text alone may be hermetic.

----------

Such a lighting+sensor or shutter is useful against snowflakes that it renders only as brilliant as remote objects of the scene, or even, it can render the nearest objects less brilliant. More generally, it gives a better sight, under rainy, foggy or clear conditions, by compensating the attenuation that results from distance.

Some vehicles can pay more than a small car to have earlier such a sight: snowcats, maybe airliners...

I wish cameras have this distance compensation when using a flash light, so that scenes appear uniformly illuminated.

Marc Schaefer, aka Enthalpy

[Enthalpy]

Here's already an illustration about the illuminating light pulse that decreases over time, so that the receiver's shutter favours light reflected by distant objects - please log in to see the attached sketch.

When the shutter (it can be electronic) opens, the reflection from near objects, like the undesired snowflakes, results from the end of the emitted light pulse, which is intentionally weaker. As opposed, the more delayed reflection on more distant objects results fro the strong beginning of the emitted pulse.

To compensate for the distance, the emitted power can depend on time high four. But to compensate an attenuation by fog, an exponential (which can multiply the previous fourth power) must be better. Refinements are possible - keeping in mind that 2*5m take only 33ns.

----------

Or increase the receiver's sensitivity over one cycle to compensate for fainter reflection by farther objects. No sketch for that.

Marc Schaefer, aka Enthalpy

[Borek]

I don't see how your setup is going to work.

Can you estimate integrated amount of light received by the camera, reflected from two identical objects, one located closer, the other more distant?

[Enthalpy]

Thanks for your interest, and my apologies for the late answer! Here are some power and signal estimates.

Light emission and sensing shall here provide a uniform brightness (as daylight would) between 5m and 150m range, without atmospheric absorption. Below 5m, the angles of view of the separated lights and camera don't overlap, and the driver uses his eyes; beyond 150m, the brightness is allowed to drop.

One 10ms frame comprises 5000 2µs cycles of illumination and sensing. As a variant from the previous sketch, 3000 cycles serve only for 150m range and beyond, the remaining 2000 share the shorter ranges.

The lights emit 200W peak optical power around 500nm=2.5eV. They must respond quickly, some 10ns if they serve at short range as well; a bit streched for LED, but modest figures for the semiconductor lasers commonly used as optical pumps for YAG. The peak electric power of 600W implies to evacuate mean 200W heat, easier in this use because the chips are spread.

----------

Around 150m range, the emitted light shall spread over an ellipse 60m wide and 30m high: that's a bright 0.14W/m² peak. A 150mm*150mm area - being one camera pixel there - receives 3.2mW and can reflect 1/4 or 0.8mW to 2pi srd, of which the D=50mm camera lens at 150m catches 14ppb or 11pW peak for the pixel.

The 3000 long-range light pulses last 1µs each. The optics and sensor converting every second photon in an electron bring a signal of 14q for each pulse, while the sensor can have 30q noise. 3000 pulses sum 42,000q signal and (by sqrt(3000)) 1,600q noise.

300m range would be a modest effort. Longer pulses and the same angular resolution would bring 10,000q signal and about the same noise. A simple Peltier can reduce the noise (or save laser cost at 150m). The angular resolution at long range and the frame rate can be exchanged for signal-to-noise.

----------

Around 5m range, the same energy density shall be kept, as would occur with Sunlight (which implies that an object of same size is brighter to the observer). The corresponding 222 pulses last 15ns each, at the same peak power. To increase the dynamic range, say if fog needs it, reduce the number of pulses and active laser chips.

Near objects reflect light also from the long-range pulses. The electronic shutter discriminates this stronger early light: a dummy read cycle dumps the early charges to a waste bin - or several dummy cycles if this improves the rejection. At 500nm wavelength, carriers are <1µm shallow hence evacuated snappily.

----------

The development takes some effort (find hobbyist astronomers), but it seems that existing laser pump diodes and CCD sensors can be used, in which case the hardware costs few k€ in small production volume, say for snowploughs, snowcats, tugboats, fire trucks, open-pit mining equipment... until the car industry drops the costs.

Marc Schaefer, aka Enthalpy

[Enthalpy]

Airliners as well could benefit from the pulsed headlights and camera, when rolling on the ground in rain or light snowfall.

They shall afford 2kW electric power in semiconductor lasers to spread 600W light on a 600m*50m ellipse at 1500m, and a 100mm lens to catch 2.8pW from a 1m*1m pixel reflecting 25% of the light to 2pi sr. Light pulses repeat every 30µs in a 15ms frame; 250 pulses specialize in far range and last 10µs each.

50% quantum efficiency at 500nm=2.5eV brings 35q signal and 30q noise per pulse, or 9000q signal and 500q noise per frame. Cooling the camera sensor reduces the noise but would worsen condensation worries. The design can be tweaked significantly; its practical limit is the sight range that an identical rain- or snowfall would permit during daytime.

Control towers have similar needs. Boats as well, especially tugs in a harbour, ferries or barges. Trains maybe.

Marc Schaefer, aka Enthalpy

[Enthalpy]

Some uses demand a quick flashlamp. For instance a home camera: to compensate the illumination strength in 0.6m range slices, it takes a flashlamp whose light varies within 4ns. Most semiconductor lasers achieve this speed, but even with many 4ns pulses, providing the necessary energy within the duration of a xenon flash would be difficult for them and expensive.

A Blumlein-fed sparkgap is a candidate. When pumping a nitrogen laser, it achieves sub-nanosecond spark and light duration at 1atm.
http://de.wikipedia.org/wiki/Bl%C3%BCmleingenerator
http://en.wikipedia.org/wiki/Nitrogen_laser
The energy of a home-built Blumlein is like 50mJ instead of 500mJ, but better materials and shape will improve that, or repeated pulses. The transverse-fed gap lases (in superradiant mode) - a curved gap will avoid it. Nitrogen is inefficient and radiates at 337nm; while quick phosphors could make white light, a different gas (xenon?) must be more efficient, with a broad spectrum.

Is an even shorter pulse achieved at a third gap? The voltage difference would increase faster at the third (etc.) gap, exceeding more the breakdown voltage, hence accelerating the avalanche. This needs to control the signal propagation far better than at 1ns lasers (at least 100ps were already achieved with a standard Blumlein). To be tried.

The first spark of a Blumlein can knowingly be triggered by an additional electrode near the gap, or by a semiconductor component... The response is not nanosecond accurate nor repeatable, so a light sensor should detect when the light is emitted to synchronize the camera's sensor.

A quick flashlamp has uses beyond the illumination-versus-distance compensation. Microsecond flashlamps show quick events, nanosecond flashlamps shall show more.

Marc Schaefer, aka Enthalpy

[Enthalpy]

Not only the headlight or flashlamp, as well the camera sensor must be quick, in order to attenuate or reject the light reflected by near objects.

----------

Some CCD sensors are topped with an photomultiplier (or rather, a photocathode and an electron multiplier). These are commonly used as a fast shutter (100ps) by reversing the polarity at the multiplier. To compensate the illumination-versus-distance, one can open them shortly and few times when light pulses come quickly back from a near object, longer and more times for the delayed light pulses reflected by far objects. This sequence fits in one CCD exposure and readout, so the sensor's reasonable throughput relates just with the frame rate.

----------

My other option, sketched below, also reads a CCD sensor at the frame rate, but this special design adds flush or dimmer electrodes (or both) near each pixel, avoiding the photomultiplier. Within one frame, before the sensor is read, the flush electrodes can drain the pixel detectors at the end of desired insensitive periods instead of transferring the charge to the accumulator-read chain, and the dimmer electrodes can divert each a fixed fraction of the accumulated charge to reduce the sensitivity when needed.

Instead of a usual irreversible transfer cycle, the dimmer electrodes only get the same potential as the charge accumulator so charges spread according to the relative capacitances, then eject the charges they hold, part to the accumulator and part (indirectly) to the drain line. Repeated dimmer operations make the sensor far less sensitive at the beginning of a frame. Different sizes of combinable dimmer electrodes, varied exposure times and number of light pulses synthesize the desired attenuation versus time.

The operation resembles a polynome evaluation in software: {[(a*X+b)*X+c]*X+d}*X+e, where a b c d e are here image signals from increasing distances, and X an attenuation - adjustable at each step in our case.

Common CCD sensors offer 30ns transfer rate in order to achieve some 50 frames per second by reading all pixels serially. Flushing and dimming simultaneously all pixels can take 30ns for the whole image, enough for a vehicle.

A home camera with flashlamp demands faster flushing and dimming. 3ns instead of 30ns are achieved with charge transfer distances sqrt(10) shorter; the electrodes can be interleaved.

Not all electrodes are sketched here. The chain, dimmers, flush are split for proper operation. Colour is possible. The drain lines can be CCD as well.

Marc Schaefer, aka Enthalpy

[Enthalpy]

While the shutter is almost classic, the operation of my dimmer isn't standard for a CCD, so I provide here the usual sketch (visible if you're logged in) with electrode potentials and charge movements.

CCD use a semiconductor, depleted near the surface excepted the charges that represent the signal; in a camera sensor, the charges result from light. The semiconductor carries, like for a MOS transistor, a thin insulator and electrodes whose potential serves to attract or repel the charges.

Usual CCD transfer all the charges. The scheme with successive electrodes triplets ABC attracts the charges under A, then  A&B, B, B&C, C, C&A', A'... to transfer all charges from A to A' and accept the next ones under A.

My dimmer instead doesn't repel the charges under the signal accumulator after sharing them with the dimmer electrodes according to the capacitance. Instead, at epoch (1) it insulates the accumulator from the dimmer by a repelling potential at an ad hoc electrode, then at (2) the potential at a second ad hoc electrode opens the way from the dimmer to the drain. At (3) charges are evacuated to the drain from the main part of the dimmer, and from the second ad hoc electrode at the same time or just after. At (4) the first ad hoc electrode and (possibly just after) the main part of the dimmer attract charges again from the accumulator, and the cycle is ready to repeat if desired.

That is, the signal accumulator is not emptied as in normal operation; the charge amount is attenuated in a controlled fashion.

The drain can have junctions, contacts... to evacuate the charges, or be a CCD.

Marc Schaefer, aka Enthalpy

[Enthalpy]

Reversing the voltage can shut off the electron multiplier; reducing properly the voltage (first calibrate and store the response curve) must reduce the gain swiftly. This as well would compensate the illumination-versus-distance.

For a mass-produced home camera, the modified CCD is probably cheaper, but controlling the gain of the electron multiplier accepts lighter investments and can also be a demonstrator for the visual effect of the modified CCD.

Marc Schaefer, aka Enthalpy

[Enthalpy]

More about the Blumlein-fed flashlamp.

Hobbyist build flat Blumleins so their nitrogen lasers fire at the ends of the second gap.
http://de.wikipedia.org/wiki/Blümleingenerator
http://en.wikipedia.org/wiki/Nitrogen_laser
For a flashlamp to emit incoherent light, I suggest to fold the propagation lines so the second gap is at an exposed edge, and roll the lines to curve the gap, discouraging the gas to lase. The joined sketch (if you're logged in) shows it cylindrical.

A Blumlein has two propagation lines pre-charged to identical kilovolts, separated by a gap. When a first gap fires, it discharges the first line within many nanoseconds. As the voltage across the second gap fires it, say when the first line is discharged, the second line discharges in the first. Because the lines are wide and thin, and the second gap wide and close to the ground, the current sees a tiny inductance and wave impedance, and reaches tens of kiloamperes within a nanosecond, zap. The spark broadens quickly through photoemission and photoionization; the fast voltage drop in the first line also brings the voltage drop at the gap well over its DC breakdown, helping to broaden quickly.

This is different and hopefully better here:

• The first gap is wide instead of a point, and is split, with a third conducting ring in between. The third conductor shall bring the fast firing known elsewhere to the Blumlein's first gap. Two big resistors keep its potential at midway; when the gap fires, by action on it or a separate trigger, the third conductor's potential swings brutally and fires other places of the gap - so my understanding. A wide spark would also reduce the wearout.
• The gap, as a closed cylindre, halfs the electric distance the the initial firing point to the most remote. The optical distance is even shorter: one diameter through a gas.
• Mirrors, for instance ellipsoidal, can strengthen the optical coupling within the cylindrical gap.
• A screen avoids the first gap's light to trigger the second gap prematurely. Felt would resist the pressure wave - or merge the screen and a mirror. But a delay line (not sketched) could bring light to the second gap to trigger it strongly if this helps.
• The cylinder helps the voltage across the second gap increase uniformly even if the first gap fires at a single place.
• The second gap is a closed cylindre, with the same advantages. The higher outer ring and a negative charge shall help photoemission.
• A smaller mirror near the centre of the second gap can strengthen the optical coupling.
• The proper gas emits more light; xenon at a few bar creates strong light as a white continuum in microsecond discharge flash lamps.

The insulator could be a ceramic if it lives longer, and the gap electrodes refractory. Here the ground plane is central to reduce the second gap's inductance; this demands a good Faraday cage, for instance at the pressure vessel. I compute with ZrO₂ (relative permittivity = 31, breakdown 30MV/m), possibly stabilized with Y₂O₃, 1mm thickness, 60mm diameter, 80mm height.

With 6.7nH/m and 52nF/m, wave impedance is 0.36ohm, and one 80mm leg takes 1.5ns, for a 3ns discharge roughly. The second line stores 0.47J at -15kV; a home camera can repeat the discharges within one snapshot. Argon would fire at 15kV for about 5mm * 10bar gap, and xenon a bit more. A uniform discharge, 5mm long and 2mm above the ground plane, would be 67pH inductive which would add 100ps rise time, but a narrow spark of 3nH fed by a concentrated 2*1ohm wave impedance would add 1.5ns.

Even in vacuum, 15keV electrons would take 140ps to cross the gap, so the nanosecond discharge observed at Blumleins allow only for a short electron avalanche and photoemission. An hypothetic alternative would make the discharge in vacuum with a fluorescent material at the anode, or in a gas at a pressure well below Paschen's minimum with a deep hollow anode or a fluorescent pressure vessel.

==============================

A bigger diameter stores more energy; as opposed, the inductance of the second gap limits any compact design. Keep for instance a gap of 60mm diameter, 5mm length, and 3.5mm above the return current: its 120pH permits two lines of 39mohm wave impedance adding 1.5ns rise time - under the most optimistic assumption that the discharge spreads evenly.

The electrodes could then be axial-radial as on the second sketch, alternating each line with ground planes. Again with 1mm thick ZrO₂, line pairs of 57 layers would fit on 73mm mean diameter, be 15mm wide, and 40mm high to reduce to 1.5ns the round trip propagation. After a non-trivial fabrication, -15kV would store 2J in the second line. A better dielectric would keep the energy but reduce the size.

The second gap might consist of many radial electrodes, say one per line element. This would reduce their inductance and allow more energy at the same speed, provided that they fire all within a short time, which is doubtful; optical coupling shall help. Fewer gaps than line elements could make a multi-armed spiral.

Or have a rectangular design as on the third sketch, with many second gaps perpendicular to the electrodes. The mirrors shall help fire all gaps, and a contaminator as well, similar to the additional wire at the first gap. This design is easy to prototype with printed circuits and vacuum impregnation.

Marc Schaefer, aka Enthalpy

[Enthalpy]

The Blumlein could feed a tube to emit nanosecond X rays pulses. The impedance variation of a line shall provide the high voltage. Sketch here under if you're logged in.

----------

Both lines start with +40kV as an example. Al₂O₃ (K=9 and 30MV/m) between r=50mm and R=53mm, 100mm long, stores 0.67J at 1.2ohm line impedance, with 2ns round trip. When the first line is about discharged, the second gap fires, creating a -20kV 17kA step at the start of the second line.

The second line tapers to r=8mm R=29mm in vacuum for 77ohm line impedance. Gaussian impedance transitions preserve pulse shapes, ask your local RF specialist. The step has -160kV amplitude, so from initial +40kV it feeds the tube with -120kV 2kA over 2ns or 0.5J.

Fields are amplified where the dielectric leaves the surrounding ground. The lower permittivity material shall alleviate this. Silicone rubber at least coexists friendly with ceramics.

----------

Pure N₂ needs about 15bar to fire at 40kV over 3mm. Its light (337nm=3.68eV) lets LaB₆ (2.70eV) and CeB₆ emit reported 7ppm electron per photon, enough to seed the avalanche. If the borides survive, this would favour the quick photoemission breakdown. ThO₂ (3.03eV) or La₂O₃ (2.71eV) filling W would be more durable, but they will see few photons.

Traces of a gas more easily ionized favours the photoionization breakdown. This worsens the gaps' firing voltage; would it accelerate the breakdown?

Electrodes of different radius favour a clean firing. The proper polarization is opposite to an avalanche particle detector... The odds through reasoning being 50-50, just experiment which polarization.

Mirrors shall favour photo-everything breakdown, while a screen separates the gaps.

----------

To emit X rays, the ancestor machine (with a Marx generator instead of the Blumlein) described in US Patent 6,166,459 has whiskers at the cathode, whose vaporization due to field emission current makes a plasma able to carry the current. Maybe the description of a natural process where the whiskers reappear at each cycle.

This process works, it may even be unavoidable, and must be considered for the present attempt. I describe now a strict field emission alternative, which is uncertain since vacuum breakdown isn't well understood.

The D=2mm cold cathode has 400 photolithographic cones of r=7µm to R=50µm tip to base. I take Cu (4.65eV), but Mo or W are possible. 120kV split as 92kV in straight 1.3mm (adjust as cones wear out) and 28kV in 400 conical paths of pi steradians achieving 3.5GV/m at the r=7µm tips. Fowler-Nordheim lets each emit 5A. Very thin LaB₆ at the tips would ease emission but must be less durable. La₂O₃ powder in W would need grains far under 14µm or be a film.

5A drop 10mV in each copper cone, and the heat spans only 500nm in 2ns. I consider electrons cool the cones: they enter at 26meV and exit at 26+10meV. Since the hotter electrons evade more easily, they could carry even more heat away. At peak 32mA/µm², electromigration isn't a worry.

The anode has a 40µm W film to stop the electrons. It could be ²³⁸U. The film could be just thinner than 29µm to improve the X-ray spectrum. The electron pulse releasing 0.5J in d=2.5mm *20µm would heat the film momentarily by 2000K, but heat exits the 20µm within the 2ns. A 2mm Al (or Cu) window hold in a cone can evacuate mean 150W heat - or make it thicker, possibly of Be, or cool by fluid. 150W permit 300Hz sustained repetition rate.

Converting 1% in photons, of which 1/10 * 1/30 have the proper energy and direction, gives 10⁹ photons per bunch. An object absorbing 9/10 and a 100,000 pixels detector 1/4 efficient leave 250 photons per pixel.

----------

The design is adaptable somewhat. Spike radii at the cold cathode adjust the breakdown voltage and the anode distance. Other pulse energies and voltages are possible, and a switchable gamma source welcome; insulating 1MV shortly in vacuum isn't that bulky.

The lines are less adaptable. Twice the 77ohm is already difficult, but crimping achieves less than 1.2ohm. More than 40kV is possible. Ferrite cores are not obvious.

----------

What uses for this source? Imaging of course, and 300 frames per second are by far not the limit. But what more?

• Study how vacuum insulation breaks down over a short time.
• Observe X-ray fluorescence, with a strong 100keV quick source.
• In backscattering or fluorescence imaging, compensate the illumination versus the distance, as previously described for light. Vary the gain through the detector's or photomultiplier's voltage.
• Any useful as the electron source of an accelerator?

Marc Schaefer, aka Enthalpy