Generating ultra-short pulses (in our case pulses of light) is not simple. There are advanced techniques to produce ultra-short light pulses, in the femtosecond or even attosecond range , but they require sophisticated lasers and equipment that are not within the reach of a “home” experimenter. Not to mention the fact that even detecting and measuring such short pulses is in itself a technological challenge.
In this post we will describe our attempts to realize a “Fast Pulser“, able to produce light pulses with duration less than 10ns.
As a light source a normal LED diode (blue 460nm) has been used, driven by a circuit for the generation of ultra-short pulses. From the scheme below it can be seen that the LED is driven by an avalanche transistor ZTX415 biased at 270V. The charge is accumulated in the capacitor C1 and then released in a very short time when a pulse arrives at the base of the transistor. On the resistor R2, to 0.1ohm, it may be taken the signal that corresponds to the impulse of current flowing in the diode LED.
In the “cover” image you see the oscilloscope trace of the current pulse: it is a spike of about 0.5V and a duration of less than 10ns, therefore extremely short. The duration and amplitude of the pulse depend strongly on the capacitance value of the capacitor C1. Increasing the value of C1 increases the duration of the pulse, while decreasing it can further reduce the duration of the pulse. In our tests we tried to reduce the capacity up to a value of 5pF.
In order to reduce the pulse duration it is necessary to reduce the parasitic inductances of the path followed by the current, in particular the resistance R2 and the connections between C1, the transistor, the LED, R3 and D1. This can be achieved by very short wiring and using a large ground area. The resistors are also to be chosen with low parasitic inductance.
R3 and R6, 50Ω value, have the function of allowing the recirculation of the current that passes through the LED, in order to contribute to the fast switching off of the pulse.
As seen from the oscilloscope trace, the positive pulse is followed by a negative pulse, this helps to empty the LED junction from the charge carriers, causing the LED to switch off quickly after the pulse.
The images below show the “fast pulser” circuit built on breadboard. The dark component on the left is the high voltage generator which, starting from stabilized 5V, generates the 270V necessary for the bias of the avalanche transistor. Note the copper base, connected to ground, on which are placed the components that produce the ultra-short pulse.
SiPM, which stands for Silicon Photomultipliers, is the most modern type of solid state photodetectors. They consist of a matrix of avalanche photodiodes, called pixels, operating in Geiger mode and connected in parallel, placed on a common Silicon substrate. We have already worked with SiPM :
The image below shows the component attached to its support with transparent adhesive tape. Its use is relatively simple and has the advantage of being more manageable than the classic photomultiplier, for this reason we decided to use it to measure and monitor the light pulse produced by the LED driven by our Fast Pulser.
The LED, together with the Fast Pulser, is placed in a “dark box”. The LED is shielded laterally and a pin-hole is placed in front to reduce the light emitted by the LED that reaches the SiPM sensor, placed in front of the LED, as shown in the image below.
How many Photons are Generated in one Pulse ?
We can easily calculate how many photons are generated in a pulse.
The energy stored in the capacitor is released during the pulse and it has the value of : E = 1/2CV2
Knowing that V = 270 V and C = 5 pF, we obtain for the pulse this energy : Ep = 0.18 μJ.
As a light generator we used a blue LED that emits a 460 nm.
From Planck rule we know that each photon carries energy calculated as : E = hν = hc/λ
Where c is the speed of the light and λ is the wavelength of the radiation, replacing the values and making the calculation we obtain that the energy of the single photon is E = 4.32 x 10-19 J
Assuming for the LED an efficiency of 10% we can calculate the number of photons produced:
N = (Ep / E) x 10% = 4.2 x 1010 photons
Naturally, not all the photons produced reach the SiPM sensor: there is the lateral shielding, the pin-hole and the distance at which the sensor is placed. Making this calculation is not easy because we do not know exactly the size of the pin-hole that was made “by hand” simply by piercing with a needle the thin aluminum foil placed in front of the LED.
We can estimate that the geometric reduction factor is between 1/1000 and 1/10000, therefore the number of photons reaching the sensor should be in the order of 106– 107 photons.
This is a huge number that immediately saturates the sensor, so it is necessary to further reduce the number of photons hitting the sensor. This can be done by interposing between the LED and the sensor some attenuation filters or polarizing filters, which have the same effect. Neutral density filters (ND filter) should be used, these filters reduce light intensity uniformly over all wavelengths.
The image on the side shows the filters we used to reduce the light intensity. These are neutral density filters of type D2, D4, D8, D16 which reduce light by a factor of 2, a factor of 4, a factor of 8 and a factor of 16 respectively.
The detection of the pulses produced by the SiPM sensor is achieved by means of a trans-impedance amplifier already described in the post Trans-Impedance Amplifier. The impedance seen by SiPM is 50Ω and the chosen operational is at high speed so as to produce short pulses at the output. It should be kept in mind that the SiPM generates pulses even in dark conditions, these pulses are due to the discharge of a “pixel” due to the thermal generation of charge carriers.
This behavior can not be eliminated even if it can be mitigated by making the SiPM sensor work at low temperature. The pulse generated in a pixel caused by a thermal electron can not be distinguished from that produced by the interaction of a photon. Since the simultaneous activation of pulses for thermal electrons in several pixels has very low probability the noise can be easily detected because the corresponding pulses have very low amplitude.
The graph below shows the pulse corresponding to the discharge of a pixel (dark pulse). In our electronics the width of a “dark pulse” is about 4mV.
We did the light pulse detection by varying the attenuation with the ND filters described in the previous paragraph. The image below shows the pulse that is produced without any attenuation. In yellow there is the current pulse on the LED while in blue the signal generated by the SiPM. You can see how the SiPM goes into saturation generating a “cut” pulse, a sign that the amount of light is excessive.
By inserting the attenuator filters between the LED and the SiPM, the amplitude of the signal produced by the SiPM can be greatly reduced, a sign that the number of photons received by the detector is much lower. The image below shows a light pulse with a limited amplitude and a duration of about 100ns. The actual duration of the light pulse is difficult to estimate because the SiPM front-end electronics inserts delays in the rise and especially in the fall of the pulse.
The image below shows the SiPM pulse produced with high attenuation filters. There is a lot of noise also due to the ringing due to the strong current pulse on the LED. A pulse (marked in white) of about 4-6mV is recognizable immediately following the current impulse on the LED.
This is an impulse produced by one or two photons emitted by the LED !
A characteristic of SiPM is that the pulses produced are “quantized” : the current pulse produced is equal to the product of the number of activated pixels times the current produced by a single pixel. Assuming that each activated pixel corresponds to the interaction of a photon, we can deduce from the amplitude of the current pulse the number of photons that hit the SiPM’s surface.
With our instrumentation it is difficult to obtain a clear representation of the quantization of the impulses, however from the image shown below we can see how the impulses are concentrated on amplitudes close to each other, corresponding to the same number of photons.
From the results obtained we can see how the tested circuit, despite its simplicity, is able to effectively produce ultra-short pulses in the nanosecond range. The fast pulser can be conveniently coupled with a SiPM detector to perform correlated photon-counting measurements.