Pump-Probe Configuration: A typical pump-probe experiment like the one shown here can have measurement bandwidths of a few terahertz.
High Speed Detector Families
High-Speed Photodetectors and Photoreceivers
How do I estimate frequency or temporal response of my system (detector, cables, connectors, scopes and spectrum analyzers)?
To maintain the fidelity of your measurements, every component in your system needs to have a bandwidth greater than the 3-dB bandwidth of your signal, or, equivalently, an impulse response faster than the fastest part of your signal. (For time-domain measurements, a good rule of thumb is to have a frequency 3-dB bandwidth greater than 0.44/t, where t is the full width at half maximum (FWHM) of the temporal pulse). For example, even a very fast (50-GHz) oscilloscope combined with our 6-ps detector will not produce a 6-ps trace. This is because the pulse width you see depends on the convolution of many bandwidths, including those of the signal, the photodiode, and the oscilloscope.Electo-Optic Sampling: The electro-optically sampled output of a Model 100X photodiode has a 5-ps FWHM (Blue). The same output measured with a 50-GHz scope has a 12-ps FWHM (Black).
To estimate the FWHM of a 5-ps pulse with a 6-ps detector and 50-GHz oscilloscope, you can sum the squares of the individual pulse responses. (This is very accurate for Gaussian pulses.) To do this, you will need to estimate the FWHM of the oscilloscope. Since the FWHM may be estimated from the 3-dB bandwidth by 0.44/f3-dB, the FWHM for a 50-GHz oscilloscope is approximately 9 ps.1 The measured signal will then have an FWHM of 12.2 ps.
Other important factors to keep in mind are the bandwidths of your cables and connectors, and the pulse-to-pulse jitter of your laser. Because the sampled oscilloscope trace is made up of data taken from many different pulses, timing jitter can broaden the measured signal. To measure signals greater than 50 GHz, you can use methods that are based on optical pump-probe configurations such as electro-optic sampling. 2 By using electro-optic sampling and exciting the detector with a 100-fs FWHM pulse, we measured a response of 5-ps FWHM from our 60-GHz photodetectors. By using a 50-GHz oscilloscope to measure the detector's output with the same excitation pulse, the pulse width was 12 ps, as expected.
1 Rush, K., Draving, S., and Kerley, J., "Characterizing High-Speed Oscilloscopes," IEEE Spectrum, 38Ð39 (1990).
2 For a discussion on electro-optic sampling, see Weingarten, K. J., Rodwell, M.J.W., and Bloom, D.M., "Picosecond Optical Sampling of GaAs Integrated Circuits," IEEE J. of Quantum Electronics, QE-24, 198Ð220 (1988).
How are these high-speed units manufactured?
All New Focus photodetectors and receivers faster than 3.5-GHz, whether optimized for the time or frequency domain, are individually built, tested, and shipped with their own pulse-response or frequency-response curves. We begin the manufacturing process by fabricating a Schottky photodiode that we designed. We then bond it into a microwave housing that was specially designed to ensure low parasitics, resulting in better measurement fidelity. Next, we actively align and bond an angle-polished fiber to the detector. Finally, the detector, with its microwave housing, is placed in a package with the battery and bias-monitor circuit board. The package is tested with either our state-of-the-art heterodyne-frequency test set or our pulse-response test set.
Why does New Focus use Schottky diodes instead of PINs or MSMs?
For our high-speed detector line, we use Schottky diodes rather than PINs or MSMs. Schottky diodes have faster responses than PINs because they have lower parasitics such as unwanted capacitances and resistances. Compared to an MSM, a Schottky photodiode typically has twice the quantum efficiency and, therefore, twice the responsivity.
We use two Schottky-photodiode configurations: back-illuminated and front-illuminated. In the back-illuminated photodetectors (Models 154X, 141X, 144X, 101X, and 102X) the input light comes in through the substrate (usually InP). The benefit of this structure is that almost all the light that passes through the substrate is absorbed in the n-type layer because the top gold contact acts as a mirror. Unfortunately, these photodiodes cannot be used at wavelengths shorter than 950 nm, because the InP substrate absorbs strongly at these wavelengths.
What's the difference between a time-domain-optimized detector and a frequency-domain-optimized detector?
Photodiodes in frequency-domain-optimized detectors are designed to produce a flat frequency response, where the responsivity varies only slightly across the operating bandwidth. Time-domain-optimized detectors, in contrast, produce clean, ring-free pulses. By using Fourier-transform methods, you can show that clean ring-free pulses result in a characteristic roll-off in the frequency domain. On the other hand, a flat frequency response results in some controlled ringing in the impulse response.Frequency-Domain vs. Time-Domain: Detectors designed for flat frequency response have enhanced responsivities at high frequencies (A). Detectors that are optimized for clean, ring-free pulses show a characteristic drop off in 3-dB frequency response (B).Time-Domain Optimized: This is the impulse response of a detector that is optimized for the time domain. You can see the characteristic frequency response in the figure above.Frequency-Domain Optimized: This is the impulse response of a detector that is optimized for a flat frequency response. You can see the corresponding frequency response in figure above.
When do I need a time-domain-optimized detector?
If you need accurate reproduction of your signal in the time domain, choose one of our Model 144X, 145X, or 102X time-domain-optimized detectors. These models provide clean, fast impulse responses with minimal ringing, and are ideal for pulse measurements with digital high-speed oscilloscopes. Moreover, they can be used in digital communications applications, where spurious ringing can degrade eye diagrams and the bit-error-rate (BER) measurement of your system. And, because these detectors are internally terminated at 50 W, you won't have to worry about any reflections between the detector and filter for standardized BER testing with SDH and SONET filters.
NOTE: Although the 50-W termination increases the NEP of the detector compared to detectors with larger termination resistances, their increase is of little consequence since the noise is dominated by the circuit following the detector.
When do I need a frequency-domain-optimized detector?
Applications that rely on transmitting signals at RF and microwave frequencies benefit from detectors with flat frequency responses and improved responsivities at higher frequencies. These applications include linear fiber-optic transmission to and from remote antennas for communication satellites, wireless cellular networks, and cable television. Since the time-domain response is not critical in these applications, the impulse response can have ringing. For these applications, we designed the Models 141X and 101X detectors, which provide especially flat frequency responses over wide bandwidths.
What is the advantage of a fiber-coupled version versus a free-space detector or receiver?
The high-speed detector/receivers have a relatively small detector area, and the fiber is bonded to the detector so that all the input light is incident on the detector's active area. The free-space detector requires the user to direct the light onto the detector's active area (which may have a diameter of no more than 25 µm). Thus, there is a chance that not all the light will be incident on the active area. Overfilling the detector will produce a response although it will be weaker and slower due to the photogenerated charge carriers migrating to the active area through diffusion rather than drift.
We do not offer free-space versions of our ultrahigh-speed detectors (>35 GHz), because these detectors are <12-µm in diameter, and illuminating (and finding) the active area can be a difficult and tedious task. The fiber-optic approach takes the guesswork out of finding the active area of the detectors. Moreover, we've designed the ultrahigh-speed photodetector modules to connect directly to test instruments with the optical signal brought in via an optical fiber. This eliminates distortion and loss from expensive connectors and mechanically rigid microwave cables that you would need to use with a directly illuminated detector.
What's the back reflection (optical return loss) on the fiber-optic detectors?
We don't specify the optical return loss of our detectors. However, we did design the packaging for low return loss. The pigtail is angle polished at 12 degrees where it is attached to the detector, reducing optical back reflection to less than -35 dB, when combined with the PC connector. We also offer a custom APC connector. The return loss for an APC connector is 60-70 dB. The charge and lead-time for a custom APC is an additional $250 and 2 weeks for the lead time.
In the multimode option of the Models 15XX and 14XX, how is the light coupled from the fiber to the detector?
In order to eliminate overfilling the 25-µm diameter detector, the multimode versions use a GRIN lens at the end. The GRIN lens has a broadband AR-coating to reduce back-reflection. For the detector models that cover the 400-1650 nm wavelength range, the multimode option decreases the usable wavelength range to 550-1330 nm. Due to chromatic aberrations in the lens, wavelengths shorter than 550 nm will focus in front of the photodiode overfilling the active area. Wavelengths longer than 1330 nm will focus behind the photodiode also overfilling the active area. Overfilling the active area creates longer tails for pulsed applications and reduces the overall frequency response.
Can I use a 1-mm core fiber with the multimode versions of the Models 15XX and 14XX?
The fiber in the multimode versions has a 50-µm core and so the coupling efficiency from your 1-mm core fiber to the internal 50-µm core fiber will be very poor. One option is to use the free-space versions of these receivers. In this case, you will have to couple the light out of the 1-mm core fiber and then focus the light onto the 25-µm diameter active area.
For unamplified detectors, what limits the minimum detectable signal?
For these detectors, the Johnson noise of the terminating resistance determines noise at the output of the detector. (These detectors are internally terminated at either 50 W or 100 W.) It is this noise, along with the responsivity, that determines the NEP we specify. In most applications, this noise is much less than noise added by subsequent amplifiers or measuring instruments, so you need to consider the entire system noise (along with the responsivity) to determine the minimum detectable signal. If you need high sensitivity, we recommend our 12-GHz photoreceivers with built-in amplifiers, or, for higher-speed AC-coupled applications, you can use our Model 142X 20-GHz broadband traveling-wave amplifier with our Model 14XX 25-GHz detector.
The photodetectors that use front-illuminated Schottky photodiodes (Models 153X, 155X, 143X, 145X, and 100X) can detect short-wavelength light even when the substrate is absorptive. Front-illuminated photodetectors are less efficient than back-illuminated versions because the light must pass through a semi-transparent gold contact and makes only one pass through the n-type material.
How do I prevent electro-static damage? How will I know if there is ESD?
All of the high-speed detector products are very sensitive to electro-static discharge (ESD). To prevent damage caused by ESD, always ground cables and connectors prior to connecting them to the photodetector. Wear a grounding strap when handling the detector or when hooking it up to an instrument. Also, remember to discharge your AC-coupled instruments before connecting your detector by momentarily connecting a load or short circuitto the input. Do not touch the inner pin of the RF connector. If the detector is damaged by ESD (or by too much optical power), there will typically be a large DC offset voltage (dark current) with no light on the detector. To check if this is the case, turn the detector on and use a voltmeter to measure the Bias Monitor output voltage with no light on the photodetector. The Bias Monitor is located on the front panel of the detector. If the output is >100 mV, then the detector is probably damaged and will need to be returned to New Focus. Repairing the detector requires completely rebuilding the photodetector and the fiber pigtail unit. (The price is one-half the price of a new detector.)
How do I avoid optical damage and how will I know if there is optical damage?
To prevent optical damage, do not exceed the maximum pulse power given in the catalog. If the detector is damaged by too much optical power, there will typically be a large DC offset voltage (dark current) with no light on the detector. To check if this is the case, turn the detector on and use a voltmeter to measure the Bias Monitor output voltage with no light on the photodetector. The Bias Monitor is located on the front panel of the detector. If the output is >100 mV, then the detector is probably damaged and will need to be returned to New Focus. Repairing the detector requires completely rebuilding the photodetector and the fiber pigtail unit. (The price is one-half the price of a new detector.)
How do I test if the battery is still good?
The Models 14XX and 10XX use one 9-V alkaline battery. To check the battery:
1. Turn on the module using the power switch.
2. Connect the Bias Monitor port on the front panel of the photodetector to a voltmeter.
3. Press and hold the Batt Chk button and observe the bias monitor output. The voltage is typically +5 V for the 14XX and 101X and -5 V for the 100X.
4. If the voltage is less than +3.5 V for 14XX and 101X (or greater than -3.5 V for the 100X), then the battery should be replaced.
How do I replace the battery?
To replace the battery:
1. Turn off the module and remove the two screws on the back panel with a Phillips screwdriver.
2. Remove the back panel.
3. Install a new battery and then carefully re-install the back panel.
4. Check the battery bias voltage as described above.
What's the time delay between the optical input and the electrical output?
We have not yet measured the time delay. However, you can model the detector as having an 8-cm long fiber to the photodiode active area. Then after the photodiode there's a 1-cm long RF transmission line to the output Wiltron K connector. Therefore, you can estimate the total delay time from the input optical connector to the output RF connector as 300 ps in duration.
How is the frequency-response measured?Heterodyne Experiment: Here two New Focus external-cavity diode lasers at 780 nm are used to generate a heterodyne beat note from DC to greater than 60 GHz.
We measure the detector's frequency response by monitoring the beat signal generated from a heterodyne configuration where two laser beams of equal power and slightly different frequencies illuminate the detector. Because the detector is sensitive only to the input intensity and not to the electric field (a square-law response), it mixes the two laser beams and generates a photocurrent that contains a signal component at the beatnote (difference frequency). Since we directly measure the frequency response, we do not use any transform techniques so you can be confident of the detector's performance.
To test our GaAs detectors, we use two of our 780-nm external-cavity tunable diode lasers; to test our InGaAs detectors, we use two tunable Nd:YAG lasers. Since these lasers are continuously tunable, any difference frequency from DC too greater than 90 GHz can be generated. We illuminate the detector with the two laser beams slightly detuned from each other and measure the detector's output with a scalar analyzer up to 40 GHz (measurement accuracy ±1.5 dB) and a spectrum analyzer from 40 to 90 GHz (measurement accuracy ±3.5 dB)
Can I get my time-domain-optimized detector, Models 1021, 1024, 1441 or 1444, characterized in the frequency-domain as well?
Yes, we are offering frequency-domain testing as an option for $250.
What equipment does New Focus use in testing the time domain?
Each of our time-domain-optimized detectors is individually tested and shipped with its own impulse-response curve recorded with an HP 50-GHz digital oscilloscope (HP Model 54750A mainframe with the 54752A dual-channel plug in). You could also use the Tektronix 11801B high-speed digital sampling scope with the SD-32 50 GHz sampling head. The excitation pulse is provided by a Time Bandwidth Products GLX-200 1.064-µm diode-pumped mode-locked Nd:glass laser. Its output has a <200-fs FWHM pulse duration. Since we directly measure the temporal response, we do not use any transform techniques so you can be confident of the detector's performance.
I need a fiber-optic collimator or patch cord, where can I get one?
One place that you can get fiber-optic collimators and patch cords is from:
E-mail: email@example.com or firstname.lastname@example.org
I would like to amplify the output of the high-speed detectors, how can I do this?
We offer RF amplifiers, Models 142X 20-GHz amplifiers with 8.5- and 18-dB gain. These traveling-wave amplifiers were specifically designed to connect directly to our photodetector modules, reducing standing wave effects and RF pickup at the input.
Can I get an RF amplifier component such as those in your 12-, 22-, and 38-GHz receivers or in your 20-GHz amplifier?
Unfortunately, we cannot offer to see these RF amplifier components, because we do not manufacture them and cannot provide service or technical support. You might want to contact one of the following companies which offer a complete line of RF amplifier products: