ADL5502 peak detect for ultrashort pulse measurement

Hello everybody,

A femtosecond laser pulse generates an electrical pulse on a photodiode.

Pulse is gaussian like with a FWHM of 590ps thanks to the rise time of the PD (175ps) - this means a Bandwith of up to 2GHz

The ADL5502 is used for RMS and peak measurements of RF signals, this means "modulated high frequency AC"

My signal is just a single short pulse every 100kHz

I do not know how to deal with the datasheet sentence: "Envelope bandwiths of 10MHz". This would somehow render the device useless for signals over 10MHz

I need to know the true speed of the peak detector and if it can follow my pulse. And if a slower photodiode can improve the setup.

A first test showed it somehow working. I used a second, phase shifted 100kHz signal to reset the peak detect so it would be triggered again at each pulse. Sadly, the height of the peak detect level was only around the 15mV level. Furthermore it was somehow fluctuating - this is why I need to know about the exact data and reproducibility.

A question related to this has already been posted, but couldn't help me:

https://ez.analog.com/rf/f/q-a/72511/detect-a-pulse-peak-value-with-rise-time-of-1ns-at-100k-repeating-occurence

Best regards, Simon

  • 0
    •  Analog Employees 
    on Dec 11, 2018 7:35 PM

    Hello Simon,

    Peak detectors are typically Schottky diode-based, where each sine-wave peak of a RF input signal will repetitively contribute to the accumulated charge on an internal capacitor. The resulting DC voltage is amplified and presented at the detector output. More than one cycle of an input waveform is required, to get an acceptable output signal. 

    For viewing single-shot events, a high-speed ADC is recommended. Low-frequency capability should extend down to well below the pulse repetition rate (100kHz in this application). With ADC's, down to DC is possible. See, or example, the AD9208 or AD9213.

  • Hello Bruce,

    Thank you very, very much for your kind reply.

    I have suspected something like this already -.-

    My idea was (of course) to use the long period (100kHz) between those pulses and sample the peak detect plateau (using a LTC2366 with comfortable SPI and 3Msps )

    AD9213 is, without question, an awesome component. But requires an awful lot of work to get it working. I have no experience with FPGA based data aquisition and also with JESD204.

    It bothers me that I need such a high sampling rate for maybe 10 samples (in 590ps). Moreover I just need the peak value or even more fundamental, the area under the pulse (integral of pulse over time).

    So I either generate a pulse train and still use the peak detector, stretch the pulse in time (chirping) to make it broader und need less samples or integrate the pulse (which is current integrated in time equaling coulombs which equals joules by using the quantum efficiency of the PD) und measuring the resulting "DC".

    I will keep you updated, the ADL5502 is a great part nontheless !