Instead of using DC bias, the DPD operates at pulsed voltages: the applied voltage is switched from reverse to forward bias. This forward bias induces a large forward current after a time delay, so called triggering time, which depends on the light intensity. This way, we measure the delay time of the large forward current instead of the small photocurrent magnitude.
Because of the large output signal, ActLight claims that DPD needs no amplifier. It can be directly connected to digital circuits.
ActLight expects the first ASIC integrated with the company’s photodiode to emerge in 2019, indicating that there are two or three OEMs already taking a closer look."
nice idea - like a capacitor that is getting reverted whilst connected with a series resistor. in that case voltage will go towards the final value asymptotically following an exponential function. having some trigger level would derive a pulse signal that is logarithmic in its conversion curve. - the circuit presented here is different because it even seems to saves the trigger circuit.
ReplyDeletea time diagram for the operation and a rough data sheet of the current technology metrics would have been interesting. i assume (without better knowledge) we might see such publishing in the next few months.
ahhh. found it in the DPD article (as linked above) - nice video demonstration with realy well selected principle animations and live demonstrations.
DeleteI cannot see how this device can do better than an APD. The delay could be induced by recharging the I region by the holes emitted from anode. In such case, does a single hole can make a significant delay change ?
ReplyDeleteWhat do you mean by better? They solve different problems, in different light ranges.
DeleteI don't see much savings though. The silicon cost has to be very small compared to packaging and testing. There are already devices on the market that convert light intensity into time. One we made (custom) at Photobit has been on the market for a long time and may still be in production. You get a lot of die per wafer, and yield is very high.
They are supposed to work in the same light range. It's shown at the end of the Youtube-video in the other article.
DeleteThat said, I join the chorus in not really seeing any improvement or savings.
You are right, they are comparing in the low light region in the video. I was thinking single photon sort of light levels. APD does not require integration, so hard to see that this device will compete with SPAD or APD. But, if you are going to integrate, it seems like an image sensor pixel would be hard to compete against. So that is why I was thinking it was competitive only in certain ranges of light.
DeleteInteresting. I'm wondering about linearity, dynamic range and process sensitivity/variability.
ReplyDeletehttp://www.freepatentsonline.com/20150221806.pdf
ReplyDeleteI found one of their patents here. If I understand well, the electrons-holes generated in the I region will modulate the potential under the respective gates and stimulate more carriers to be emitted inside the I region. Am I right ? By using SOI and thin silicon layer, maybe the capacitance can be reduced to make a consequent gain ...