Quick question on avalanche photodiode (APD)
We are using an avalanche photodiode in our project in order to detect the reflected light pulses from objects. However, we are only getting a response for objects at about 50 cm away. Our desired distance is to be able to receive a signal from about 5 m away. One of the main issues I believe is affecting the response of our APD is that we did not apply a reverse bias across it. The typical reverse bias voltage recommended from the data sheet is 150 V. Initially, we did not want to apply this bias for fear of burning the APD as it was very expensive and we can only afford 1. However, we believe that not applying this bias is hindering the distances it can detect.
Our questions are:
1) Is it NECESSARY to apply this reverse bias in order to receive adequate reception at further distances?
2) If that is the case, are there simple(and safe) options in order to do this? Would making a boost converter to convert our 5 V supply to 150 V be a possibility? Are there any relatively inexpensive IC's that can convert 5 V to 150 V?
If it helps the data sheet of our APD can be found here(we are using the S5343): http://sales.hamamatsu.com/assets/pd.../S5343_etc.pdf
Thank you for any help you guys can provide.
You must apply the reverse bias otherwise you would be better off with a normal photodiode. Normally you would run within a volt of so of breakdown. The optimum signal to noise ratio depends on various factors such as DC current from ambient light. I have designed various rangefinders over the years with several different manufacturer's APDs and haven't blown one up yet. I always bias them through a series resistor - I cannot remember what values but I seem to think 100k+. You need to add some decoupling which risks a surge current but the time is limited. I also made a system which ramped up the bias voltage to breakdown and then backed of by 1V to avoid temperature compensation. My rangefinders were eye safe and good for 1km+.
I have used a simple switching regulator IC to generate the high voltage with a discrete design of linear regulator to control the high voltage but I have also done some other designs but cannot look them up at the moment.
Simply check you data sheet and make sure you don't exceed the maximum ratings, taking care of maximum current and power dissipation. You should be fine.
Keith
---------- Post added at 21:47 ---------- Previous post was at 21:46 ----------
Just to add, if you only want 5m I don't really see why you need an APD, although I don't know your light source. I have managed 500m with a PIN photodiode.
Thank you for the advice Keith. We chose to use an avalanche photodiode due to our relatively weak light source. We are using a 5 mW 650 nm laser. However, if you believe we could get away with using a regular photodiode, that would save us a lot of trouble instead of trying to figure out a way to generate the 150 V necessary. It seems like generating this 150 V will be a small project in itself...
Chee
What measuring technique are you using? You should manage 5m with a PIN photodiode with 5mW depending on the optics diameter. However, it is best to do some calculations first. You need to know the reflectivity of your target. The diameter of the receiving optics is very important. Also, make sure the receive optics images the full size of the laser spot.
Keith
We are implementing a time-of-flight method using an ACAM GP21 TDC chip.
The receiving end consists of a 4 in diameter Fresnel lens that focuses the light onto the APD at the moment. The focal length is ~77 mm if I remember correctly.
I did some initial calculations earlier in the year. Basically, assuming about 80% absorption by the object, the reflected power is roughly 1 mW. However, if we further assume a divergence angle of 45 degrees, at 5 m, only about 0.62 microW are received by the APD. This calculation is what led us to choose using an APD over an ordinary photodiode.
An additional problem is although we can see the full laser image after focusing the lens, at further distances it is not visible anymore. This may again be due to the low power laser or poor reflectivity of the object. Ideally, we want our laser to work for a wide range of objects independent of how reflective it is, but if worse comes to worse I believe we can adjust our specifications to only be compatibly with highly reflective objects. We are in the process of having our department's machine shop build us a mounting system so even if we cannot see the laser we can at least fix the lens to its focal length away from our APD.
I can actually attach our report if you happen to be interested in more component details.
Thanks again,
Chee
---------- Post added at 21:30 ---------- Previous post was at 19:31 ----------
I just read another post which is very similar that you commented on: https://www.edaboard.com/thread173146.html
(We actually are using the same APD as him).
It seems that using an APD is much more complex than we had imagined considering:
1) The high biasing voltage required
2) The temperature compensation you mentioned
We may have to change our design to use a regular photodiode instead. If indeed a regular photodiode does not need such a high bias voltage and is much more stable with temperature variation it may prove to be the much simpler solution. The main concern is how sensitive an ordinary photodiode can be. If it can sense objects at a maximum of 5 m away, then it should be satisfactory.
-Chee
Chee,
What sort of accuracy do you want? As you are using a pulsed time of flight I am surprised you aren't using one of the Osram high power pulsed lasers. You can get them up to 90W I think. That will give you more light! You would need to choose the repetition rate and pulse width carefully to stay eye safe.
0.62uW sounds reasonable. The whole design is not trivial though. You will want a fast transimpedance amplifier with as much transimpedance as possible and some more gain stages.
One problem you will encounter is a disturbance due to firing the laser. Less of a problem with a 5mW laser than a 90W one but you will still probably get a false received pulse. This problem can be solved with a length of optical fibre between the laser and the laser optics. You might also want to arrange for some optical feed through to the receiver as a calibratio/start pulse.
I have never user Fresnel lenses for receive optics. I have always been concerned about the imaged spot size. APDs are usually large and you don't want a large photodiode anyway as the capacitance affects overall performance. The largest I have used is 1mm diameter and the smallest 0.2mm.
Good luck.
Keith
Accuracy within 10 cm should be pretty good actually. We have been prohibited from using high power lasers and in our design specifications have limited the power to 5 mW and visible light rather than infrared. May I ask if you have experience working with the TDC chip that I mentioned? We are actually having a lot of difficulties with it and any help we can find is welcome. Thanks again
Chee
I have designed with one off the shelf TDC but I cannot remember which. It will be quite a few hours before I can check. Even so, I mainly do analogue design so I didn't do the software to drive the TDC in this case.
Keith
---------- Post added at 05:32 ---------- Previous post was at 05:31 ----------
For 10cm accuracy you need a fast, well defined pulse edge. A high speed transimpedance amplifier will be important as well as high speed comparator.
Keith