Noise in photodiode and Transipedence amplifer
Does anybody has calculated the noise from photodiode and transipedence amplifier?
I am picking a photodiode and Transipedence amplifier to work together.
Because my input signal is extremely low, so I need some low-noise photodiode and low-noise Transipedence amplifer.
I picked the Hamamatsu S6967, and analog device AD8015.
http://www.sales.hamamatsu.com/asset...506-02_etc.pdf
http://www.analog.com/static/importe...ets/AD8015.pdf
The NEP of photodiode S6967 is 2*10e-14 (W/Hz 1/2), cut-off frequency is 50M, the sensitivity is 0.6W/A, that means its' current noise is 85pA, or 12 fA/Hz 1/2 .
But, the lowest noise Transipedence amplifier which I can find, AD8015, has input current noise 26.5 nA to 100MHz or 3pA/HZ 1/2 .
That means, the noise generated from amplifier is nearly 3000 times higher than noise from photodiode.
So, am I calculating wrong ? Or this is true that when you designing high sensitive low noise optical receiver, the only thing you should pay attention to is choosing a low-noise Transipedence amplifier, and the noise from photodiode can be ignored.
The noise generated in a photodiode is just shot noise, the amount of shot noise is proportional to the area of the junction. To maximise your signal to noise you need to capture as many photons as you can, there is also optical shot noise associated with the light, this is your absolute noise floor - you can never do better than this. What you need to do is work out what size your signal is (what is your signal, a sinusoid on a DC background or just a fluctuating DC level or something else?) in comparison to the background, then how long your measurement period is (the very best you can do is detect a signal larger than the square root of the number of photons collected over your measurement period - this is due to the properties of light).
Sorry if I appear to have gone off in a tangent from your question, but if you're trying to detect very small fluctuations in an optical signal this is what you need to be thinking about.
Did you consider an APD? Hammamatsu make them, as they do photomulitplier tubes (PMTs). Both devices have a gain mechanism in them. APDs need around 100 V, PMTs around 1000 V. There is noise associated with the amplification in both the APD and the PMT. I think for really low noise, you should look at a PMT.
Depending on what you are doing, it may be possible to use a lock-in amplifier. Then you can start recovering signals well below the noise floor. I once built a lock-in amplifier out of a couple of low-noise op amps and a few FET switches. I guess it cost around S30 in parts, but I was able to recover signals 60 dB below the noise floor. You can pick lock-in amps up off of ebay quite cheaply. The DSP based lock in amps are quite nice as its possible to have very long time constant filters, which would be virtually impossible to implement with analogue components.
Dave
Thanks ,Old Nick.
you mean when they calculate the NEP of a photodiode they only consider about the shout noise of photodiode, which is equal to 2*q*(Ip+Id))*BW, (q is the charge, Ip is the photo current, and Id is dark current) the thermal noise is totally ignored.
if it is correct, that also leads to my previously question： does the thermal noise generated from load which is 2*kB*T/R is 3000 times higher than this shot noise , and then makes the shot noise can be neglected？
my input is a 50 MHz PRBS (Pseudo Random Binary Sequence).
---------- Post added at 22:46 ---------- Previous post was at 22:45 ----------
thanks.Dave . before your reply, I did not even know there is something called PMT and lock-in amplifier.
But my project is for medical use. The patients can not wear the big stuff or high voltage photodiode.
Actually I ask one guy in optical department, if there is some low voltage APD which operated in below 10v. They said, to their knowledge ,some APDs can work at 20v, but 1Ov, no idea.
Beside if you decrease the bias voltage , the gain suffers.
So, Do you know if there is a APD can work around 10v, and still more sensitive than a pin photodiode? Or maybe I should ask Hammamatsu directly
Hi, I don't have the data sheets for the devices you talk of, but I'm pretty sure you can ignore any noise generated in the diode. We never contemplate pd noise when designing our cameras (I design modulated light CMOS cameras). Lock-in amps (LIAs) are only useful if your signal of interest is periodic, they can dig out surprisingly small signals from noisy backgrounds. The optical shot noise floor is still the limit though, if it digs out a signal 10^-6 below the background it has taken at least 10^12 photo electrons to make that measurement.
If you tell me more about your application and what the signal is you plan to measure etc. that would help a lot. If you're doing this at a Uni, we maybe even able to lend you some detectors if they are suitable. We have a linear array with programmable well depths which was designed for measuring extremely small fluctuations (less than part in a million) on a large DC background. Under the correct experimental conditions this camera is limited by optical shot noise, and thus can match the sensitivity of 256 PDs and LIAs running in parallel.
Hi,
I've used APDs, PMTs and lock-in amp for various medical uses. I did my Ph.D. in medical physics
Ph.D. Thesis of Dr. David Kirkby
When I was using APDs, there were none running at 10 V. I've no idea if that has changed. 100 V was typical.
The gain is a very sensitive function of the voltage. I forget the formula - you will find it in my PhD. above, but I'm sure the Hamamatsu data sheet will have it too. Essentially, I think if the APD is designed to work at 20 V, there is little chance of it working very well at half that voltage. There is a temperature dependance of gain, which can be removed by either holding the APD at a constant temperature, or more easily by having a power supply which changes the output voltage to compensate for the temperature change. Hamamatsu used to sell such PSUs, and probably still do.
BTW, that 19" rack system you see in the above photo was NOT the subject of my Ph.D, but something I worked on after it. This measured blood oxygenation of babies during labor. This has a PMT running at about 1 kV, and a pair of lock-in amplifiers (black boxes near the bottom). The coupling between the lasers and the babies head, as well as the head to the PMT was all via optical fibre (i.e. electrical insulator). So there is no electrical connection. The whole unit had to be work from medical grade power supplies, which is most easily achieved by using a medical grade isolation transformer, then running everything off of that.
The particular instrument you see, despite running at 1 kV, had been tested independently for safety and met the necessary regulations in the UK for medical use. So depending on the practicalities of what you are doing, you should not dismiss using 1 kV for medical use. X-ray tubes work at a lot more than 1 kV!
I would certainly ask Hammamtsu about the APDs. My knowledge is 10 years old, so I assume things have advanced since then. I know they were (probably still are) the leading experts on APDs.
In the interest of full disclosure, I should state I have worked on research projects sponsored by Hamamatsu. I've never worked for them directly, and have no connection with them for the last 8-10 years.
Dave
If the signal is actually extremely low, you most likely should to use an amplifier with lower input noise current. On the other hand, the calculated photo diode noise numbers are only representing the dark current generated shot noise. If some DC current is present in your application, noise is considerably higher.
P.S.: You didn't yet mention the expected photocurrent respectively signal power.
Hi Dave
Thanks very much!
There is some difference between my application and the X-ray which can work at 1 kV. The device which I finally want to have is something portable, like a hearing aid. So...
For APD, I checked some datasheets from Hamamatsu, like you said, most of APD are designed work with 100-200v. and I really agree with you, there is little chance an APD can works below 10v, and still get better performance than an PIN photodiode.
I will start another Thread to described my application in detail.