My objective :

converting laser pulses from photo diode ranging from 10nA - 100mA (60dB dynamic range of incident power) to digital format, for measuring pulse width exactly

pulse width 10ns and repeating at a rate of 20us

EDIT: Modifications: photo diode current ranging 100nA -10mA

My problem statement

what i tried is a conventional approach using a TIA and limiting excess current at TIA input and using 3 additional opamp stages as i have to see the 10nA as 50mV , so i can fix a threshold of 25mV , i can limit the excess current to avoid saturation of opamps,the problem is the more number of opamp stages, they came because of GBWP as i require atleast 120dB of gain at 100MHz(1/10ns), this is done(simulated model) and i was able to achieve calculation of pulse width, but i did not like this !

My Idea !

what i am thinking is to boost the 10nA current to atleast 1uA and any way current limiting is taken care now i can detect pulse at low gains itself ! and i can overcome the stability problems at high gain

am i thinking wrong ? is boosting such a low current is possible ? any way i should be ready keeping the fact that "this will boost my dark and noise current also " in mind


sorry that i have made some serious claims, may be due to underestimating the design problem

here below i have the schematic and results obtained,and the results were not at all up to the mark as expected in the comments

schematic: enter image description here

10nA input perfectly output is preserved enter image description here

1uA input enter image description here

1mA input enter image description here

my simulator shows me some noise analysis results which looked impressive, but i have to find out how they are calculated

noise analysis results

enter image description here

enter image description here

At the end the pulse width is not all maintained, my current implementaion failed due to saturation of opamp i believe, any how doors open for the comments and criticism

  • \$\begingroup\$ Very much related to this question also asked by the OP: electronics.stackexchange.com/questions/229866/… and, you haven't appeared to taken on-board the noise gain problem I gave in that answer. \$\endgroup\$ – Andy aka May 13 '16 at 9:47
  • \$\begingroup\$ i have to at any cost process the 10nA current how to achieve it even in the noise gain problem is the big question which is not answered there even :( i cannot modify the specs they are stringent, i can at maximum bring the lower limit to 50nA and bring the max limit to 10mA, a help from experts like you is much needed for a beginner like me \$\endgroup\$ – kakeh May 14 '16 at 3:18
  • \$\begingroup\$ I have answered the noise gain problem. I've told you that you can't get a TIA to do what you want i.e. convert a fast 10 nA pulse to a fast mV signal - there has to be a trade off and you appear to not be able to accept this. If you were able to talk about the bigger picture in your question then maybe someone could see a solution but ressurecting the same question is not going to help you. \$\endgroup\$ – Andy aka May 14 '16 at 8:47
  • \$\begingroup\$ i was looking at few log amplifiers which are achieving 100dB dynamic range like these, linear.com/solutions/1598,but again my pulse being very short width i have to find a very fast BJT in feedback stage or a schottky diode \$\endgroup\$ – kakeh May 16 '16 at 2:43
  • 2
    \$\begingroup\$ You are not helping yourself by limiting everyone elses perception of the bigger picture. If you feel I answered the question then please accept it. \$\endgroup\$ – Andy aka May 16 '16 at 9:26

1 - 10 nA to 100 mA is 70 dB, not 60.

2 - I am astounded that you're getting acceptable results over this dynamic range. Running a TIA to give 50 mV for 10 nA will produce 500 kV for 100 mA, and frankly I'm having problems believing that you have designed and built a TIA with that sort of clipping/overload performance. You also have an effective transimpedance of 5 Mohm. Granted that you're using a much lower transimpedance followed by voltage amplification, getting adequate speed in this setup is not something I'd want to try, and the dynamic range of the output chain remains at wildly improbable levels. Assuming, just as a SWAG, 1 nsec rise and fall times, a perfectly compensated TIA with a single pole response will need 350 MHz bandwidth, rather than 100 MHz, and the subsequent amplifiers will also need that bandwidth or better. I suggest that you take a long hard look at your test setup, from your test emitter all the way down, and make sure you're not fooling yourself about your results.

3 - I also don't think I believe your results on the basis of noise. Assume, just for instance, a 10k feedback resistor on your TIA. This will boost a 10 nA pulse to 100 uV. At the same time, Johnson noise from the feedback resistor combined with a 350 MHz bandwidth will give you 240 uV rms noise, which is not what I call a good SNR for measuring pulse width. Note that going to a 1k resistor gives 10 uV signal and 75 uV rms noise, so you want your transimpedance as high as possible. This, of course, will bang head-on with your amplifier speed and stability.

Like I say, I'm very skeptical about your claimed performance. Unless you post a detailed circuit showing how you got it, I think you're kidding yourself.

| improve this answer | |
  • \$\begingroup\$ 1. i thought people will raise this, the dynamic range is in respect to photo diode not TIA, 2. you are right, but i am using a current protection(under eval) i wont let a 100mA to pass to TIA even 3. i have a noise suppression using differential FET pair as like one suggested by bob electronicdesign.com/analog/…, through which i can reduce noise up to few uV, as soon as i complete circuit fully i would post it, my main concern is dumping my present approach and choosing a new one, by the way the OPA i am using is LHM6629 4stages, \$\endgroup\$ – kakeh May 13 '16 at 4:41
  • \$\begingroup\$ @WhatRoughBeast. I agree with your synopsis, but will not vote until the OP has reviewed and revised his circuit. Something is not right here, or the OP has found a 'perfect' TIA. \$\endgroup\$ – VTNCaGNtdDVNalUy May 13 '16 at 4:41
  • \$\begingroup\$ @kakeh - Dude, I admire your cojones, but following Bob Pease's advice scaled up to sub-GHz frequencies is not remotely simple. Commercial FETs that work at 1 nsec around an op amp loop? Are you kidding? Furthermore, your op amp current noise over this bandwidth is going to be nearly 50 nA rms which is not what you call compatible with your 10 nA pulse amplitude. Again, unless you provide a schematic, I will remain deeply, deeply skeptical. \$\endgroup\$ – WhatRoughBeast May 13 '16 at 4:53
  • \$\begingroup\$ please find the update dear \$\endgroup\$ – kakeh May 16 '16 at 10:58
  • \$\begingroup\$ 2N4416 is a also a JFET with response time of 2.5ns, which can serve my purpose \$\endgroup\$ – kakeh May 17 '16 at 3:44

To put a little perspective on Andy's concerns, 10nA is 10nC/second, or 6.2 * 10^10 electrons/second. Or 62 electrons/nanosecond.

Or 620 electrons per pulse.

I suspect you need to be looking at avalanche techniques - photomultipliers or avalanche diodes for optical detection, which will pose problems of their own in terms of overload characteristics, recovery time.

| improve this answer | |
  • \$\begingroup\$ please find the update dear \$\endgroup\$ – kakeh May 16 '16 at 10:58

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.