7

Objective :

Converting laser pulses from photo diode ranging from 10nA - 100mA to digital format, for measuring pulse width exactly.

Pulse width 10ns-150ns and repeating at a rate of >20us up to 1second.

Observations :

I have seen pretty encouraging application notes like this for a wide dynamic range TIA requirements, but my doubt is how fast they can be? The normal diodes can't be very fast as 2ns (considering my photo diode rise time,) how about a GHz BJT as feedback element or a schottky diode in feedback? ( I did not find people trying such.) I have found a few integrated designs like LOG112 but I could not find high frequency log amplifier modules.

More than bandwidth, it is the pulse response of such log amplifiers, for example MAX4206 matches all requirements but the response time is very poor, my pulse being 10ns the rise time of photo diode is 2ns.

EDIT:

As suggested I have also seen very high bandwidth log amps like SDLVA which have ns response time but oriented for different applications and are very much advanced than my requirements and too costly, surprisingly achieving only 5ns response time!

Problem Statement :

Is it possible to sense ns pulses using log amplifiers? If so how? I want to construct from off the shelf components only.

EDIT:

Here I have one flexibility. I have two channels - that is two diodes, so I can break this 10nA-100mA range in to two, one being 70nA-70uA from one channel and 70uA-70mA in another channel, so even if the log amp covers this range it would be satisfactory, I understand how difficult its to sense 10nA even if 100nA is sensed that is a good job.

JRE
  • 67,678
  • 8
  • 104
  • 179
kakeh
  • 592
  • 3
  • 11
  • 31
  • Yes, but could you please separate your observations and parameters from actual questions. You have them jumbled into a wall of text that is difficult to understand. –  May 16 '16 at 03:35
  • The LOG112 is only good to 10KHZ even if it has 5 decades of dynamic range. If you find such an IC it is likely to be expensive. You could try GHZ op-amps wired as a log amplifier. 10nA and 10nS are pushing the limits of off-the-shelf products. –  May 16 '16 at 03:45
  • 3
    If you want real blazing logging performance check out SDLVAs as the first result from a Famous Search Engine states "Analog Devices successive detection log video amplifiers (SDLVAs) include an operating range of 0.1 GHz to 20 GHz", log video amplifiers give you the amplitude of a signal (irrespective of frequency - well, to some degree anyway) and can have output rise times of 5nS or less, ranges of 50+ dB is pretty common too. – Sam May 16 '16 at 11:03
  • @Tom i have already seen them, i dont require that high end one, my requirement is 5-100Mhz, even if go, i would be wasting somany bucks, i require 19 such :p, each is > 1000$, i will look if any low range are there, thanks – kakeh May 16 '16 at 13:06
  • @kakeh You should add the constraints on cost and the desire for multiple (identical) amplifiers to your question text. – user2943160 Jun 06 '16 at 03:09
  • @kakeh further, please add specifications for pulse width (minimum/maximum expected widths and timing accuracy) and more details on signal levels (is 10nA to 100mA the required range of peaks? the signal's high and low level?). – user2943160 Jun 06 '16 at 03:31
  • @Tom unfortunately, SDLVAs appear to require a UHF to microwave carrier to operate correctly and they would not be suitable for this purely-pulsed application. – user2943160 Jun 06 '16 at 03:40
  • Heck, after looking into this a bit, the problem as stated may be straight-up physically impossible. – user2943160 Jun 06 '16 at 03:53
  • find the edit on current range flexibility – kakeh Jun 06 '16 at 04:34
  • Are the pulses random or is there some pattern to them? – user110971 Jun 06 '16 at 05:02
  • absolutely random, sometimes only 4-5 pulses only may come with a repetition ranging 1Hz-50Khz ,ultimate aim is to digitalize them, so a log amp will proceed by a comparator – kakeh Jun 06 '16 at 06:19
  • What is your input/output range? Have you considered a multi channel approach? – Voltage Spike Jun 06 '16 at 16:46
  • I don't understand. You say you want to measure the pulse *width*. Do you need to measure the amplitude also? Or just the presence/absence of a current that can be from 10nA to 100mA? If you need to measure the amplitude (with log response, then), what is the max error you can allow (absolute and gain errors)? Tell us a little bit more about what's after the amp you want to design. – dim Jun 06 '16 at 20:34
  • The analog Devices ADL5513 is a log amp with 80dB of dynamic range and works from 1MHz up to 4GHz and costs <$12 from Digikey and has the feature that "The ADL5513 has 20 ns response time that enables RF burst detection to a pulse rate of beyond 50 MHz", seeing as a single pulse is just a real short rf burst, something like that may do the job, for $12 it's something that should be cost effective to prototype, although you may need to add a transimpedance gain stage to boost the 10nA signal, but you'd probably only need 100x gain and some fast opamps can probably do that – Sam Jun 06 '16 at 23:34
  • @dim as i have said, i just want to do digitalization so this log amp will be followed by a comparator only – kakeh Jun 07 '16 at 02:41
  • @Tom i have already said that log amplifiers are very slow for my system, my pulse being 10ns and you have to take a distributed energy pulse, in that case it will be around 8ns only , the ADL5513 wont even perform properly, i thought of simulating, sadly they dont have any simulation model :( , more over my pulse is not like RF burst, the rate of repetetion is 1Hz-50KHz – kakeh Jun 07 '16 at 02:48
  • @kakeh I it is just followed by a comparator, why do you think you need a log amp? A regular amp would do. And actually, no amp at all bwould also do. Just put a high speed comparator directly. No? – dim Jun 07 '16 at 05:14
  • @dim you can have a look at the difficulty level here http://electronics.stackexchange.com/questions/229866/selecting-a-tia-and-using-it-for-60db-dynamic-range – kakeh Jun 07 '16 at 05:58
  • I may misunderstand the whole picture, but I'm sorry, I still think the approach is flawed since the beginning. At no point you need 60dB of **dynamic range** since ultimately, you simply want to identify a **level**. The fact your signal can be from 10nA to 100mA, does not imply you need 60dB of dynamic range. For example, in your case, if the signal is clipped because the amp saturates, you don't care. The resulting level will still be identified correctly. I still think you just need a comparator, and everybody was misleaded because of the way you wrote your both questions. – dim Jun 07 '16 at 07:32
  • @dim i basically want to use this kind of opamp for pulse width measurement and pulse power measurement, for pulse width i need only pulse shape to be maintained, this particular log amp i need it for power calculation also, more over with normal TIA you can hardly cover a range of 15-20dB at high frequency due to noise considerations, so a log amplifier is inevitable if want to cover good dynamic range, as this confusion arised i would put it detail that it should be able to help in pulse amplitude measurement even, dont mix two questions – kakeh Jun 07 '16 at 07:38
  • 1
    Ok, so you don't have just a comparator after the amp, you also have some other device for the power measurement part (ADC, or whatever) – dim Jun 07 '16 at 07:46
  • Just a thought: If you can allow the power measurement part to be quite inaccurate (if you just need it to have a rough idea of the order of magnitude of the signal strength), you can just use a diode as a very simple logarithmic current-to-voltage converter. It is very inaccurate, depends on the temperature and stuff, but depending on your application, it could be acceptable. – dim Jun 07 '16 at 09:22
  • @dim i have tried it, but it did not respond for lower currents like nA-uA – kakeh Jun 07 '16 at 10:02

1 Answers1

2

Another way to look at it is if you can get the high speed and low sensitivity and IF the device doesn't saturate or slow down if you load it heavily then you need not have to have the log aspect of it.

Look at the Analog devices ADN2880 which is a SiGe based 2.5 GHz Integrated TIA.

And the HMC6590 (which might be too costly for you) also is blazing fast (faster in fact) and can saturate.

placeholder
  • 29,982
  • 10
  • 63
  • 110
  • yeah i have also thought of this kind of solution, ADN2880 kind of solution as stage one and , i want the log amplifier to be followed by a TIA which will be very useful to cover the dynamic range if i fix my limits to 1uA-4mA lets say(as ADN says 300nA is noise itself ), i may not digitalize properly the lower currents, if suppose the TIA like ADN2880 is followed by a log amp that will even stop the second stage amp to go in to saturation even, when i try to amplify 4.4mV(4400V/A is ADN) by 10V/V and results 40mV which helps me to put a ref of 25mV atleast, 4.4V should not override the stage2 – kakeh Jun 11 '16 at 03:47
  • @kakeh You can't follow the log amp with a TIA, you are getting the speed out of the photodiode by the fact that you are operating it in that fashion. – placeholder Jun 11 '16 at 03:52
  • i did not understand, do you mean a slow log amp will under perform and distort if its followed by TIA ?? if that is so, its true and i want the log amp to be fast enough to be followed by a TIA, correct me if my perception is wrong – kakeh Jun 11 '16 at 03:59
  • i also did not get why the filter is used un-necessarily, it seems like a LPF, but i need a HPF of 4MHz, as mine being a pulse waveform the filter would bring distortion i feel – kakeh Jun 14 '16 at 11:26
  • @kakeh Your Photodiode will be biased at a constant voltage and will generate current. If you put the log amplifier first then it should be a current mode amplifier. The TIA converts from current to Voltage through the capacitor in the feedback loop. It is a tricky circuit to get to run fast. – placeholder Jun 14 '16 at 15:49