In search of a high speed op-amp for photo diode applications

thanks for your timely reply,

for your info , the rise and fall timei both of the photo diode are 5ns 

so with this i fell we need a op-amp which should be able to change the output state with in or less than 5ns

how can OPA657 will help me ? even if i see the slew rate it doesnt meed my slew rate requirement !

my application is having arround 19 photodidoes so in order to miniaturize it i should at-least need a quad-opamp :) any way first of all i need a best low noise high speed op-amp

when it comes to power supply i can have 5V or 12V both are possible and i want to go for a single supply op-amp, is there any advantage going with +5v and -5v

the level of output swing can be 5v or 12v, i feel 5v is enough, again is there any advantage if we go for higher voltages ? i feel it requires very high slew rate isnt it ?

thanks for the analysis,

but in the above calculation what is 0.35 that you used in calculation of BW, any way why did you use BW in calculating SR instead of using the actual frequency of signal ??

just after this i want to compare the output with an ref and also want to give it to an ADC, i want to reduce the op-amp count to 2 per channel one for TIA and other for conversion to finite digital pulses, for not to make the circuit more bulky,

i want the gain of the TIA to be more so that the nA current can atleast be converted to few mv, will it be sufficient ?
just after which i can use comparator for digital conversion

your suggestions in this aspect would help me a lot in designing a better circuit, is there any TI tool to simulate a photodiode and see the response from TIA and also other opamp stages, that would help me a lot

thanks

Hello,

The OPA659 is +/-3.5V to +/-6.5V...that translates to 7V (min) to 13V (max)

If you plan to take this to an ADC next, then you don't need any voltage output larger than 5V, since that is the max. ADC range. If your ADC is a 3.3V ADC then 5V is also too much. If your ADC requires a fully-differential input then you will need to use one of our fully-differential amplifiers (FDAs) like the THS4541 to convert the single ended signal from the TIA to differential.

If you require nA to be converted to mV then you will need a transimpedance gain of 1MOhm. This will require a JFET/CMOS input amplifier like the OPA659 I suggested, otherwise the current noise of a bipolar amplifier will dominate.

See this article on the tradeoffs when designing transimpedance amplifiers

http://www.ti.com/lit/an/sboa122/sboa122.pdf

You can use TINA TI to simulate the transimpedance amplifier + photodiode circuit. The model for the amplifier is available on TINA, however for the photodiode you will have to create your own model which should include an input current source and a capacitor equal to the specified photodiode capacitance.

http://www.ti.com/tool/TINA-TI

Daram,

  If you have a signal that has sharp transition edges but a relatively lower frequency and you want to preserve transition edge, then that's the fastest moving component in your system so the edge rate determines the required bandwidth. See the article below:

http://www.edn.com/electronics-blogs/all-aboard-/4424573/Rule-of-Thumb--1--The-bandwidth-of-a-signal-from-its-rise-time

You could use the OPA659, however that part does have higher noise (because it is designed for a very high slew-rate) and higher quiescent current. You may be able to find a lower noise amplifier to meet your needs. The OPA690 may work for  you. The VCA821 another option . It is a difference amplifier, which could make things really simple for you.

You will be very hard pressed to find multiple channel, JFET high-speed, high-voltage  amplifiers.

-Samir

please help me choosing another op-amp which is followed by this stage of TIA where its supposed to do difference between two signals and then digitalization based on comparision
TIA->DifferenceAmp->Comparator

when i searched i found most of them to be fully differential, any way my ADC doesnt need such input, can i ignore the other output which is out of polarity ?

after further search i found LM6172 being one of suitable for my case ,  i can build a difference amplifier and also a comparator stage and more over its available in dual package, am i right ? and it also matches the bandwidth requirement and also slew rate and further voltage levels also

but the confusion is its bipolar VFB, what does bipolar here mean exactly ?? will it not be as low noisy as FET ?

or can i use the same FET input op-amp for further stages also ? but it would make design bulky as they are not available in quad or doublet

Hello, The LM6172 is a good choice for making a difference amplifiers. Bipolar amplifiers have a BJT input stage and usually have lower voltage noise performance compared to JFETs.

thanks for your valuable suggestions, i have to now dump OPA659 from my design because of poor response at 100MHz , what i ignored is the gain bandwidth product which is a very important parameter, i need to amplify a current atleast 50nA for which i need 120Mohm of gain(or atleast through multiple stages), but with OPA659 the 100MHz signal is almost attenuated, because my photo diode recieves 100Mhz signals i have to go for atleast 100dB gain, for which only useful one i found is THS4509 but the problem here is it can maximum give me a flat gain of 60dB at range of 5MHz to 100MHz, how to overcome this problem of gain getting narrowed down, i am currently using two stages of THS4509 each of gain 1Kohm, i am really in need of solving this typical problem, is there no solution for this ?? i have seen people going for FET based design to achieve this, does TI has any high gainbandwidth product Opamp?? i have also tried the other FDAs of TI but no one helped

Let me put it very clearly ,

the signal i get from photo diode is arround 50nA so i should atleast have a gain of 120dB that is 12M of transimpedance gain to process such a small signal,

very important thing is that signal comes at a freq of 100MHz because its a pulse of 10ns

you may ask me what is your plan on noise, i will put a pair of FET at photo diode output to avoid excessive noise, leave it aside lets comeback

what happens if i put a 12M gain to my TIA, with OPA659 and even THS4509 what i observed is if i go for huge gains my signal at 100MHz will die

see the response with THS4509

Circuit:

Note the circuit i attached will have mechanism to handle excess input voltage noise(kindly find any mistakes if there )

Response at Rf being 800

result at Rf being 1K and 1K

so just imagine what happens if i put a feedback of 12M to primary TIA itself, the response will be very poor,

so typically i need  a >100dB gain at 100Mhz so i can say my GBWP to be 10^10 !

any way the opamps you suggested will not be able to satisfy these needs, 

in a nutshell i want the pulse of 10ns to retain its shaped even after going through TIA stage and amplifier, provided the pulse is minimum 50nA to be processed

coming to your questions:

1. What is rise time ?

<2ns

2. do i require dual channel opamp ?

if it satisfies above parameters i would love to have one

3. powersupplies ?

not an issue

4. required level of output swing?

as because the minimum that want to detect is 50nA even if a high current 100mA should also to be maintained its pulse shape, i feel any way at this current opamp will be saturated and it will not affect pulse shape i believe, so a 0-3V would do the job, or 0-1.5

to put it very precise: the objective is to convert a 10ns analog pulse to digital 10ns pulse . 

Note: please find the attached ckt alsoDEMO_THS4509_TIA.TSC

With 12pF of input capacitance from the photodiode + 2pF of input capacitance from the opamp (just assumed a low number here for a hypothetical opamp), for 100MHz of closed-loop TIA BW, even with 5kOhm of feedback resistance an opamp with a gain bandwidth of 4.4GHz is needed. Please see this blog post and the link at the end for an excel calculator which makes the calculations I just described.

What you need to know about transimpedance amplifiers – part 1

Other Parts Discussed in Post: TINA-TI Transimpedance amplifiers (TIAs) act as front-end amplifiers for optical sensors such as photodiodes, converting the sensor’s output current to a voltage. TIAs are…

By Samir Cherian over 9 years ago in Technical articles > Analog

Your restriction here is going to always be the GBP of the amplifier and not the slew-rate. The fastest opamp we currently have is the LMH6629 and the OPA857. I have some test data with the OPA857 with a 10pF input capacitor and that has a bandwidth of around 110MHz with 5kOHm feedback resistance. So I would recommend this as the 1st stage TIA. For the second stage I would recommend either the LMH6629 or OPA847 in a non-inverting configuration.

This is the best suggestion I can think of right now.

-Samir

Hi sameer ,

here i picked LMH6629 and used the same TIA configuration as suggested but i was unable to see the output, its just noise only 

please find the attached circuit and help me TIA.tsc

Hi Daram and Samir,

The problem with the LMH6629 TINA-TI simulation file was that the current source was set to off:

Here is the AC response with the current source state changed / corrected (set to "None" instead):

Here is the new simulation file for reference:

TIA modified 5_17_16.TSC

Regards,

Hooman