7
\$\begingroup\$

I've been asked to produce a photodiode transimpedance amplifier with a 10MHz bandwidth. Current range of 1 nA to 100 uA (5 decades current) -> 100dB Dynamic Range.

Photodiode is a Silicon APD with junction capacitance 6pF.

Logarithmic compression is not suitable since 1.) this is imaging, and 2.) full bandwidth required for all signal strengths.

I believe output referred noise requirement, Vrms is therefore < 1nA * Rf. I understand bias current should be in the 1-10pA range. Other parameters influencing noise/BW are GBW, Input Noise Voltage Density, Input Capacitance.

The problem is, I have not found a single low-bias op-amp that seems capable. Is this performance even possible to achieve with commercial components?

Calculation tool I'm using for bandwidth/total output noise (let me know if this is bad)

Full-scale output of 1 or 2 V is reasonable. For max input 100uA this implies Rf ~= 20Kohm. Therefore, the noise at output should be <= 1nA * 20K = 20uVrms.

For example, I've looked at the following op-amps: ADA4817, LTC-6268, OPA657 all have ~1000 uVrms order noise, but meet bandwidth requirement.

And these lower GBW products: AD8651, ADA4807, AD8655, OPA2301, OPA2365, just barely make the bandwidth, but still show output noise on the order of 100-200 uVrms.

Also, could I possibly use an external input stage? Like some sort of FET buffer to drive the input of a higher bias current op-amp? Discrete transistor circuits are not my forte.

Thanks for any help you can offer.

\$\endgroup\$
  • \$\begingroup\$ I'm not going to do the sums, so this might be right off beam, but have a look at Texas OPA211. Rather high current noise, but might fit. \$\endgroup\$ – Neil_UK Oct 7 '15 at 14:10
3
\$\begingroup\$

Grin, Well your list includes all the good TIA opamps that I know of.
Are you allowed to change the gain resistor at the lower light levels?
If not then I would say maybe a bigger feedback R to get up to 10-20 V at the maximum light level. I would also direct you to Phil Hobbs. Here's his website.
http://electrooptical.net/ You can scroll down for his take on low noise PD front ends.
His book will also be useful for you.

\$\endgroup\$
  • \$\begingroup\$ Well the output is eventually scaled to drive a 16-bit ADC (theoretical SNR ~98dB) so I need the same gain across the full current range. However, I will run through the numbers to see if a really high Rf can achieve the full dynamic range, before being scaled down by a resistor divider I guess. \$\endgroup\$ – Keegan Jay Oct 7 '15 at 14:30
  • \$\begingroup\$ A bigger feedback R is typically a gain. The signal goes as R and the noise as sqrt(R). The other thing to try, (in Phil's book) is to bootstrap the capacitance of the diode. This can reduce the noise gain of the TIA. \$\endgroup\$ – George Herold Oct 7 '15 at 14:43
  • \$\begingroup\$ Ah yes, that makes sense R vs. sqrt(R)... Thanks very much for mentioning the Phil Hobbs resource looks extremely helpful. \$\endgroup\$ – Keegan Jay Oct 7 '15 at 14:47
  • \$\begingroup\$ Phil's a nice guy, if there is any money available for some consulting you could drop him an email and ask his opinion. If there is no budget for that then he hangs out at sci.optics and sci.electronics.design (usernet) and if you ask this question there you are liable to get his opinion for free. Oh I see you are a student, maybe send him an email. Feel free to use my name. \$\endgroup\$ – George Herold Oct 7 '15 at 14:59
  • \$\begingroup\$ Accepting because this is the only answer with resources suggesting alternative approaches to a simple op-amp TIA. \$\endgroup\$ – Keegan Jay Oct 22 '15 at 19:39
2
\$\begingroup\$

ADI have a nice wizard for this: -

enter image description here

It can be found here and importantly you can select which op-amp from the page above. The list is pretty much all the op-amps made by ADI and there are plenty that come with the tag "recommended". It will simulate them all AND give you the option of a two stage design.

I checked the choice of AD8651 and this isn't recommend so I chose the AD8056 to try it out with your requirements. It said that the SNR would be 70.2dB. Give it a go but I suspect you'll need a 2 op-amp solution.

\$\endgroup\$
1
\$\begingroup\$

The Johnson noise of your feedback resistor itself is ~57uV.

It will be rather difficult to meet your requirement.

\$\endgroup\$

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.