This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

LOG114EVM: Adding a photodiode intensity preamplifier to shift the range

Part Number: LOG114EVM
Other Parts Discussed in Thread: OPA2107, LOG114, LMP7721, LMC6001, OPA627, OPA365

Dear experts;

In the frame of developing a low-light sensing experimental board, I recently acquired a LOG114EVM development board. With included components and the reference current source locked at 500nA, the default range is stuck between 10nA and 10mA, for a floor output voltage Vout of 1.15V.

My application uses a low dark current photodiode VTB1013BH, sensitive in the green region (0.3A/W) in photovoltaic mode. Dark current is at most 20pA while full illumination outputs 1uA (5 decades). To obtain a full response from the LOG114EVM board, I need a current amplification of 1000. Precision low bias current OPA2107 and 1% resistors are available, what kind of circuitry do I need to  feed the LOG114EVM a 1000 scaled current from the photodiode in photovoltaic mode?

  • Hi Maxime,

    you could build a TIA which is fed by your photodiode. Choose a feedback resistance of 10M. Then connect the output of this TIA to input of LOG114 via a 10k resistor. This gives you an input current of 1mA when the photodiode outputs 1µA.

    But I think the input bias current of OPA2107 might be too high for this job. I would take an OPamp with a much lower input current. Another issue is the offset voltge of TIA and LOG114. To handle 20pA photodiode current the total offset voltage must be less than 200µV. So, you would need an additional scheme to trim the input offset voltage of LOG114. This would mean to modify the circuit of EVM a bit.

    Kai
  • Thank you for your reply.

    A TIA with high resistance providing U=Id x R is indeed a good solution. If a resistor is set after this adaptation stage, an increased current may be observed.

    Concerning the input voltage offset, refering to this application note of the EVM, it looks like an independant source may be used on the «high current correction input» pins to fake a photodiode:

    I wonder how to connect the TIA output though, and given the «high current correction» is substential above 1mA, using the other connecting configuration would lead to:

    If I replace Vref by the output voltage Vs of the TIA, in the following configuration:

    Could it work? Here  I = Vs/Ri, where Vs = Id x Rt. The OPA2107 datasheet displays a 10pA bias current, half the value of the dark current of the photodiode.

    I was also considering a single transistor as current amplifier.

  • Hi Maxime,

    I am certain that applying a low input bias current op amp in a TIA configuration and the series output resistor that Kia suggested will result in more predictable current amplifier performance than a bipolar transistor circuit. The series resistor approach discussed in the EVM section 3.4 can be used as a simple voltage to current converter at the TIA output. Note that there is additional circuitry on the LOG114 EVM to improve log conformity when the input is greater than 1 mA. If you keep the input to 1 mA and less it should not be needed. You are going to have to keep track of the current polarities of the input and output currents to make sure they keep the op amp output in the TIA and LOG114 operating regions.

    The TIA op amp will need to have ultra low input bias current to keep its error much lower than the minimum 20 pA dark current. The OPA2107 is a legacy JFET input op amp, and its input bias current can be much higher than 20 pA, especially at higher temperatures. I suggest using a femptoampere input op amp such as the LMP7721 for the TIA, and its inherently low bias current will be a fraction of the minimum photodiode current. You can find more information about the LMP7721 here:

    www.ti.com/.../lmp7721.pdf

    There is an EVM for the LMP7721 as well:

    www.ti.com/.../snou004.pdf

    You are going to be dealing with very low input current levels at both the TIA and LOG114 inputs. It is very easy to obtain poor results if extreme measures aren't taken to provide an ultra clean PC board environment. All solder flux residue and any other contaminants such as finger oils, etc. must be cleaned off the board. You will have to give consideration to the solder flux and how you can completely remove it. We use solder having water soluble flux, and then use two separate ultrasonic cleanings using clean DI water each time.

    Regards, Thomas
    Precision Amplifiers Applications Engineering
  • Dear Thomas, 

    Thank you for your advices, considering the LMP7721 EVM board as a TIA for the LOG114 seems appealing, but several points need to be cleared:
    1) If a gain resistance Rf is taken at 1M, LMP7721 output voltage Vs should range between 20uV to 1V, well enough to be discriminated by further circuitry.
    2) The log114EVM board embeddeds a level shifter polarizing Vbias (2V) on the non-inverting inputs: this implies the current will flow backwards if refering to eq 4.
    3) If I add a precision summing circuit (with OPA2107), adding Vbias to my LMP7721 output Vs, a resistor of 100 Ohms converts Vs+2V into I1, then the current should flow towards the LOG114 in the designed range (200nA to 10mA)

    I cannot discard the Vbias on the LOG114 board as the power supply is asymetric.

  • Hi Maxime,

    See my responses yo your questions:

    1) If a gain resistance Rf is taken at 1M, LMP7721 output voltage Vs should range between 20uV to 1V, well enough to be discriminated by further circuitry.

    I would try and maximize the TIA output voltage swing range for the 20 pA to 1 uA input current. The transimpedance gain could easily be increased to almost 5 M (V/A) and then the output swing range swing would be closer to 0 to 5 V. Yu would be able to utilize more of the op amp's dynamic range.

    No op amp has an output that swings exactly to the output rails so you have to factor in reasonable lower and output levels that are within the linear output swing range. The lighter the load you place on the op amp the closer it will swing to the rails.

    Be sure you understand which direction the TIA output will swing in relation to the photodiode current flow direction. If the input current flow is into the op amp circuit, and then through the gain set resistor to the output, the output will move negative. If you use a single positive supply the op amp output will not be able to swing below 0 V. If the diode is connected reverse from this the current will flow from the output, through the gain set resistor, and through the photodiode to ground. The output will move positive in that case.

    2) The log114EVM board embeddeds a level shifter polarizing Vbias (2V) on the non-inverting inputs: this implies the current will flow backwards if refering to eq 4.

    The EVM is set up for a single +5 V supply and the +2 V Vbias would be a problem; the LOG114 input current must flow into I1, 0r I2, and cannot be reversed without adding a current reversing circuit. That adds more complexity which I think you would want to avoid.

    If you plan to use the LOG114 EVM the TIA output would have to be established with a minimum voltage of 2 V. Then, the output could rise upward from there. Unless you can get enough of A TIA output range from 2V to almost 5 V, it would be necessary to use an op amp that can swing higher than 5 V. The LMC6001 Ultra, Ultra-Low Input Current Amplifier is usable with a supply up to +15.5 V, but you wouldn't need to use a supply to be set that high.
     
    3) If I add a precision summing circuit (with OPA2107), adding Vbias to my LMP7721 output Vs, a resistor of 100 Ohms converts Vs+2V into I1, then the current should flow towards the LOG114 in the designed range (200nA to 10mA)

    If you could show us how you intend to connect the photodiode to the TIA that would be helpful for discussions.

    My thought was to apply the +2 V to the TIA non-inverting input which would rise the TIA output by +2 V. Doing so will add some reverse bias to the photodiode reducing its junction capacitance and dark current.

    Regards, Thomas
    Precision Amplifiers Applications Engineering

  • Hi Maxime,

    as I said earlier I would modify the circuit a bit. I would use a dual supply voltage of +/-5V for the LOG114. This would ease many things. Then you could use the offset adjust scheme I suggested in this thread:

    e2e.ti.com/.../2713011

    Also, the TIA should see a much higher supply voltage. The higher the feedback resistor can be chosen the lower the offset voltage issues are. That's why I recommended a 10M feedback resistor (Rt in your scheme). It transforms your photodiode current range of 20pA...1µA into the voltage range of 200µV...10V. To be able to handle this full range the total offset voltage must be smaller than 200µV. But the LOG114 has an offset voltage of up to +/-4mV. That's why I suggested the offset adjust scheme. (Keep also an eye on the offset voltage of TIA.)

    The voltage to current converting resistor (Ri in your scheme) can be set to 10k then. This converts back the voltage range of 200µV...10V into an LOG114 input current range of 20nA...1mA.

    You can choose a supply voltage for the TIA which is higher than the supply voltage of LOG114. The 10k current converting resistor (Ri) should provide enough protection for the LOG114. An alternative is to protect the input of LOG114 by an additional protection scheme as discussed in section 7.3.7 of datasheet of OPA627. But I think an additional protection scheme shouldn't be necessary.

    Kai

  • Dear Thomas.

    Thank you very much for your involvement.
    The photodiode D1 in photovoltaic mode connected on a short BNC cable is grounded on the cathode.
    Biasing the LMP7721 non-inverting input in TIA configuration would indeed ensure an always positive input on the I1 LOG114EMV side.
    If the LMP7721 evaluation board has Vref as Vcc (+5v), the Vbias obtained across the voltage divider R1 and R2 should equal 2.5V.
    Given a 2M Rf gain resistor, the TIA should output 2.50004 ... 4.5V, well inside the rails.

    On the LOG114EVM side, the input voltage should drop by Vbias(2V) due to the level shifter, still usable.
    The selected voltage-to-current precision resistor Ri of 1K will feed the 4'th LOG114 pin a current between 500uA ... 2.5mA.

    Please find enclosed a schematic of the slightly modified evaluation boards as intended to be used.

  • Hello Maxime,

    There are a couple of issues with circuit that I have found upon my first review. Please consider these items:

    1. The photodiode connections need to be reversed. The Vbias voltage which will attempt to appear at the LMP7721 summing node will be instead forced to the forward bias voltage level of the photodiode. Even if the photodiode produces current it will be in the direction such that LMP7221 output voltage will try to swing below the output level established by Vref. If the diode connection is reversed the diode becomes reversed biased and the LMP7721 output will move positive in response to the generation of photodiode current.
    2. The LMP7721 Vbias is set too low. Vbias as is establishes the LMP7721 output level at +1.25 V, which is too low relative to the LOG114 VCM_in voltage of +2.56 V. The LOG114 requires the input current flows into the input pin. With VCM_in higer than LMP7721 output voltage the current will attempt to flow in the wrong direction. The goal is to set Vbias to a level such that no current flows through the LOG114 input V-I resistance when the photodiode current is zero. Then, as the photodiode does produce current the LMP7721 output moves positive and sources current to a LOG114 input via the series output resistor.
    3. Your schematic shows both a 100 Ohm Rout and 1 k Ohm Ri between the LMP7721 output, and the LOG114 input. Do you have a particular reason for using two resistors?

    Regards, Thomas

    Precision Amplifiers Applications Engineering

  • Hello Thomas;
    1) Yes, I should reverse the photodiode's anode and cathode connection on the inverting input to have a current flowing towards the LMP7721

    2) The Vref on the LMP side is the Vref fed on the board. You may notice the J3 and J6 connectors are shunted, thus, Vbias should equal 2.50V. The LMP7721 output voltage should at least overcome the Vbias of the LOG124EMV input (2V, eq). The VCM_in voltage of the EMV board is equal to Vbias delivered and regulated at 2V by the OPA365

    3) Rout seems to be embeded on the LMP7721 board (100 O, figure 4) Keeping only Rout would drain too much current on both sides, adding Ri reduces this current in the linearity range of the LOG114.

    Regards

    Maxime

  • Hi Maxime,

    I made some changes to the current amplifier and LOG114 to see if I could set it up such that everything functions electrically and within their linear ranges. The TINA TI Spice simulation circuit I show below looks like it might to be getting close to achieving what you need. 

    The TIA gain resistor was reduced to 1 Meg in order to keep the LOG114 logout within its linear operating range. Also, there really isn't any reason to run the LOG114 input current beyond 1 mA for a full scale TIA input current of 1 uA. Doing so does away with the additional linearization circuitry when the LOG114 input current exceeds 1 mA. It simplifies things and can be accommodated with the LOG114 EVM board.

    I didn't really know how you wanted to set up the two additional op amps A4 and A5 so I used the resistors you had in your diagram.

    If you would like to run some easy simulations on the TINA TI Spice circuit, I have attached the simulation file for your use. You can obtain TINA TI for free from TI.com if you don't already have. Just do a search on TINA TI. The program is intuitive and easy to use; especially, when you have a ready built circuit.

    Regards, Thomas

    Precision Amplifiers Applications Engineering

    LOG114_Cur_amp_02.TSC

     

  • Hello Thomas.

    Thank you very much for the file enclosed. This design will be tested as soon as we receive new components. This question may now be closed and flagged as resolved. The dedication of the TI community is very much appreciated.