Transimpedance circuit issue

Thread Starter

matkt051

Joined Sep 22, 2021
7
Hi Forum,

I am currently looking at improving the existing Photodiode circuitry to measure light at low levels (close to 0.01 Lux).I can understand the complexities of measuring light at these levels, however this is not what I am after.

Initially the design used an OSD15-5T Photodiode. Now, I was thinking of using one with a bigger surface area so that more light is seen by the photodiode. After wiring up the circuit (with the bigger photodiode - OSD50-5T), I can see that the output voltage of the Transimpedance amplifier goes negative when I blank/darken the sensor and it hits the positive rail when I open it. Please note that the diode is connected with 0 bias voltage (Cathode facing the -ve input of the Opamp).

Can the photodiode current change polarity when it goes dark (my initial thought is NO)?

Or

Can this be fixed by adding a bit of bias to the +ve terminal?



The TIA has bipolar rails (+/-5V). Because our system only uses the 0-5V range for the A/D converter, is it better to have the OPAMP connected to a single positive rail rather than Bipolar rails (+/-5) as it is currently?

Thanks
 

DickCappels

Joined Aug 21, 2008
10,152
Can the photodiode current change polarity when it goes dark (my initial thought is NO)?
No.

Can this be fixed by adding a bit of bias to the +ve terminal?
No. That would move your black reference level.

The behavior suggest peaking or other AC coupling in your amplifier. Get rid of the peaking or AC coupling and you should be fine.
 

Thread Starter

matkt051

Joined Sep 22, 2021
7
Can the photodiode current change polarity when it goes dark (my initial thought is NO)?
No.

Can this be fixed by adding a bit of bias to the +ve terminal?
No. That would move your black reference level.

The behavior suggest peaking or other AC coupling in your amplifier. Get rid of the peaking or AC coupling and you should be fine.
Hi Dick
Thanks for your reply. Can you please expand a little bit more on peaking and how it would affect the circuit ?
Many thanks
Mat
 

Papabravo

Joined Feb 24, 2006
21,159
You might find the following informative. I did an LTspice simulation of the circuit and I found it both interesting and informative.


1632349101984.png

1632349240461.png
 
Last edited:

Thread Starter

matkt051

Joined Sep 22, 2021
7
You might find the following informative. I did an LTspice simulation of the circuit and I found it both interesting and informative.


View attachment 248604

View attachment 248606
Hi Papabravo,
I am attaching a snippet of my circuit here FYI.
the circuit is built in the software TINA.
the diode is OSD50-5T. The TIA Opamp is OPA192.
The output of this Opamp is connected to a voltage follower.
The problem that I am seeing is that with the fluke DMM that I have it only reads to 3decimal places and I can see that when I cover the Photodiode, the output voltage goes to -1mV. this is what is being amplified by the other gain stages and hence giving -ve Voltages that are read as 0 by the ADC in the MCU as the ADC only operates from 0-5V.
this is why I was asking if adding a small bias of approx. 100-200mV would work.

I did an AC analysis of the circuit and found that the output is quite stable and no ringing can be seen. Is there a way to simulate this in TINA? I'm not sure tbh.

thanks
 

Attachments

Papabravo

Joined Feb 24, 2006
21,159
What is the reason for a bipolar supply?
I don't think the number of digits on your meter has anything to do with the problem.
 

DickCappels

Joined Aug 21, 2008
10,152
Pay special attention to the video lecture that was posted by @Papabravo in post #4 and pay special attention around 8:00 and past that a bit. It looks like your amplifier is having a hard time recovering from "overload", which is why the speaker has those diode-connected JFETs in the feedback loop.

Also, C4 is not desirable and may be a big part of your problem -having it there will increase the noise gain of the amplifier and make the amplifier "peaky" so it tends to overshoot the voltage towards it is heading. C3 should be adjusted for a fast enough response and at least large enough to remove the peaking effect of the photodiode's capacitance has on the amplifier's frequency response.

A bipolar power supply is a good idea since the input and sometimes the output will be near ground. but for bypassing, I suggest putting a small (0.01 to 0.1) multilayer ceramic capacitor between the power supply terminals of the amplifier chip and then larger (10 uf to 100 uf) capacitors from each power supply connection to ground.
 

Thread Starter

matkt051

Joined Sep 22, 2021
7
Pay special attention to the video lecture that was posted by @Papabravo in post #4 and pay special attention around 8:00 and past that a bit. It looks like your amplifier is having a hard time recovering from "overload", which is why the speaker has those diode-connected JFETs in the feedback loop.

Also, C4 is not desirable and may be a big part of your problem -having it there will increase the noise gain of the amplifier and make the amplifier "peaky" so it tends to overshoot the voltage towards it is heading. C3 should be adjusted for a fast enough response and at least large enough to remove the peaking effect of the photodiode's capacitance has on the amplifier's frequency response.

A bipolar power supply is a good idea since the input and sometimes the output will be near ground. but for bypassing, I suggest putting a small (0.01 to 0.1) multilayer ceramic capacitor between the power supply terminals of the amplifier chip and then larger (10 uf to 100 uf) capacitors from each power supply connection to ground.
Thanks Dick for replying.
The value of C4 is the maximum as specified by the datasheet. To match this, is it better to select a capacitor (C3) that is of the same or slightly higher value than C4? For a bandwidth of approx. 50Hz, I calculated the C3 to be 47nF (calculated value was 38nF)
The reason for 50Hz BW is because the light that we are measuring is DC and is not modulated. Also with 50Hz, this would filter any mains frequencies as well.

I tried shorting the +ve pin of the OPAMP to GND but it was of little or no use. then I connected the -Ve voltage rail of the OPAMP to GND. However, now the output voltage was +ve (as expected) but it was stuck at 0.005v when in darkness and goes up with more light. I thought that the output may be saturating looking at this. therefore, I connected the -Ve voltage rail to -2.5V. this time the output was still stuck at 0.005v, however some of the other TIA outputs were measuring 0.002V.
My next trial is to connect the +ve of the Opamp to 100mV bias voltage and see how it is so that the minimum voltage seen by the A-D converter is 100mV (at darkness).
any other suggestions are welcome.

N.B The previous diode that we were using was an OSD15-5T. The reason for using the larger diode was so that we could measure low light levels better.

thanks
 

Thread Starter

matkt051

Joined Sep 22, 2021
7
What is the reason for a bipolar supply?
I don't think the number of digits on your meter has anything to do with the problem.
I did ask the same question when I started looking at this issue and I was told that the reason was because in the old design, the TIA outputs would not go down sufficiently to measure the low light levels. Hence adding bipolar supplies made it work.
 

Papabravo

Joined Feb 24, 2006
21,159
I did ask the same question when I started looking at this issue and I was told that the reason was because in the old design, the TIA outputs would not go down sufficiently to measure the low light levels. Hence adding bipolar supplies made it work.
I think that for any given combination of components it would be helpful to quantify the voltage and current levels we are talking about
 

Thread Starter

matkt051

Joined Sep 22, 2021
7
I think that for any given combination of components it would be helpful to quantify the voltage and current levels we are talking about
Hi Papabravo.
you are right. I should have clarified the voltage and current requirements before.
The maximum voltage seen/read by the ADC is 0-5V DC and hence the TIA output should be within 0-5VDC.
doing a few measurements and calculations, I figured out that the maximum Photo diode current to be sensed is approx. 60uA (52uA to be precise) and the minimum is more than 15nA (this is the maximum dark current generated by the device with typical value of 5nA).
There is no need to measure the dark current. therefore, I would assume that the minimum current to be measured is 20nA.
Since we are measuring non modulated light levels, the maximum BW needed for this Amplifier would be less than 1k (As most of the signals are around the DC level).
There is no need for a dual supply as the final voltages read by the ADC is 0-5V (maximum Ref voltage of the ADC is +5V). However, if a negative rail is connected (so that the output hits the 0V rail) then this would have to always stay positive or else the ADC will start seeing negative voltages.

To summarise:
Vin Max = 5V
Vin Min = 0V
IPd Max = 60uA
IPd Min = 20nA
BW = 1k
Photodiode to be used: OSD50-5T
Opamp: OPA192

the whole idea to use larger photodiodes was that it would be able to measure a lot more low light level (0.01 Lux) reliably than the existing design.
Hope this helps.
Thanks
 

DickCappels

Joined Aug 21, 2008
10,152
Thanks Dick for replying.
The value of C4 is the maximum as specified by the datasheet. To match this, is it better to select a capacitor (C3) that is of the same or slightly higher value than C4? For a bandwidth of approx. 50Hz, I calculated the C3 to be 47nF (calculated value was 38nF)
The reason for 50Hz BW is because the light that we are measuring is DC and is not modulated. Also with 50Hz, this would filter any mains frequencies as well.

N.B The previous diode that we were using was an OSD15-5T. The reason for using the larger diode was so that we could measure low light levels better.

thanks
TIA means transient ischemic attack to some of us. I don't remember the acronym (TIA) being used in electronics. Since this is an international site, it is best to only use slang or abbreviations that are very commonly used. Back to your circuit:

C4 is not helping you. You will need clamp diodes if recovery from overdrive is the problem.

One last note: Can you use LTspice? We have many experts on that particular embodiment of SPICE on this site.
 

Thread Starter

matkt051

Joined Sep 22, 2021
7
TIA means transient ischemic attack to some of us. I don't remember the acronym (TIA) being used in electronics. Since this is an international site, it is best to only use slang or abbreviations that are very commonly used. Back to your circuit:

C4 is not helping you. You will need clamp diodes if recovery from overdrive is the problem.

One last note: Can you use LTspice? We have many experts on that particular embodiment of SPICE on this site.
Hi Dick
Thanks for raising that point. In most circuits I have seen they have used that acronym . That is why I used it here.
I can use LTSpice , however it doesn’t have the opamp used in this circuit. Not sure about using an alternative LT part as the simulations or performance may not match.
that is why I used TINA
I can check if there is an alternative OPAMP from Linear Tech that can be put in place of the OPA192
 

Papabravo

Joined Feb 24, 2006
21,159
Hi Dick
Thanks for raising that point. In most circuits I have seen they have used that acronym . That is why I used it here.
I can use LTSpice , however it doesn’t have the opamp used in this circuit. Not sure about using an alternative LT part as the simulations or performance may not match.
that is why I used TINA
I can check if there is an alternative OPAMP from Linear Tech that can be put in place of the OPA192
You can also look for and OPA192 model. I would not be surprised if you could find it with minimal effort. It was in @Bordodynov 's library. I've taken the liberty of extracting the relevant pieces.

For LTspice users. Libraries of models, examples, etc (ltwiki.org)

The 18 Mbyte file is the one you want.

Edit: found it. Put the ".asy" file in ...\lib\sym\Opamps and put Sborka.lib in ...\lib\sub\Sborka.lib

1632430127298.png
 

Attachments

Johnfoxwell

Joined May 23, 2021
14
Can the photodiode current change polarity when it goes dark (my initial thought is NO)?
No.

Can this be fixed by adding a bit of bias to the +ve terminal?
No. That would move your black reference level.

The behavior suggest peaking or other AC coupling in your amplifier. Get rid of the peaking or AC coupling and you should be fine.
I built one of these many years ago as part of an automatic photodiode test set. If you have this circuit on a bench then you have a really good proximity sensor. It will pick up, by capacitive coupling, any and all stray signals that exist close by. The only way I found to eliminate this AC coupling is to put the circuits in a metal box. Also you will have massive coupling from mains powered lights. You can try to filter these out but the signal levels are so high that a CR filter will have little or no effect.
As to offsets you should look at the op-amp specifications.
Finally, using 10M resistors can cause offsets via leakages on the circuit boards. I remember that humid weather caused problems with offsets. Drying out the circuit in an oven and then varnishing the circuit solve most offset problems.
 
Top