1
\$\begingroup\$

Below is the beginning schematic for a circuit I'm working on. We have a previous revision of that product (Rev A), and two sets of recent product (Rev B1 and B2). Rev B1 was made 2 weeks prior to Rev B2. All three products are tested on the same calibration system with the same settings.

When I measure pin 1 from the OpAmp, I get three different values for each product. I've unsoldered the 5.1Mohm resistor and measured it and it is correct.

According to the manufacturer, the components on Rev B1 and B2 are of the same batch but made a different times. Technically, all three revisions should have the same components.

Rev A (Older) - 1.21V/1.20V/1.21V

Rev B1 (Manufactured 2 weeks ago)- 1.15V/1.15V/1.1V

Rev B2 (Manufactured last week) - 1.05V/1.05V/1.05V

My question is: do photodiodes have tolerances? I am inputting the same amount of light into each product, but I get a different output from each version.

Technically, all revisions should be the same. At the very least I would expect B1 and B2 to have the same readings. It just seems very odd to me that every group of product has the same components, but they're all different readings. This is the very first measurement that takes place, and only gets amplified down the circuit.

Any reasons for why this is happening?

Photodiode Datasheet

Front end

Ignore R13.

\$\endgroup\$
  • \$\begingroup\$ Do you always measure with the same amount of IR Intensity? Are these the forward voltages? The diode is placed in reverse direction where you work with the current, if i get it right? So maybe your resistor is to high (depending on your light intensity). Is the diode soldered correct? I have saw similar uses for this diode, there you always had a potentiometer to adjust it, maybe consider that too. \$\endgroup\$ – Sider Aug 18 '15 at 20:03
  • \$\begingroup\$ Measured the same way for all 9 products back to back. The voltages are measured right out of pin 1 on an oscilloscope. Regarding the potentiometer, it's on the next OpAmp, TR1. \$\endgroup\$ – Luke Aug 18 '15 at 20:46
  • \$\begingroup\$ We currently use the POT to compensate for the differences. It just seems odd that I'm having differences between each product from the photodiode. I can understand if the resistor had a tolerance, but it doesn't. So this leaves each set of photodiodes outputting a different voltage for the same amount of input light. \$\endgroup\$ – Luke Aug 18 '15 at 20:48
  • \$\begingroup\$ Yes, like everything else, photodiodes have tolerances. So does everything else in the device. When you list three output voltages for each revision - is that taken from three different units of each revision, or one unit tested three times. If the latter, I suggest you grab different units and test for variability. \$\endgroup\$ – Scott Seidman Aug 18 '15 at 21:48
  • \$\begingroup\$ The most likely source of variation is the optical coupling. What are you doing to ensure the same fraction of the light reaches the photodiode in each test? \$\endgroup\$ – The Photon Aug 18 '15 at 23:27
2
\$\begingroup\$

Operating a photodiode in photovoltaic mode as you are doing is usually quite acceptable. I have done that in many systems. It tends to be slow but you are doing static measurements where speed is not an issue.

The data sheet is not as explicit as it could be but as indicated by SeanJ there is a significant unit to unit variation of sensitivity as is common.

The data sheet does not indicate a maximum but if we assume that the typical value of 45uA is halfway between the minimum (40uA) and the maximum sensitivity it would indicate that there could be a 20% variation (40uA - 50uA). You are measuring ~15%, that's within spec.

With high value feedback resistors you need to be careful about PCB cleanliness, leakage across the resistor would reduce the output signal. Try cleaning the PCB with alcohol and see if the output changes.

Another test you could do is inject a current into the amplifier and measure the output. You can inject a current by using one of the 5.1Meg resistors together with an AA cell (you need to inject a negative current). You should get the same output as the battery voltage. This will test for PCB leakage and correct value resistor. If you cover the photodiode to avoid any photocurrent you could probably do that without unsoldering the diode or resistor - if there is a problem due to leakage, unsoldering a component could cause a change disturbing the test.

Do you get a correct zero when there is no light on the diode? If you don't that could also indicate PCB leakage.

\$\endgroup\$
  • \$\begingroup\$ I will do both of the tests you gave me. Thanks for the extra ideas. \$\endgroup\$ – Luke Aug 19 '15 at 15:27
1
\$\begingroup\$

As you know, a photodiode is designed to operate in reverse bias.

On the datasheet,

Dark Current 2 - 30 nA

The dark current or leakage current can be 2 - 30nA

Reverse light current min 40  max 45 (Ee = 1 mW/cm2, λ = 950 nm, VR = 5 V)

The datasheet says the photodiode characteristics may vary.

Calibraion is done in fibre by splitting the signal, e.g. 90/10 splitter. The 10% signal is connected to the PD and the 90% signal measured using a power meter or OSA or vice versa. As the light is varied in intensity, the curve parameters are determined for the PD.

\$\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.