TLC27L2: Log photodetector problem at 2v

Hi Charles,

a schematic would be helpful :-)

Kai

Hello Charles,

As Kai has mentioned, could you please provide a schematic? Also, are there any other tests being done on the system, preemptively this sounds like an EOS edge condition.

Best,
Jerry

Kai,

Here you go.  (The two detector circuits are identical.)

Thanks, Charles

Hi Jerry,

The test of log conformity is really the only one done on this circuit.

The op amp doesn't appear to be damaged - we can repeat the test multiple times and the circuit works fine as long as the (simulated) photocurrent is high enough.  The problem is deviation from good log performance with low photocurrent.

Regards, Charles

Hi Jerry,

I put a schematic in the thread (I tried to add it to my first post but evidently it didn't work).

The test of log conformity is really the only one done on this circuit.

The op amp doesn't appear to be damaged - we can repeat the test multiple times and the circuit works fine as long as the (simulated) photocurrent is high enough.  The problem is deviation from good log performance with low photocurrent.

I put a schematic in the thread.

Regards, Charles

Charles,

I just reread your post and noticed that you are powering it below the minimum supply voltage. You can not expect reliable performance operating below the minimum supply voltage, as there may not be enough voltage present to bias the transistors. Even if you are getting the expected performance with 95% of the parts, we cannot advise/support operation outside of the datasheet conditions.

Does the circuit pass when supplied with 3V? By supplying it with 2V Vcc and having your "Signal Zero" which I'm assuming is your common mode, you are violating at least two specs, the input common mode range and the supply voltage range. Your input common mode should be Vcc-1.5V, which would be 0.5V, not 1.25V. 

Best,
Jerry

Jerry,

You asked what is connected to the photodiode input, and what is connected to the op amp output.

For testing, the input is a variable d.c. voltage (with respect to signal zero) in series with a 1 meg resistor to make a simple variable current source, fed into the op amp inverting input / transistor collector.

The op amp output has a 10K resistive load.

You also pointed out that the circuit violates the op amp's common mode spec.  Thank you for spotting that.  (I inherited the circuit and had never looked that closely at it - in part because we've made many thousands of this device so I assumed the design had to be good!)  It's amazing we haven't seen more problems.

Regards, Charles

Hi Charles,

If you decide to swap parts, here is a list of 20 parts with the following characteristics:

• 2 channel
• Rail-to-rail input
• IQ/channel less than or equal to 10uA
• Capable of 2V supply or less
• Catalog parts

You can also sort or filter them by bias current, since that may be a care-about in your application.  I haven't examined your case in close detail, but am guessing that may be a specification of interest based on your 1MOhm test impedance and the part you currently have in place.

Regards,
Daniel

Hi Charles,

what is the detector capacitance of your photodiode?

Kai

Thank you, Daniel.  I'll take a look.

Bias current is certainly a concern, as is the common mode voltage range.  And also the package and pin-out, as we need enough clearance around the critical pins for guard rings.  Lots to think about.

Regards, Charles

Kai,

The photodiode capacitance is 730 pF.

Note, though, that the failures occur when testing without the photodiode connected.  For test we just have the voltage source in series with a 1M resistor inputting a test current, and the capacitance is likely on the order of 5 pF.

Regards, Charles

Hi Charles,

I would probably do it this way:

charles_tlv2379.TSC

charles_tlv2379_1.TSC

Kai

Forgot this one :-)

Kai

Jerry,

I finally got around to re-testing the circuit with four ICs that had previously failed the log conformance test with the 2v supply.  This time the power supply was 5v, and the "Signal Zero" was 1.25 volts (as before).  In theory that should have taken care of both the low power supply voltage and high common-mode voltage problems.

Interestingly, the circuit performed identically at 2v and at 5v, so neither the low power supply voltage nor the high common-mode voltage appears to be the cause of the deviation from good log performance at low (simulated) photocurrents.

What else could be going on?  I have trouble thinking it's excessive bias current (it would have to be way out of spec. [unless it was damaged by ESD]).  Low open-loop gain?

Regards, Charles

Hey Charles, 

Did you try out Kai's suggestions above?

I am thinking that since you have already powered the device below the minimum supply voltage, those 4 ICs have been damaged.

I know you said that they didn't appear damaged "The op amp doesn't appear to be damaged - we can repeat the test multiple times and the circuit works fine as long as the (simulated) photocurrent is high enough.  The problem is deviation from good log performance with low photocurrent." But sometimes devices can appear non-damaged and have small sectors of the ICs damaged. 

I recommend repeating the 5V test with new ICs to verify that your setup is correct, additionally exploring some of Kai's suggestions above. 

All the best,
Carolina 

Caro,

No, I haven't tried Kai's suggestions.  I'm trying to minimize changes to the extant product, which has only 2v and 5v supplies.  Also, changing the reference voltage to 500 mV limits the range of the device in a following stage.  The TLV379 op amp could be a good replacement for the TLC27L2CP, though.

I don't follow the low voltage concern.  If powering a device below the minimum supply voltage will damage the part, then wouldn't it be damaged in the act of powering it on?

I think the circuit is safe from the conventional avenues of damage: the 1.5v reference is derived from the 2v supply so the power supply voltage is always greater than the input voltage.  And the op amp output is not loaded down at all - it goes through a 10K resistor to the summing point of the following amplifier.

Regards, Charles

Hey Charles, 

I was thinking that you had applied an input signal to the device once it had been powered at the lower voltage. Am I misunderstanding you?

All the best,
Carolina

Caro,

The input signal is a current (a voltage source in series with a 1 meg resistor).

At worst case the voltage source is outputting 3.75 volts (2.5v above the 1.25v signal zero), to provide a 2.5 microamp simulated photocurrent.  That's well below the 5 mA maximum input current.  (And nearly all the current should be flowing through the transistor, not the op amp.)

Regards, Charles

Hi Charles,

I think it has to do with the input offset voltage of OPAmp, here simluated by the help of "V3" :-)

charles_tlv2379_2.TSC

Kai

Kai,

That's a very interesting idea.  Thank you for replying and doing the simulation.

I'm working on a simple tester to measure some op amp parameters, including input offset voltage.  In a few days I should know if you've found the problem, and will let you know.

Regards, Charles

You are welcome :-)

Kai