Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Using a photodiode with low shunt resistance?

Status
Not open for further replies.

Artlav

Full Member level 2
Joined
Nov 26, 2010
Messages
144
Helped
2
Reputation
4
Reaction score
2
Trophy points
1,298
Activity points
2,723
Greetings.

I'm trying to get a reading off an InAs photodiode.
To do it i'm using an transimpedance amplifier with the PD in photovoltaic mode.
pdcirc2n.png


Now, with a silicon PD everything works fine, with Rf up to GΩ range.

But the InAs PD produce a varying current of about 50µA, whether it's lit or not.
This current keeps changing, from the light level and/or temperature.

My best guess is that this is caused by low shunt resistance - R0 is ~10^11Ω range for silicon, but only about 400Ω for the InAs one:
zbr.png


I've tried to compensate for that current by routing it away:
pdcirc2.png


But that does not do a lot, since the current is extremely variable - shining the light on the PD apparently causes the photocurrent to go up, while the parasitic one goes down, so the reading rises for a moment and then rapidly falls out of range.

So, the questions are:
-Is such behaviour explained by the low shunt resistance and it's high temperature sensitivity, or there is something else at work?
-How do i compensate for that properly?

Summarily, how do you get high-sensitivity readings off an InAs photodiode?
It should be useful well into pA range, according to the descriptions.
 

Which device are you using? What temperature are you running it at?

Keith
 

I think silicon pdiodes are typically for visible, and InAs are typically for IR, so would also be interested to know what source you are using. Also curious if it is an AC or DC measurement you are trying to make. There's an app note from Hamamatsu that describes two effects, a charge build up at room temperature, and a background noise current due to background light. They recommend limiting the FOV of the photodiode or throwing two pieces of black tape on the front of the diode to reduce background noise current. **broken link removed**
 

Knowing the device is important to decide how it should be operated. Some InAs photodiodes are only rated for 1V reverse bias. Usually they are operated at zero bias and either peltier or liquid nitrogen cooled. I guess you must need the long wavelength sensing otherwise you would have chosen something easier to use such as InGaAs (although some of those can work at 2.6um).

Keith.
 

The problem turned out to be not as acute as i first thought - i forgot to remove the NULL resistors from the op-amp's pins thus skewing the results.
The PD works rather nicely well into MΩ's of Rf, but the basic issue remains - there is a parasitic current of about 85nA, and it drifts around.
And i need it to be usable within 10-100pA of photocurrent.

Which device are you using? What temperature are you running it at?
PD33fsi, an 3.4um peak InAs photodiode, local manufacture.
Runs at room temperature, designed for it.

what source you are using.
Scattered sunlight would likely be the primary source.
Tested with a matching LED, and it is also sensitive to human body heat a little.

Also curious if it is an AC or DC measurement you are trying to make.
In place it would have to do a reading about 1000 times a second, with absolute values.

But with 10MΩ i'm basically getting a thermometer (or whatever-causes-that-drift-ometer) - the drift overwhelms the light.

There's an app note from Hamamatsu that describes two effects, a charge build up at room temperature, and a background noise current due to background light. They recommend limiting the FOV of the photodiode or throwing two pieces of black tape on the front of the diode to reduce background noise current. **broken link removed**
Well, it would be looking into the background through a tiny hole, so it's either works or not fixable.

Also, if i read that right it applied to the InSb detectors, not InAs?

Thanks for the document, there is some interesting things in it.
 

I think, fundamentally, the problem with your low shunt resistance is current noise. If you look at the Johnson current noise equation, it increases as resistance goes down. This is going to get amplified by your TI amp and show up as noise/offset that changes with temperature. I read about using a transformer to couple the signal, making it look larger in resistance to the amp, but you lose DC. You might be stuck cooling the detector. You can get through hole modular cooler controllers from Wavelength Electronics, or roll your own. I think InSb and InAs have similar problems, which is why they group them together in the app note but I can't be sure.
 

I try to imagine, how it should be possible that a photo diode generates a DC current without receiving light. As far as I understand pysical laws, this isn't feasible, provided you didn't "invent" a perpetual motion machine of the first or second kind.

Either there's still background light, or a rather trivial explanation applies, it's OP offset voltage converted to current by the low shunt resistance. In the latter case, a thorough offset null adjustment would remove the apparent diode current. OP noise voltage, an AC quantity, will be still there and converted to noise input current of course.
 

Thermal agitation of charge carriers in a circuit causes a small current to flow. Is not perpetual, it's Johnson noise and has been known for about 100 years. You could null the offset, but it will just change with temperature.

---------- Post added at 11:56 ---------- Previous post was at 11:51 ----------

Also you might want to throw the same value of RF on your non inverting input to reduce bias current amplification, but it should only be in the pA in this application so probably isn't your main problem right now.
 

Please be precise.

Johnson noise is an AC voltage/current. By it's nature, you can't pull a power from it, because you'll supply the same amount of noise power in the opposite direction.

The original post is discussing a DC error current. If a DC current would be sourced from a photo diode without an external cause, it's perpetual motion. Thermal power can be converted to electrical energy only with temperature difference across the generator.
 

it's OP offset voltage converted to current by the low shunt resistance. In the latter case, a thorough offset null adjustment would remove the apparent diode current.
Which turned to be exactly the solution for that part of the problem. Luckily, the op-amp have the feature.

That leaves the temperature drift.
I'm seeing ±2nA of slow drifting, and a lot of what looks like white noise below that.
In a big closed box, with a stable light source inside.

The noise is certainly thermal - just breathing lightly into the box (~0.1 m³) drops the reading off the scale, and i can't imagine the PD heating up as much as a tenth of a degree from that.

I guess there is no helping that other than refrigeration?
Say, put a second PD near, with crystal covered, as a reference?
 

FvM:
Yeah, I always looked at Johnson noise as an additive noise source, but I always use AC signals so I'm not sure any more about this producing too much effect on the output. I always pictured little current sources in parallel with the resistance, but I could be wrong. Typically it's in the pA range anyways. I guess I don't really understand the application, as you say, it is a DC current, but there is really no AC band limit on the input light until the feedback resistor and photodiode capacitance roll the signal off. In that case, isn't his measurement just a point measurement of the DC value in an AC signal?

Artlav:
At DC, I do know that the input noise voltage of the op amp gets transferred to the output with a gain of 1+(RF/Rshunt), which at high photodiode shunt resistances is a small contribution. Since your ro of the diode is a very sensitive temperature sensor, and it is low, I think what you could be seeing is your input noise voltage.

---------- Post added at 13:25 ---------- Previous post was at 13:24 ----------

If cooling is out, and you have a microprocessor, I think the effect should be repeatable enough to compensate for with a temperature sensor/lookup table. Especially if it is related to the diode.

---------- Post added at 13:43 ---------- Previous post was at 13:25 ----------

Actually the 2uV/C input offset drift of the amp will come into play with the low shunt resistance. So it's likely a combo effect from the diode and the op amp. A little more difficult to compensate for in software.
 

So you suggest it's the voltage noise of the op-amp that is the limit, not current noise?
Let's try a different amp.
If i understand this right i should look for a bipolar one, like AD797?
Any recommendations?

About software corrections - not going to help much. At desired levels the output just shoots off the scale from even a small temperature change - nothing for the software to work on.
 

The noise is certainly thermal - just breathing lightly into the box (~0.1 m³) drops the reading off the scale, and i can't imagine the PD heating up as much as a tenth of a degree from that.
This is all about temperature drift, which in fact can't be exactly distinguished from low frequency noise. The thermal drift of the OP as such isn't severe, it mainly becomes a problem in combination with the low shunt resistance. Just as an example, the nominal shunt resistance at 20°C results in a DC gain of about 2500 in your circuit. The most severe temperature effect is caused by the shunt resistance's temperature dependency, even with the rather low AD820 offset voltage.
 

The AD797 does look like a better amplifier, but you got me with the "shoots off scale" comment. If the amplifier is pegging out at 5V (~5 uA of current) with a small temperature change, then I must be missing something. If we were relying on the noise gain transfer function of the amplifier and the input voltage noise, this would mean that the input voltage noise is around 2 mV, which is about 15k times the value I would expect you to see. Unless there's that much IR in the heat from your breath.
 

but you got me with the "shoots off scale" comment. If the amplifier is pegging out at 5V (~5 uA of current) with a small temperature change, then I must be missing something. If we were relying on the noise gain transfer function of the amplifier and the input voltage noise, this would mean that the input voltage noise is around 2 mV, which is about 15k times the value I would expect you to see. Unless there's that much IR in the heat from your breath.
That happens in case of target Rf=2.2GΩ, not the testing 1MΩ shown in the diagrams above for the earlier issue.
The drifting is in ±2nA range with the box left alone.

If i point a hair drier towards it, then it would go up off the scale for a moment from all the IR, then down off the scale when the heat reaches it.
With a breath it just goes down off the scale without noticeable rising, and does not stay long below.
 

That happens in case of target Rf=2.2GΩ, not the testing 1MΩ shown in the diagrams above for the earlier issue.
The drifting is in ±2nA range with the box left alone.
Good to know. 2.2G/400 equals 5.5*10e6 DC gain, the OP is operating more or less open loop. This effectively can't work.

I don't think, that the InAs diodes have been ever intended for operation with such a high DC gain.
 

I would recommend you step back to the 1 M resistor, or even 500k, and use two stages, one transimpedance, one voltage gain. Add a second resistor to the transimpedance stage on the non inverting terminal of the same value as your feedback resistor. You can do this before changing your transimpedance op amp to the lower noise one you selected. This will let you know what sort of noise level you can expect, and should dramatically reduce your sensitivity to temperature fluctuations (and instability). The noise will still be there, but maybe you will be able to compensate for it once you make those changes.
Failing that I'd cool the diode and move back to a single stage.
 

I would recommend you step back to the 1 M resistor, or even 500k, and use two stages, one transimpedance, one voltage gain.
I don't really understand how that could work.
Essentially, i'm zooming in onto a randomly moving thing.
With less zoom it stays within sight, so it's movement can be observed.
With more zoom it would just pop in and out of the screen.
So how would it matter if it's the screen that is getting magnified again or the first magnification gets bigger?
 

The randomly moving thing is directly proportional to the input referred voltage noise amplified by the gain formula above. That formula shows how the input voltage noise will be amplified by the ratio of the feedback resistance and the shunt resistance of the photodiode. The shunt resistance changes with temperature, so in order to desensitize your amplifier to fluctuations in ambient temperature I recommend using less "temperature dependent gain" by reducing the ratio between the photodiode shunt resistance and the feedback resistance. Instead, use voltage gain after you perform the current to voltage conversion. It should provide you with a solution that depends less on temperature, hopefully enough so that you can compensate for temperature fluctuations.
 

Ok, tried that.

Put 1MΩ as Rf, and added a voltage amplifier as a second stage, shown below, thus getting roughly the same gain as with Rf=50MΩ:
pdcirc_va.png


The result was less than encouraging - the magnitude of drifting remains essentially the same.
The only difference is that with a voltage amp stage there is slightly more noise.

Did i do it wrong, or did i do the wrong thing?
 

Status
Not open for further replies.

Similar threads

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top