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.

help : photodiode with TIA

Status
Not open for further replies.

natnoraa

Full Member level 1
Full Member level 1
Joined
Jun 19, 2011
Messages
98
Helped
12
Reputation
24
Reaction score
11
Trophy points
1,288
Activity points
2,108
Hi,

I have designed a circuit as shown in the last link.

However, i'm getting a very small bandwidth at Vout of about 6MHz (ac analysis). or maybe it isn't small?

the waveform of Vout (transient analysis) looks fishy as its offset is about 300mV. Is there anything wrong with the circuit itself?

The compensating capacitor was being calculated and doing some trial and error.

I would like to have a stable circuit with the highest bandwidth it can afford for some analyses.

Please advise. thanks! :)

http://focus.ti.com/lit/ds/symlink/ths4031.pdf

THS4031 datasheet.

**broken link removed**
image of design
 
Last edited:

I don't see an image of your design. You can post it on here rather than trying to host externally.

Keith
 

design.jpg
result.jpg

hi,

sorry about that. i have attached the schematic and the transient analysis results as well. I think my knowledge of analog should be quite low.

actually i'm using a pulse current source parallel with a capacitor to simulate a photodiode.

the photodiode has a 15pf capacitance and for the photocurrent from the photodiode, it's up to me to define right?

in case you're wondering, this is the photodiode i would like to implement
**broken link removed**

thanks in advance
 
Last edited:

The output offset is due to the opamp bias current through the feedback resistor. The photodiode capacitance is specified at 150V reverse bias so you are going to have to be careful to protect the opamp. If you don't reverse bias it then the capacitance will be larger. That is a specialised four quadrant photodiode but I guess you have picked it for a reason.

I would suggest using AC analysis to tune the compensation capacitance (maybe you have already done that). Personally I prefer to have a minimum of 1pF to minimise the effect of strays, but that may require a lower feedback resistor.

Keith
 

However, i'm getting a very small bandwidth at Vout of about 6MHz (ac analysis). or maybe it isn't small?
the waveform of Vout (transient analysis) looks fishy as its offset is about 300mV.

The waveform looks basically as expectable. What do you expect?

The 300 mV offset is caused by the nominal THS4031 bias current of 3 uA.
A closed loop bandwidth of about 6 MHz results from the OP GBW and the C1/C2 ratio.
 

Hi,

Is it necessary to offset the dc bias? ultimately this output is going to be amplified by another amplifier to reach a 3V input for a TDC.

Noted, and actually no specific reason. I chose that because of its detection of 1064nm wavelength light.

My design spec is a laser pulse of width from 10nm to 40nm. 1064nm wavelength laser and a detection of 5m to 4000m laser rangefinder. Any suggestions?

Yes i have. Base on the calculation of Cf = sqrt [ (Cin)/(2sqrt2(Rf)(fgbw) ], i have derived a capacitance of about 0.41pF for a stable and highest bandwidth output. However i have tuned it to 0.5pF because the bandwidth is at the highest before dipping with increasing compensation capacitance.

That being said, i was wondering whether the ac analysis result makes any sense? the starting point is a quite a -ve dB, doesn't it say that noise is very dominating here for the SNR? or am i wrong? the BW is about 6MHz and what's the implication for bandwidth actually?
result2.jpg
 

natnoraa, for frequencies in the MHz range the photodiode capacitor determines mainly the gain which is app. 30 dB at 5 Mhz.
Looking at the open-loop gain characteristics in the data sheet this corresponds rather good with the bandwidth observed by you.
 

thanks for the prompt analysis. the photodiode's cap is 15pF and the common mode cap from the op amp is 1.5pF, thus i've combined them to 16.5pF as the total Cin. does it even make sense? :/

---------- Post added at 16:59 ---------- Previous post was at 16:54 ----------

Hi,

I haven't had much expectations but reading from the different journals, maybe i'm quite skeptical about the oscillations.

From my results, i see oscillations of about 0.7mV for Vout and thus i was actually suspecting something might be amiss. Maybe it's not afterall :)

6MHz and 10MHz closed-loop BW, what are the implications really?
 

What about reducing the gain by reducing the 100 k feedback resistor?
 

What about reducing the gain by reducing the 100 k feedback resistor?

Hi,

I have actually thought of the possibility. But the output will be even smaller and i really need some nifty amplifier to amplify the output of this TIA to about 3V? It is easy to recalculate and recompensate the capacitance again, but i'm concern about the implications of bandwidth like i have questioned earlier.

so from the output, i can just -300mv due to the offset of the OPAMP to get the actual range of output?
 

i see oscillations of about 0.7mV for Vout
There are no oscillations. The small pulses are simply a feedforward signal caused by the OP output impedance and finite gain. The purpose of feeding a pulse train to the TIA isn't quite obvious, I'm still curious about the expected output waveform.

If the application is laser TOF, I would expect a considerably faster amplifier.

It seems reasonable to compensate the bias current. For the THS4031, that hasn't an internal bias current compensation, the most simple way is to connect a 300k || bypass capacitor circuit to the positive input, reducing the error to offset current induced magnitudes.
 

There are no oscillations. The small pulses are simply a feedforward signal caused by the OP output impedance and finite gain. The purpose of feeding a pulse train to the TIA isn't quite obvious, I'm still curious about the expected output waveform.

If the application is laser TOF, I would expect a considerably faster amplifier.

It seems reasonable to compensate the bias current. For the THS4031, that hasn't an internal bias current compensation, the most simple way is to connect a 300k || bypass capacitor circuit to the positive input, reducing the error to offset current induced magnitudes.

I see, to answer your question, i fed it a pulse train because i'm trying to simulate the laser pulses the photodiode will receive. Hope i'm doing it correctly.

For the time being i shall use this amp for simulation first, will take note of your advice.

Okay, noted again as well for offsetting the dc bias.

One question, can the design of this front end module be closed now? Can i safely say that the output is at least stable enough for me to feed to the back end module with the post AMP and a TDC+microcontroller?
 

Can i safely say that the output is at least stable enough for me to feed to the back end module with the post AMP and a TDC+microcontroller?

I now understand, that the term "oscillation" meant the 10 percent overshoot. It's in fact indicating a limited phase margin. Increasing the compensation capacitor is the only way to improve it. The real circuit behaviour may be slightly different, so you should check the pulse response in operation.
 

I now understand, that the term "oscillation" meant the 10 percent overshoot. It's in fact indicating a limited phase margin. Increasing the compensation capacitor is the only way to improve it. The real circuit behaviour may be slightly different, so you should check the pulse response in operation.

Hi,

thanks again for the reply. So oscillation is present since there's a 10 percent overshoot in my results? If i were to increase the compensating capacitor, the BW decreases and it takes longer for the circuit to be stable. Can't seem to find any solution to this particular circuit. advice appreciated.
 

If you meant 10ns to 40ns pulses (not nm) then I think you need a lot more speed out of the receiving amplifier. Why use the four quadrant sensor? What about the effect of not using 150V reverse bias? What laser power are you using and have you estimated the returned signal power? I suspect you will have considerable signal to noise ratio issues. All the laser rangefinders I have worked on have had photodiodes no larger than 1mm and often using avalanche photodiodes - the signals at 4km are usually quite small off non-cooperative targets.

Keith

---------- Post added at 12:04 ---------- Previous post was at 12:03 ----------

I forgot to mention, your AC analysis is flawed. It should show a very high 'gain' (=transimpedance) but then I don't see an AC source in your schematic.

Keith
 

If you meant 10ns to 40ns pulses (not nm) then I think you need a lot more speed out of the receiving amplifier. Why use the four quadrant sensor? What about the effect of not using 150V reverse bias? What laser power are you using and have you estimated the returned signal power? I suspect you will have considerable signal to noise ratio issues. All the laser rangefinders I have worked on have had photodiodes no larger than 1mm and often using avalanche photodiodes - the signals at 4km are usually quite small off non-cooperative targets.

Keith

---------- Post added at 12:04 ---------- Previous post was at 12:03 ----------

I forgot to mention, your AC analysis is flawed. It should show a very high 'gain' (=transimpedance) but then I don't see an AC source in your schematic.

Keith

hi keith,

what do you mean by 10ns to 40ns? are you referring to the rise time of the laser pulses?

I have no specific reason for using the quadrant sensor, just that it's suited for 1064nm laser wavelength, which is one of my design goals. Maybe you have one to recommend?

As i am simulating a photodiode with a current source and capacitor only, i am not able to add in the effect of the 150V reversed bias. even if i do, the circuit became like this

circuit.jpg

and the results are still the same from what i see.

The returned signal power is a suitable random pick first. i have a formula to work it backwards to find the transmitting laser power as a function of range and other variables.

Yes i understand the SNR issue and the need for a APD but the requirement given to me was a silicon PIN photodiode. maybe i can suggest for an APD with the very high gain due to the multiplication effect. 1mm square area you mean?

Sorry, i hid the ac source from the schematic. it's actually a 0.1u AC amplitude as seen from the attached picture above.
 

Attachments

  • circuit.jpg
    circuit.jpg
    24 KB · Views: 121

If i were to increase the compensating capacitor, the BW decreases and it takes longer for the circuit to be stable. Can't seem to find any solution to this particular circuit. advice appreciated.
Reduce the transimpedance resistor and recalculate the compensation or go for a faster OP. In fact, I also don't see another solution to improve the phase margin with given diode capacitance.

I have no specific reason for using the quadrant sensor, just that it's suited for 1064nm laser wavelength.
Effectivly any Si photodiode will perfectly work around 1100 nm.

what do you mean by 10ns to 40ns? are you referring to the rise time of the laser pulses?
That has been also my previous question. Why do you use 20 ns pulses in the simulation (respectively in your real setup) but the amplifier can't reproduce them, just an averaged magnitude?
 

Reduce the transimpedance resistor and recalculate the compensation or go for a faster OP. In fact, I also don't see another solution to improve the phase margin with given diode capacitance.


Effectivly any Si photodiode will perfectly work around 1100 nm.


That has been also my previous question. Why do you use 20 ns pulses in the simulation (respectively in your real setup) but the amplifier can't reproduce them, just an averaged magnitude?

I will go with the former advice.

Yep, that was what i understood. But i just wanted one which has increased responsivity at 1064nm. I shall choose another so that it has lower reversed bias and re-simulate. Need to recalculate the Cin as the Cd would be different. Shall seek help again after i'm done later. Should i add in the dc bias voltage source in series to the parallel current source/capacitor?

That's a very good question.I think there was a 180 degree phase shift since the time when the pulse stays low ( 20ns ) becomes the output pulse being high and vice versa =)



---------- Post added at 19:54 ---------- Previous post was at 19:46 ----------

 
Last edited:

You mentioned a 10nm to 40nm laser. That doesn't make sense - I assumed you meant 10ns to 40ns.

As mentioned before, most silicon PIN diodes are fine for the laser you are using. Depending on the target (you haven't said if it is co-operative or not) and optics you might need a laser of around 1kW to get 4km I would guess. Eye safety could well be an issue.

I would also suggest a diode rated for operation at less than 150V. You can get cheap 1mm diameter pin diodes with just a couple of pF at maybe 15V which will be a lot more convenient.

Normally you would be looking at a FET input opamp to keep the current noise down with the high transimpedance. You also need to think about the effect of ambient light which can easily saturated the amplifier.

Keith
 

You mentioned a 10nm to 40nm laser. That doesn't make sense - I assumed you meant 10ns to 40ns.

As mentioned before, most silicon PIN diodes are fine for the laser you are using. Depending on the target (you haven't said if it is co-operative or not) and optics you might need a laser of around 1kW to get 4km I would guess. Eye safety could well be an issue.

I would also suggest a diode rated for operation at less than 150V. You can get cheap 1mm diameter pin diodes with just a couple of pF at maybe 15V which will be a lot more convenient.

Normally you would be looking at a FET input opamp to keep the current noise down with the high transimpedance. You also need to think about the effect of ambient light which can easily saturated the amplifier.

Keith

Hi,

Oh sorry and thanks for pointing that out. yes i meant a 20ns pulse with 10ns high and 5ns on rise and fall respectively.
I've changed to **broken link removed** now for the Si photodiode

the op amp i have changed to https://focus.ti.com/lit/ds/symlink/ths4601.pdf as well.

With a 0.5uA of photocurrent simulated, i would expect a received optical power of about 26uW/cm^2. The transmitted power i shall work backwards.

The response curve is now:

response.jpg

The transient analysis result is now:

tran.jpg

without the dc bias offset now as the values are relatively small from the datasheet.

I will probably keep the optical bandpass filter to the future work to restrict only to 1064nm wavelength of light to pass through.

However, i would like to understand how to look at the noise results. is this considered to be good or bad?
 

Status
Not open for further replies.

Similar threads

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top