Photodiode amplifier designed by wizard oscillating

Hi,

I used the Photodiode Circuit Design Wizard (http://www.analog.com/designtools/en/photodiode/) to come up with the attached circuit.

The circuit simulation opamp oscillates on the output. I guess that chances are that it will do so in the real world as well.

Please could someone advise; why and what to do about it?

I attached a ZIP which includes 2 files;

1. TransientAnalysis.asc (generated by the wizard)

2. TransientAnalysis_delay.asc (modified by me; i simply delay the pulse from the photodiode by 900ns) which shows oscillation after approx. 200 ns.

TransientAnalysis.zip

If you want to check out the design in the wizard (http://www.analog.com/designtools/en/photodiode/), you can load the file included in this ZIP into the wizard:

CBSphoto.zip

/HEK



fixed hyperlinks.
[edited by: HEK at 9:40 PM (GMT -4) on 12 Sep 2018]
Parents
  • +1
    •  Analog Employees 
    on Sep 14, 2018 11:34 PM over 2 years ago

    OK figured out what is going on here.  The LTC6268-10 is a G=10 stable amplifier and the circuit is violating that.

    In a photodiode circuit you have the feedback capacitance (Cf), the photodiode capacitance (Cd), and the op amp input capacitance (Coa)   For stability purposes the noise gain of the circuit is (Cd + Coa + Cf) / Cf. 

    The LTC6268-10 is a G=10 stable amplifier.  This means (Cd + Coa + Cf) / Cf should be 10 or greater.  There may be a little bit of play where you can push it under this number, but I think this is a pretty good rule of thumb.  Normally Cd >> Cf and this is not an issue, but for this particular circuit, Cd and Cf are about the same magnitude, giving a noise gain of about 2, which makes the circuit unstable.

    There are two lessons here for us at ADI regarding our photodiode wizard:

    1) The photodiode tool should be smart enough to prevent you from getting into this situation in the first place

    2) Our model of the LTC6268-10 poles and zeroes in the wizard could be improved so that if this were to happen, we'd see the ringing in the time domain plot.

    I'll file a ticket for us to work on these things.

    In the meantime, I think it will be difficult to find an optimal feedback cap for the LTC6268-10 with a 402 ohm feedback resistor and a Cd of 5pF.  Something like the attached might work, where the extra 80 ohm resistor helps to increase the circuit's total noise gain.  The LTC6268-10 datasheet apps section also has some commentary on min and max Cf tradeoffs.

    Matt

    TransientAnalysis 2.zip

  • Hi Matt,

    thank you for your explanation!

    I understand and see that you needed to lower Cf in order to achieve stability.

    I will try to be careful with the layout here and follow the layout recommendations of the LTC6268-10 datasheet to minimize spurious Cf capacitance, and I will add a resistor to GND on negative input as well and cross my fingers!

    /HEK

Reply Children
No Data