Photodiode Amplifier Problem

[emran_shd]

Hi! i am working on the design of photodiode transimpedance amplifier of 100MHz using OPA657 Opamp. The first time when i switched on the power to the circuit, there were high frequency sinusoidal oscillations of about 10Vp-p. These Oscillations were removed after i tuned the feedback capacitor to 1pF. But the problem is that oscillations of 100 mV p-p having frequencies from 100 MHz to 400 MHz still remain there. These oscillations occur at the I/O pins and Supply pins of opamp. What could be the cause of these oscillations and How can i remove these oscillation when tuning the feedback capacitor and resistor has no effect in removing these oscillations.

 

[keith1200rs]

Not an unexpected problem although we would need more information such as component values and photodiode characteristics to be able to be specific.

Keith.

 

[emran_shd]

The photodiode used is AEPX65 having a capacitance of 4pF at +5V bias voltage and responsivity of 0.45A/W at 750nm. The 10Vp-p oscillations are removed with feedback resistor of 2.7k and feedback capacitor of value 1pF (These are also the values obtained from simulation of the circuit for 100MHz bandwidth). The capacitors used at supply pins are of values 33pF, 68pF, 1nF and 22uF. The load at output is of 50 Ohms.

 

[keith1200rs]

With those values I would not expect oscillations, although it is not totally stable, I would say.

Keith.

 

[emran_shd]

thanks sir for your quick replies.

With those values I would not expect oscillations, although it is not totally stable, I would say.

Then what could be the cause of the observed oscillations. does it have to do something with the pcb layout? How can i improve the stability of the circuit if it is not totally stable?

 

[keith1200rs]

Certainly PCB layout is very critical. It also depends on what bandwidth you need. Adding more capacitance should stabilise it but if that reduces the bandwidth too much then you will need to reduce the feedback resistor or use a different opamp.

Keith

 

[FvM]

Did you notice that OPA657 is only stable with close loop gains >= 7? This means, that there's also an upper stability limit for the feedback capacitor, depending on the photodiode capacitance. 1pF seems somewhat high. The fact, that you observed oscillations at the supply pins seems to indicate insuitable placement of the bypassing capacitors. Personally, I would rather use 100 nF and high capacitance ceramic or tantal as suggested in the OPA657 datasheet. 1nF is possibly useful, the pF values are of little use. As the datasheet tells, 0.25" should be a maximum distance of 100 nF to supply pins, a solid ground plane is required.

 

[emran_shd]

Certainly PCB layout is very critical.

What should be the PCB layout considerations for good performance. I am designing the amplifier for 100MHz bandwidth. should i use ground plane? i can only have a single sided PCB.

Adding more capacitance should stabilise it but if that reduces the bandwidth too much then you will need to reduce the feedback resistor or use a different opamp.

I have already tried this. Adding More Capacitance increases the amplitude of oscillations.
Which opamp is suitable for 100MHz bandwidth with as much transimpedance gain as possible other than OPA657.

 

[keith1200rs]

Maybe the THS4271 with a very good PCB layout. I don't think I would bother if you are not going to use a double sided PCB with a ground plane.

Keith.

 

[FvM]

The feedback capacitance must be considerable below to achieve 100 MHz bandwidth with Rf of 2.7k. OPA657 is basically OK, unity gain compensated OPA656 should work as well. The problem is however in the PCB layout I think. I doubt that single side will give good results.

 

[emran_shd]

@ FvM

Did you notice that OPA657 is only stable with close loop gains >= 7? This means, that there's also an upper stability limit for the feedback capacitor, depending on the photodiode capacitance. 1pF seems somewhat high.

I read somewhere that It is possible to use op amps that are not unity-gain stable in transimpedance circuits provided that the noise-gain at the intersection of the noise-gain curve with the open-loop op amp gain is greater than the minimum gain specified for the op amp.
I have roughly Calculated the noise-gain at the crossover of noise-gain and open-loop curves and it is just above 7. so i think you may be right, i need to lower the feedback capacitance to increase the noise gain and hence the stability of the circuit. I shall also have a look at the placement of supply capacitors. thanks

 

[FvM]

A feedback factor of 1/7 respectively noise gain of 7 should be sufficient for stability. But the bandwidth requirement suggests a lower value, e.g 0.3 or 0.5 pF, I think.

 

[emran_shd]

I have changed the pcb layout and with the new pcb layout the sinusoidal oscillations on the output are gone but now there exist sinusoidal bursts in the output with random time spacing. Please see the attached images. The images contain the same output waveform with decreasing time scale.
















What could be the cause of these sinusoidal bursts in the output and how they can be removed?

 

[FvM]

Besides the 60 mV scale, a timebase information would be quite informative.

The spikes/bursts are obviously electrical interferences from the enviroment and unlikely to be generated from the amplifier. It may be the case, that the amplifier still has an oscillating pulse response, but this should be better checked with an optical test signal.