OPA656:

bhavani korasikha

Part Number: OPA656
Other Parts Discussed in Thread: OPA657, OPA695, OPA830

Hiii

My photodiode current is very less (5Micro ampere to 10 micro ampere). So the standard transimpedance amplifier can't give the optimum performance since signal is mixing up with noise..

I came to know about Bootstrap technique. So I have designed simple bootstrap amplifier as shown below..

Theoritically at inverting terminal of OPA656 which is in feed back(buffer amp), should be virtually grounded since its non inverting terminal is directly connected to inverting terminal of other OPA656..

The voltage across capacitor 4.7UF should be Voltage across photodiode minus offset voltage of OPA 656 i.e, 25.75V theoritically..

But when I have connected this circuit on breadboard, at Inverting terminal of Buffer amp 4.4 V is observed. Observed Voltage across the capacitor is 21.6V.. 4.7UF/50V is used..

My photodiode capacitance is 920pF

Bias voltage applied is 26V..

Why OPA656 only??

Due to its low input noise density: 7nV/root HZ.
Also input capacitance is less around 3.5pF.
Low input bias current 2pA..(since photodiode current is very low, bias current should be low).

Where I have gone wrong in the design?

Why the terminals are not getting virtually grounded?? what could be the reason?

Is there any other parameter that I have to consider while designing this type of circuit??

I would be grateful If I got the solution..

Thank you

over 3 years ago

0 Michael Steffes over 3 years ago

Guru 48395 points

Well Bhavani,

The bootstrap is used to broadband a high C source, but normally we use as low a voltage noise transistor as possible - If I look at your circuit, it quite possible could be oscillating as the output of the bootstrap finds its way back to the V+ input pin via the diode C, Here are some sims,

1. The basic transimpedance with just the OPA656 is very nice, is this 224kHz about what you want?

2. The integrated output noise is poor of course as the voltage noise is peaked up by that large source C

3. Putting in your bootstrap, little more broadband but that little bump around 6MHz is a concern

4. Looking at the spot noise shows you might be unstable with that very large peak around 6Mhz,

What are your design goals?

0 bhavani korasikha over 3 years ago in reply to Michael Steffes

Prodigy 70 points

This resolved my issue. Now I came to know why it is oscillating..

Thank you

0 kai klaas69 over 3 years ago

Guru 140200 points

Hi Bhavani,

you might want to do it this way :-)

Kai

0 Michael Steffes over 3 years ago in reply to bhavani korasikha

Guru 48395 points

But what to do?

Here I took an existing OPA657 bootstrap circuit using the BF862 low noise N-channel JFET, Looks surprisingly good - the JFET is abut 10pF input C so maybe with that 0.2pF feedback C (parasitic) for 200kohm looks to work pretty well in sim. Your 10uA to 20uA current with your original 1Mohm feedback is going to overdrive the output of the OPA656 or 657 so I reduced it to 200kohm,

Next I ran an input current noise sim with the OPA657 then OPA656 dropped in. The OPA657 is of course a little better,

Don't know your bandwidth needs but the next stage would be an active filter normally to control the integrated noise,

And then running the OPA656 response, actually a little peaking out there?

And then here is the OPA657 file in V9

OPA657 with BF862 bootstrap.TSC

0 kai klaas69 over 3 years ago in reply to Michael Steffes

Guru 140200 points

And the noise figures with and without JFET bootstrapping:

Kai

0 bhavani korasikha over 3 years ago in reply to kai klaas69

Prodigy 70 points

Hi Kai

Thank you so much

Initially I have tried below circuit.

Many issues I have faced like at gate terminal unwanted +5V is seen. I have reduced the feedback resistor to 100K, 10K also. But the voltage at inverting terminal(gate of BF 862) is same.

And some other problems like, circuit is oscillating sometimes.. One observation is that BF862 what we have is with marking code of W28U but for NXP make of BF 862 it is 2AP.. I don't know whether the part what we have received is exact or not..

That's why I have tried with another approach

I have connected another OPA656 in feedback , want to see what it will give as output.

Could you please tell me where the mistake has been occured??

I will test again the circuit and will reply you..

+1 kai klaas69 over 3 years ago in reply to bhavani korasikha

Guru 140200 points

Hi Bhavani,

I think you biased the JFET improperly. Try to reduce the drain resistance. Michael has chosen 332R :-)

Reducing the drain resistance will also increase the drain source current which is good for the noise. For JFETs the noise decreases when increasing the drain source current because then the channel resistance decreases and by this the associated resistor noise (Johnson noise).

Kai

0 Michael Steffes over 3 years ago in reply to kai klaas69

Guru 48395 points

Yes, I was trying to bias at an avg IDSS at 15mA to get approximately a zero Vgs.

The model is giving 40mV on the source with this set up current,

0 bhavani korasikha over 3 years ago in reply to Michael Steffes

Prodigy 70 points

Thank you Michael

There is peak occurred when you simulated with OPA656. Is it the reason behind why my circuit is oscillating ?

OPA657 part is not available. I will try to get that part.

Also we have OPA830, OPA695. Can I use any one of these??

0 bhavani korasikha over 3 years ago in reply to kai klaas69

Prodigy 70 points

Hi Kai

Useful information. I will try with 332R and update you.

+1 Michael Steffes over 3 years ago in reply to bhavani korasikha

Guru 48395 points

The OPA656 did not look like it was oscillating, just a little lower phase margin,

The OPA830 and OPA695 would not be suitable to your application

0 Michael Steffes over 3 years ago in reply to bhavani korasikha

Guru 48395 points

You have never said what speed you need, but here is the input current noise sim through 100kHz, looks pretty good to me,

0 kai klaas69 over 3 years ago in reply to Michael Steffes

Guru 140200 points

This would be the phase stability analysis of the OPA695:

bhavani_opa695.TSC

But unfortunately the OPA695 suffers from huge input bias current.

Kai

0 Michael Steffes over 3 years ago in reply to kai klaas69

Guru 48395 points

Kai, you are using a current feedback device here in the OPA695, that would not be suitable to a 200kohm transimpedance gain - look at the inverting input current noise. JFET is the right was to go.

0 Michael Steffes over 3 years ago in reply to Michael Steffes

Guru 48395 points

But, if I modify your LG sim to the OPA656, it has about 39deg phase margin, which is why there was a little peakin the response, but not oscillating, And, the set up here is not suitable to the CFA OPA695 - it is a not a differential input voltage as the error signal, but the inverting error current.

0 kai klaas69 over 3 years ago in reply to Michael Steffes

Guru 140200 points

Thanks Michael, you are right, of course :-)

Kai

0 bhavani korasikha over 3 years ago in reply to Michael Steffes

Prodigy 70 points

Hi Michael

Bandwidth requirement is yet to be found.

Photodiode input light is of 492nm wavelength and the pulses are of 2ns rise time.

Please tell me what could be the front end bandwidth requirement??

0 bhavani korasikha over 3 years ago in reply to kai klaas69

Prodigy 70 points

Hi Kai

I have tried with 330R and with OPA656. But the issue is not resolved. At inverting terminal of OPA656 +4.65V is observed..

I trying to purchase some other JFETs like BF861C since BF862 is obsolete.

0 kai klaas69 over 3 years ago in reply to bhavani korasikha

Guru 140200 points

Hi Bhavani,

the 2SK2394 might be used as an replacement for the BF862.

Kai

0 bhavani korasikha over 3 years ago in reply to kai klaas69

Prodigy 70 points

Thank you Kai. I will try with 2SK2394

0 Michael Steffes over 3 years ago

Guru 48395 points

Not sure how to reply to the most recent point Kai made, but here is where I see the reply key,

Here is a source follower test for noise using that ON semiconductor 2SK2394, looks pretty good. This set up provides a zero Vgs voltage on the upper FET. I don't think your really need that and just using a 1kohm to the -5V supply on that upper fet source gives you a delta V across gate to source but operates at much lower current with only slightly higher noise -

0 Michael Steffes over 3 years ago

Guru 48395 points

Not sure how to reply to the most recent post, but here is a source follower sim of the ON semi 2SK2394 - looks pretty good, and the model file, And now I can't find how to insert saved snips or sim files, tried insert below, but that doesn't seem to be it.Oh well, if you want the 2SK2394 TINA model, need to fill me on attaching files,

0 Hasan Babiker31 over 3 years ago in reply to kai klaas69

TI__Genius 15555 points

Hello Bhavani,

Can you start a new thread and link this post? Because the issue has been marked as resolved a while ago, I believe it is causing issues with attaching files.

Best,

Hasan Babiker

0 bhavani korasikha over 3 years ago in reply to Michael Steffes

Prodigy 70 points

Hi Michael

New thread related to this question is started. You please try to insert the files in the new thread.

Because of the Thanksgiving holiday in the U.S., TI E2E™ design support forum responses may be delayed from November 25 through December 2. Thank you for your patience.

Amplifiers

Amplifiers forum

OPA656: