OPA857: OPA857

HI Marcela, 

We haven't directly characterized the switching time between the gain settings, but I would expect it to be less than 1us. You will see a transient disturbance on the output signal when you switch gains. The bandwidth limiting will depend on what your input capacitance is. Assuming you are driving this with a photodiode, can you share the capacitance? I don't fully understand your last question. You can use the output pin single-ended or you can take the difference between the out pin and reference voltage. 

Regards, 

Hi Jacob,

Thanks for the fast reply.

The input capacitance is expected to be ~1.5pF, including parasitics. Rload is 500ohms. This results in a BW of 105MHz to 125MHz per data sheet. We need to reduce the BW  down to somewhere in 60 to 80 MHz range. What is the solution you would recommend for BW reduction?

The last question refers to the output load, if single-ended, then we take only the Vout signal and the GND as reference. Probably we will need to AC couple the load? I would expect that the value of the transimpedance gain does not change, even if single-ended, since the chip is not a true differential amplifier. Please confirm.

Thank you,

Best regards,

Marcela 

Hi Jacob,

I would appreciate a response to my previous questions.

I need to know also the input impedance of OPA857 in transimpedance mode.

Thank you,

Best regards,

Marcela Faina

Sr. Eng

PS. This might have been posted in a different thread as well, please disregard the other one if true

Hi Marcela,

Sorry for the delay. The input impedance, in trans impedance mode, will be the trans impedance gain resistor (5kΩ or 20kΩ) in parallel with 2pF.

For a 1st order RC filter, place additional capacitance across the feedback. For 80MHz and 20kΩ of TIA gain, that's a 1/(2*pi*20k*80M)=0.1pF. feedback capacitor. For an additional filter zero and a faster roll off, since you are using one side single ended, you can also add an output capacitor to ground after the internal 25Ω. I get an output capacitance of 1/(2*pi*25*80M)=80pF.

Capacitors in both of these spots can cause peaking once they exceed a certain point, but since this is an integrated amplifier, it is difficult to surmise from the information given exactly what that point is. The feedback capacitance is probably fine up to several pF, but you may want to leave room for extra external resistance to isolate, and reduce the required size of, a load filter capacitor if it is needed.

Best regards,

Sean

No actually for a transimpedance stage, you are looking into the low impedance of the summing junction that ramps up with frequency as the loop gain roll off. 

Thanks Michael,

I expect a low input impedance, indeed, in TIA configuration, albeit not specified in the data sheet. 

The reason to ask is that I need to increase the dynamic and I plan to use a shunt resistor at input to halve the photo-current, so it has to be of the same value as Zin.

What is the order of magnitude of Zin for this OPA857? Is it ~100-200 ohms or lower than this when Tz=5k?

Thanks,

Best regards,

Marcela

No you can't shunt off signal with a resistor, at DC the input impedance is <1ohm. 

It just sound like you have a lot of dynamic range to deal with. one way to do that is make say 2parallel different fixed gain Zt stages, and then feed those into a 2X video mux and pick which gain is active, switches in 4nsec, would need seperate detector + amp stages which might not work. 

http://www.ti.com/lit/ds/symlink/opa875.pdf

Hi Michael,

Thank you for the feedback. I still do not understand the input impedance being <1ohm in DC, since it sounds like a shortcircuit to GND. What about  Zin in AC then? 

Could you please elaborate more on this?

My plan was to add a resistor equal to the Zin in series with a fast diode to GND, in order to divert half of Iphoto from the TIA to GND. This is something that works very well with a discrete TIA circuit (transistors, etc...). 

If this is not a viable solution, then connecting one photodiode to 2 different OPA857 with 2 different Tz would work?

Thanks,

Best regards,

Marcela

Well we can take the TINA reference design and move the sense meter to the input pin and generate the dB ohm plot shown here, so that -26dBohms is 0.05ohms at DC. That is loop gain which rolls off with frequency raising the apparent input Z to100ohms at 300MHz. 

Thanks Michael,

Then I could have a shunt resistor in series with a cap connected to IN for current division? In the application, the optical signal (photocurrent) will consist of very short pulses.

Thanks,

Marcela

A shunt resistor at the input will not do anything but raise the noise gain,it will not shunt off your signal current 

What is the range of these pulse amplitudes of current.

Way back at the beginning you talked about filtering, that should be a post RLC filter not in the feedback C - you need to tune the feedback C to get the desired 2nd order broadband response. Maybe Bessel, or Butterworth? 

The expected photocurrent range is roughly 3 orders of magnitude, ~750:1, from 0.6uA to ~450uA. My plan was to use high Tz for low input and then benefit from the selectable Tz to get to lower gain when the first level of 60uA is reached. But even with the lower Tz, saturation will occur at 240uA.

I was hoping that I could divert half of the current (only in low Tz mode) so that there is no saturation at high end of photocurrent. There is some room for the noise to increase when the current is high, since the signal is high as well.

Perhaps some kind of active component, or a diode instead of resistor (with very low capacitance) that would be connected to GND by the same CTRL signal  that switches the Tz gain?

Any suggestion on how to achieve this would be greatly appreciatedBest  regards,

Marcela

so are you looking for the edge position in time or is the absolute amplitude also needed? If you are switching gains, how fast is required. 

Yes, both are required, edge position and amplitude information as well, or rather pulse width. In fact, the saturation is not allowed since it would stretch the detected pulse.

The switching time should be below 1us, and I believe this can be met by OPA857, as you have mentioned in a previous answer.

Thanks,

Hello Marcela, 

So I got to wondering if I could make my own switched gain transimpedance stage - one cool new part for this it the OPA837 unity gain stable VFA with a unique disable - it has switches out the inverting input to make the impedance looking back in very high disabled - I can use this as a switch in a transimpedance Tee network to switch between two (or more) gain settings. The attached is a quick run at this, probably needs more work and maybe the OPA846 is not the right part but conceptually looks interesting - a possible filter is in here also. You could put another gain setting with this approach in parallel with the first and use the dual Switched gain transimpedance 1st pass Aug28_2019.docxOPA2837. So yes, this OPA837 essentially give you 0 output impedance when on, and inifinite with off (not really, but closer than MOSFET switches). 

Hi Michael,

This is very creative indeed, thanks, lots of nice ideas!

My problem is that I have several identical channels that have to go in one small packaged hybrid circuit and the real estate is at a premium.

At this stage, I would like to please confirm with you that the OPA857 would work fine with switching Tz. I understand that the current division I have proposed for the high end of input current might not work.

In fact, the max Iph~450uA is within the 2x overload (for Tz=5k) and therefore, although saturation will occur, the recovery could be very fast.

Thank you for confirming all this,

Best regards,

Marcela

After a quick look at your paper for switching Tz,shouldn't the 250ohm connected to GND decrease the Tz gain to 1/3, so low gain would be 4.5k/3, ~1.5k instead of 12.5k? So that the low impedance of 250ohm will drain a good amount from the input current and therefore the toal Tz would be lower?

Thanks,

Marcela

The signal is a voltage after the first transimpedance R, that then creates a current in the grounded resistor that gets gained up by the remaining feedback R. 

Ch. 7, Photodiode Amplifiers, Jerald Graeme. 

Hi Michael,

Could you please respond to my previous question on OPA857 working with switched Tz and overload issue?

Thank you,

Actually I cannot, that is a product group question. 

Who could please answer from "Product Group"?

Thanks,

Marcela

Hi Marcela, 

If I understand correctly, you are just trying to confirm that the part can switch and also recover from overdrive? It is perfectly fine to switch between the gains of the device, and the overdrive recovery is quite fast on the OPA857 at less than 25ns typically. Does that answer your question? 

Regards, 

Hi Jacob,

Some of my questions have been answered, thanks.

I still need to better understand:

1. The input impedance value of the OPA857 (is this really 0ohm?)

2. A viable approach to deviate some of the photocurrent  from the TIA input at high end of the signal (very short pulses, 250uA to 450uA) in order to avoid saturation.

Thank you and best regards,

Marcela Faina

Hi Marcela, 

The OPA857 has a fairly high input impedance (it is a MOSFET input), so the input impedance of a transimpedance configuration will essentially measure the value of the feedback resistor terminated to the common mode (in this case 5k or 20k). Although its not the most elegant solution, you can add a parallel resistor to the feedback network to reduce the gain. However that will reduce the dynamic range of the gain switching fairly significantly. 

Regards, 

Hi Jacob,

Isn't the input impedance of the transimpedance equal to the feedback resistor divided by the gain, since the input configuration is of shunt type?

If  yes, should the Rin equiv be rather in the low range (~ 10 to 100 ohms)?

Thanks,

Best regards,

Hi Marcela, 

Maybe I am misunderstanding your circuit configuration, but a typical OPA857 configuration has a high impedance source (like a photodiode) driving directly into the amplifier input. The amplifier input itself is high impedance, so the low impedance path you see is the connection of the feedback resistor to the output of the amplifier. Since there are no other low impedance paths, that becomes the input impedance. 

Regards, 

This seems to be coming up regularly, 

So in the transimpedance application you are looking into the virtual ground of a high gain feedback LG. I recall that then is the Rf/(LG+1) where at DC that is relatively low then rises with frequency as the LG rolls off. 

Hi Jacob, Michael,

I agree with Michael on the input impedance of OPA857, which is expected to be low, since it drains the input photocurrent from the high impedance source (photodiode).

And indeed the feedback resistor at the input is devided by ythe loop gain.

What is the order of magnitude of this low input impedance? Ohms, 10ohms, 100ohms? It is not specified in the data sheet or application exemples. If there is no theoretical value, perhaps we can measure it on the evaluation board for OPA857? 

Thank you,