OPA857: OPA857

Hi Marcela,

using matched OPA857 would only make sense, if you would use matched photodiodes as well and if you could control the parasitic stray capacitances of the assembly (PCB layout, etc.). Parasitic stray capacitance can play a major role in respect of BW!

I would think about increasing the bandwidth by the method shown in section 8.2.2 of datasheet, because then the external resistor and its precision will define the final transimpedance and manufacturing tolerances of OPA857 will play a smaller role. A tight bandwidth control could be carried out in a second stage containing a passive low pass filter, again using a precise R and C.

Only an idea...

Kai

Hi Kai,

Thanks a lot for the prompt reply.

We use indeed a photodiode array which ensures very well matched characteristics. The assembly is a hybrid circuit where the parasitics are well controlled as well. We need a minimum output voltage in worst case of illumination and therefore we cannot reduce the Tz gain by the method you have described.

Are there any chip testing capabilities within TI if we were to require matched OPA357 dice? If yes, what are the implication on the cost/lead-time?

Thank you,

Hello Marcela,

To see the effects of the transimpedance tolerance on the bandwidth of the OPA857, I would recommend simulating the AC response in TINA. You can then add a large resistor between pins IN and OUT. This will lower you transimpedance gain and increase your bandwidth. You can adjust the external resistor so that the transimpedance gain is at -15%. 

Unfortunately we don't offer the testing service you are looking for. If you need the bandwidths to match as close as possible, I would measure the bandwidths before adding the parts into your design.

Best,

Hasan Babiker

Hi Hasan,

Thanks for the reply. In fact, the PSpice model allowed us to change the value of the Tz gain from 20k to 17k and 23k and therefore simulate the impact of Tz gain tolerance of +/-15% on the BW. This has resulted in BW change from 121MHz to 87MHz respectively. The same result was obtained when a 115k resistor was added in parallel to the TIA to reduce the Tz by 15%, as you have suggested.

Engineers love when the simulation/practical results follow the theory :).

Could you please provide some insights about the tolerance of the Rout?

Thanks and Best regards,

Marcela

Hi Marcela,

you could add an additional gain stage to guarantee the desired minimum output voltage in worst case of illumination. This would also allow to carry out some pulse shaping.

Kai

Hi Kai,

Thanks, I wish I had the real estate for 3 gain stages within the hybrid, but this is not the case...

Any thoughts on the Rout tolerance, please, yourself or other TI apps engineers?

Thank you

Hello Marcela,

I do not have the information on the tolerance of the Rout resistors. I will look into this more and get back to you.

Best,

Hasan Babiker

Hi Marcela,

I'm not a TI's employee, but from my understanding of datasheet, the precision of built-in resistors should correspond to the specified "DC transimpedance gain error", which is +/-15% maximum. This makes sense as the absolute resistance values of die can greatly vary, usually, while the ratio of resistances is often very precise.

Kai 

Thank Kai,

If this is the case, it means that the Rout tolerance is already included in the Tz gain tolerance of +/-15%, and it makes sense.

I am waiting for confirmation from one of the TI apps engineers.

Best regards,

Thanks,

Hello Kai,

Thank you for your comment. I have confirmed that it is exactly as you have stated, where the tolerance is +/- 15% but the ratio of resistances are similar.

Best,

Hasan Babiker

Hi Hasan,

Could you please clarify which one of the following applies, is it a) or b)?

a) The tolerance of the Tz gain (+/-15%) includes within it the variation of Rout or

b) The tolerance of the Rout is +/-15% and I need to consider it, additional to the Tz tolerance of +/-15%. 

Thank you,

Hello Marcela,

It is b), the tolerance of Rout is +/- 15% as well. However the variation should match that of the Tz gain. For example, if the transimpedance gain is at -15% (13.65 Kohm), Rout will also be at -15% (18.75 Ohms).

Best,

Hasan Babiker

Hi Hasan,

My understanding is the following:

For OPA857, the numbers would come down to the following pairs:

Nominal, Tz=20k, Rout=25ohms (single ended)

-15% tolerance: Tz=17k, Rout=21.25ohms

+15% tolerance: Tz=23k, Rout=28.75ohms

When calculating the Tz effective in the load, the impact of Rout variation is negligible (<1%) compared to the impact of Tz variation of ~15%. 

Please confirm,

Thanks,