AnsweredAssumed Answered

Using ADA4817 as TIA stage for fast pulses

Question asked by pret on Apr 26, 2015
Latest reply on Jul 10, 2015 by krisf


I have certain amplifier design comprised of 2 stages, the first one is TIA configuration based on National's LMH6624 and 60.4k feedback (gain) resistor.

Recently the amplifier TIA stage has been updated replacing the LMH6624 by ADA4817, this is due to better noise floor of ADA4817 and broader bandwidth capabilities of one. Due to better noise characteristics the gain of ADA4817 has been raised (beedback resistor 274k comparative to 60.4k of the old design).

In both cases the OPA are powered by +/-5V supply.


The old design (with LMH) achieves design goal and used to be utilized throught years, the new one - samples where built and tested - there was a gain of nearly 2dB in sensitivity (which is considerable improvement in our case). However, there are few things in ADA4817 TIA behavior I'm trying to optimize.

I build PSpice circuitry (within Cadence Orcad environment) with both circuits - one is the legacy one with LMH6624, another one is the new one with ADA4817 (downloaded and installed the model of ADA4817) and run both to compare their performances.

What is really weird is that in simulation the ADA4817 TIA stage performs worse then LMH6624 with similar inputs streamed (100-200nsec puses with 4 ns rise/fall and 10usec period). According to the simulation the bandwidth of ADA4817 TIA suffers relatively to LMH6624.

The attached sumilation graph shows in red the ADA4817 TIA responce while the green depicts LMH6624 response.


The amplitude of the pulse generated by ADA4817 TIA stage is larger then that of LMH6624 by approx. 15-20% (which is ~ 1.7dB), however it is obvious that the BW of ADA suffers noticeably relatively to LMH. I realize that rising feedback resistor by 4 on ADA limits the BW, however BW of the OPA itself is better then that of LMH as well as lower input capacitance.

Any chances there might be an issue with ADA4817 PSpice model ?


Thanks in advance, Alex