0
\$\begingroup\$

I am embarrassed to say I have been wrestling with this problem for some time, and am near to giving up. My TIA PCB always outputs high. The bread board version works well enough, albeit with slightly different components. The current schematic looks like:

TIA Schematic

The opamp is a LMH6629, all of the resistors and caps are 0805. The PCB for this looks like: TIA layout

I have tested each of the lines, and so far as I can tell, they have the voltage I would expect (excluding the output). I have checked for shorts for days now. I can't seem to figure out the cause. Any trouble shooting or design advice would be helpful.

Improve this question
\$\endgroup\$
  • 1
    \$\begingroup\$ Compare the input impedance spec of the LMH6629 vs MCP629. I'd almost guarantee that's your problem right there. That 1M feedback resistor just won't work with the LMH's "low" input impedance. \$\endgroup\$ – brhans Sep 28 '16 at 14:47
  • 1
    \$\begingroup\$ The 15 to 30 uA input bias current isn't going to be particularly helpful either. \$\endgroup\$ – The Photon Sep 28 '16 at 15:37
  • \$\begingroup\$ As Photon says, 15uA bias current kills functionality. Its got PNP inputs, so the output can't go low enough. If it managed to reach balance, there'd be 15 v dropped across that 1M feedback resistor. In this circuit, you need FET, MOSfet or CMOS op-amp. \$\endgroup\$ – glen_geek Sep 28 '16 at 15:41
  • \$\begingroup\$ The bias current is the proximate problem but note also that the current noise is huge as well 2.6pA/sqrt(Hz) - it blows away the fine voltage noise performance of this amplifier by more than three orders of magnitude. \$\endgroup\$ – Spehro Pefhany Sep 28 '16 at 15:49
  • \$\begingroup\$ I have tried a 1K feedback resistor to little success. After reading your comments, I have ordered an opamp with higher input impedance and lower input bias current and current noise. I will try that soon, and reply back. Thank you very much for the help. \$\endgroup\$ – SillyInventor Sep 28 '16 at 18:10
1
\$\begingroup\$

As @ThePhoton rightly points out, the bias current is the proximate problem. Tens of uA * M\$\Omega\$ = tens of volts output offset.

Note that the current noise is huge as well 2.6pA/sqrt(Hz) - it blows away the fine voltage noise performance of this amplifier by more than three orders of magnitude with a 1M feedback resistor.

You can still salvage this by adding a JFET voltage follower to the input (inside the loop). Then the input current noise effect will be reduced by a factor of gm*1E6:1, and total voltage noise will only be slightly increased (assuming a decent JFET).

Improve this answer
\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.