I'm working on the above mentioned circuit. I'm a hobbyist with little electronics training so don't laugh if something terribly dumb comes out of my keyboard please.

Anyway, a different question on stackexchange had a comment that opamps don't need a negative rail in order to invert a signal. I'd always thought a negative rail was required to handle a negative signal so this was like a bolt of lightning. instead of messing with voltage inverters etc, this just uses plain old single supply. Circuitlab seems to like it.

Anyway, I'm concerned with pH 4-10 (just enough to sneak in standard pH solutions for calibration). I've add enough gain so that 4-10pH @ 40C almost hits the entire span (which is plenty more resolution that I really need).

What circuitlab can't really help with (or at least I don't know to to do it) is handling noise and instability. How do you think this would hold up in the real world? I figure I'll need an RC filter on the output and I'm just reading up on those. Like the description says, if it takes 2 minutes to go from 0 to 1.8V that's actually Speedy Gonzalas territory in this application.

So, is this simply not going to work the way I think or is there modifications I can do to make it work (better)?



simulate this circuit – Schematic created using CircuitLab

P.S. BBB = BeagleBone Black. 3.3v supply, 1.8v ADC and 1.8v analog supply limited to about 10mV which is why the voltage follower/buffer to power the opamps.

Edit: Spotted a tip about matching input resistances between the two inputs. Reduced R7 and R8 to 1k and 4.3k to get 811 ohms on - and 779 ohms on +. Should I add the extra 30ish ohms to the + input or do you think the 30ish ohms makes no difference?

  • \$\begingroup\$ Where are you getting your estimate of the pH sensor's output impedance? \$\endgroup\$ – Matt Young Sep 12 '17 at 15:55
  • \$\begingroup\$ A post someplace that said it would be between 10 and 50 megaohms. Also this brochure omega.com/Green/pdf/pHbasics_REF.pdf says they require teraohm input impedance equipment to read. \$\endgroup\$ – user3758018 Sep 12 '17 at 16:08
  • \$\begingroup\$ Your schematic shows milliohms, aka 9 orders of magnitude lower. Your input impedance is 10k, which is 9 orders of magnitude lower than teraohms. This circuit will not work. Will come back and write a proper answer later, if somebody else doesn't beat me to it. \$\endgroup\$ – Matt Young Sep 12 '17 at 16:16
  • \$\begingroup\$ Ahha! Thank you for pointing that out, completely missed the small "m". I was wondering why the voltage divider seemed to not be affected by the source impedance. I was thinking it was one of those mysterious electronics principles I can't wrap my head around. \$\endgroup\$ – user3758018 Sep 12 '17 at 22:35
  • \$\begingroup\$ @user3758018 If the PH sensor really has 5 Mega-Ohms output impedance then youy probably need to buffer it with an op-amp that has very low input bias current. Even the MCP6231 may have too much (1pA * 50Meg = 50mV). 50mV could be quite a bit of error depending on the full scale output range of the sensor. \$\endgroup\$ – user4574 Jan 26 '19 at 18:47

Well now that I've been made aware of the m vs M error I can see this simply won't be practical. While I can get resistors of appropriate values to make the above work for gain and offset, I probably can't be sure what source impedance I'll get from probe to probe, even from the same manufacturer so proper gain is likely to be big problem. Looks like it's back to two op-amps.

Still, I can use this without gain as a buffer then a second amp for the gain and still not need a negative rail.

| improve this answer | |

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.