3
\$\begingroup\$

I have a computer engineering background with minimal experience with real EE work. I want to interface to a ph probe and was wondering what the best way to do this is.

It seems to me there are 2 basic options for doing this. The first is fully differential, where the ph probe connects to Vin in the diagram below. The output can then be read from the output of the first 2 stages via differential ADC. Is this correct?? Are there any flaws with connecting a ph probe directly to this circuit??

Second stage can be excluded and is just if you want a single ended signal ended with additional gain

enter image description here

This second design is simply a single ended gain+offset configuration. First stage does some gain, second stage offsets and has a bit more gain.

enter image description here

Which one is ideal for minimizing signal loss/noise? My thought would be that the first circuit would be better at noise rejection since it is differential, and feeding this into a differential ADC, some common mode noise would be rejected.

Are there better configurations for doing this?? I am interested in reading range of 3-10ph only. In this case, is one better than the other?

Some notes on ph probe:

  • Extremely high impedence opamps are required(eg femptaamp input bias current).
  • -420mv is 0ph, 420mv is 14 ph(may have that switched around).
\$\endgroup\$
  • 3
    \$\begingroup\$ This seems a lot like other questions you have already asked. Have you tried any of the advice you have already been given? If so, what were the results? \$\endgroup\$ – Olin Lathrop Oct 21 '11 at 23:50
  • \$\begingroup\$ Well I am on a limited budget for this school project and I can't just go out and buy a bunch of stuff. I need to design a reasonable circuit in the first place...I have limited experience in this domain...I think my question is valid and not the same as the other questions. \$\endgroup\$ – user623879 Oct 22 '11 at 0:21
  • \$\begingroup\$ Are you running long wires to this extremely high impedance probe that will pick up tons of noise? The voltage range you are measuring is (420+420)*(10-3)/14 = 420 mV? And you're feeding this to a 3.3 V range ADC, so you need a total gain of about 7? \$\endgroup\$ – endolith Oct 22 '11 at 1:31
  • \$\begingroup\$ Don't try to build an instrumentation amplifier (in-amp) because the mismatch in the components (op-amps, resistors) will yield terrible results. Instead, grab a INA121. It has very high input impedance (10^12), low offset voltage 200uV. The resistors in the in-amp have been laser trimmed to match. \$\endgroup\$ – NothinRandom Jul 24 '13 at 17:34
2
\$\begingroup\$

You cannot just connect the probe across inputs to the inamp and expect good results.

Operational amplifiers need a path to ground from each input. So the second circuit is better.

the 741 is totally unsuitable, but the LMC6041 in the notes should work well for this circuit. (2 fA is typical rating though, use the guaranteed value of 4 pA for design unless you want your circuit to only mostly work... 4pA through 300Mohm sensor is 1.6mV offset.)

LMC6041 will be happy with a single supply, so if the -5V supply is inconvenient have a look at this circuit: What is the purpose of this op amp?

I suggest you stop simulating and actually build the circuit. LMC6041 is $2 from a distributor or ask National, they'll give free samples for school projects. You don't need to hook up the ADC yet, just get the probe and signal conditioning amplifier working. Read the output with a voltmeter. Just like software, breaking the problem into small pieces is the way to make progress.

\$\endgroup\$
  • \$\begingroup\$ what is functionally incorrect about the first one? I am considering using ti.com/product/ina116 which is equivalent to the first opamp circuit. Would it work if I pulled one end of the ph probe to ground? You are right on building the thing...I just don't think my project sponsor will like it if i mess up and have to start over...lol \$\endgroup\$ – user623879 Oct 22 '11 at 18:54
  • \$\begingroup\$ It would work to put one end of the probe (and one of the in-amp inputs) to ground. But it is unnecessary to use an in-amp where you don't have a big common-mode voltage to remove. If you ground one end of the probe then you have a single-ended circuit and a standard op-amp should work fine in non-inverting configuration. \$\endgroup\$ – markrages Oct 22 '11 at 18:58
  • \$\begingroup\$ Sorry, but what if I connect the ground end of the probe through a large say 1Mohm resistor to ground. Is not the first configuration more desireable for eliminating noise, or is it better to do this on the output of the opamp via low pass filter \$\endgroup\$ – user623879 Oct 22 '11 at 19:07
  • \$\begingroup\$ It will help eliminate noise due to the added 1Mohm resistor. Better to not add the noise in the first place. Why not build the circuit and see what kind of noise problems you get? \$\endgroup\$ – markrages Oct 22 '11 at 19:13
  • \$\begingroup\$ marked down cos using an instrumentation amplifier is the correct method. normally you use an intergrated circuit IA and not do it discreetly. The INA116 is suitable. TI have a good range of IA that are low noise and have wide bandwidth and excellent PSRR/CMRR. \$\endgroup\$ – smashtastic Oct 24 '11 at 19:20
0
\$\begingroup\$

I like a hybrid of the two. Firstly, the first is an instrumentation amplifier build. You can't use off the shelf components to build an IA with anywhere near the CMRR of a commercial IA -- at least without a whole bunch of hard work. Spend the $6 on a real IA. If you'd like, use a high resistance to ground to deal with bias currents on each input. A modest gain IA stage to deal with CMRR, followed by modest gain opamps with appropriate filtering is often a good approach.

\$\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.