30V opamp attenuator with high impedance for ADC

[markbng]

Hi,

Thanks for reading my message. I am making a high impedance measuring device that measures input voltage in the range of 0v up to 34V. Now I generate 36V for the input opamp and this first stage is a 1x buffer. Is there an opamp solution that only needs a single voltage supply (3v3) and that can handle 0v-34v and has an attenuation of 12x. The input must have a very high impedance (+ or - input of the opamp), so a resister divider is clearly no option. Do you know how to solve this in an easy way?
Thanks in advance for your feedback.

Kind regards,
Mark

 

[dick_freebird]

First, "how high is high"?

"Beyond the rail" op amps exist, I believe I've seen an
interesting one from Linear Tech recently (engineer
magazine ad, somewhere) but I doubt that range is
going to encompass proper operation and 30V input
above rail. Still it might be worth your look.

Your alternative is a local boost supply (or pair, if you
also want accuracy near ground). Your choice (?) of
3.3V as the sole supplyis unfortunate, you would have
more and better options at 5V or 12V for power
conversion.

 

[crutschow]

What is the minimum input impedance that you can tolerate?

 

[schmitt trigger]

What is the minimum input impedance that you can tolerate?

Exactly my question, how high is "very high input impedance? 100 Meg?
Anything higher than that and the board leakage will be enough to throw you out

 

[BrunoARG]

If I were you I wouln't hesitate in generating 36V, if you add a resistive voltage divider in the input opamp (supplied with 3V3 or so) as part of the input impedance you will attenuate the measured voltage and won't saturate the opamp. Maybe a rail-to-rail opamp would be a good choice to avoid clipping since the supply voltage is so low.

 

[markbng]

Now I am using a OPA171 as an input buffer (Vin on the + input of the opamp). After this opamp is a resister divider that brings it into the ADC range. The OPA171 has an IIB of 15 pA. I am more than satisfied with that current, but I don't like it that I have to make an additional 36V DC voltage with a step up converter (which possibly will introduce a disturbance on the signal). So that's why I wanted to avoid the 1x opamp on the input.

 

[crutschow]

Okay, you are satisfied with a 15pA bias current at the input, but what is the maximum you can tolerate?
What is the voltage source?

 

[FvM]

The answer to the original question is simple: You can use a buffer with high supply voltage or a voltage divider. Apparently there's no third alternative.

For the voltage divider variant, you can e.g. use 1 GOhm resistors, which impose some restrictions regarding leakage current induced errors, also availibility of precise resistors is limited, and still get 30 nA input current. Or 300 nA with 100 MOhm, which would be a more convenient resistor value.

 

[markbng]

The problem is that it has to be a voltage meter with a VERY high impedance, since the voltage source is from energy harvesting. I guess the max current is about 50pA (I know, that's very low).

 

[KlausST]

Hi,

50pA energy harvesting? with max 30V it gives 1.5nW. I have some doubts.


I compleetely agree with the posts before.

One other point that comes into my mind: With that high input impedance you might have probles, because humidity and solvent residuals may affect your results.
Additionally you may need something to protect your input of the OPAMP from ESD or other high voltage. If so, look for "active shielding" where you sorround your input signal with the output voltage of your buffer.
Maybe use additional low leakage diodes from output to input of buffer. This introduces verly low capacity (instead of one diode to GND and another to VCC), but may affect stability of buffer.

Good luck

Klaus

 

[FvM]

The problem specification sounds somehow dubious, but if you actually need this low input current level, look for electrometer amplifiers like OPA129.

 

[acannell]

The problem is that it has to be a voltage meter with a VERY high impedance, since the voltage source is from energy harvesting. I guess the max current is about 50pA (I know, that's very low).

How often do you have to make the measurement? If you only sample the voltage once a second and for 1 us each time, that can reduce the impact the measurement has on the voltage source.

 

[schmitt trigger]

I'm going to repeat myself.
With common board materials (FR4), and without active guarding, the actual board leakage will be orders of magnitude higher.

There used to be an EXCELLENT app note by National Semiconductor that thoroughly explained this.

Personally, I once saw an electrometer design, whereas the board had a teflon insert for the opamp's input terminals. The opamp itself was a TO39 package (metal case with a bottom glass-frit seal), to prevent humidity absorption in the DIP8's plastic moulding.

 

[acannell]

I'm going to repeat myself.
With common board materials (FR4), and without active guarding, the actual board leakage will be orders of magnitude higher.

There used to be an EXCELLENT app note by National Semiconductor that thoroughly explained this.

Personally, I once saw an electrometer design, whereas the board had a teflon insert for the opamp's input terminals. The opamp itself was a TO39 package (metal case with a bottom glass-frit seal), to prevent humidity absorption in the DIP8's plastic moulding.

The OP might not understand what you are saying here..are you saying that because guarding is required and FR4 might not work that he cant use an opamp?

Guarding for leakage, especially at DC, is very simple to implement and is covered well in the gigaohm transimpedance examples in the OPA series of opamps datasheets. Using other board materials is possible too, especially if this is a prototype or a one-off, then you can pick whatever material you want and since the front end circuit is simple and low parts count, just make it by hand and it wont be expensive. You can also strategically locate slots in the board to block leakage, depending on what the circuit ends up being. These are all basic things to do that should not be reasons to abandon an approach that requires them.

It sounds like the OP may not have yet done his homework though and doesn't know exactly what impedance/measurement duty cycle he can afford to have. Those things need to be known before getting into designing the measurement front end.

 

[schmitt trigger]

The OP might not understand what you are saying here..are you saying that because guarding is required and FR4 might not work that he cant use an opamp?

.

No, I'm not saying that. I'm only stating that for picoamp leakeage, proper board techniques are required.