Measuring input impedance 2

[Rakestraw]

I have been asked to measure the actual input impedance of an analog circuit that is designed to accept micro volts as the input which is directly fed into a high impedance op-amp buffer stage. The typical frequency at the input is about .1 HZ, so I am treating it as DC so I am thinking that I am really just trying to measure the input resistance? The maximum input voltage to this circuit is 2 Vdc. In order to measure the impedance I used a laboratory DC reference as a voltage source with a resistor in series across the input to my circuit, I figured I would measure the voltage drop across the resistor to calculate the current and measure the voltage across the input to my circuit and then use these two numbers (R=V/I) to calculate the input resistance.
The problem is that the input Op-amp LMC6462 has an input impedance that is specified at >10 Tera Ohms and our best volt meter has an input impedance of 10 Giga ohms, so the meter loads the circuit and my measurements are incorrect.

Does anybody have any Idea how I can overcome this problem? Or a better way to measure the input impedance without renting or purchasing additional equipment.

Thanks !

 

[CarlPugh]

There have been some articles written on this type problem. I believe that Bob Pease of National Semiconductor wrote an article on this.
The approach was to use a capacitor with extremely low leakage and charge it to some voltage, say 2 volts. Remove the charging source and then check how much the voltage had dropped after say two days. From this the leakage resistance of the capacitor can be calculated.
Then charge the capacitor to two volts and connect it to the device whose resistance is being checked.
Check the capacitor voltage after say one day.
From this data, the total resistance discharging the capacitor can be calculated, and from the total resistance the resistance of the device can be calculated.
If the device under test also is a voltage source, then the calculations get much more complicated. However the calculations can be done using this approach.
The article also mentioned that the ambient temperature could affect the voltage on the capacitor. ?When the capacitor got hot the dielectric expanded and the voltage increased, and when the temperature got cold, the dielectric contracted and the voltage decreased?
Good Luck

 

[dspDad]

Careful about your assumptions.
This ic has 0.005 nA leakage and 3 pf input capacitance.
The impedance at 0.1 Hz is about 10e9 which is a small fraction of the 'dc' resistance of 10e13.

Frequently, at these levels, the dc offset is controlled with a 100 megOhm or so resistor to ground which 'sets' the dc input resistance, and the input impedance is then 3 pf in parallel with 100 mOhm over all operating conditions.

Because the innate impedance is so high, several 2nd order effects need to be controlled. The boards have to be washed to remove grease, and the humidity may need to be specified especially if the boards are not fiberglass. The resistance will vary somewhat with temperature and voltage. If the ic output is used to measure its input, then other temperature and voltage coefficients have to be accounted for.

I missed Bob Pease's article mentioned above. He covers interesting territory.

 

[Rakestraw]

Thanks Mr. dspDad;

I believe my circuit topology is not the same as what you are talking about, you can see the input circuit at

http://www.geocities.com/rakestraw69/Sch.PDF

Our problem is that this circuit is specified to work with various transducers that spec impedance ranges from 20 meg ohms to 1.5 gig ohms in addition to a length of 1 meter of RG174 coax (??pf). Higher impedance transducers are creating problems as the voltage at the output of the buffer is not steady enough. I am thinking that the op-amp impedance is to high to get the transducer current that I need to stabilize the voltage? I am considering re-specifying the op-amp to a lower input impedance. I inherited this circuit and the headaches to go with it.
Thanks !!

 

[LangdonShaylor]

dspdad has some good points about the board cleanliness but there are other ways to make it less critical. On the pcb a "guard-ring" must be incorporated. this is just a ring of copper surrounding, but not touching, all circuitry from the op-amp input pin to the BNC centre pin. This ring must be connected to the output of the op-amp, which must be in a unity gain voltage follower configuration. The idea behind this is that the surface leakage current is obviously dependent on the potential difference between the signal line and physically adjacent areas. Incorporating the guard-ring, connected to the op-amp o/p means that the potential difference is reduced to just the input offset voltage of the op-amp.
To actually measure the input impedance the easiest way is apply your input signal, eg 2v in your case, for which you'll get 2v o/p from your op-amp and then insert a resistance in series with the input to decrease the o/p to 1v. Your input impedance will be equal to the series resistance. You will be using your circuit as the i/p to the voltmeter, which could be an old AVO for example, and the measuring test gear wont load your input signal. To double check increase your input signal to regain the 2v o/p from the op-amp. Theoretically you should need to exactly double your input voltage. If not it will show up any other non-linearities in either the physical construction of the circuit or the op-amp itself.
With such high impedance circuitry pick up of external noise sources is also a big problem.
Best of luck.

 

[dspDad]

Rakestraw,
The input circuit url was cutoff, and won't open.
It reads :

http://www.geocities.com/rakestraw69/Sch...

Can you repost it?

Grease includes oil and sweat even from recently washed hands. If you aren't fastidiously careful, the 'measured' impedance could mysteriously drop orders of magnitude after touching the backside of the board, for example.

 

[Rakestraw]

Sorry, Yahoo has some rules I did not follow
try

http://www.geocities.com/rakestraw69/

then click on the "schematic" link

 

[elfgriper]

Rakestraw,
Some of our lo current circuits also include a can over the circuit connected to the output to sheild against
stray voltage fields. Iv' seen mechinical movement of the sensor or the connecting coax (looks like theres 2.5v across your coax) also induce noise into these types of circuits and sometimes close proxmity to a switcher supply or poorly filter Vcc can cause noise.
Good Luck -elf

 

[dspDad]

Thanks, I got into the circuit that time.

Is this the input to your proposed measuring system or the input to the instrument?

The input circuit is pretty unremarkable. Measuring the input impedance and changing the ic type to a jfet or lower impedance op-amp proably won't help unless very low level noise is an issue, in which case quieter components are available.

What is the frequency of interest?
How small is the signal?
Is it ac or dc coupled? i.e. is the dc level of interest?
Is the dynamic impedance of the probe the same as the static impedance. Piezoelectric transducers, for example, are mostly capacitive and have fairly low ac impedance.

I will assume ac from 0.01 to 0.1 hz give or take, and signals at least in the 10's of millivolts or more.

With either of the proposed inputs shown, there is only 3 db power supply rejection for your probe reference at frequencies below 1000 Hz. Any ripple on your power supply is immediately coupled back to the input.

A quick fix might be to change r1-c1 to 10megOhm and 10 uF, giving a 3 db rolloff around 0.001 Hz.

There is no decoupling shown from power to ground on the op-amp which is important both for noise and sometimes for stability.

As mentioned by LangdonShaylor, guard rings are necessary, a ground plane is almost indispensable, and power supply routing will be important.

DspDad