Resistance measurement

[Gargy]

Hello!

I need your help. I want to design a circuit for measuring resistance. The problem is that this device should be accurate from 1Kohm to 1Gohm. Plan is bigger load, better accuracy.So basicly I have two options - fixed voltage and measuring current through load or vice versa - current source and measuring voltage drop on load. Load can be 20m away form the circuit itself.

First option (fixed voltage) - i can use voltage boosters from linear technology which can generate stable 80V DC, but current on high load will be very small and sensing this current (100nA or less) in long cables can be difficult and can produce errors.

Second option (current source) - maybe 1 uA current source, but than I need 1000V voltage source on high loads.

Anybody has an idea? And sorry for my grammar, I'm from Slovenia.

Regrads,
Gargy

 

[keith1200rs]

I would say fixed voltage & measure current. You will need a 4 wire measurement system due to the distance. Of systems I am aware of that are designed for open circuit/short circuit testing, they tend to use up to 100V for the test to try to get a sensible current to measure. You need to be very careful about leakage currents everywhere. 1G ohm is asking quite a lot. You can get less than that on a PCB in a humid environment.

Keith.

 

[Gargy]

4 wire measurement is not an option. I thought that 4 wire is "only" for very small resistance measurements - to eliminate the resistance from the cable and contacts.

I saw some high resistance instruments on the net. It has selectable voltage (10V, 100V, 1000V). So my guess to use 1uA current source and than select the voltage range to get the best resolution/accuracy possible.

I also looked after transresistance amplifiers, but tha current range is just too big to us only one set-up. Maybe different gain resistors?

regards,
Gargy

 

[keith1200rs]

4 wire measurement is not an option. I thought that 4 wire is "only" for very small resistance measurements - to eliminate the resistance from the cable and contacts.

True, but you are wanting to measure 1k ohms over 20m of cable. I depends on the accuracy you require and whether the cable is fixed so you can simply subtract a fixed offset.

Keith.

 

[Gargy]

This resistance represents (after converting) the moisture in wood. So wet (fresh) wood has around 6KOhm (which is roughly 94% humidity) resistance and 1% dryer (93%) is already at 7KOhm. So as you see, few ohms does not play a big difference.

But resistance in wood is not linear. I made a mistake - dry wood has 5GOhms. If the difference in dry wodd is only 0.1% I get the resistance change of 100MOhms or more. So if I have 5% accuracy in resistance measurement it will be enough.

Regards,
Gargy

 

[Gargy]

Hello again!

So for last couple of years I dropped the high-resistance circuit and worked on new automation system.
This part is now finished and tested so I'm returning to this circuit.

After searching over Internet I came to the lock-in amplifiers. Somehow I understand the principle especially the part where I can get the original signal out of noise, which can be generated over cables.

My plan for now is using AC signal (sinus) with +/- 10V, going to the wanted resistance which is in voltage divider (10Mohm in series). From voltage divider to voltage buffer (OPA129 or similar) and then into lock-in amplifier (reference signal will be the same as on resistance). The result should (!) be a DC voltage with corresponding value to the voltage drop on the unknown resistor.
with use of ADC I can get this value to my PLC and with some filter I can get really clean, noiseless value.

Reference signal frequency and ADC sampling rate can be really low. 1Hz is more than enough.

Will this concept work?

Gargy

 

[ZekeR]

Hi Gargy,

Lock-in amplifiers are indeed good for measurements where you send a signal and detect the system's response, as you have here. However, they are really only good for two purposes:
1.) Avoiding DC offset (If there are environmental factors creating a DC stimulus, e.g., a DC voltage existing within the wood).
2.) Avoiding low-frequency flicker noise within the electronics.

Otherwise, the added complexity of lock-in makes it more difficult to do clean measurements. I don't know if the wood has any residual voltage sources due to chemical processes (you may want to check), but I wouldn't think there would be one. And flicker noise shouldn't be a problem. Therefore, lock-in doesn't seem to be a good idea for resistance measurement.

However, lock-in is perfect for capacitance measurement, which perhaps you should consider (although in this context it wouldn't be called lock-in, but just switch-cap). Most MEMS humidity sensors rely on the capacitance of a material changing, rather than resistance. You may find this easier to measure. There are many chips today that have built-in capacitance measurement circuits, or you can build your own using an LTC1043.

 

[Gargy]

Hi ZekeR,

capacitance is a possibility in wood moisture measurement, but because of its complexity, temperature deviation, species dependency and some other stuff, it's not a good way to measure woods moisture content. Resistance is the far more accurate value and there are already tons of different tables and calculations for temperature drift and species.

The base signal has to be AC, because DC starts to electrolyze pins which are hammered down into wood. Also my sensor has to be up to 16m away from the actual wood block so I have to run cables from sensor to the measurement pins. At this point I'm using parallel coaxial cables with shield connected to the GND. It still produces some error, but it can be mathematically subtracted from the actual value. But those cables are hard to repair (near to impossible) so my goal is to use basic Teflon or silicone wire.

Lock-in seemed to be good option because you can get the main signal out of the noise, which can be generated in wires. Also you can get really small values and from my calculation and measurements my voltage should be from 9.8V to 4 mV. Also the output of the LIA is DC which is ideal for ADC and for low sampling rate. The time windows for measurement can be 10 or 20 seconds so no need for speed here.

I'm open for suggestion for this kind of measurement. I looked for picoammeter but I'm afraid of the noise...

Gargy

 

[ZekeR]

Gargy,

1.) If you won't be measuring the same piece of wood for years, then you probably don't care about electrolysis of the nails. However, you may still find AC stimulus useful for other reasons.
2.) If the wood impedance is extremely high, then the frequency you apply must be very low to permit parasitic capacitances (for example, dipoles existing within the wood) to fully charge before measuring.
3.) When operating at low frequencies, you don't need to use the full "lock-in" approach; you can simply measure voltage when positive bias is applied, and then measure voltage when negative bias is applied. Taking the difference between the two voltages will tell you the resistance. (This technique is formally known as "correlated double-sampling," but the big name makes it sound much more intimidating than it really is.) You will still want a long integration time when measuring each voltage, to filter out noise.
4.) You should use multiple resistors for the divider. 50Meg and 50k will probably do the job. Switch into the 50k range using a relay.
5.) When using a shielded cable for the high-impedance node, instead of grounding the shield, you may consider using an actively-driven shield (by connecting the shield to the output of the buffer). You have to ensure the buffer remains stable, such as by connecting the shield through a 1k resistor, but this could help your measurement settle faster.
6.) Be careful with the high-impedance node--leakage currents can easily throw your measurement off. Guard-ring the node with the buffer's output wherever possible (same idea as actively-driven shield), as well as guard-ringing the 50k resistor and relay high-Z terminal. If you build your 50M resistor out of multiple 10M resistors, you can guard-ring the intermediate nodes with divided-down versions of the buffer output.

 

[FvM]

The term "lock-in amplifier" doesn't exactly match the application, but a complex impedance measurement with phase sensistive rectifier is perfectly suited. In addition a measurement circuit that is insensitive to cable capacitances, e.g. voltage source/current sense. That's how all AC impedance (LCR) meters work.


After compensating all external influences (cable, test fixture) you get a complex, frequency dependent impedance of your device under test. If you calculate a parallel resistance, it will be probably different from measured DC resistance and frequency dependent, too. So you need to select a measurement frequency that gives significant results and find out how it can be mapped to known DC resistance values.