I put together a wheatstone bridge to measure conductivity/resistance of semisolids and gels. I found that trying to measure conductance directly with DC led to weird inconsistent effects due to ion migration (I think), and so I started using AC, and I was getting pretty good numbers.

My understanding of the mechanics of the wheatstone bridge for these kinds of measurements is that it is essentially a "balancing" act for the current running through the galvanometer. Tweaking the rheostat too far in either direction allows current to flow through because of Kirchoff's and Ohm's Laws. The schematic of my device is posted below.

Here's the weird part: My bridge is unidirectional. I start with the potentiometer (used as a rheostat) at zero. I see a current register on the galvanometer. I can zero it out carefully by increasing the resistance of the potentiometer, but if I blow a little past the point at which the resistances are equal... nothing happens. Galvanometer continues to read zero. So current doesn't return as the potentiometer resistance gets arbitrarily large, unless I dial it back to the point before it went to zero. I've tested several known resistances, finding the zero point in one direction, and the device gives me an accurate resistance reading. I just have absolutely no clue why it's asymmetrical this way.

Explanation of the Schematic and Circuit:
I used an 100V EL-Wire inverter as my AC source rather than run a current limited line voltage or use a bench power supply because it's very important that the circuit be portable. But, EL-wire inverters are very unstable in low or no load conditions and will fail with just pure resistance. I learned this by making magic smoke. I looked up other methods of pure or semi-pure sine wave inversion from DC, and since most that I found were basically variants of 555 timer astable multivibrators, I decided to keep using EL-wire inverters, since that's pretty much what they are, but I decided to also put a simulated load on there since a capacitative and resistive load is necessary to keep the frequency and peak voltage within the tolerance of the device. At first I tried EL-wire itself, but it seemed to severely limit my current and the galvanometer always read zero. I then used a 0.022μF capacitor and a 1500Ω resistor to limit current and keep the magic smoke inside the inverter and to stabilize the output, and that seems to work fine as a sort of "ballast" stabilizing the inverter, but I wonder if it could be causing the phenomenon... it shouldn't. The capacitor causes current to lead voltage but this should make no difference to the application, since the current is still there.

Thanks in advance for the help.

 

I don't understand how you are sensing AC.
Isn't that a DC galvanometer?

 

I don't understand how you are sensing AC.
Isn't that a DC galvanometer?

Is the circled "G" only a symbol for a DC galvanometer? If so, that's not what I meant to imply. I was using an AC galvanometer for this application. I believe DC would have read zero for a digital meter or the needle would just vibrate for analogue. In any case, the "galvanometer" is really just my multimeter, set to AC current. I get a non-zero, stable reading until I zero it.

 

I get a non-zero, stable reading until I zero it.

What is "it" in your sentence? Is it the meter (pressing the "delta" button on the meter to zero the meter's output, or are you balancing the bridge so the meter reads zero?

 

ou may want to use the "alternate polarity"method. http://www.tek.com/blog/making-ultra-high-resistivity-measurements

In some materials the applied voltage affects the resistance.

 

What is "it" in your sentence? Is it the meter (pressing the "delta" button on the meter to zero the meter's output, or are you balancing the bridge so the meter reads zero?

The latter, I balance the bridge, I don't zero the meter. Sorry that I was unclear.

ou may want to use the "alternate polarity"method. http://www.tek.com/blog/making-ultra-high-resistivity-measurements

In some materials the applied voltage affects the resistance.

Oooh thanks, bookmarked, I'll take a look at that. Always interested to learn new things. If you can't already tell, I'm a chemist, not an electrical engineer. I do know and worry about electrolytic effects and issues that might make the sample I'm testing non-ohmic, but for what I'm working with, the resistance is pretty low, on the order of hundreds of kiloohms, give or take a megaohm between two wildly different samples.

Here's the other thing, and I'm sorry I neglected to mention it, because it could be huge, but I was trying to keep it brief. The device did work as expected for a little while, in addition to giving expected results, but then stopped after a couple of tests. (This was when I was using the EL wire as a "ballast.") When I switched to the capacitor and current limiting resistor is when this effect sprang up. I have, incidentally, checked the wiring twenty bajillion or so time to make sure that it's wired as shown in the diagram. The thing that's so mystifying to me is that it does work right now, when I test for known resistances using your basic carbon-film resistors with negligible impedances. It just doesn't work in a way that makes sense. Just to clarify, since I'm not sure that I was clear, this is the flow of events:

1. I put a known carbon film resistor (from few hundred ohms to a couple of Kohms) in place of a sample (here I'm just testing if the device works at all).
2. I turn on the bridge, and I see a non-zero current value on the ammeter.
3. I start turning potentiometer knobs, increasing the resistance until the current reads zero.
4. If I continue turning the potentiometers and increasing resistance, it continues to read zero, even if I climb up to the maximum of 1.6 Mohms.
5. If I dial the the resistance back to where it first read zero, it's the same spot, and if I continue to reduce the resistance, the current reading climbs up from zero.
6. I measure the resistance of the point where the ammeter first read zero current, and it agrees with the nominal (and independently measured with my multimeter) value of the resistor (give or take, obviously).
7. I try a different resistor, same results.
8. I have doubts about my basic scientific comepetence, followed by an existential crisis
9. I post about it on All About Circuits, and hope I'm not missing something incredibly obvious. (Happens way too often.)

 

So, here's the solution to my problem, and why the bridge was behaving wonky: It turns out that when the ammeter is set mA, raising the resistance often drops the current to below the detection limits, and my max resistance isn't high enough to notice this. If you use μA, it gets very touchy and slight tweaking of the potentiometers throws it out of balance, but it works as expected. I guess I didn't think I had limited the current that significantly, but I was using a different current limiting resistor when I was putting everything together, and I guess I just forgot to do the math for the new one and assumed I'd still be in the mA range if below the max resistance of my series rheostat, and the ammeter I had at the time didn't detect μA: Long story involving equipment changes, confusion, and general stupidity.

But otherwise it's a great success. It's incredibly sensitive, and I suspect which even more sensitive ammeters I could get hair-width resistance accuracy. Though I also discovered that I also need much better pots. In any case, this one is solved, though I'm incredibly annoyed with myself right now. Basically every single question I've brought to AAC I've solved myself, and it's usually due to me missing something.

 

Just tel ling someone about it, usually works. I did this numerous time when helping someone debug code. I'd ask them to tell me about how there program works, I usually wasn't paying attention, but generally they would find the problem while telling me about the code.