Originally Posted by arnyk
It is true that some may lack the technical discernment required to appreciate technical work at a certain high level.
Thanks for the answer Arny. I am puzzled by this sarcastic remark though. The devil is in the detail when it comes to measurement accuracy. As I will explain, the circuit "at high level" is trivial. And your post lacks much of the detail necessary to ascertain the accuracy of your measurements.
The graph is self-evident to most people who are aware of the technical issues.
Nothing of the sort. There was a professional audio interface involved in the measurement, but since it is well known that those devices are not calibrated as voltmeters, a regular laboratory Fluke voltmeter and precision resistors were used for that purpose. I also have long had two Hewlett Packard AC voltmeters on hand for the purpose of confirming the results.
Passive, following good measurement practice.
Um, this was done 14 or so years ago. No idea. My lab's inventory includes several complete series of of precision resistors of various wattages. I picked something that was appropriate, no doubt.
I appreciate that but don't you have the fixture still Arny? You asked me to send you the wires so I assumed you have the fixture and could easily look up the resistor spec.
Let's see what we can tell from the general comments you are making. The typical DIY circuit for measuring small resistances such as that of our wire (measured in milliohms or thousands of an ohm) is like this:
RL is our small resistance or the 12 gauge wire in this case. As the graph explains, if we know the current going through our load and the voltage across it, then we can simply use Ohms law to compute our load resistor: R = Voltage/Current.
Your fixture of course is a bit different. You don't have the battery on the left, nor does a sound card measures current. The former you did without by using the sound card output as the voltage source. To measure current, you use a "sense resistor" and measure its voltage. If one knows the resistor then dividing the voltage across from it by that value gives us the current. Which we can then plug into above formula to get our load resistance. This is what that looks like from this link which shows how you can build one with a voltmeter: http://www.instructables.com/id/Simple-Low-Resistance-Tester-Milliohmmeter/
R1 is the sense resistor. Hopefully now everyone can follow along. I asked you how much current was going through the sense resistor and you said:
That is problematic when it comes to accuracy. If very little current is going through our 12 gauge wire and it has very little resistance itself and we multiply those two together, we get a well, very, very small voltage.
Measuring small voltages is prone to accuracy error. Professional milliohm meters us as much as 20 amps with 2 amps being the typical value to make sure sufficient voltage develops across the load to measure its voltage accurately. Your sound card is designed to sample a fixed voltage level of 1.5 to 5 volts. If you reduce its maximum to say, 1 millivolt, you throw away 99.9% of its range and relying on the accuracy of 0.1% of its A/D.
Since you used a passive circuit, just the resistor with the sound card being the voltage supply, you had no choice but to pick a large resistor and hence, small current. This is why the "right" answer to my question would have been an active circuit with its own power source and preferably op-amps on output and input to buffer the sound card and not let its internal impedance impact the test.
Note that having a multimeter would not have helped you calibrate this device as your multimeter would not have been capable of measuring sub 1 ohm correctly or else, you wouldn't need to build this circuit. The accuracy of meters once you get below 10 ohms is poor as the lead resistance and contact can be in the order of 0.5 ohm.
Probably around 1%.
Unless you can provide the full details and analysis of how many effective bits you had to measure that low voltage, you can't state the accuracy.
That's one of the advantages of using an audio interface rather than a meter. I had ready access to a wide band FFT of the voltages applied and measured. They were very clean.
Your used a sampling rate of 96 Khz meaning your bandwidth was just 48 Khz. That doesn't match the definition of any engineer's "wide bandwidth."
As to it being clean here is your transfer function graph:
I see that you used 20 run averaging. You data must have been pretty volatile to require that much averaging. And where did you plug in the sense resistor value in SpectraLab?
It is true that for some jealous people, no good deed goes unpunished.
I had an old school education in testing and measurement. Give me a single precision resistor or precision voltage source and I can follow my formal training to leverage it into a wide variety of accurate measurements. Remember, my first electronics courses were based on vacuum tubes, with solid state added because it seemed to be the coming thing! ;-)
I don't understand the reference to "no good deed." You are saying that you were helping me say that cheap wires are under spec? If so, I don't need that kind of support. I am interested in the correct data. If it turns out that low cost wire does the job I would be as happy as anyone else. I am not after any certain outcome.
As to the second comment, that concerns me again. You say you want a voltage source and a resistor. Your sound card is not a proper voltage source as it is not designed to drive very small loads. Its output impedance is likely 50 to 100 ohms making it a very poor "voltage source." That limits your choice of that sense resistor.
As to needing precision resistor, that is wrong Arny. No precision is required at all! Why? Because we can simply measure its resistance with our multimeter. Unlike our load, the sense resistor is a much higher value like the 220 ohm one in the above circuit. Our standard multimeter can measure the resistance of a 220 ohm resistor with very high accuracy. And that is all we care about. Whether the manufacturer rates the resistor as 1% or 10%, matters not because that simply says how much variation can be in the value of the resistor from its advertised value. We can measure it and know what it is. If it is 230 or 222 ohms matters not. As long as we know its value we can divide the voltage across it by the measured resistance and get our current.
There is one potential problem though. If you push too much current through that resistor as to get a higher voltage across our load (and hence increase the accuracy of that voltage measurement), the resistor will heat up and that changes its value from what we measured with our multimeter. This is why I asked you about the temperature coefficient of the resistor.
It is for these reasons and others I have not mentioned that you don't want to build your own fixture unless you are aware of all the things that can throw off its accuracy. Measuring it with a sound card adds additional problems. This is why I am using a proper milliohm meter. Using PC for measurements is fine as a hobbyist but not as authoritative data to put forward.