ΩHere in our plant we have a 1/8 DIN panel mount ohmmeter which is used to measure the resistance of long lengths >30,000ft of wire on spools. measurements are generally <2kΩ. The meter is a Newport Electronics meter and they are one of the only manufacturers I can find of a 1/8 DIN panel mount ohmmeter. The problem is that the measurement on the spools is very jumpy. When measuring a fixed resistor, there is no jumpiness. Handheld DMMs do not have this problem. My first thought was that the spool was acting as a giant inductor, picking up stray noise; but no HV is around the area. I tried countless different RC combinations to try and filter it out, but nothing seemed to work. My second thought was that; with the spool acting as an inductor, the meter would have to build up a magnetic field in the spool in order to get an accurate resistance measurement. I thought maybe the panel meter didn't have enough umph to push through the inductance. The panel meter measures with 4.5v @ 40μA when shorted, working out to .18mW. a Fluke meter measures with 1V @ 1mA = 1mW (5X as much as the panel meter), which is why (I assume) it doesn't have the same problem: seems to support my theory about the low power.
So, I set out to find a 1/8DIN panel mount ohmmeter with at least 1mW output at short. It's alot harder to find than you'd think. Actually, I never found one. So I decided it was time to find another solution. I got a 5W 500Ω potentiometer, a 24V DC power supply & a Red Lion IMP 0-50mA Process Meter. +24V goes through the pot (set to ~480Ω, to limit current to 50mA (24V @ 50mA = 1.2W, 6,667X as powerful as the installed meter)), to the test leads, to the process meter, back to ground. simple. The not-so-simple part is the nonlinear measurement curve created by adding the fixed 480Ω. The meter has 9 scaling points and I plotted the segments along the nonlinear curve and got the accuracy (according to my "ohm-ranger", not a calibrated standard) to within 1.5% from 0-2000ohms. It took me weeks, working in my spare time.
I just now, beaming with pride, went to install it, and saw the calibration sticker next to it. This whole time I thought it wasn't a cal item. All our cal is 3rd party. When the calibration guy comes next month, he's got no way to calibrate the thing. you have to manually change the scaling points which is an all day affair. That, and due to the linear segments falling along the curve, any error is not constant; so simply adjusting a trim pot might fix an error at one reading, but totally mess up the rest of the readings.
This is a terrible solution. a kludge. it cannot be. What can I do?
And to take a step back, is my logic even right? do I really need a high(er) power ohmmeter?

P.S. installing an anolog meter is ruled straight out.

 

Here's a drawing for you.

I thought about increasing the value of the current limiting resistance to decrease the amount of effect that the nonlinearity has on the measurement, but there's still going to be an issue with not bein able to calibrate it. If there were a process meter with the ability to enter and equation rather than scaling segments, this would be alot better.

 

The meter is a Newport Electronics meter and they are one of the only manufacturers I can find of a 1/8 DIN panel mount ohmmeter.

The meter is not a proper ohmmeter, it is intended to be use with a PT100 element only.

Handheld DMMs do not have this problem.

Then mounts one up and uses it instead. The 9V dry battery are not expensive and will work for months.

 

The meter is not a proper ohmmeter, it is intended to be use with a PT100 element only.

Yes, it looks like the meter I linked to is not the one we have installed. I can't find the proper one on their site now, but the specs are the same: 4.5V @ 40μA.

Then mounts one up and uses it instead. The 9V dry battery are not expensive and will work for months.

You know I joked about that with a coworker. I never seriously considered it until you mentioned it. I don't see any practical reason why not. Only the aesthetic factor of having a fluke meter bolted down to a panel might not go over well. If they don't like it then they are welcome to pick up where I left off trying to find a legitimate panel mount meter. Thanks for the idea!

 

Only the aesthetic factor of having a fluke meter bolted down to a panel might not go over well.

Mounts the meter "behind" the panel with a perplex window cut out for viewing the display.

 

Even better, buy a 9V battery clip and attach it to the battery clip inside, run wires outside and use a 9V power supply. Worked for me in a pinch.

 

Buy or build a constant current source. An MCL1304 is a 4ma CC diode. Put you 24v PS in series with the CC diode and your unknown resistance. Assuming a 4ma diode, and a 2,000 coil, E = IR = 0.004 * 2,000 = 8 volts. Because current is constant, E will be directly proportional to R. Now, you will no longer need to find an ohmeter, you will be looking for a voltmeter instead. Much easier to find, like $7.00 at All Electronics.

 

So, that would give me a meter with a max reading of 6000ohms, which is fine and 96mW power which should definately be plenty. I see the tolerance is +/- .6mA, or 15%; would that be fixed? If I get the diode, and test it, and it's putting out 4.4ma, would it always put out 4.4mA? Or would it put out 4.4mA sometimes and other times 3.5mA. If It's fixed then I can always adjust for that with scaling, but if it's continuously variable then it won't be a very accurate meter. Thank you for this good idea.

BTW, nobody has commented yet on whether or not my logic is even sound regarding "low power meter" vs. "high power meter". Does "charging the magnetic field" of a spool of wire really require more power to obtain a steady reading than it would for a fixed resistor? I've been going on this theory for a while now and it's kinda eating at me as to whether it's correct or not.

 

BTW, nobody has commented yet on whether or not my logic is even sound regarding "low power meter" vs. "high power meter". Does "charging the magnetic field" of a spool of wire really require more power to obtain a steady reading than it would for a fixed resistor? I've been going on this theory for a while now and it's kinda eating at me as to whether it's correct or not.

An ohmmeter usually supplies a constant DC current to the DUT (device under test.) Because V=IR this creates a voltage drop which the meter measures. Even if there is such a massive inductance (which I doubt), it will simply cause the resistance to start at a high value and settle at the correct value. Even with a 1H (massive) inductor, I can't imagine this taking more than a few seconds. I'd look for a more obvious problem: starting with the meter; is it known to be good (have you tested it), is the power supply or battery okay, are the leads good, and so on?

 

An ohmmeter usually supplies a constant DC current to the DUT (device under test.) Because V=IR this creates a voltage drop which the meter measures. Even if there is such a massive inductance (which I doubt), it will simply cause the resistance to start at a high value and settle at the correct value. Even with a 1H (massive) inductor, I can't imagine this taking more than a few seconds. I'd look for a more obvious problem: starting with the meter; is it known to be good (have you tested it), is the power supply or battery okay, are the leads good, and so on?

This panel mount meter is one of two identical meters we have and they both exhibit identical symptoms. I verified them against a calibrated fluke DMM on wide range of fixed value resistors, which they measure just fine, no jumpiness. steady with fixed resistors, jumpy with spools of wire. They have internal DC power supply, powered by 110VAC.
So do you think the spools are acting like antennae, picking up EMF from elsewhere? if so, why does the fluke not have the same problem? Or is it something else I've not thought of?

 

Hi Strantor,

Is it poosible that the panel meter, which has a grounded frame and is power from an AC source, has a limited common mode input voltage range that is being exceeded? Noise that is being pickup by the coil of wire and is common to both wires coming from the coil can cause irratic readings.

This is difficult to filter out with an AC powered meter that has a high input impedance. Battery powered meters don't have a problem as they are naturally isolated from ground.

To confirm this idea, can you isolate the meter from ground and try it, and if that doesn't work then also power it from a battery via an inverter to get 110VAC.

Regards,
Ifixit

 

The MCL1304 diode will not arbitrarily or randomly change its particular value. The span of the panel meter can be tweaked so it will directly in ohms. If that is not acceptable, look at this:

http://cds.linear.com/docs/Datasheet/134sfc.pdf

Pay particular attention to page 7, last circuit on the right. With only a few parts a deadly accurate temperature compensated current source can be made.

 

The MCL1304 diode will not arbitrarily or randomly change its particular value. The span of the panel meter can be tweaked so it will directly in ohms. If that is not acceptable, look at this:

http://cds.linear.com/docs/Datasheet/134sfc.pdf

Pay particular attention to page 7, last circuit on the right. With only a few parts a deadly accurate temperature compensated current source can be made.

Jaguarjoe,
Why did you recommend this specific component? They are not selling it direct from their website. I was trying to find a replacement on mouser.com or newark.com but I can't find anything like it. All the current limiting resistors are in the DO-XX package and I'm not sure they are the same thing. none go as high at 30V. I know you specified that part because one or more specs that you know I need, but what are those critical specs? What should I be looking for? I was also thinking that something around 8mA would be better. that would give me a 0-3000Ω range which would make the measured resistance more accurate.
Thanks!

 

Nobodies got that 134 part but don't despair. Here is an LT3092:

http://cds.linear.com/docs/Datasheet/3092fb.pdf

Its available from Digikey for about $3, in stock, able to buy in single quantity.

It doesn't require a special temperature compensating circuit, and it can provide the current you need. It'll go up to 200ma but across a 2k coil, you'd get 400 volts! You can still set it for 8 or 10 or whatever milliamps by selecting the value of 2 resistors. The formula for them is given on pagres7 & 8 of this datasheet. The chip needs about 1.5 volts of overhead so you'll have to subtract that from the 24. Just to be safe, I'd use 2 volts. You can use a higher voltage supply but don't exceed the max rating of the chip which I think is 40 volts.

The only hitch is that it comes in the small SOT23 package, but that's not insurmountable. Breakout boards for that size chip are available from places like Sparkfun:

http://www.sparkfun.com/products/717

 

Nobodies got that 134 part but don't despair. Here is an LT3092:

http://cds.linear.com/docs/Datasheet/3092fb.pdf

Its available from Digikey for about $3, in stock, able to buy in single quantity.

It doesn't require a special temperature compensating circuit, and it can provide the current you need. It'll go up to 200ma but across a 2k coil, you'd get 400 volts! You can still set it for 8 or 10 or whatever milliamps by selecting the value of 2 resistors. The formula for them is given on pagres7 & 8 of this datasheet. The chip needs about 1.5 volts of overhead so you'll have to subtract that from the 24. Just to be safe, I'd use 2 volts. You can use a higher voltage supply but don't exceed the max rating of the chip which I think is 40 volts.

The only hitch is that it comes in the small SOT23 package, but that's not insurmountable. Breakout boards for that size chip are available from places like Sparkfun:

http://www.sparkfun.com/products/717

Looks good, thank you! are you saying that it will output more voltage than it's input? I had been assuming that it would increase voltage to maintain a specific current up until it reached the input voltage, but the drawing shows an op-amp (?) inside. If I use a 24V p/s will it really go up to 400V? I didn't see anything in the datasheet about the max voltage it will put out.

 

Maybe the panel mount meter was doing something clever like only powering the ohmeter for a short time for each reading. Easy enough to check with a scope.

 

Looks good, thank you! are you saying that it will output more voltage than it's input? I had been assuming that it would increase voltage to maintain a specific current up until it reached the input voltage, but the drawing shows an op-amp (?) inside. If I use a 24V p/s will it really go up to 400V? I didn't see anything in the datasheet about the max voltage it will put out.

No, it won't put out more voltage than what's put in. Real world current sources have what's called compliance. That's the driving voltage behind the current being put out. For you, it would be 24 volts minus the 2 volts for the chip. You can go up to the maximum input voltage that the chip allows, 40 volts. Ideal current sources have infinite compliance. The drawing of the chip's innards is only a block diagram, don't it literally. Even the schematics for most chips have magical components- like transistors with 3 emiters.

 

Maybe the panel mount meter was doing something clever like only powering the ohmeter for a short time for each reading. Easy enough to check with a scope.

Could be; like a pulse/ sample. If it were pulsing then I could see inductance having an effect on the reading