Resistance of wire seems to go down when part of a circuit?

[jwallis]

Hi all,
I'm a software guy who wishes he knew more about hardware and have played with circuits some but am very frequently stumped or find myself saying "I don't know how this is possible, but this is indeed happening" which can get pretty frustrating.

So here's my latest thing:
I have a circuit in my 33 year old car that is getting very hot and I'm afraid will start a fire. It's the connection to the 450W radiator fan. At 12V, it seems like it should be carrying 37.5A, but the correct fuse, per the manual and everyone online, is 16A, which itself doesn't make any sense (comments welcome).

My actual question however was I'm thinking the age of the wires + the crusty and slightly burned GBC fuse holder are causing increased resistance. I was thinking this would cause higher current but as I write this I realize in theory the higher resistance should lower the current. Regardless, it's getting hot, so I wanted to measure the resistance of the components in the circuit. I measured parts of the circuit and then had (what I thought was) a bright idea to just measure the resistance from the battery to the fan, expecting it to be high. I thought I would then clean parts and replace old connectors, re-measure and see a lower resistance. I needed a length of wire so I measured its resistance first. Below is what I saw and why I am confused.

 

[Wireaddict]

Howdy, welcome to the forum! It appears that your fan is drawing too much current; how do you know the fan motor is rated at 450W? A 12V motor needs to be rated at less than 192W to keep 16A fuses from blowing (actually 80% of that or about 154W for reliability). Does the fan turn freely & also are the bearings loose so the armature can drag against the field poles? These conditions would cause overcurrent as would arcing brushes & shorted armature sections.

Signs of heat stress on wiring & the fuse holder also indicate likely overheating. As for the discrepancy between resistance readings, did you make both resistance checks on the same resistance scale, preferably R X 1 or lower or with an autoranginging meter?

Another thought, is the engine hot when you tested the wiring resistance? If not, I'd expect the relay contacts to be open between the fuse & the connector so the meter should read infinite or if the fan motor's in the circuit during the test & the relay contacts are open, current would flow from the chassis common, through the fan to the connector through the ohmmeter to the positive battery terminal & bypass the open relay contacts & fuse. In this case the meter may be pegging the left stop pin or otherwise giving an erroneous reading. Try disconnecting one of the battery terminals. Hope this helps.

 

[kellys_eye]

When measuring such low resistances you have to be aware of the effects of poor contact resistance - even pressing the test meter probes 'harder' can make a difference to your readings.

At high currents ALL connections need to be very clean and very secure. The fuse holder itself is often an area where contact resistance is neglected and trouble can be found. Dirty relay contacts, the in-line connector etc, ALL must be scrupulously clean.

If you have an accurate and reliable test meter you should test the fan current consumption using the shortest leads to a battery supply as you can get if only to eliminate the fan motor itself as being a problem.

Are you positive the fan is rated at 450W?? Seems an awful lot!

 

[BobK]

For ten feet of wire to have a resistance of 2 Ohms it would have to be 33 AWG.

Bob

 

[hevans1944]

For ten feet of wire to have a resistance of 2 Ohms it would have to be 33 AWG.

Bob

Literally "hair thin" in other words. And if you did manage to find some 33 AWG wire, it would likely be "magnet" wire, insulated and difficult to strip the insulation from.

 

[jwallis]

Thank you all so much for your replies and warm welcome, this seems like a great community!

Howdy,

I believe it's 450W per various part suppliers (here and here) and the part number still legible on the fan.
It turns pretty well - without wobble, noise or any reason to think the bearings are going out or that things are dragging, although it is at least 20 years old.

I have a digital autoranging meter.

The engine is not hot. I am jumping the temperature sensor to manually power/close the relay. It reads infinite, then when I close it, goes to 0 Ohms. The fan motor is disconnected for the test (step 3).

the effects of poor contact resistance...

Tell me about it, this is why I was hoping the culprit was the burned fuse holder to fuse connection. When I measure resistance from one side of the fuse holder to the other side with the fuse in, it jumps around some. I'm going to get a new fuse and clean the fuse holder very well in hopes this fixes it... I was performing this test as an experiment/test of my knowledge in hopes of seeing actual results: resistance measures high, clean the parts, resistance measures lower.

I am also trying (so far without luck) to pull the relay and check it for heat/corrosion damage.

I tried to measure the current to the battery assuming that it took < 10A because yeah, 450W is a lot...
Blew the fuse in my 10A meter. Hopefully didn't hurt any internals at the same time. That's a great tip though if I can find a meter than can handle higher currents.

Literally "hair thin"

The wire is old speaker wire, probably 18ga. I stripped insulation off the ends and they were a little green so I spread the strands and used a blade to scrape in one direction, flip over do the other side, then spread the strands laterally and did the same thing. Crap, I'm wrong though, there was more on the spool so it was greater than 10'.

That being said, when I just connect the two probe leads to each other it reads 1.5 Omhs.

 

[jwallis]

Ok, I just realized that my van has some customizations, and is using a later model fan than typical for the year. The documentation I could find for the later model years says to use a 30 A fuse, which is certainly a lot closer than 16A. Still doesn't quite work out with a 450W fan, and I plan to ask the vw community (thesamba.com). Thanks for making me think harder on that one...

I have been using this 16A fuse as it's how it came from the previous owner 5 years ago. Now this is a dumb question, but maybe the fact that the fuse is too small is causing it to get hot? I'd really hope it'd just blow! That fuse holder has always shown some signs of heat damage but I never gave much attention until I smelled it the other day. I'm wary of putting in a higher-rated fuse!

 

[kellys_eye]

Fuses protect the CABLING, not the equipment, so for a fan requiring 450W (37.5A) you would need 8awg for the main feed to/from the relay/fan.

The relay CONTROL wiring can, of course, be much smaller.

a 40A fuse would be appropriate for 8awg.

 

[jwallis]

Interesting, I had never thought of it that way but it makes sense. I was just told there should be a separate 50A strip fuse next to the relay, which is behind the dash and very hard to get to. Going to check them out now...

 

[hevans1944]

On measuring low-valued resistances with the ohmmeter function on your digital multi-meter: don't even think about doing this! When resistance values begin to approach one ohm or less, the resistance you are trying to measure becomes less (often much less) than the total resistance in series with the measurement test leads. This "external" resistance includes the resistance of the test leads, resistance of the connections to the multimeter, and resistance of the connections between the test leads and the unknown resistance. This "external" resistance, which has nothing to do with the small resistance you are trying to measure, becomes the major contributor to what is essentially an unreliable and usually erroneous resistance measurement. You have virtually NO control over the "external" resistance, nor any knowledge of what value it adds to your resistance measurement.

There are several ways to attempt to avoid this situation, which includes shorting the test leads together to record the "external resistance" prior to making a resistance measurement, and using the "relative" button on the multimeter to "zero" the meter with the test leads shorted together just prior to making the resistance measurement. However, as the resistance you are trying to measure becomes smaller and smaller, all the simple techniques ultimately fail. The only procedure that produces guaranteed results for any value of resistance you choose to measure is the Kelvin four-terminal resistance measurement. See figure below.

To make a resistance measurement of Rsubject, you must place an ammeter in series with the resistance you want to measure, apply a low voltage to that series connection, and record both the current read on the ammeter and the voltage developed across the resistance you are trying to measure. That measured voltage divided by the measured current equals the unknown resistance you are trying to measure. With a sensitive voltmeter and a stable current source, very low resistances can be accurately measured, despite values of Rwire that are several times larger than Rsubject. Note the Rwire values do not have to be equal, and generally never are.

The battery can be your car battery (engine off, so battery is not charging), and the Rwire resistance can be real resistances that serve to limit the current through Rsubject. You can use just one real resistor in series with the battery to represent both Rwire resistances. Value isn't critical, but it should be low enough to provide sufficient current through Rsubject to allow the voltage developed across Rsubject to be accurately measured. I would start with about 120 ohms (providing 100 mA into a shorted Rsubject) with a 12 V car battery. Then the voltmeter will measure Rsubject with a sensitivity of 100 milli-volts per ohm. Make sure you use another independent multimeter to measure the current through Rsubject.

Try the above method on your fuse holder block with fuse inserted. From what i have read in previous posts, the fuse holder should probably be replaced. Try measuring random lengths of wire of various diameters to get a "feel" for how very low resistances correspond to various wire diameters and wire materials. If you can find some (a discarded old toaster perhaps?) measure the resistance of various lengths and sizes of nichrome heating elements.

Good luck making a smooth transition from software weenie to hardware weenie. A hotdog that can be both is a rare treasure to work with.

Virtually all digital multimeters use a synthesized constant-current source to apply a constant current to the test leads during the ohmmeter function. The multimeter then measures the voltage this constant current source produces in the unknown resistance connected to the test leads. As mentioned earlier, however, this unknown resistance also includes "external resistance" from test lead wires and test lead connections over which you have no control and even less knowledge about. The only reliable measurement of unknown resistance must use the Kelvin four-terminal method. Not even a Wheatstone Bridge is better.

 

[jwallis]

Thank you, I wonder if you were a professional teacher. Most people who have advanced subject knowledge don't have patience for beginners, so I am very appreciative : ).

I am glad to hear you say such things about trying to measure small resistances with an amateur-grade meter. It has been so challenging learning, after the fact, the many gotchas like floating values on high impedance digital inputs, or that voltage dividers only working for low current applications. Software seems more deterministic, whereas with hardware, if you have a resistor, it might be "close enough" to what you need to do the job... and I never before appreciated the luxury of being able to completely break something and have it *not* go up in smoke.

It took me reading this about 6 times, but I think I understand now. I only have one meter, but maybe I could measure the current, then measure the voltage drop across Rsubject.

 

[kellys_eye]

A decent clamp ammeter (DC) is essential for currents over 10 or 20A. Whilst these types of devices aren't as commonly used as ordinary multimeters they are now quite a lot more affordable than they were a few years ago.

You can also purchase (or build) a milli-ohm meter designed, as the name suggests, for low ohms measurement.

http://www.scullcom.uk/category/projects/milliohm-meter/

But all this is speculative as most average users of tests equipment wouldn't or couldn't justify the expense for such rare/occasional usage.

As you suggest, measuring the volt drop across the load is perhaps easier method but accuracy may be an issue for generic test meters.

Knowing and measuring the 'problem' is probably less useful than simply fixing it - and knowing all the areas where problems arise and how to deal with them is more useful too.

As the OP already seems to understand, the crucial aspects of joint cleanliness and security makes all the difference and any problems can often be cured without knowing the actual 'reason' behind the heat issue - although most members on forums such as this are more curious for the technical reasons than the actual solution!

 

[Chemelec]

My Low Ohm Meter.
http://chemelec.com/Projects/Low-Ohm/Low-Ohm.htm

 

[hevans1944]

Thank you, I wonder if you were a professional teacher. ...

No, I was offered a "job" teaching basic electronics at Lowry AFB, Denver, CO after graduating from the full electronics course there in 1964. Turning that offer down was maybe the stupidest decision I had made up to that point in time, but I was anxious (at age 19) to "get into the field" and actually work on the weapons system I was trained to maintain and service. Something about 20mm Gatling guns with radar controls for aiming and tracking excited my interest.

... It has been so challenging learning, after the fact, the many gotchas like floating values on high impedance digital inputs, or that voltage dividers only working for low current applications. Software seems more deterministic, whereas with hardware, if you have a resistor, it might be "close enough" to what you need to do the job... and I never before appreciated the luxury of being able to completely break something and have it *not* go up in smoke. ...

Software is always deterministic, but so is hardware... sometimes in more subtle ways. I was attracted to software running on embedded processors (back then, minicomputers) in the late 1960s as a replacement for discrete hard-wired logic. It is very difficult to make changes in hard-wired logic, but I thought that software could allow a solution to that problem by presenting a lot of "what if" programmable alternatives. As you note, when software breaks there is usually no release of smoke and you just re-write the code and try again. This statement can and often does fail with embedded systems. Software errors can cause an embedded controller to go berserk and create monumental damage to the peripherals and the equipment the peripherals operate.

... It took me reading this about 6 times, but I think I understand now. I only have one meter, but maybe I could measure the current, then measure the voltage drop across Rsubject.

That will work. Since nothing much is changing during either of the two measurements, so you can use your one and only multimeter to measure the voltage drop across the series current-limiting resistor, whose value is presumably well-known, and calculate the current in Rsubject. Then you move one of the probes and adjust the range (or let the autoranging function do it) to measure the (much smaller) voltage drop across Rsubject. Calculate its resistance from the two measurements. Some gotchas: any decrease in battery voltage while making the voltage measurements will lead to errors. Any increase in value of the series current-limiting resistor, because of self-heating, while making voltage measurements will lead to errors. Be quick making the two voltage measurements, and disconnect the battery power between measurements. You should easily be able to determine very low resistance values to an accuracy of better than ±5%, which is all you need for troubleshooting analysis.

A decent clamp ammeter (DC) is essential for currents over 10 or 20A. Whilst these types of devices aren't as commonly used as ordinary multimeters they are now quite a lot more affordable than they were a few years ago.

I agree they are more affordable, now that Hall-Effect current sensors are so ubiquitous. I disagree that they are essential. Convenient, certainly. Essential, no. Current shunts are readily available with millivolt per ampere "burdens" or you can make your own by paralleling wire-wound precision power resistors. Of course there is the extra work of opening the circuit to insert the current shunt, so a clamp-on DC ammeter has the advantage for that. Be careful about "zeroing" them. The ones I have used tend to drift and are sensitive to orientation in the Earth's magnetic field. Still, given my druthers, I would rather use an accurate clamp-on DC ammeter than screw around trying to open a high-current circuit to wire in a shunt.

... But all this is speculative as most average users of tests equipment wouldn't or couldn't justify the expense for such rare/occasional usage.

Agreed. I have never felt the "urge" to construct a milli-ohm meter, because it is so easy to cobble up a test rig on the spot with sufficient accuracy for what I need.

... Knowing and measuring the 'problem' is probably less useful than simply fixing it - and knowing all the areas where problems arise and how to deal with them is more useful too. ...

The essence of problem solving is first knowing what problem you have. Sure, cleaning terminal connections is advisable, but knowing that a connection is faulty is even better.

... As the OP already seems to understand, the crucial aspects of joint cleanliness and security makes all the difference and any problems can often be cured without knowing the actual 'reason' behind the heat issue - although most members on forums such as this are more curious for the technical reasons than the actual solution!

Agreed. Sometimes having full understanding of the technical reasons for a fault isn't even practical, especially if that knowledge contributes nothing to a solution. For example, knowing that thermal fuses are subject to poor internal connections... for whatever reason... makes it prudent to either (1) replace the fuse with a known good fuse, or (2) measure the internal resistance of the fuse and its holder and replace it if too large a value is measured. If I had a box of spare fuses, I know which choice I would make first. OTOH, if the fuses cost upward of fifty bux each (as some high-speed fuses used to protect semiconductors do), maybe some careful resistance measurements are in order.