Clifton Laboratories 7236 Clifton Road  Clifton VA 20124 tel: (703) 830 0368 fax: (703) 830 0711

E-mail: Jack.Smith@cliftonlaboratories.com
 

 

To search within the Clifton Laboratories site, enter your search term below.
 

 


Home
Up
Updates
Current Products
Prior Products - no longer available
Documents
Book
Software Updates
Softrock Lite 6.2
Adventures in Electronics and Radio
Elecraft K2 and K3 Transceivers

 

Carbon Composition Resistors - Not Like Fine Wine

(A greatly expanded version of this web page appears in the American  Radio  Relay League's Mar/Apr 2008 QEX magazine.)

Perhaps the most common trait of ham operators is that of a pack rat. It's hard to throw away something that may be used some day. When searching in the garage today for a hose nozzle, I ran across a coffee can with a couple hundred 1/2 watt International Resistance Co., carbon composition resistors. These resistors were new around 1960, as I recall acquiring them not long after that date, and are unused. Hence, they should not have drifted due to conditions inside equipment. However, the parts have been stored in the garage for 20 years, so they are exposed to Northern Virginia's high humidity and temperature.

I measured a selection of these resistors today and found that unlike wine, carbon composition resistors do not get better with age.

Seven 150K, 10% carbon composition resistors, still in the original package.
Reverse side of "Grip Reel" holder.

I used a General Radio 1658 Digibridge to measure the part values, at 1 KHz test frequency. I also compared a few samples with other ohmmeters (two Hewlett Packard 4-wire ohmmeters) and found agreement in the 0.2% range or better, so I believe the data is accurate.

The figure below shows the  results graphically. The blue arrows indicate the tolerance limits for the particular resistor, whilst the red circles show the mean value. If the resistor is within its tolerance, the red circle will be between the two blue arrows. The value on the vertical axis is the nominal resistance.

For example, consider the 1 MΩ resistor (1000 KΩ), the topmost entry. The tolerance arrows are at 95% and 105%, since this is a 5% part. The red circle is at approximately 109%, indicating measured mean of all similar value parts is 9% high. Since 9% high is outside the Ī5% tolerance, the red circle is not within the blue arrow tolerance bounds.

Looking at the data in this fashion, rather than a table, allows us to see relationships that are difficult to tease out of raw numbers. For one thing, the eye is immediately drawn to the fact that all the parts measured high, between 105% and 112% of nominal value.  In other words, they have all aged in the same direction, and at about the same amount. Further, the data is ordered by increasing nominal value. There is little correlation seen between the error and nominal value, i.e., it seems that parts > 500 KΩ have about the same measured value distribution as parts < 500 KΩ. However, closer examination shows there may, and emphasis on may, be a correlation with higher value parts having a greater increase. There isn't enough data to be sure of this, one way or the other. It might seem reasonable that higher resistance values would be more affected by humidity, but if that's the case, the data is not overwhelming in either direction.

One additional point is that there seems to be little difference between 5% and 10% partsóboth seem to have aged about the same amount. This suggests there was no difference in construction, but rather parts were sorted by value into 5% and 10% groups. (If so, this means that 10% resistors will likely be devoid of parts around the nominal value, as they were removed and sold as 5% tolerance parts.)


Don't believe me that the plot allows you to see relationships easier than a table? Here's the table, so you can judge for yourself.
 
Nominal KΩ Tolerance  Number  Measured KΩ High Limit KΩ Low Limit KΩ
3.3 10%             6 3.475          3.63 2.97
3.9 10%             1 4.238          4.29 3.51
5.6 10%             4 5.888          6.16 5.04
6.8 10%             5 7.403          7.48 6.12
8.2 5%             2 8.723          8.61 7.79
13 5%             7 14.05        13.65 12.35
36 5%             4 38.22        37.80 34.2
39 5%           12 42.93        40.95 37.05
56 10%             4 60.41        61.60 50.4
82 5%             6 89.1        86.10 77.9
120 10%             7 127.68      132.00 108
150 10%             7 162.68      165.00 135
180 10%             3 193.12      198.00 162
220 10%             4 238.3      242.00 198
330 10%             3 352.3      363.00 297
390 10%             8 417.48      429.00 351
560 10%             6 616.97      616.00 504
680 10%             3 757.8      748.00 612
820 10%           10 884.93      902.00 738
1000 5%             4 1090.9    1,050.00 950

 

Of course, means do not tell the entire story. A more thorough analysis would look at the standard deviation, for example, to see whether the means were influenced by a few parts being outliers, i.e., with major value shifts.

I did not compute the standard deviation, because I entered the data by hand and it was too burdensome to compute the higher order statistical parameters. However, I did discard a couple of resistors that were clearly bad, e.g, a 3.3 KΩ part measured 4.2 KΩ.

Let's look, however, at the distribution pattern for 39 KΩ and 820 KΩ parts. I selected these because there were enough parts (12 and 10, respectively) to provide a moderately meaningful distribution.
 

The data shows that all measurements are on the high side of nominal value, and that in both samples there are outliers, although the 820 KΩ outlier at 1.065 MΩ is extreme.

The 39 KΩ is a 5% part, so the highest in-tolerance value allowed is 40.95 KΩ, and none of the parts measured within that tolerance. The 820 KΩ resistor is a 10% part, so the highest in-tolerance value is 902 KΩ, and we see seven of the 10 parts were in tolerance, with three out.

None of these resistors have been used in equipment, so the changed values are due to normal aging. (At least I assume the parts were in tolerance when new. They are manufactured by one of the major component houses in business at the time and are physically appear new.) It's well known that carbon composition resistors drift up in value with age, and that drift is accelerated by heat.

The lesson to be learned from this is that some partsócarbon composition resistors in particularódo not age well, and if you decide to keep these around the shack for historical restoration, by all means check their values before installing.

I'll also mention in passing that I have a variety of 1% metal film resistors of similar vintage and all check within the 1% tolerance. Carbon composition resistors are notoriously unstable, and the modern carbon film and metal film parts are much better in terms of long term stability, temperature stability and being within tolerance. It's rare, at least in my experience, to measure a 5% tolerance carbon film resistor and find it more than 2% or 3% from nominal value.

 
01 June 2009 - 1% Metal Film versus 5% carbon film 10 Ohm Resistors

I recently purchased several hundred new 10 ohm ľ watt resistors, in both 5% carbon film and 1% metal film types.

As a matter of curiosity, I sampled 25 from each lot and measured the resistance values with a General Radio 1658 Digibridge, with the results reflected in the plot below.

The black and white "retro" look to the graph is a consequence of my spending part of the last two weeks reading copies of the Bell System Technical Journal from the early 1920's through the mid 1950's. The plot is my attempt at duplicating the style of draftsman produced plots, lettered with Leroy lettering tools. (I still have a Leroy lettering  tool in the closet along with several  templates.)

The manufacturing process for both carbon film and metal film resistors results in a resistor blank lower than the  target value. It is then cut, generally in a spiral, with either a diamond  tipped tool or a laser until it reaches the target value. As a matter of manufacturing efficiency, of course, some savings are achievable if the trim stops when a part is within the tolerance band, rather than carrying through until the part is produced with zero error. Hence, it is rare to find a modern carbon film or metal film part at or above the nominal value, as is reflected in the measured data.

The worst case parts are at -0.73% (1% tolerance) and -2.69% (5% tolerance) with the mean values being closer than than 0.5% (1% tolerance) and 1.9% (5% tolerance) to the nominal value.