SO.... What Capacitance are your Resistors...?

[jaunty]

Its *TOUGH* if you don't trust your test equipment *OR* any of your resistors.  However if you put one of your 1M supposedly 400nF resistors in series with an ordinary NE-2 style neon bulb, and power it from between 150V & 200V DC (which doesn't need to be well regulated), with 47uF decoupling across the supply, *IF* your measurements are correct, the flash rate can be expected to be between 1 and 10 Hz.  (it depends on the RC time constant, the supply voltage and the breakdown and holding voltages of the NE-2 bulb)   If, as we expect, you see a steady glow, the resistor's not got much capacitance.   As a final check, try adding a 470nF 200V cap across the 1M resistor and watch it flash.

i'm EXTREMELY confident with my test equipment.  it's you who isn't confident. Also they aren't 1M resistors - ALL of them do that - all the chinese ones. The european and american ones DO NOT do that. As I already said. it's the difference between these results that interests me. And something nobody seems to be willing to address...

[edavid]

i'm EXTREMELY confident with my test equipment.  it's you who isn't confident. Also they aren't 1M resistors - ALL of them do that - all the chinese ones. The european and american ones DO NOT do that. As I already said. it's the difference between these results that interests me. And something nobody seems to be willing to address...

Perhaps a difference in inductance.  Why don't you make a voltage divider with your "400nF resistor" and a 400nF capacitor?  If you plot the voltage ratio over a range of frequencies, you'll get a good idea what's going on.

[jaunty]

Garbage in, garbage out. I just "measured" a bunch of resistors using my cheap ProsKit MT-5210 LCR meter, which is reliable when measuring _capacitors_ on the Capacitance range. Depending on the resistor type (carbon comp, metal film, wirewound, etc.) and the Range setting on the meter, I too can obtain "capacitance" readings in the 0.4 uF or even greater range. What does this prove? Garbage in, garbage out. It proves that test equipment, when not used properly, will give misleading readings.

I haven't tried doing the same thing with my Fluke 83 or 87 multimeters using their Capacitance settings .... what would be the point?

Sure, if you want to crosscheck and determine if the "400nF" reading for a _resistor_ is accurate, just build some simple oscillator circuit with other components of known values and measure the frequency. I think you'll find that the "400nF" reading is an error, and the lesson to be learned is .... don't use a DMM on capacitance range to measure resistors!

Sure- but that's not the issue. Perhaps you misread. Perhaps i'm not being articulate enough (?)
The issue WAS - WHY does this particular brand read as having a 'capacitance' (however you want to interpret that - it's immaterial) that is 100x higher than the reference resistors. And what does that mean for a circuit?

So - well what if you DID use your fluke 87 and get the same results? what then? (The TX1 is at LEAST as good - probably better than a Fluke 87)

[Jay_Diddy_B]

Hi,

I just tried a couple of my meters.

First a Fluke 87

I took at 1 Meg 1/4 metal film resistor and a 0.47uF film capacitor.

I measured them individually, in series and in parallel.

If I measure just the capacitor I get a reading of 450nF which is close enough.

If I measure the 1 Meg resistor I get a reading of 140nF

In parallel I get 0.6uF and in series I get around 0.1uF

All of which seems about right for series and parallel capacitors.

However, the Fluke 87 Manual says:



Fluke 289

I measured the capacitor first. I got a reading of 452nF.

I tried to measure the resistor on the capacitance range, I got an unstable reading.

I tried the series combination. The Fluke 289 reads the correct value of capacitance, even with a 1 Meg resistor in series.


This just shows, that some meters don't work with a parallel resistor when measuring capacitance.

I will try the same components on an LCR meter, HP4274A, but it is too late....

Regards,

Jay_Diddy_B

[Ian.M]

@Jaunty:  Why the hostility?  I provided an extremely simple method of proving whether or not your resistors have significant parallel capacitance.  If they don't, your capmeter (you said its a Tek TX1 multimeter on nF range) is quite simply lying to you because its measurement method cannot tolerate parallel resistance (i.e leakage current).   

Why a 100:1 difference from your other resistors?  Well that's going to be down to your capmeter's autoranging being fooled slightly differently,

Is it going to make a difference in your circuit?  Maybe.  If its very high impedance, high gain OPAMP circuit, or is a HF or higher RF circuit, the inductance of spiral cut film resistors may become an issue,

However you haven't told us what trouble you are having with your circuit.,  In fact all we know is: "(the circuit in question is a controller board with lots of op amps, some 4011BEs and an 8048 processor...)"

[jaunty]

@Jaunty:  Why the hostility?  I provided an extremely simple method of proving whether or not your resistors have significant parallel capacitance.  If they don't, your capmeter (you said its a Tek TX1 multimeter on nF range) is quite simply lying to you because its measurement method cannot tolerate parallel resistance (i.e leakage current).   

Why a 100:1 difference from your other resistors?  Well that's going to be down to your capmeter's autoranging being fooled slightly differently,

Is it going to make a difference in your circuit?  Maybe.  If its very high impedance, high gain OPAMP circuit, or is a HF or higher RF circuit, the inductance of spiral cut film resistors may become an issue,

However you haven't told us what trouble you are having with your circuit.,  In fact all we know is: "(the circuit in question is a controller board with lots of op amps, some 4011BEs and an 8048 processor...)"

no hostility meant at ALL Ian -  you were the one suggesting that I didn't trust my equipment - it's not at all true.i most certainly do. I don't think there is any reason NOT to. It's the best and most consistent meter i could find ... understanding or interpreting results is another thing altogether though. as you know....

the big question for me is "are these chinese resistors actually creating a problem for me?" and creating some kind of cumulative issue in the circuit itself.... and so this spurious issue with this 'capacitance' or whatever it translates to in real world terms seemed to me something worth investigating - because I'd never seen or considered it before.

As for the issues in particular with a circuit - well I won't trouble anyone here with that - since I have another forum I go to for help - since its' a replica of a board from a synthesizer from the early 80s... so it's a good place to source such problems - i was just at my wit's end and thought maybe this weird behaviour of the resistors might bear exploration (hence me being here)... that's about it...

[Ian.M]

A typical 1/4W axial leaded film resistor has a body length L of 6.5mm, diameter D of 2.3mm, with metal end caps each of up to 25% of the total body length. 

This can be approximately modelled as a parallel plate capacitor consisting of two plates of area (Pi/4)(D^2) = 4.15 mm^2 or 4.15E-6 m^2, separated by a dielectric of thickness D/2 = 3.25mm or 3.25E-3 m.

Alumina ceramic, which is commonly used as the substrate for film resistor bodies has a dielectric constant of 9 - 10. 

Z5U ceramic (as used in high value ceramic caps) has a dielectric constant (relative permittivity) of up to 18000.  Its the highest dielectric constant ceramic that's commonly available.

You'll find the formulae for calculating the capacitance here: http://www.rfcafe.com/references/electrical/capacitance.htm
Here's an online calculator for it that will accept mm and mm^2 and output in pF:
http://www.calctool.org/CALC/eng/electronics/parallel_plate

Plug in the numbers and you will see that a 1/4W thin film on alumina resistor should have a stray capacitance of no more than 0.11pF.  If the manufacturer FUBARed it and used the much more expensive Z5U ceramic for the substrate, it *MIGHT* have as much as 200pF of capacitance.  In practice, the ends aren't parallel plates, but little cups, so most of their surface is further apart, and the substrate diameter is significantly less than the body diameter so the capacitance will be significantly less.

If your resistors are different dimensions, plug in the numbers for them.  You wont get more than a few pF for any normal body ceramic and are unlikely to get into the nF range even if you used the dielectric constant for Z5U ceramic

It is therefore obvious that your meter cannot be trusted to measure the stray (shunt) capacitance of a resistor, no matter how much faith you have in its accuracy when used for other purposes.

[bson]

Keysight 1733C handheld LCR meter, 1.2k resistor (Vishay-Dale 1W):

R=1200.1 ohm @ 1kHz
Z=1200.1 ohm 0.00deg @ 1kHz
Par C = 3-5pF @ 10kHz

The parallel C is so small it's likely just noise and meaningless for an axial component.  The meter would need to be carefully calibrated with a short consisting of a section of the lead to factor out these kind of error sources.  Or it's really just measurement voltage noise.  10kHz is the lowest frequency it would even give me a reading on!

4nF?  Almost certainly a test setup error.

[Seekonk]

This is the political season so it only follows that the  hard questions are not answered.
Instead the the question they wished was asked is answered. 

The voltage divider suggestion is a good idea.  I would use the two types of resistors in a non inverting R1/R2 op amp circuit and then test with R2/R1 comparing frequency response. This would make it independent of probe effects.  It certainly appears that some quality of this resistor is able to fool a simple capacitance meter. If this is a real real it should show up in the frequency response of an op amp. If it is capacitive I would think the model would be a spiral capacitor. I would heat both resistors with a torch to burn off the enamel and see how many spirals it has. It might give some clue. The ceramic may have some unknown quality.  Wouldn't be surprised if it was filled with some waste from another process, like they did with sheetrock.

[jaunty]

you would make a great engineer... but a lousy scientist.
yes if you refer to a common design model for such resistors and limit your understanding to such published data tables, then you can refute anything you want. So the problem must be my crappy meter...
even if i weren't using a 'high end' meter and it was just a cheap $100 chinese affair and not a $400 meter- it wouldn't make any difference. The result would still be completely objective... (as mentioned previously) i am simply curious what gives these resistors such a different property to the others...

A typical 1/4W axial leaded film resistor has a body length L of 6.5mm, diameter D of 2.3mm, with metal end caps each of up to 25% of the total body length. 

This can be approximately modelled as a parallel plate capacitor consisting of two plates of area (Pi/4)(D^2) = 4.15 mm^2 or 4.15E-6 m^2, separated by a dielectric of thickness D/2 = 3.25mm or 3.25E-3 m.

Alumina ceramic, which is commonly used as the substrate for film resistor bodies has a dielectric constant of 9 - 10. 

Z5U ceramic (as used in high value ceramic caps) has a dielectric constant (relative permittivity) of up to 18000.  Its the highest dielectric constant ceramic that's commonly available.

You'll find the formulae for calculating the capacitance here: http://www.rfcafe.com/references/electrical/capacitance.htm
Here's an online calculator for it that will accept mm and mm^2 and output in pF:
http://www.calctool.org/CALC/eng/electronics/parallel_plate

Plug in the numbers and you will see that a 1/4W thin film on alumina resistor should have a stray capacitance of no more than 0.11pF.  If the manufacturer FUBARed it and used the much more expensive Z5U ceramic for the substrate, it *MIGHT* have as much as 200pF of capacitance.  In practice, the ends aren't parallel plates, but little cups, so most of their surface is further apart, and the substrate diameter is significantly less than the body diameter so the capacitance will be significantly less.

If your resistors are different dimensions, plug in the numbers for them.  You wont get more than a few pF for any normal body ceramic and are unlikely to get into the nF range even if you used the dielectric constant for Z5U ceramic

It is therefore obvious that your meter cannot be trusted to measure the stray (shunt) capacitance of a resistor, no matter how much faith you have in its accuracy when used for other purposes.

[PedroDaGr8]

I think some people are getting hung up on the number being wrong. What I gathered from this thread is that Jaunty does NOT think the number is real. He doesn't think that his resistor actually has 0.4uF of capacitance.

What he is asking is:

1) WHY does this one read 0.4uF, not 4nF like the other resistors he has. Not that the number is real, but why is it different from the rest (including ones with similar resistance values).
2) If he can figure out why it is different, is there a possibility that this difference is the reason for his circuit messing up.

[jaunty]

Ahhh yes! Political Season! You're talking about the Superbowl right ?? (haha - just a little joke - as it seems more and more - we seem to treat 'politics' like a team sport.. without consulting actual 'issues' - my team for the win!)

At any rate  - TOUCHÉ! THANK YOU for that acknowledgement!! Yes - I thought perhaps a little surgery might be in order... i also wonder if there's a 'cleaner' way to possibly intervene - like using liquid nitrogen to break it open.. (?) - I am worried the heat might affect (or destroy) some interesting aspect of the construction.

This is the political season so it only follows that the  hard questions are not answered.
Instead the the question they wished was asked is answered. 

The voltage divider suggestion is a good idea.  I would use the two types of resistors in a non inverting R1/R2 op amp circuit and then test with R2/R1 comparing frequency response. This would make it independent of probe effects.  It certainly appears that some quality of this resistor is able to fool a simple capacitance meter. If this is a real real it should show up in the frequency response of an op amp. If it is capacitive I would think the model would be a spiral capacitor. I would heat both resistors with a torch to burn off the enamel and see how many spirals it has. It might give some clue. The ceramic may have some unknown quality.  Wouldn't be surprised if it was filled with some waste from another process, like they did with sheetrock.

[Seekonk]

I have operated 1/4W CF resistors at currents that made them glow red. After an hour the enamel was burned off, but resistance had changed less then 1%.  Don't think a torch would do any damage.

[The Doktor]

Jaunty, don't know where you're located, but I'm in NJ, USA. I've got a few month old DER-EE DE-5000 and lots of free time. I'd be happy to measure a few of each type if you wanted to send them to me. If you're not in USA, you can probably get a similar offer from somebody closer, just post a request in this thread.

Ed

[jaunty]

yes - sure- i'm in california... shoot me your address (there's PM on this thing right??) - would be fun - who knows... i'm hoping to uncover the big chinese resistor conspiracy (haha)...

Jaunty, don't know where you're located, but I'm in NJ, USA. I've got a few month old DER-EE DE-5000 and lots of free time. I'd be happy to measure a few of each type if you wanted to send them to me. If you're not in USA, you can probably get a similar offer from somebody closer, just post a request in this thread.

Ed

[edavid]

yes if you refer to a common design model for such resistors and limit your understanding to such published data tables, then you can refute anything you want. So the problem must be my crappy meter...
even if i weren't using a 'high end' meter and it was just a cheap $100 chinese affair and not a $400 meter- it wouldn't make any difference.

This is a misconception you have... that meters having a C function, can be measured on a single scale from "crappy" to "high end".   For the measurement you are talking about, the best DMM is not as good as the crappiest LCR meter.  A simple RC bridge would also be much better than your DMM.  The TX1 is just incapable of making an accurate measurement in this situation.

[wraper]

So the problem must be my crappy meter...
even if i weren't using a 'high end' meter and it was just a cheap $100 chinese affair and not a $400 meter- it wouldn't make any difference. The result would still be completely objective... (as mentioned previously) i am simply curious what gives these resistors such a different property to the others...

The problem is not in the meter indeed. But between the chair and the meter. There is no problem in the multimeter because it is incapable of measuring anything other than capacitors on the capacitance range. How it will fail to do this depends. A bit different resistance or a bit of inductance in the resistor can make it fail to measure with very different results. When this "something" is on the border in which auto-range starts acting, you might not get any stable results at all but just display flashing some different nonsense all of the time.

[jaunty]

yes if you refer to a common design model for such resistors and limit your understanding to such published data tables, then you can refute anything you want. So the problem must be my crappy meter...
even if i weren't using a 'high end' meter and it was just a cheap $100 chinese affair and not a $400 meter- it wouldn't make any difference.

This is a misconception you have... that meters having a C function, can be measured on a single scale from "crappy" to "high end".   For the measurement you are talking about, the best DMM is not as good as the crappiest LCR meter.  A simple RC bridge would also be much better than your DMM.  The TX1 is just incapable of making an accurate measurement in this situation.

gosh-i'm not certain why it would matter...  you're saying it's measuring a difference that doesn't actually exist...? that if a meter sees a difference then the circuit will not?

[jaunty]

The problem is not in the meter indeed. But between the chair and the meter. There is no problem in the multimeter because it is incapable of measuring anything other than capacitors on the capacitance range. How it will fail to do this depends. A bit different resistance or a bit of inductance in the resistor can make it fail to measure with very different results. When this "something" is on the border in which auto-range starts acting, you might not get any stable results at all but just display flashing some different nonsense all of the time.

and yet... that's not true. if it WERE true then it would not be repeatable, would it? - and one would be getting spurious DIFFERENT results from the SAME samples - and not give a consistent reading depending on the resistor. It blows my MIND the lengths people will go to in order to support their own beliefs... rather than have to accept something that runs counter to their training and experience.

[wraper]

and yet... that's not true. if it WERE true then it would not be repeatable, would it? - and one would be getting spurious DIFFERENT results from the SAME samples - and not give a consistent reading depending on the resistor. It blows my MIND the lengths people will go to in order to support their own beliefs... rather than have to accept something that runs counter to their training and experience.

Try them with another (even crappiest meter). Then talk something about repeatability. Same resistors give the same result when failing to measure them properly. Unstable results will be seen only if auorange will start acting.

[wraper]

Also, measure the DC resistance of each resistor (writing the value until the last digit the meter shows) you are measuring. This could tell something why they are measured differently under the C range.

[jaunty]

Wraper - I think you're getting a bit too obsessed with the 'capacitance' thing a bit too much. Perhaps it's my fault for titling it so precociously lke i did.

[The Doktor]

Jaunty, just sent you my info.

Ed

[The Electrician]

DMMs measure capacitance by applying a (relatively) constant current and measuring the time taken to charge to a certain voltage.  When this method is working properly, one can see a ramp of voltage across the capacitor as the DMM is carrying out the measurement.  This ramp is repetitive and looking at it with a scope can tell us whether it's working in this case.

Here's what I see across the DMM leads in capacitance measuring mode with nothing other than the 10 megohm scope probe connected to them:



With a 1 meghom resistor connected, I get this waveform across the DUT and the DMM leads:



There's no ramp.  The DMM is struggling to make a measurement and just can't do it.  The result of the measurement is meaningless.

Jaunty, what do you see of you look at the DMM's waveform with the two classes of resistor you're measuring?

[edavid]

yes if you refer to a common design model for such resistors and limit your understanding to such published data tables, then you can refute anything you want. So the problem must be my crappy meter...
even if i weren't using a 'high end' meter and it was just a cheap $100 chinese affair and not a $400 meter- it wouldn't make any difference.

This is a misconception you have... that meters having a C function, can be measured on a single scale from "crappy" to "high end".   For the measurement you are talking about, the best DMM is not as good as the crappiest LCR meter.  A simple RC bridge would also be much better than your DMM.  The TX1 is just incapable of making an accurate measurement in this situation.

gosh-i'm not certain why it would matter...

It matters because you are trying to measure a small capacitance in parallel with a resistance.

you're saying it's measuring a difference that doesn't actually exist...?

I'm saying there's no evidence that it's measuring a difference in capacitance.

that if a meter sees a difference then the circuit will not?

You haven't shown your circuit (why not?), but yes, it's quite likely.