Maker Pro
Maker Pro

Creating a higer wattage resistor

S

Skeptic

Jan 1, 1970
0
Responding to John Fields:



I was speaking statistically. True, in your example the resultant value
would still be at the 5% tolerance value. However as unusual as finding a
resistor 5% high, finding two consecutive resistors 5% high would be a very
rare occurrence. It would be much more likely that the second resistor
would be closer to the marked value or even on the low side tending to
compensate for the 5% high side error of the first resistor.



I have requested the statistical distribution of resistors from resistor
companies but haven't been able to get it. Nevertheless I think it's
probably close to normal distribution centered around the marked value with
the tolerance set at the 3 standard deviation points. What I am saying is
that using two resistors of equal value and with the same tolerance will
move the 3 standard deviation points of the distribution down to 3.5% which
is due to the error of one resistor tending to cancel out the error of the
other.





Responding to Abstract Dissonance:



Yours is an interesting way of looking at the problem but you must remember
that actual value of the resistor is the average of all the little segments
you can divide that resistor into. Rather than having a distribution around
the marked value, all the little resistor segments have a distribution
around the actual value. It is the population of completed resistors that
are controlled to be distributed around the marked value.
 
J

John Fields

Jan 1, 1970
0
i.e. 10 resistors of 100 ohms in series will dissipate the same power as 10
resistors of 1 ohm in series.
---
Huh???
---


This does not take into account the wattage of
the but I explained that in the original post.

---
Huh??? The wattage of the _what_?
---
Now, to the problem at hand... why is it better to do it in parallel than in
series?

---
It isn't. Electrically it's the same.
---
Well, Ra = n1*n2*Rb.

If we are working with the same number in each case, i.e., n1 = n2 = n (its
still called the square in ither case though)

then

Ra = n^2*Rb.

Rb represents the resistance in series and Ra in parallel.


So, what happens as we use 10 resistors in parallel of resistance R? then
our resistors in series need to be R/100 ohms to get the same equivilent
circuit.

I guess it doesn't matter though as large resistances are easy to come by
too. Ultimately the equation fails at both ends though but if your trying to
do something like 100 resistors in parallel then you have a factor of 10000
if you want to convert that circuit to one in series. (you could use that to
your advantage if you happen to be working with extremly large or small
resistances to convert to the other topology and get resonable results)

If, say, R is 100ohms and you have 100 in parallel but decide you need to do
it in series then you need 100 1/100 ohm resistors. (since 100/100 = 1 =
100*1/100)...

But on the other hand if you have 100 100 ohm resistors in series and want
to convert that to a parallel(for whatever reason) then you need 100 100000
ohm resistors for parallel. (sinec 100*100 = 1000 = 100000/100).

---
No. One hundred 100 ohm resistors in series is 10,000 ohms, while
one hundred 100,000 ohm resistors in parallel is 1000 ohms.
---
So ineffect going from parallel to series will decrease the resistance you
need while going from series to parallel will increase it(obviously).


If we look at the above resistors and look at the power they are dissipating
then they have to be the same because resistance adds... doesn't matter the
applied voltage or current because they are equivilent topologies. (they are
effectively the same bulk resistance) It doesn't matter how many resistors
you use or how they are configured because if they are equivilent to another
configuration then they will dissipate the same total power. BUT they will
not necessarily dissipate the same power PER resistor.

Thats the mistake you seem to be making. Its easy to show a counter example:

------R------
| |
--- R ---- ---------
| |
------R------

The first resistor does not dissipate the same power as the other
two(individually).

The first is 2 the power dissipated in either of the others.

---
No, it's 4 times the power dissipated in either of the others:
---

Why is this
important? It should be obvious.


---
What's obvious is that you're, at best, confused.
---

This goes to the part when I talk about
combining the two configurations)




Anyways, What I said is this:

---
You're insane.

Here:

100V--+--------------------+
| |
GND>--|--------+ |
| | |
+-[100R]-+ [1R]
| | |
+-[100R]-+ [1R]
| | |
+-[100R]-+ [1R]
| | |
+-[100R]-+ [1R]
| | |
+-[100R]-+ [1R]
| | |
+-[100R]-+ [1R]
| | |
+-[100R]-+ [1R]
| | |
+-[100R]-+ [1R]
| | |
+-[100R]-+ [1R]
| | |
+-[100R]-+ [1R]
| |
+-----------+
ARRAY A ARRAY B


1. What is the total dissipation of array A?

2. What is the dissipation of each resistor in array A?

3. What is the total dissipation of array B?

4. What is the dissipation of each resistor in array B?

If you read carefully what I say you will realize it makes perfect
sense(although I'm not saying its completely clear). You have to understand
that I didn't say that the power changed when the topology changed.... The
first sentence is an obvious statement. The second is is obvious too. I'm
saying that if you want the same equivilent circuit(same bulk resistance)
then you have to reduce the "new" resistors by a squared factor.

---
No, you don't. It's a simple division.

For the same number of resistors in both arrays:

Rp
Rs = ----
n

Where: Rs is the value of a single series resistor,
Rp is the value of a single parallel resitor, and
n is the number of resistors in either array

---
 
R

redbelly

Jan 1, 1970
0
phaeton said:
1) This is all hypothetical, and I'm not about to actually do this.

2) I realize that doing this will be ugly, immoral, and is possibly
illegal in some states or religions.

3) I realize that it's cheaper, easier and all around better just to
use the right part.

4) I also realize that if one of the 'legs' fails, it means increased
current through the other legs (at least until the rest of it burns
up). In other words, if this fails it could ruin something else too,
burn down the house, eat all your cheese and impregnate your favourite
girlfriend.


Suppose you need a 10 ohm 5W resistor, and you don't have one on hand.
Can you 'create' one by connecting ten 100ohm 1/2W resistors in
parallel? Or fifty 500 ohm resistances in parallel? Or two-hundred
and fifty 2K5 ohm resistances in parallel? Or 1250 125K resistances in
parallel? On and on until you're holding up traffic by soldering
stuff out in the street?


:p

-phaeton

p.s., it's friday. everyone lighten up.

As others have said, yes, you can do this.
Two other options:
1. Wire "n" resistors of R/n ohms in series (as was said elsewhere in
this thread).
2. Take 4 resistors of R ohms each.
Make two 2-resistor series strings.
Wire those two strings in parallel.
Now you have a total resistance equal to R, and it will dissipate
4 times the power of a single resistor.

Regards,

Mark

p.s. Extra credit problem: figure out how to wire 9 (or 16) resistors
of R ohms, to get a net resistance of R with 9 (or 16) times the power.

p.p.s. Using series / parallel combinations of resistors is also a
quick and dirty way to get a more precise value from lower-precision
resistors. I.e., get within 1% of a design value when all you have is
a box full of 5% resistors.
 
J

John Fields

Jan 1, 1970
0
p.p.s. Using series / parallel combinations of resistors is also a
quick and dirty way to get a more precise value from lower-precision
resistors. I.e., get within 1% of a design value when all you have is
a box full of 5% resistors.
 
A

Alan B

Jan 1, 1970
0
Note: many snips follow, not annotated as such, in the interest of
attempted brevity.

n is the number of resistors that you plan on using.... should be clear from
the context of my original post.

It certainly is not clear what you are trying to say. Hence my followup.
Nope... look you can have 2 possible configurations: All n of the identical
resistors in parallel or all in series.

Okey dokey, we will work from that premise.
Obviously your eating the wrong pie because your forgetting that you can
have constant current.

Attempts at obfuscation will not help you. The proof is in the pudding,
which in electronics, is the math. Read on!
I didn't know I was going to have to demonstrate it in such simple terms. I
thought most people here would be educated enough to see the what is going
on and I wouldn't have to take baby steps.

You are trying to explain away your inability to answer questions by
insulting the audience for asking them. I asked you to show your work in
order to see if you are capable of basic algebra. You apparently are not.
Sheesh... I think you are being a little dense.

I think you are being a little evasive. You've made what I consider to be
a vague and misleading statement about power transmission. Please answer
the questions.
I will, for the sake of your enlightenment be more precise and clear.

Well, you've failed to do that. Your post is only more vague and
misleading. If you could confine yourself to terse statements that answer
particular and direct questions, we might get somewhere.
I
figured most here would have basic knowledge of elementary circuit analysis
but i guess I was wrong ;/

Your condescending attitude is not helping your case. The point is, your
explanations make no sense. *You* might know what you are talking about,
but I'll wager few others do. I pointed out the error here:

I showed this to be clearly wrong, and you have responded with insults as
to my intelligence. Why? Perhaps because you are hoping that the error
will go away if you respond with enough new, obtusely presented, argument.
Well, it hasn't, and you've made many more in your insult ridden response.
Suppose we have n1 resistors each with resistance Ra and wattage Wa in
parallel. I'm not going to try and draw it because its obvious and if you
can't picture it in your brain then maybe you don't need to be in
electronics.

More insults, no rectification of previous errors.
the current going through each resistor is I/n1. This is basic kirchoff's
current law.

Pedantically, it is not; you've modified it and not shown how you arrived
at your derivation. Basic Kirchoff's current law would be:

0 = I1 + I2 + I3 + ... + In
This can be seen from another much simpler method.

What follows is *simpler?*!!
Its quite simple and any 3rd grader should have no problem with the math and
electronics to this point.

More insults. Well, better suck up your pride, because you are about to
become rather embarrassed.
For some real world examples to make all the egg heads understand? Ok...

More insults. You must be very insecure.

******* CONTENT ALERT! BEGINNING TO GET TO THE POINT! ********
Given 10 resistors of 100 ohms in parallel how can we get an equivilent
circuit(same resistance and wattage dissipation) in series?

Simple. Add the inverses of the individual resistances, and invert. This
is a basic formula, and works even if the resistances might happen to be
not all identical.

Rs = 1 / (1/Rp1 + 1/Rp2 + ... + 1/Rpn)

If the denominators happen to all be identical, as is the case in our
mutually accepted premise, then you may simply use 'n' as the numerator and
reduce:

Rs = 1 / (n/Rpi) = Rpi/n

That's extremely simple, isn't it? Now you know the equivalent series
resistance. For dissipation, add the individual legs.

Pd = PdR1 + PdR2 + ... + PdRn

Simple, and it works, again, even if the resistances are not identical. In
the case of identical values,

Pd = n*PdRi

******* CONTENT ALERT! CRUX OF THE MATTER REACHED! ********
Ra = n^2*Rb.

Aha! I've found your hypothesis! You buried it well!
Rb represents the resistance in series and Ra in parallel.

But, I thought we were solving for series resistance? So I guess you mean
to say:

Rb = Ra / n^2

So as a simple example, if Ra = 1000 and n = 10, then you say:

Rb = 1000 / 10^2 = 1000 / 100 = 10

making Rb = 10, but by the "product over sum" method (which is taught at
the basic level, and should be understood by most readers of the group):

Rs = Rp^n / Rp*n = 1000^2 / 1000*10 = 100

or from the reduced inverse-sum form:

Rs = Rp / n = 1000 / 10 = 100

and by either equation the series resistance is 100. I've shown the math
behind my work; you've posted an equation that appears to be nonsense, and
supported it only with more vague wording and error-filled calculations.
But on the other hand if you have 100 100 ohm resistors in series and want
to convert that to a parallel(for whatever reason) then you need 100 100000
ohm resistors for parallel. (sinec 100*100 = 1000 = 100000/100).

Fascinating. Rather than bury these glaring math errors within convoluted
meandering of faulty logic, I'd simply calculate:

Rp = Rs * n = 100 * 100 = 10,000

Whereas you would:

Ra = Rb * n^2 = 100 * 10,000 = 1,000,000
So ineffect going from parallel to series will decrease the resistance you
need while going from series to parallel will increase it(obviously).

What is obvious is that you have no idea what you are talking about.
Thats the mistake you seem to be making. Its easy to show a counter example:

Hmm? Did I make a mistake somewhere? Please point it out. I am open to
constructive criticism, albeit possibly curmudgeonly.
The first is 2 the power dissipated in either of the others.

More gibberish. I will now tell you what I expect from you. I expect
apologies for your numerous insults, and a retraction of your nonsensical
parallel-to-series conversion equation:

Ra = n^2*Rb

I don't think that's too much to ask.
 
A

Alan B

Jan 1, 1970
0
making Rb = 10, but by the "product over sum" method (which is taught at
the basic level, and should be understood by most readers of the group):

Rs = Rp^n / Rp*n = 1000^2 / 1000*10 = 100

I made a mistake here. Product over sum apparently only works for n = 2.
 
A

Alan B

Jan 1, 1970
0
making Rb = 10, but by the "product over sum" method (which is taught at
the basic level, and should be understood by most readers of the group):

Rs = Rp^n / Rp*n = 1000^2 / 1000*10 = 100

Here's a breakdown of the brain fart. For two parallel resistances:

Rs = 1 / (1/i + 1/j)

Multiplying all terms by the product of the denominators gives:

Rs = ij / (1/ij/i + 1/ij/j) = ij / (j + i) <product over sum>

For three legs, using the same technique:

Rs = 1 / (1/i + 1/j + 1/k)

Rs = ijk / (jk + ik + ij)

and so on.

Example

Rs = 1000^3 / (1000^2 + 1000^2 + 1000^2) = 333.3

being identical to the simplification in my post of

Rs = Rp / n = 1000 / 3 = 333.3

Sorry for the error.
 
A

Abstract Dissonance

Jan 1, 1970
0
yes, as I have have said several times but you seem to pick and choose what
you want to read. Electrically they might be the same but not necessarily
practically. I pointed it out but you don't care to use your brain to much
and only find obvious mistakes that are unimportant in the logic.

simple mathematical mistake. 100 10 ohm resistors in series. The
calculation error doesn't defeat the logic as you seem to think. You have
similar mentality to people who point out spelling and grammar mistakes as
proof of a logical error.


Sure. So? I said this in my previous post how its a difference of the
square.

No, I'm making simple calculation mistakes. I explained everything correctly
in the original post. Just like you I made some oversights and mistakes
because no one is perfect and I wrote this in a hurry at 1AM. Forgive me
for not living up to your standards of perfection.

Yep. So stop replying to me then.
Here:

100V--+--------------------+
| |
GND>--|--------+ |
| | |
+-[100R]-+ [1R]
| | |
+-[100R]-+ [1R]
| | |
+-[100R]-+ [1R]
| | |
+-[100R]-+ [1R]
| | |
+-[100R]-+ [1R]
| | |
+-[100R]-+ [1R]
| | |
+-[100R]-+ [1R]
| | |
+-[100R]-+ [1R]
| | |
+-[100R]-+ [1R]
| | |
+-[100R]-+ [1R]
| |
+-----------+
ARRAY A ARRAY B


1. What is the total dissipation of array A?

2. What is the dissipation of each resistor in array A?

3. What is the total dissipation of array B?

4. What is the dissipation of each resistor in array B?

If you read carefully what I say you will realize it makes perfect
sense(although I'm not saying its completely clear). You have to
understand
that I didn't say that the power changed when the topology changed.... The
first sentence is an obvious statement. The second is is obvious too.
I'm
saying that if you want the same equivilent circuit(same bulk resistance)
then you have to reduce the "new" resistors by a squared factor.

---
No, you don't. It's a simple division.

For the same number of resistors in both arrays:

Rp
Rs = ----
n

Where: Rs is the value of a single series resistor,
Rp is the value of a single parallel resitor, and
n is the number of resistors in either array

---

What this
means is that you might end up having to use extremly low resistor values
if
you plan on using a series circuit. To do the job. The logical converse is
obviously true in that you have to use "extremly" large resistor values in
parallel as compared to the equivilent series topology. You can see this
in
the two examples I gave. In the parallel topology they the resistors are
way
larger than in the series. (Ofcourse you might be able to get away with
choosing different numbers in each topology to make it easier to do)

So.. the point is that you cannot use your brain to decipher the logical
argument and can only see immaterial mistakes. So... So keep on point them
out of it makes you better but you won't learn anything.
 
A

Abstract Dissonance

Jan 1, 1970
0
Alan B said:
Note: many snips follow, not annotated as such, in the interest of
attempted brevity.



It certainly is not clear what you are trying to say. Hence my followup.


Okey dokey, we will work from that premise.


Attempts at obfuscation will not help you. The proof is in the pudding,
which in electronics, is the math. Read on!

yes, and I have a degree in applied mathematics. Sure I still make stupid
calculation errors... but the logic is solid and I prefer to work in the
abstract while you think its important to work in the concrete. Maybe I did
make some calculation errors but it is not important. I'm not going to keep
on trying to explain something that is perfectly clear(ok, maybe not but use
your brain and don't expect to be spoon fed everything).
You are trying to explain away your inability to answer questions by
insulting the audience for asking them. I asked you to show your work in
order to see if you are capable of basic algebra. You apparently are not.

Nope, I explained it perfectly clear. You jsut don't like the insult and
hence you try to divert attention away from my proof that follows. I'm kinda
getting sick and tired of explaining things so I'll stop here. Everything I
said in the first post holds and if you dont' like it then tough. I have
better things to do with my time than explain minutiae. Maybe I wasn't clear
but so what. I'm not hear to teach you anything.
 
A

Abstract Dissonance

Jan 1, 1970
0
Alan B said:
I made a mistake here. Product over sum apparently only works for n = 2.

TO FUCKING BAD!

Its ok for you to make mistakes but not me?
 
A

Abstract Dissonance

Jan 1, 1970
0
Alan B said:
Here's a breakdown of the brain fart. For two parallel resistances:

Hmm.. why so many mistakes? Ateast you wern't doing it at 1am. If you want
to bitch bout triffle mathematical calculation errors then so be it but
expect me to do the same. Again, the everything I said in the original post
logically holds. In the second post I may have made some calculation errors
but hte formulas still hold. Anyways, I got better things to do.
 
A

Alan B

Jan 1, 1970
0
TO FUCKING BAD!

Its ok for you to make mistakes but not me?

If you're that sensitive, perhaps you shouldn't haul off and insult people
who are trying to understand you. It gets you slapped.
 
A

Alan B

Jan 1, 1970
0
yes, and I have a degree in applied mathematics. Sure I still make stupid
calculation errors... but the logic is solid and I prefer to work in the
abstract while you think its important to work in the concrete. Maybe I did
make some calculation errors but it is not important. I'm not going to keep
on trying to explain something that is perfectly clear(ok, maybe not but use
your brain and don't expect to be spoon fed everything).

The logic is not sound, and you are not properly supporting your position
with mathematics. And you really should stop with the insults, or you are
going to get slapped some more.
 
A

Alan B

Jan 1, 1970
0
Nope, I explained it perfectly clear. You jsut don't like the insult and
hence you try to divert attention away from my proof that follows. I'm kinda
getting sick and tired of explaining things so I'll stop here. Everything I
said in the first post holds and if you dont' like it then tough. I have
better things to do with my time than explain minutiae. Maybe I wasn't clear
but so what. I'm not hear to teach you anything.

'Nuff said.
 
E

ehsjr

Jan 1, 1970
0
Abstract said:
No, you are not understanding what I'm saying.


No one can understand what you are saying.
I'll re-word it for you:

Rp = Rs*N^2

Where N is the number of identical resistors, each of
resistance Rs in series, or Rp in parallel.

Is that what you are trying to say?

Ed
 
J

John Fields

Jan 1, 1970
0
So.. the point is that you cannot use your brain to decipher the logical
argument and can only see immaterial mistakes. So... So keep on point them
out of it makes you better but you won't learn anything.

---
So... You can't make a point without tripping all over yourself,
your math stinks, and your language skills are... well, let's just
say they're lacking...

Then, on top of all that you expect everyone to try to find a needle
of sense in the haystack of your post while trying to decipher your
"immaterial mistakes"?

LOL, it sounds to me like a lot of work for no reward at all!
 
A

Alan B

Jan 1, 1970
0
Rs = 1 / (1/i + 1/j)

Multiplying all terms by the product of the denominators gives:

Rs = ij / (1/ij/i + 1/ij/j) = ij / (j + i) <product over sum>

Rs = ij / (ij/i + ij/j) = ij / (j + i) <product over sum>

Moral of the story, boys and girls, always check your work.
 
T

tlbs101

Jan 1, 1970
0
Skeptic said:
...
I have requested the statistical distribution of resistors from resistor
companies but haven't been able to get it. Nevertheless I think it's
probably close to normal distribution centered around the marked value with
the tolerance set at the 3 standard deviation points. What I am saying is
that using two resistors of equal value and with the same tolerance will
move the 3 standard deviation points of the distribution down to 3.5% which
is due to the error of one resistor tending to cancel out the error of the
other.
...

Carbon resistor manufacturers used to do it this way:

They would create a large batch of resistors of value X.
The resistors in the batch would then be measured for value X, and the
ones with the tightest tolerance would be pulled and labeled as tight
tolerance (1% 2%).
Next the 5% resistors would be pulled, then the 10% resistors, and
finally the 20% resistors.

So for 5% and looser tolerance resistors, rather than see a nice
gaussian distribution , you ended up with a 2-lobed distribution curve,
with 'dip' at the base value. If you had a large sample of 1000 Ohm,
5% resistors, they would typically be close to either 1025 Ohms or 975
Ohms. If you had a batch of 1000-Ohm 20% resistors you would be
hard-pressed to find a single 1000 Ohm resistor in the bunch

I have seen this in my experience.

I don't know if manufacturing processes have changed over the decades
to increase target accuracy, or if this practice described is still
performed. I don't know if this applies to metal film resistors,
either.

Tom
 
R

redbelly

Jan 1, 1970
0
John, yes, that goes without saying. But why should that be a problem?
You're at least the second person in recent months to suggest to me
that 1% resistance measurements are not easy to come by.

Because of your post, as well as an earlier comment some months back by
Phil Allison, I've dug out my DMM's manual to look up the accuracy. It
turns out to be 0.3% to 1% for resistance measurements, depending on
just where the reading is. (If you want, see my "p.s." below for how I
calculated this)

So it's 1% or better. At least the manufacturer claims it is, in the
specs. Are most DMM's not nearly this good?

Regards, and TIA for any information,

Mark

p.s. Details of figuring the meter's accuracy:

For my meter (Extech model 380282), the stated resistance accuracy is
0.2% of the reading, PLUS 3 least-significant digits. That works out
to between a best-case of 0.3% for a reading of eg. 3.999 k, and at
worst 1% for a reading of eg. 4.01 k. Scale changes occur at 4, 40,
400, etc.
 
P

Peter Bennett

Jan 1, 1970
0
So for 5% and looser tolerance resistors, rather than see a nice
gaussian distribution , you ended up with a 2-lobed distribution curve,
with 'dip' at the base value. If you had a large sample of 1000 Ohm,
5% resistors, they would typically be close to either 1025 Ohms or 975
Ohms. If you had a batch of 1000-Ohm 20% resistors you would be
hard-pressed to find a single 1000 Ohm resistor in the bunch

I have seen this in my experience.

I don't know if manufacturing processes have changed over the decades
to increase target accuracy, or if this practice described is still
performed. I don't know if this applies to metal film resistors,
either.

Tom

Last time this subject was mentioned, I measured a handful of 10K 5%
resistors from my employer's stock - mostly Philips SFR25 - but both
metal film and carbon film, I think (brown body and light blue body).

From a sample of 50 or so, all but two were within 1 % (that's +/-
0.5% of the average value) I think there was one each barely above and
barely below that 1% spread. The average value was just under 10K, by
my Fluke DVM - I concluded that my DVM read just a hair low...

It seems from that test that current manufacturing techniques allow
the makers to produce resistors of any desired value to a very close
tolerance, so the ancient practice of searching your resistor stock
for a specific (non-standard) value won't work any more.


--
Peter Bennett, VE7CEI
peterbb4 (at) interchange.ubc.ca
new newsgroup users info : http://vancouver-webpages.com/nnq
GPS and NMEA info: http://vancouver-webpages.com/peter
Vancouver Power Squadron: http://vancouver.powersquadron.ca
 
Top