Point absolutely missed, despite an extraordinary number of paragraphs. In answer to a number of posts.

Hello again,

So what was YOUR point then? Keep it short please.

 

Reiterating for the third time....

You may have missed this. It may have been missed in your frequent bloviated posts here.

You can't get more significant digits by multiplying numbers together. In my example, if you have a 1 gram weight – one significant digit – you can only KNOW the accuracy of the weight in ounces to ONE significant digit, no matter how accurate the conversion factor is.

But again, the facts in the matter at hand is a 10 - 20% variation in LED current isn't going to make a bit of difference, especially when the calculations are based on the average value of Vf with no standard deviation given.

 

Published where?

Who has said or implied that we can ALWAYS get away with 0.035?



Yes. SOMETIMES. But you keep using it when it serves ZERO purpose, which implies that you think that we need it ALL THE TIME.

Hello again,

Your second sentence makes sense so I will make a note of when I think it is relevant and try to remember to point out that it is not required in those that do not need it.

The first sentence does not seem to line up with what my reply to him was. It was simply that HE posted a number that had high precision, then said we don't need it. If we didn't need it, then why did he have to post it?
That's a little ironic, but the reason I point this out is that sometimes we do need the extra precision.
However, I'll try to remember to point this out so you can both feel more comfortable with it.

Something like this perhaps:
"My resistor value calculation came out to 2.345678k but we only have to use 2.3k for the actual value."

So that shows the distinction between a calculation using the standard 16 digit floating point built into most CPU's these days (and already reduced in precision) and the actual value or 'good enough' value.

It's also interesting following some of your ideas that sometimes we can detect errors using LESS digits too. For example, sometimes even after rounding, we get two different results, say:
1.2 and 1.3
and that allows us to see that somebody did not get it right or they rounded differently.
However, that does not cover all the cases.
If we post 1.234567 and they got 1.234567 then they both round to 1.2 so if we post 1.2 instead we still can tell there is an error if they post 1.3 units.
But if we got 1.234567 and they got 1.244567, they would never know they made a mistake somewhere if we just post "1.2".

I can't help but wonder also, how hard is it to round 1.234567 to 1.2 for any of us here?

 

@MrAl, I long ago blocked you on Electrotech for your long-winded mostly pointless answers for my sanity. Where is the block button here?

You might notice I have not been alone in criticizing your wordy replies.

 

Reiterating for the third time....

You may have missed this. It may have been missed in your frequent bloviated posts here.

Hello again,

Thanks for repeating the same wording again then also mocking my posts. Thanks
You don't seem to even understand that IF in fact someone does not understand you, repeating the same words NEVER helps. Can you at least understand that?

Sorry to say, YOU are still not getting MY point at all.
My point is about the calculation results, not the practical application.
Not only that, but the practical result is EASY to obtain from the calculation result.
If the calculation result is 1.23456 and the practical is 1.2, then all you have to do is round the previous to 1.2 and I assume you know how to do that.
As far as the statistics goes, it STILL does not matter that the LED current varies by 10 percent, 20 percent, or 100 percent. That's because every calculation comes with a set of assumptions, and when we post a value like 1.23456 we did that with a set of assumptions that DOES NOT include that 10, 20, or 100 percent variation in ANYTHING.

Note that the capitalized words are to put emphasis on those not shouting or anything.

Let's look at a sarcastic example.
We have a resistor of 1k and a voltage source of 1.01v. We connect them so we get 0.00101 amps current flow. I know you can do that.
Now the resistor is heated up from a hot plate or something and gets hotter and hotter. The temperature rises from 70F to 200F.
The resistance goes up to 2k (poor temperature tolerance). What is the current through the resistor now?
The current is obviously high, but does that make the result of 0.00101 amps wrong? NO. We assumed room temperature 70F for that one.
Now if we want to include the second calculation, we can do that too, but for the first calculation we don't do that because we have a different set of assumptions.

One more little interesting side point...
In the first calculation above I typed exactly "0.00101 amps".
Well, what if I typed "0.001010 amps". Do you have any idea at all why I included that extra zero '0' at the end of it? That would possibly seem like too many digits in your viewpoint. So why the extra zero?
I am asking so that you might think about other uses for extra digits.

 

@MrAl, I long ago blocked you on Electrotech for your long-winded mostly pointless answers for my sanity. Where is the block button here?

You might notice I have not been alone in criticizing your wordy replies.

You don't seem to understand a lot of things and are unwilling to TRY to understand, and your continuously mocking attitude does not help any either.
You are free to block anyone you wish don't let me stop you

 

Hello again,

Apparently, some members can't understand that different people have different opinions on the same issues so I suggest we just halt this conversation. I'm going to stop replying for the sake of everyone.

 

Hello again,

Your second sentence makes sense so I will make a note of when I think it is relevant and try to remember to point out that it is not required in those that do not need it.

The first sentence does not seem to line up with what my reply to him was. It was simply that HE posted a number that had high precision, then said we don't need it. If we didn't need it, then why did he have to post it?
That's a little ironic, but the reason I point this out is that sometimes we do need the extra precision.
However, I'll try to remember to point this out so you can both feel more comfortable with it.

Something like this perhaps:
"My resistor value calculation came out to 2.345678k but we only have to use 2.3k for the actual value."

So that shows the distinction between a calculation using the standard 16 digit floating point built into most CPU's these days (and already reduced in precision) and the actual value or 'good enough' value.

It's also interesting following some of your ideas that sometimes we can detect errors using LESS digits too. For example, sometimes even after rounding, we get two different results, say:
1.2 and 1.3
and that allows us to see that somebody did not get it right or they rounded differently.
However, that does not cover all the cases.
If we post 1.234567 and they got 1.234567 then they both round to 1.2 so if we post 1.2 instead we still can tell there is an error if they post 1.3 units.
But if we got 1.234567 and they got 1.244567, they would never know they made a mistake somewhere if we just post "1.2".

I can't help but wonder also, how hard is it to round 1.234567 to 1.2 for any of us here?

The notion of using mismatches way out in the sig figs of a calculation to detect errors only makes sense if both calculations MUST use the same values for EVERYTHING out to at least that resolution. That is seldom the case in engineering problems.

This one is a good example. You used a two points on the graph (1.8 V, 0 mA) and (2.3 V, 50 mA) to get your slope and intercept. You then report the LED voltage to seven sig figs. Why does the other person have to use the same points from the graph that you did to seven sig figs? If they don't, then a mismatch out that far means nothing! It does NOT detect an error.

If I take the graph from the data sheet, expand it to nearly fill the screen and do a screen capture, and then bring it into Paint and determine the intersections and use those to get the equation for the line, the voltage across the LED comes out to be 1.807946 V. Comparing this to your result of 1.828846 V, and if a mismatch of 0.000001 V indicates that one of us made an error, how egregious must the error have been to result in a difference that is more than 22000 times that?

Even just changing x-values of the two endpoints of the line by a single pixel results in an answer of 1.804470 V, which is a mismatch of nearly 3500 times the threshold that you say indicates that someone made an error.

So I still maintain that giving results to that resolution is not only pointless clutter, but it implies knowledge and information that simply doesn't exist.

 

The notion of using mismatches way out in the sig figs of a calculation to detect errors only makes sense if both calculations MUST use the same values for EVERYTHING out to at least that resolution. That is seldom the case in engineering problems.

This one is a good example. You used a two points on the graph (1.8 V, 0 mA) and (2.3 V, 50 mA) to get your slope and intercept. You then report the LED voltage to seven sig figs. Why does the other person have to use the same points from the graph that you did to seven sig figs? If they don't, then a mismatch out that far means nothing! It does NOT detect an error.

If I take the graph from the data sheet, expand it to nearly fill the screen and do a screen capture, and then bring it into Paint and determine the intersections and use those to get the equation for the line, the voltage across the LED comes out to be 1.807946 V. Comparing this to your result of 1.828846 V, and if a mismatch of 0.000001 V indicates that one of us made an error, how egregious must the error have been to result in a difference that is more than 22000 times that?

Even just changing x-values of the two endpoints of the line by a single pixel results in an answer of 1.804470 V, which is a mismatch of nearly 3500 times the threshold that you say indicates that someone made an error.

So I still maintain that giving results to that resolution is not only pointless clutter, but it implies knowledge and information that simply doesn't exist.

Well hello again,

Oh, that's very interesting because that reveals exactly what is going on here.
You are generalizing to every case, while I am generalizing to specific cases. In fact, we actually agree on all of this. Ironically, that means that since Jon agrees with you then he also agrees with me.

It's the classic example of the first person going general while the second person goes specific, or the first person goes specific while the second person goes very general. This almost always leads to a disagreement that simply is not really there. From your posts and my posts we can see that if we both go completely general, we agree. If we both go specific, we still agree.

Specific statement: "R1=10 Ohms".
Generalized reply: "R1 can't possibly be 10 Ohms in every single circuit on earth".

One statement is specific, the other is general, they seem to be at odds but in reality, they are both right.

I hope this helps

 

Well hello again,

Oh, that's very interesting because that reveals exactly what is going on here.
You are generalizing to every case, while I am generalizing to specific cases. In fact, we actually agree on all of this. Ironically, that means that since Jon agrees with you then he also agrees with me.

It's the classic example of the first person going general while the second person goes specific, or the first person goes specific while the second person goes very general. This almost always leads to a disagreement that simply is not really there. From your posts and my posts we can see that if we both go completely general, we agree. If we both go specific, we still agree.

Specific statement: "R1=10 Ohms".
Generalized reply: "R1 can't possibly be 10 Ohms in every single circuit on earth".

One statement is specific, the other is general, they seem to be at odds but in reality, they are both right.

I hope this helps

How am I generalizing to every case????

I have said that there are times when reporting results to seven sig figs but that this is not one of them. That is not generalizing at all -- that is very specific to THIS thread and THIS circuit.

But let me be more specific and explicit. Reporting the value of the voltage across the LED in the circuit in this thread to seven sig figs is pointless clutter that serves zero purpose at all. It does not enable the detection of errors because of mismatches in the last reported digit.

 

How am I generalizing to every case????

I have said that there are times when reporting results to seven sig figs but that this is not one of them. That is not generalizing at all -- that is very specific to THIS thread and THIS circuit.

But let me be more specific and explicit. Reporting the value of the voltage across the LED in the circuit in this thread to seven sig figs is pointless clutter that serves zero purpose at all. It does not enable the detection of errors because of mismatches in the last reported digit.

Hello again,

I think we pretty much agree on all of this, but you seem to be making a big deal out of this with phrases like "pointless clutter".
It's hard for me to believe that in a post with maybe 1000 alphanumeric characters or more, that even 5 extra digits is "clutter".

The previous text, up to the point just before the "T in "The" in the beginning of this sentence, has 850 characters in it. If I added 1.2k that would be 4 more characters, if I added 1.23456k that would be 8 characters. 4/854 is 0.005 or 0.5 percent, while 8/858 is 0.009 which is about 0.9 percent. I don't think that 4 more characters is clutter. It's also unintended but a little interesting that if we round up, we get 1 percent and 1 percent. So using your preferred character reduction idea, they would equal the same 'clutter'.

Also, in experiments, we have to list the assumptions and get as accurate as we can. If one of the assumptions is R1=10 Ohms, we can't say that we did not get the same results because when we change R1 to 20 Ohms we get a widely different number. R1=10 was an assumption that must hold so that other labs can verify the results in the same way. If we change R1 to 20, then the original lab has to change their value to 20 also, and then we again have a basis for comparison.
In this thread 1.8 and 2.3 were assumptions, and so you can't say "what if they used different numbers". We can also ask, "what if they did". The point is, when you use those assumptions you get the same results, and the results can be carried out to any number of digits. Then if person B does not get the same result as person A, there is reason to seek out why they got different results.
We agree that in some cases even 1 digit will suffice. If I get 3 and somebody else gets 4, then that says we did something different and maybe messed up a value used for a circuit element. On the other hand, if I get 3 and somebody else gets 3, it may look like we got the same results but looking even one more digit we might see 3.0 and 3.4, or even 3.0 and 3.9 if they don't round properly. If we do go one more digit we might see 3.1 and 3.1, indicating that we both got the same result, but going one more digit we might see 3.10 and 3.14, which are not the same so something may be wrong. Along that same line, we may see 3.11 and 3.11, but then going one more digit we might see 3.110 and 3.114, which are not the same. If we continue this pattern, we might eventually see 3.11000 and 3.11000, but going one more digit, we might see 3.110000 and 3.110004, which amazingly can still indicate an error in someone's calculation. If we don't use the extra digits, the student goes on to assume they did the problem correctly. As we agree, usually 3.1 and 3.1 would be good enough, but you can see why that's not always true. If we generalize and say that in most cases we don't need more digits, we lose the ability to catch SOME errors (not all).

If I had to generalize about what we are looking at here, I would say that approximation theory seeks less information while high precision numerical analysis seeks more information. They both have their place, but understanding the difference might not be immediately apparent.

Since this thread went way off on a tangent all its own, I suggest we might continue this discussion in PM's or just go on to something more productive. We can always come back here and post the results of the discussion if you like. I also find that some people are overly sensitive to disagreement and that frustrates them. In a PM we would be the only two that might be frustrated