This thread is not about discrimination, nor about disdain for the group, but rather an observation I found interesting - please keep it PC.

Any time an engineer says "perfect", "exact", "100%", and (depending on the circumstance) "best", more often than not, that person is from/in India.

It seems many Indian engineers speak in absolutes and require the "perfect solution" or the "exact answer". Anytime I see this in a post, or hear it said, I immediately think they are from India.

We know that everything in engineering has a certain degree of uncertainty, error, and tradeoffs associated with it, yet some ask for absolute answers.

Perhaps this is a cultural/language thing, though I've met Indian engineers that do not use this kind of language when referring to a technical problem, so I'm at a loss.

Anyone have some insight as to why this might be?

 

I tend to to think those seeking exact numbers tend to be students in general.

I would love to see the designs using realist resistors and capacitors values.
I did recommed that at the other forum to an instructor who was seeking input in general.

Even the calculations could be held to some standard ....like scientific notation to two decimal places.

The only problem with establishing standards is everyone agreeing to enforce them.

 

As Mahatma Gandhi said

“Your beliefs become your thoughts,
Your thoughts become your words,
Your words become your actions,
Your actions become your habits,
Your habits become your values,
Your values become your destiny.”

Max.

 

In my case, it was a beginner thing. Desperately grasping for any parts of the education that I could trust to be known constants.

Eventually I got into precision analog where I could see the results to 4 decimal places and my trust in the math developed. It really does work, but with 20% resistors and a 1960's vacuum tube volt meter it all seemed like guesswork.

 

#12, are you saying that you did precision guesswork??

 

Like manipulating questionable data with infinite precision?

 

I tend to to think those seeking exact numbers tend to be students in general.
I would love to see the designs using realist resistors and capacitors values.

I think you're right.... when you're starting out, you tend to think that everything needs to be as exactly as the theory says it should be. And of course, in the real world, it isn't. Trouble is, knowing which bits you can let slip by, and which need to be accurate.
I used to work with a guy who did a lot of schematic drawing, but who wasn't an engineer.. at all. He could never understand that some resistors values needed to be calculated accurately, and some could almost any value at all. (Within reason of course ... but 'pull up' values aren't that fussy, for example!).
'How do you know which values are the important ones?'... he would ask.....scratching his head...

The test books don't always help. In the attached image - taken from a National Semiconductor analogue data book from the 1980s - you can see that R13 is marked as 93.1K. In series with R14 ---a 10k variable resistor!

Why on earth specify 93.1K when you adjust the overall value by such a huge amount? Doesn't seem to make any sense...

Never could figure that one out......

 

In my case, it was a beginner thing. Desperately grasping for any parts of the education that I could trust to be known constants.

Eventually I got into precision analog where I could see the results to 4 decimal places and my trust in the math developed. It really does work, but with 20% resistors and a 1960's vacuum tube volt meter it all seemed like guesswork.

I took a slide rule class in engineering. Our instructor who really disdained calculators said, "A calculator is a thing that will give you the wrong answer to 15 decimal places." I never forgot that.

 

........ "A calculator is a thing that will give you the wrong answer to 15 decimal places." .........................

Garbage in, garbage out. You would reach the same wrong answer with a slide rule, but not as precisely.

 

..............Why on earth specify 93.1K when you adjust the overall value by such a huge amount? Doesn't seem to make any sense...

Never could figure that one out......View attachment 73440

Look up "trimming instrumentation amp to achieve maximum common mode rejection." (Resistors must be matched to within .1% or better)

 

In my case, it was a beginner thing. Desperately grasping for any parts of the education that I could trust to be known constants.

Eventually I got into precision analog where I could see the results to 4 decimal places and my trust in the math developed. It really does work, but with 20% resistors and a 1960's vacuum tube volt meter it all seemed like guesswork.

One of my college electronics instructors tells about how he paid his way through college working at Ohmite painting stripes on resistors. All the resistors came off the same assembly line, 20%, 10%, and 5%. The only difference in the quality was how accurately they were measured. Without testing, the manufacturing process could pretty well assure a tolerance of 20% but not more. So his job was to go through huge piles of nominally 20% resistors, and sort the ones that fell within 10% in one pile, and 5% in another.

Now....here's the trick. At the end of the week, all the unsorted resistors got shoved off the bench into the 20% pile....no fourth stripe. If you BOUGHT a big pile of 20% resistors that were manufactured on a Friday, there would be X number of 10 and 5% resistors....you'd just have to measure them yourself!

Eric the Old

 

Look up "trimming instrumentation amp to achieve maximum common mode rejection." (Resistors must be matched to within .1% or better)

I'd like to see someone try and match to 0.1% easily by rotating a 10k pot...even a 10 turn one!
And that circuit is an audio preamp.. not even an instrumentation amplifier!
Sorry, but putting 93.1K resistor into that circuit is an example of why some 'theory' books are not helpful when it comes to teaching 'practical' engineering - especially to beginners...

On the other hand, from the same data book, National are rather less precise - and more 'real world' - when advising on how to deal with RF pick up in audio circuits, by suggesting prayer as an engineering technique

 

There you go. 100K trimmed to .1% Let me know if I can do anything else for you.

Again, I suggest that you look up, "trimming instrumentation amp to achieve maximum common mode rejection."

 

Sorry about the rather flip answer but you asked for it. Now for an answer that will do you some good. Attached is a app note from Jensen. They make the BEST audio transformers. They require resistors to be trimmed to .01%

 

Let me know if I can do anything else for you.

You're very kind. Thank you.

 

Now for an answer that will do you some good. Attached is a app note from Jensen. They make the BEST audio transformers. They require resistors to be trimmed to .01%

I think you've rather missed the point on that data sheet!....

Jensen - obviously keen to promote the advantages of galvanic isolation, and the inherent balancing actions of audio coupling using transformers - use the "0.01%" resistor matching analogy to illustrate just how difficult (and essentially futile) it is to achieve in this scenario.
Unfortunately, they rather let down their technical analysis by stating that....

"A BALANCED input or output uses two signal conductors which have equal impedances to ground"

And then presenting a "Typical Balanced Input" schematic where the impedances of the two inputs are different by 100%..
They then go on the illustrate a driver using polarised (presumably electrolytic?) capacitors where you you be hard pushed to match by better than 20%

So, all in all, not a very 'accurate' paper..... except for the suggestion to use transformers to achieve the desired results, rather than messing around trying to 'match' things with 0.01% resistors.

Which rather adds to my original point......

 

One of my college electronics instructors tells about how he paid his way through college working at Ohmite painting stripes on resistors. All the resistors came off the same assembly line, 20%, 10%, and 5%. The only difference in the quality was how accurately they were measured. Without testing, the manufacturing process could pretty well assure a tolerance of 20% but not more. So his job was to go through huge piles of nominally 20% resistors, and sort the ones that fell within 10% in one pile, and 5% in another.

Now....here's the trick. At the end of the week, all the unsorted resistors got shoved off the bench into the 20% pile....no fourth stripe. If you BOUGHT a big pile of 20% resistors that were manufactured on a Friday, there would be X number of 10 and 5% resistors....you'd just have to measure them yourself!

Eric the Old

The problem with sorting resistors is that the 20% resistors would have a bimodal distribution containing no resistors at the nominal value. The tighter tolerance resistors have been selected out and sold at a higher price.

Now consider the consequences in a circuit in a production line. You end up with circuit performance in two batches, neither at the required designed value,

 

The test books don't always help. In the attached image - taken from a National Semiconductor analogue data book from the 1980s - you can see that R13 is marked as 93.1K. In series with R14 ---a 10k variable resistor!

Why on earth specify 93.1K when you adjust the overall value by such a huge amount? Doesn't seem to make any sense...

Never could figure that one out......View attachment 73440

I can't see the point to the series either... OTOH... I often place small trimmers and fixed resistors in parallel to nail down a precice figure. The remaining question, is how to deal with the paralleled tolerances, when your tweaked resistance is created from the combination...

 

Hopefully the following will help connect some dots.

Per the Jensen app note, achieving the common mode rejection of a transformer, in a differential amplifier, requires matching the positive and negative gains very closely. The schematic below is an audio differential amplifier. In the highlighted box, R13 and R14 form an adjustable 100K resistor. Adjusting R13/14 allows you to match the positive gain to the negative gain. This adjustment directly affects the amplifier's ability to reject common mode noise.

Rogs, the Jensen app note is a conceptual paper. The differential amplifiers in it simply represent a function. In real life they would be very close to the above schematic.

 

Hopefully the following will help connect some dots.

Per the Jensen app note, achieving the common mode rejection of a transformer, in a differential amplifier, requires matching the positive and negative gains very closely. The schematic below is an audio differential amplifier. In the highlighted box, R13 and R14 form an adjustable 100K resistor. Adjusting R13/14 allows you to match the positive gain to the negative gain. This adjustment directly affects the amplifier's ability to reject common mode noise.

Rogs, the Jensen app note is a conceptual paper. The differential amplifiers in it simply represent a function. In real life they would be very close to the above schematic.

.......

.....I'm not really clear why you suggested the Jensen paper as an 'answer that would do me some good' with my understanding of the details of CMRR techniques, if it is just a 'conceptual' paper?.......with schematics that contradict the text.!......

I'm most grateful for your insights into how CMRR is best addressed, but really, there's no need for you to trouble further. I do get it!!.....

The point I was trying to make is that the inclusion of a stupid E96 values like '93.1K' into a textbook schematic is not a good way to introduce beginners into the best way of selecting resistors.

The idea of adjusting a CMRR offset, accurately and easily to within 0.1%, with a single turn variable resistor of 10% of the resistor value is nonsense.

A ten turn version? .. OK, you've shown that could work. But that's way over the top for this kind of audio application.

And if a 10k variable is to be used, then I would have thought a 95.3K resistor -- (which is an E48 as well as an E96 value) -- would be a more logical choice? Closer to the 'centre' value of the adjustable range.

But as I say, the inclusion of circuits in text books with those kind of obscure values, where the 'real world' solutions are much simpler, does not help students early on.....

Even Jensen themselves have addressed the simplification of the CMRR problem. You might like to take a look at this paper:
http://www.jensen-transformers.com/an/ingenaes.pdf
Hopefully, it may 'do you some good', to use your own phrase..........