Accuracy !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

[Telequipment]

I have noticed on some other channels accuracy of voltages measurements on even 60000 count meters, seem to stir interest,maybe unhealthy interest,& i wonder WHY,surely in hobbyist electronics it doesn't have to be that accurate does it, take a 5 volt measurement 5.000007, surely the .0000007 , what does it matter, are some getting paranoid about this, doesn't near enough count anymore, personally I'm not bothered , as long as it's near enough and works, take a resistor I was looking for a 100 ohm in my box I found one 97.5 ohm, good enough, the circuit works, are there people out there that will only put in a 99/100 ohm resistor in, after all it's a hobby isn't it.

[Rerouter]

it depends on what your doing, if you know your meter is accurate to those counts (remembering counts reference precision not accuracy) and for instance your confirming or calibrating some other bit of equipment those few counts may be important, for instance a frequency standard of 10.000000Mhz as opposed to 10.000070Mhz is critical to very high bandwidth applications, otherwise you have frame alignment issues, 

same for resistance, perhaps your making a mega-ohm bridge to measure even higher resistances, those few ohms difference can really count

please recall most of us applying out skills in some way at our respective jobs and we are not all kiddies playing around with an arduino dev board and a linear power supply, some people actually need high accuracy measurements,

[Psi]

yep, most of the time it really doesn't matter.

However there are exceptions, even for a hobbyist.

Suppose they were trying to measure a 0-20V signal using a 5v MCU adc, something thats pretty common. They might use a 4x voltage divider so 20V = 5V.

If they used 5% resistors, say 2k and 6k
then the worst cases are
2100R  5700R =  3,7x     10V input becomes 10.8V
1900R  6300R =  4.35x    10V input becomes 9.2V

Which is pretty damn terrible.

[amyk]

take a resistor I was looking for a 100 ohm in my box I found one 97.5 ohm, good enough, the circuit works, are there people out there that will only put in a 99/100 ohm resistor in, after all it's a hobby isn't it.

If it was a 100R 5% it's perfectly within tolerance, on the other hand that would be defective for a  2% or less. It definitely depends on the application.

[AndyC_772]

I have a 6.5 digit meter, and yes, most of the time it's overkill for many purposes.

The extra resolution over and above a hand held meter is sometimes useful, though. For example, a while ago I had a job which involved measuring the resistance of power cables, and so the ability to accurately measure very small resistances was essential. I could even see how the resistance changes with temperature as the cable warmed up.

It's that resolution that's most useful for me, rather than the absolute accuracy.

[Telequipment]

Oh I see, well I'll just keep playing with my toy circuits and having fun

[SeanB]

In many cases 3 digits are fine, some 4 digits are fine but you always want to check with something that has more digits than your device. Then you can be sure that the error is in the device under test, not part of the uncertainty of the source. If you try to check a meter with 2 decimal point of reading, and the source has an error that is 5 least significant digits at the 2 decimal setting then either the meter being checked is inaccurate or the source is, but you cannot tell which. If the source is accurate to 3 digits then you can tell that the meter being checked is accurate.

Then again most people are happy to 1 decimal or even to no decimal point. It depends, you would not like to pay for food where they say 10kG with an accuracy of 1 kg, either you get 9kg or 11 kg, and as the manufacturer will not like to lose all will be 9kg.

[Telequipment]

In many cases 3 digits are fine, some 4 digits are fine but you always want to check with something that has more digits than your device. Then you can be sure that the error is in the device under test, not part of the uncertainty of the source. If you try to check a meter with 2 decimal point of reading, and the source has an error that is 5 least significant digits at the 2 decimal setting then either the meter being checked is inaccurate or the source is, but you cannot tell which. If the source is accurate to 3 digits then you can tell that the meter being checked is accurate.

Then again most people are happy to 1 decimal or even to no decimal point. It depends, you would not like to pay for food where they say 10kG with an accuracy of 1 kg, either you get 9kg or 11 kg, and as the manufacturer will not like to lose all will be 9kg.

Hi Sean mine is a 22000 count, which doesn't mean anything to me that's what i bought,  it works ok.

[Bloch]

after all it's a hobby isn't it.

Or even work.

 


And I do blame Dave for it 


No more serious there are a some ... that do not know then or that details that are important.


If there are someone to blame is maybe the "maker movement" vs instead of skilled / schooled.

[c4757p]

Sometimes it is your interest. My favorite hobby area is measurement. I don't just buy accurate and precise equipment, I actually like to design and build it. In that case, I don't have them because I naively think I need them to make Blinkenlights, I have them because I just like them.

[KJDS]

It's been very rare in 25 years of professional RF design that I've needed better then 0.1% accuracy.

[AlfBaz]

To some of us, be it a hobby or work, we have a keen interest in the very wide field of electronics and having sensitive high precision gear lets us see things you normally can't. Clearly its not needed for day to day measurements but its nice to have it when you need it.

Here's a poor example but it's recent so...

I had my laptop sitting on top of a 6.5 digit meter doing a trend plot of a resistance. at first glance the plot looked very erratic, jumping up and down all over the place. I turned the laptop off and surprise, surprise it calmed down substantially. Nominal resistance of 17.980k, the part that looked erratic was jumping a mere 0.3 of an ohm. Would never have seen that with lower precision

[Bloch]

Nominal resistance of 17.980k, the part that looked erratic was jumping a mere 0.3 of an ohm. Would never have seen that with lower precision

 


So that was the lesson learn ?

  The higher Accuracy the more difficult it will be to measure the right value. So how often do 99.9999987654321% need to know that that a 17980 ohm is jumping peak +-0,15 ohm ?

[Amarbir[Lynx-India]]

To some of us, be it a hobby or work, we have a keen interest in the very wide field of electronics and having sensitive high precision gear lets us see things you normally can't. Clearly its not needed for day to day measurements but its nice to have it when you need it.

Here's a poor example but it's recent so...

I had my laptop sitting on top of a 6.5 digit meter doing a trend plot of a resistance. at first glance the plot looked very erratic, jumping up and down all over the place. I turned the laptop off and surprise, surprise it calmed down substantially. Nominal resistance of 17.980k, the part that looked erratic was jumping a mere 0.3 of an ohm. Would never have seen that with lower precision

Damn ,
      Laptop Noise Trickling To The Precision 6.5 Circuit 

[c4757p]

Not everything is about what you need. Sometimes it's fun to learn about how things work. I love seeing how little things like that affect measurements. Hobby electronics especially isn't just about getting something done by a deadline.

[Rufus]

after all it's a hobby isn't it.

For a proportion of people here it isn't just a hobby. Don't know what that proportion is, might be an interesting topic for a poll.

[AlfBaz]

Nominal resistance of 17.980k, the part that looked erratic was jumping a mere 0.3 of an ohm. Would never have seen that with lower precision

 


So that was the lesson learn ?

  The higher Accuracy the more difficult it will be to measure the right value. So how often do 99.9999987654321% need to know that that a 17980 ohm is jumping peak +-0,15 ohm ?

It amazes me that engineering ever got beyond using a rock as a hammer.

The religious fervour with which some people approach certain subjects is beyond me. I provided precious little in my post with regards to why I was doing such measurements, so for you to go off on some half cocked tangent without having any data to back your misguided assertions, leads me to believe that you are a person with little to no regard to free and open discussion about technical matters. If that is the case then how about abstaining from threads that whish to explore matters that you disagree with.

To me this thread is about exploring situations that require precision measurements and I attempted to participate citing the caveat

a poor example

[c4757p]

We've got two extremes here who always chime in. There's the "you can't do anything without two 8.5 digit multimeters and a four-channel 1 GHz oscilloscope made in the last two months" brigade, and the "you're wasting your money if you buy more than a beer, a hammer, a 9V battery and a length of nichrome to make resistors" brigade. Anyone who advocates moderation invites one or both.

[AlfBaz]

you're wasting your money if you buy more than a beer...

Sometimes I think I would be of saner mind if I converted my lab into a home brew setup

[ejeffrey]

It's been very rare in 25 years of professional RF design that I've needed better then 0.1% accuracy.

I'm working on a system right now that needs better than 0.01% amplitude accuracy microwave pulses (.001% would be better, but is really hard), _but_ we calibrate them in-situ.  Even if we had an instrument that could measure to that accuracy, we would never be able to get it correct at the DUT due to connector and cable losses.  So we use a high resolution DAC and mixer, and find a way to measure when the amplitude is correct.  We do care about stability, and we do have to recalibrate every day to get to that level, but it works.

The point is, sometimes accuracy matters, but sometimes you can find a better way to measure it than brute force.  When possible that is almost always preferable.  The only real exception is frequency.  Frequency is so easy and cheap to make accurate that if you care about it you might as well always start with 9 digit accuracy.

To be sure, there are some applications where you really need an instrument that can measure 6.5 digits.  In real engineering the trick is always to figure out what you actually need, and what you can get away with.

[Telequipment]

See here we go again, All you pro's out there missing the point, glad your not doctors, you'd scare me to death, my point IS, for a hobbyist  like me, who enjoys the hobby making things that works is the fun, I at my level just cannot see the point of measuring down to 6,7,8,9,10,11,digits or what ever else. AND as note a" HOBBYIST"  making all these measurements down to .000001 or whatever is just every so slightly Anal. That's all, reel your necks back in and calm down a bit, so some of you are pro's that's fine. I'm not, I think by some of the replies ,some of you should take a step back and take a good look at yourselves. for some of us it's a hobby isn't it.

[G7PSK]

It seems to me that there are some in the world who worry about the number of digits on the meter and not the accuracy. A case of never mind the accuracy as long as its got more digits than an Arab states gas meter, just watch the numbers go around is the name of the game. 99% of what I do can be achieved with an analogue meter quite satisfactorily, and most probably applies to many out there as well, but we are now in the digital world and the more digits there are the better it sells, I expect that a three and half digit Fluke is more accurate than many of the six or more digit cheapo Chinese meters. Most accuracy at the end of the day will depend on the user any way.

[Telequipment]

It seems to me that there are some in the world who worry about the number of digits on the meter and not the accuracy. A case of never mind the accuracy as long as its got more digits than an Arab states gas meter, just watch the numbers go around is the name of the game. 99% of what I do can be achieved with an analogue meter quite satisfactorily, and most probably applies to many out there as well, but we are now in the digital world and the more digits there are the better it sells, I expect that a three and half digit Fluke is more accurate than many of the six or more digit cheapo Chinese meters. Most accuracy at the end of the day will depend on the user any way.

I quite agree, yes I have a fancy meter bought for me , by a sister in law who has far to much money,but she's my sister in law and is lovely so a welcomed gift, the meters I use for everything  are analogue A Farnell, and a AVO

[Dave]

There is one thing you aren't taking into account: the challenge. I (and probably many other forum members) like pushing the boundaries of what's possible. If you want to create precision circuitry, you need the right tools for the job. This is where 6.5 digit multimeters and rubidium frequency standards come into play.

[c4757p]

See here we go again, All you pro's out there missing the point, glad your not doctors, you'd scare me to death, my point IS, for a hobbyist  like me, who enjoys the hobby making things that works is the fun, I at my level just cannot see the point of measuring down to 6,7,8,9,10,11,digits or what ever else. AND as note a" HOBBYIST"  making all these measurements down to .000001 or whatever is just every so slightly Anal. That's all, reel your necks back in and calm down a bit, so some of you are pro's that's fine. I'm not, I think by some of the replies ,some of you should take a step back and take a good look at yourselves. for some of us it's a hobby isn't it.

Did you miss the part where we said sometimes measurement is the hobby? Obviously you don't need that sort of thing if that's not what you're interested in, but don't try to say precision is useless because hobby.