Test Equipment Anonymous (TEA) group therapy thread

[bitseeker]

A neat alternative to the 34401A for a 6.5-digit bench DMM, if you don't mind the color scheme, is the Keithley 196. Starburst red LED display!

No need to ask. Of course, I had to have one, especially when one comes along at an impulse buy price.

[Berni]

My 6.5 digit DMM of choice is my Keithley 2015. For some time you could get a really good deal for them on ebay.

[bd139]

Unfortunately, at least here in the UK, no more. They're 500 smackers plus now.

[Berni]

I looked back and i got mine for 390 USD in October 2016. It came in like new condition with the protective film still on the display.

[Specmaster]

There was a time when everyone knew what the voltage, current or resistance was and there was never any question about it, it was what it was precisely. When was that, it was when everybody could afford to relinquish their analogue meters and go digital and we all had 3.5 digits and that new meter also gave us other things to worry about. That was because most of the service manuals/charts had their voltage references taken and stated as being measured with a meter @ 20,000Oopv and voltages we were reading were much higher because the new meters had an input impedance of around 10,000Mopv had a significantly lower loading on the circuits under test.

Then someone came along with a 4.5 digit meter, which is not too bad as it happens, then came a 5.5 digit, a 6.5, 7.5 and 8.5 digits and suddenly no one really knows what the voltage is anymore. A case in point, I'm sitting here looking at my 3.5digit DMM's connected to a voltage source, I've connected 10 up and they are agree with each other, 1.948V fine. Then I connect the same voltage source (untouched) to my 4.5 digit DMM's and now that extra digit is constantly changing, its now 1.9485, no wait its 1.9482 no wait its 1.9488 and so one.

Next I swap out the 4.5 digit meters for my 5.5 digit ones and WTF, I now have the last 2 digits constantly changing so what is the actual voltage that I'm reading?? The true constant here is the 1.943v which remain steadfast anything after the 3 just really cannot be relied upon as it only takes a gnat to fart and the reading changing again and again and again.

Its much the same story when switched to measuring resistance but this time the only digit that changes is the last on the 5.5 meters.

So the results that I'm getting is that the more digits the meter has, the more unsure of the actual value you become but tend to to go with either the first 3 or 4 digits after the decimal point so it would seem that the extra digits become just points in a game of brag or top trumps rather than serve a useful function. How many times have I either read it or heard people say that that they would have to have their meter professionally calibrated, especially so if it were to be 6.5 or higher digits because it cause them to worry about those last digits really were? As if they really matter in the grand scale of things 

Now then the real big question is if there was to be a 6.5 digit DMM that came within my price budget would I go for it? You'd better believe I would so I can join in the game of top trumps and besides there is something hypnotic about seeing all of those digits light up, even if they aren't giving me or the equipment I'm working on any real advantage   

[bd139]

Lets just say we need 6.5 digits for dynamic range AND resolution at the same time.

A contrived example without going into offset drift, picoamp bias, chopper amps, regulator standby currents and all the cool stuff that needs a 6.5 digit meter (or differential voltmeter but bugger that thread fork):

Consider a 20mV signal on a DC bias of 20V. What resolution can you measure that to on a 4.5 digit meter (assume 22 based count on both)?

4.5 digits: 20.020V     -> 20.??mV
6.5 digits: 20.02005V -> 20.05mV

Adding count uncertainty or say +/- 2 counts:

4.5 digits: 18 - 22mV
6.5 digits: 20.03-20.07mV

[Specmaster]

Lets just say we need 6.5 digits for dynamic range AND resolution at the same time.

A contrived example without going into offset drift, picoamp bias, chopper amps, regulator standby currents and all the cool stuff that needs a 6.5 digit meter (or differential voltmeter but bugger that thread fork):

Consider a 20mV signal on a DC bias of 20V. What resolution can you measure that to on a 4.5 digit meter (assume 22 based count on both)?

4.5 digits: 20.020V     -> 20.??mV
6.5 digits: 20.02005V -> 20.05mV

Adding count uncertainty or say +/- 2 counts:

4.5 digits: 18 - 22mV
6.5 digits: 20.03-20.07mV

Fair points, I hadn't considered the higher voltage ranges, perhaps it had never occured to me that there ever would be a need to be that precise (assuming the meter is able to stop bouncing about on the last few digits) on say 20v or 100v ranges etc as the actual percentage difference is so small in comparison to the really low voltage ranges. For example ,amy low powered devices have a cutoff voltage below which they don't function, such as handheld DMM's where as little .1v can be the difference between running and switch off. Somehow I just couldn't see that level of value being as critical at the higher end of the scale.

[djos]

I finally found a half decent DSO, that I could afford, 2nd hand - it was half the cost of a new 1054z so I'm happy ... anyhoo, it's a UNI-T UTD2102CEL 100MHz, 1GS/s unit in retina searing Red and Black!  It's a huge upgrade from my virtually useless Velleman WFS210 Wifi POS. 



 

[Specmaster]

Nice unit, why not add a splash of colour to the lab? 

[Cerebus]

There was a time when everyone knew what the voltage, current or resistance was and there was never any question about it, it was what it was precisely. When was that, it was when everybody could afford to relinquish their analogue meters and go digital and we all had 3.5 digits and that new meter also gave us other things to worry about. That was because most of the service manuals/charts had their voltage references taken and stated as being measured with a meter @ 20,000Oopv and voltages we were reading were much higher because the new meters had an input impedance of around 10,000Mopv had a significantly lower loading on the circuits under test.

Then someone came along with a 4.5 digit meter, which is not too bad as it happens, then came a 5.5 digit, a 6.5, 7.5 and 8.5 digits and suddenly no one really knows what the voltage is anymore. A case in point, I'm sitting here looking at my 3.5digit DMM's connected to a voltage source, I've connected 10 up and they are agree with each other, 1.948V fine. Then I connect the same voltage source (untouched) to my 4.5 digit DMM's and now that extra digit is constantly changing, its now 1.9485, no wait its 1.9482 no wait its 1.9488 and so one.

Next I swap out the 4.5 digit meters for my 5.5 digit ones and WTF, I now have the last 2 digits constantly changing so what is the actual voltage that I'm reading?? The true constant here is the 1.943v which remain steadfast anything after the 3 just really cannot be relied upon as it only takes a gnat to fart and the reading changing again and again and again.

Its much the same story when switched to measuring resistance but this time the only digit that changes is the last on the 5.5 meters.

So the results that I'm getting is that the more digits the meter has, the more unsure of the actual value you become but tend to to go with either the first 3 or 4 digits after the decimal point so it would seem that the extra digits become just points in a game of brag or top trumps rather than serve a useful function. How many times have I either read it or heard people say that that they would have to have their meter professionally calibrated, especially so if it were to be 6.5 or higher digits because it cause them to worry about those last digits really were? As if they really matter in the grand scale of things 

Now then the real big question is if there was to be a 6.5 digit DMM that came within my price budget would I go for it? You'd better believe I would so I can join in the game of top trumps and besides there is something hypnotic about seeing all of those digits light up, even if they aren't giving me or the equipment I'm working on any real advantage 

That sounds more like something wrong rather than an intrinsic characteristic of using a high resolution/long scale meter. I'd take seeing that as an indication of either a horrifically noisy environment, poor measurement technique or something wrong with the meter.

For what it's worth, I rarely see last digit instability on a 5 1/2 digit meter unless it represents a genuine feature of the signal and my measurement technique is generally quite sloppy with the exception that I do use 4 wire techniques for resistance measurements as a matter of course as I have kelvin clips and keep them handy.

If the last digit or two is changing for me on something that ought to be rock solid I take that as an indicator that I need to change my measurement technique or hook the scope up to make sure that what I think ought to be present actually is (e.g. HF instability showing up on a signal that ought to be DC). If I'm on 2 wire ohms, switch to four wire, if I'm measuring voltage, add a guard connection to suck up common mode currents and so on. I'd suggest a read of the Keithley "Low Level Handbook" to brush up on measurement artefacts and their sources and see if that isn't your problem - the name's a bit of a misnomer, as the issues and techniques covered apply as much to precision measurements as low level ones.

[Specmaster]

I was measuring the output from a cheap basic bench power supply which I thought would be pretty much what an average hobbyist would be likely to use. I agree if I was using something a little more sophisticated from say HP or TTi etc then the last digit would be more stable. That can be demonstrated if I connect my AD584-M instead of the power supply, set it on 10v it does improve dramatically so the 8840A reads 10.0076 and the 3478A reads 10.0075/4 and the discrepancy there is probably down to the 8840A having massive input impedance advantage (10,000M) over the 3478A's 10M.
 

[djos]

Nice unit, why not add a splash of colour to the lab? 

Cheers, It should meet my amateur needs quite nicely think .... at least till upgraditis kicks in! 

[med6753]

There was a time when everyone knew what the voltage, current or resistance was and there was never any question about it,

Snipped....


Source: AD584-M -  9.99691V

Results after half hour warm-up. Hardly worth fussing over.

It's the variability of the source that drives you nuts.

[Specmaster]

It's the variability of the source that drives you nuts.

You are spot on, if the source is varying, even slightly, with the higher resolution meter, the more the least significant digits are going to keep altering in an effort to reflect the changing input. Whereas with fewer digits, that change has got to be much bigger in order to alter the least significant digit. 3.5 meter, no change in display, 4.5 meter sometimes the last digit will go up or down by 1, 5.5 meter, last 2 digits were changing all the time on the power supply. Its like a proverb, man with 3.5 and 4.5 digits, knows the voltage, man with 5.5 or more digits is unsure of the voltage due to last digits changing.

[Specmaster]

I'm in the garage tomorrow looking some angle brackets so I can fix my function generator and LCR meter to underside of my shelf. I'd like to mount the TF930 up there too but I'm worried about spoiling the resale value of it later if I upgrade it.

Looks like I've started a trend. You must post pix 

As promised here is my revised (for now) bench, what you can't see is dotted around the room in any suitable space I can are 4 CRO's, signal generator, numerous, microscopes, handheld DMM's, analogue meters, portable LCR devices, soldering kit, jigs, isolating transformer, variac, power meter, parts racks, calibrators and tools etc. I'd love to have a room big enough to take a proper 2 or 3m bench and shelves then I could have all the kit out and ready to be used, maybe this week its my turn to win the lottery....

[med6753]

It's the variability of the source that drives you nuts.

You are spot on, if the source is varying, even slightly, with the higher resolution meter, the more the least significant digits are going to keep altering in an effort to reflect the changing input. Whereas with fewer digits, that change has got to be much bigger in order to alter the least significant digit. 3.5 meter, no change in display, 4.5 meter sometimes the last digit will go up or down by 1, 5.5 meter, last 2 digits were changing all the time on the power supply. Its like a proverb, man with 3.5 and 4.5 digits, knows the voltage, man with 5.5 or more digits is unsure of the voltage due to last digits changing.

Quite frankly the 5.5 digits is overkill for me. I could get along fine with 4.5 or even 3.5 digits. But it's nice to have....just in case. 

6.5 digits? Only if drops from heaven for free.   

[med6753]

I'm in the garage tomorrow looking some angle brackets so I can fix my function generator and LCR meter to underside of my shelf. I'd like to mount the TF930 up there too but I'm worried about spoiling the resale value of it later if I upgrade it.

Looks like I've started a trend. You must post pix 

As promised here is my revised (for now) bench, what you can't see is dotted around the room in any suitable space I can are 4 CRO's, signal generator, numerous, microscopes, handheld DMM's, analogue meters, portable LCR devices, soldering kit, jigs, isolating transformer, variac, power meter, parts racks, calibrators and tools etc. I'd love to have a room big enough to take a proper 2 or 3m bench and shelves then I could have all the kit out and ready to be used, maybe this week its my turn to win the lottery....

Looks good!  You learned well. 

[Specmaster]

It's the variability of the source that drives you nuts.

You are spot on, if the source is varying, even slightly, with the higher resolution meter, the more the least significant digits are going to keep altering in an effort to reflect the changing input. Whereas with fewer digits, that change has got to be much bigger in order to alter the least significant digit. 3.5 meter, no change in display, 4.5 meter sometimes the last digit will go up or down by 1, 5.5 meter, last 2 digits were changing all the time on the power supply. Its like a proverb, man with 3.5 and 4.5 digits, knows the voltage, man with 5.5 or more digits is unsure of the voltage due to last digits changing.

Quite frankly the 5.5 digits is overkill for me. I could get along fine with 4.5 or even 3.5 digits. But it's nice to have....just in case. 

6.5 digits? Only if drops from heaven for free.   

Same thing here, nothing I do actually warrants more than 3.5 digits, the others are just nice to have and TBH, I'd love to have a 8.5 digit one as well just for the sheer hell of it and I'd still stand by my other post about for most applications 3.5 or 4.5 meters are adequate.

Interestingly I have just done a 2 hour test of these meters using my AD584-M and there was only 2 meters that did not change their readings, the Fluke 27 handheld and a HP3466A, the others all changed by 1 digit so not to shabby.   

[GreyWoolfe]

I'm in the garage tomorrow looking some angle brackets so I can fix my function generator and LCR meter to underside of my shelf. I'd like to mount the TF930 up there too but I'm worried about spoiling the resale value of it later if I upgrade it.

Looks like I've started a trend. You must post pix 

As promised here is my revised (for now) bench, what you can't see is dotted around the room in any suitable space I can are 4 CRO's, signal generator, numerous, microscopes, handheld DMM's, analogue meters, portable LCR devices, soldering kit, jigs, isolating transformer, variac, power meter, parts racks, calibrators and tools etc. I'd love to have a room big enough to take a proper 2 or 3m bench and shelves then I could have all the kit out and ready to be used, maybe this week its my turn to win the lottery....

Nice.  Love the 2 HP 3466As.  I have 2 myself.  Don't forget to post this in the what's your bench look like thread.

[bitseeker]

It's the variability of the source that drives you nuts.

You are spot on, if the source is varying, even slightly, with the higher resolution meter, the more the least significant digits are going to keep altering in an effort to reflect the changing input. Whereas with fewer digits, that change has got to be much bigger in order to alter the least significant digit. 3.5 meter, no change in display, 4.5 meter sometimes the last digit will go up or down by 1, 5.5 meter, last 2 digits were changing all the time on the power supply. Its like a proverb, man with 3.5 and 4.5 digits, knows the voltage, man with 5.5 or more digits is unsure of the voltage due to last digits changing.

TEA Proverb #382: Man with more digits knows when he needs to buy a more stable power source (or replace the filter caps).

[mnementh]

It's the variability of the source that drives you nuts.

You are spot on, if the source is varying, even slightly, with the higher resolution meter, the more the least significant digits are going to keep altering in an effort to reflect the changing input. Whereas with fewer digits, that change has got to be much bigger in order to alter the least significant digit. 3.5 meter, no change in display, 4.5 meter sometimes the last digit will go up or down by 1, 5.5 meter, last 2 digits were changing all the time on the power supply. Its like a proverb, man with 3.5 and 4.5 digits, knows the voltage, man with 5.5 or more digits is unsure of the voltage due to last digits changing.

Quite frankly the 5.5 digits is overkill for me. I could get along fine with 4.5 or even 3.5 digits. But it's nice to have....just in case. 

6.5 digits? Only if drops from heaven for free.  

*Thumps chest*

Church!!!

It's the variability of the source that drives you nuts.

You are spot on, if the source is varying, even slightly, with the higher resolution meter, the more the least significant digits are going to keep altering in an effort to reflect the changing input. Whereas with fewer digits, that change has got to be much bigger in order to alter the least significant digit. 3.5 meter, no change in display, 4.5 meter sometimes the last digit will go up or down by 1, 5.5 meter, last 2 digits were changing all the time on the power supply. Its like a proverb, man with 3.5 and 4.5 digits, knows the voltage, man with 5.5 or more digits is unsure of the voltage due to last digits changing.

TEA Proverb #382: Man with more digits knows when he needs to buy a more stable power source (or replace the filter caps).

Or get rid of PWM-regulated lighting. 


In other news...


     

It's evolving. 


mnem
*Toddling off to watch Tomb Raider with the wife*

[bitseeker]

It's evolving. 

Can we make test equipment do that?

Alice: "Hey, how come your 34401A has one more digit than mine? "
Bob: "It just evolved last night."
Alice: "How'd it do that?"
Bob: "I'm not sure. I did leave it monitoring the output of my 1kV power supply for a while yesterday."

[Berni]

As long as said test gear doesn't 'evolve' in the form of suddenly saying on boot up:
"Selftest Clock board: FAIL (222)"

[tggzzz]

Oh dear.

I think I'll have to get an 8800A, just so I can fondle those switches.

I really wish people wouldn't post geekporn pictures.

[bitseeker]

I really wish people wouldn't post geekporn pictures.

Shhh, don't say stuff like that. It increases the probability of the opposite occurring.