Fluke 17B accuracy?

[IanB]

Here is an interesting picture:

http://www.candlepowerforums.com/vb/showthread.php?324810-Good-cheap-multimeter&p=3780807&viewfull=1#post3780807

I didn't expect that of a Fluke meter, even if it is a low end model.

[Kiriakos-GR]

Is this a trap question ?   

All the DMM units are just products, some can fail prematurely.
This is where the Warranty plan comes handy if you have one.

[vl400]

My 17B seemed pretty close to other meters I had available (https://www.eevblog.com/forum/index.php?topic=4068.msg54357#msg54357), but as mentioned in the link you posted his test (and mine) was just one unit.

[Uncle Vernon]

I didn't expect that of a Fluke meter, even if it is a low end model.

Expect what of a fluke meter? We have three low cost devices in a dubious test situation. What is to say any of them are correct?

The error is less than 1% the quoted accuracy of the 17B is  DC Volt : 400mV / 4V / 40V / 400V / 1000V , +/-1.0% + 10

The TI chip accuracy is quoted at 1%. I doubt very much the millispec part is used, I can see the rest of the parts are no better than 1%. What is the accuracy repeatability of the rest of the device?
Sure it's checked against a laboratory device. That is before shipping storage, operator manhandling etc. Were the temperature, humidity etc identical to those When was the calibration pot was last tweaked (knocked)?

Seriously if a couple of sub $100 meters and a mail order reference board are that close in a knock about workshop test what is the problem? 
The Fluke could be spot on, there could be error each way or thing could be as pictured.

Anything requiring better accuracy shouldn't be using this class of device.

[RCMR]

My 17B concurred almost exactly with my other two budget DMMs when I compared it.  Just as (in)accurate.

I'm hoping that the difference is the Fluke will *stay* accurate for longer than the other two.

[IanB]

I didn't expect that of a Fluke meter, even if it is a low end model.

Expect what of a fluke meter? We have three low cost devices in a dubious test situation. What is to say any of them are correct?

I expect the Fluke meter to be better.

The error is less than 1% the quoted accuracy of the 17B is  DC Volt : 400mV / 4V / 40V / 400V / 1000V , +/-1.0% + 10

IMHO, that accuracy is terrible and is setting the bar absurdly low. I have meters costing $2.99 that can do better than that. Why should anyone expect so little of a meter?

The TI chip accuracy is quoted at 1%. I doubt very much the millispec part is used, I can see the rest of the parts are no better than 1%. What is the accuracy repeatability of the rest of the device?
Sure it's checked against a laboratory device. That is before shipping storage, operator manhandling etc. Were the temperature, humidity etc identical to those When was the calibration pot was last tweaked (knocked)?

I have the same voltage reference device. It is designed with care and guaranteed to be accurate within 0.01% for 6 months. I have checked mine against calibrated HP bench meters to confirm this. It is stable and repeatable from day to day and week to week.

Seriously if a couple of sub $100 meters and a mail order reference board are that close in a knock about workshop test what is the problem?

Mainly I don't consider an error of 40 mV in 5 V to be very acceptable, given how easy it is to not have an error of 40 mV. It's not a matter of low cost meters or being within specs, it's a case of meter vendors being given a free pass by customers with low expectations when they could do better.

[vtl]

I've got the Fluke 17B as well. On DC volts it reads 4.96 on a 5V 0.2% TI REF02 chip.

Link here:
http://www.ti.com/lit/ds/symlink/ref02.pdf

This is pretty crap since my $6 meter reads better than that. The Fluke 17B has a shitload of trimmers on the PCB but I haven't been able to find a service manual. Does anyone have one?

[alm]

This reading is within specs, as stated. The specifications are clearly documented and anyone being surprised by them should have done better research. Sure, the specifications are quite loose, but this is their low-cost model only intended for sale in Asia (China?). It's not their fault that people have been importing them with hopes of getting a Fluke 110 series meter for cheap.

[vtl]

The Fluke 17B is actually 0.5% not 1% so these meters are out of spec

[alm]

0.5% plus 3 digits.

[amspire]

The Fluke 17B is actually 0.5% not 1% so these meters are out of spec

0.5% + 3 digits means that 4.95v to 5.05v would be within specs, so 4.96v scrapes in.

[saturation]

This model was targeted as general purpose electrician level DMM, it meets the spec it was designed for in the Asian market.  Despite that, Fluke name on it would be some assurance that it would live up to its spec sheet and build quality.

However, seeing the abuse DMM get by workers around me, not electrical mostly physical, I can see why pro electricians, HVAC and other techs choose US sold Fluke even if its not rated as IP54 or similarly ruggedized, as its field proven to survive daily grinds.  Generally the accuracy for electrician use its in the 0.5-1% for DC and 1-3% for AC, but the US sold ones [ even if made in China] are made with better housings.

Agilent's move in the DMM field has produced quality but similarly expensive DMM.

[alm]

4.95 to 5.05 assuming the reference is exactly 5V. If you would do proper error propagation including the uncertainty of the reference, the confidence interval would be larger.

[Kiriakos-GR]

4.95 to 5.05 assuming the reference is exactly 5V. If you would do proper error propagation including the uncertainty of the reference, the confidence interval would be larger.

I agree with alm , plus I do not believe that those big spenders they payed the full price, so to buy a reference like the LT1021-10 with with an initial tolerance as low as ± 0.05%.
This move removes the uncertainty of the reference from the equation.

[grenert]

I can't believe anyone is complaining about a multimeter that IS IN SPEC.  Also, trying to claim that one meter is more accurate than another based on ONE READING on ONE SCALE for ONE FUNCTION is absolutely ridiculous.

If you got lucky and your $2 throwaway is more accurate on this one measurement, then be happy.  The reason the Fluke is more expensive is because most likely it will meet all of its specs, survive your abuse, and probably continue to be within spec for a very long time.  If you're whining that you want more accuracy, then you need to choose a meter with better stated accuracy, not hope that the Fluke name magically guarantees better accuracy than specified.  The specs on cheapo meters are usually just lies, which the meter may or may not meet.

[IanB]

I can't believe anyone is complaining about a multimeter that IS IN SPEC.  Also, trying to claim that one meter is more accurate than another based on ONE READING on ONE SCALE for ONE FUNCTION is absolutely ridiculous.

If you got lucky and your $2 throwaway is more accurate on this one measurement, then be happy.  The reason the Fluke is more expensive is because most likely it will meet all of its specs, survive your abuse, and probably continue to be within spec for a very long time.  If you're whining that you want more accuracy, then you need to choose a meter with better stated accuracy, not hope that the Fluke name magically guarantees better accuracy than specified.  The specs on cheapo meters are usually just lies, which the meter may or may not meet.

Bullshit.

It's my belief that meters in that class have a single reference for their DC volts scales and all measurements are ratio-ed against that. Once the basic adjustment is made all the rest is about linearity and stability. I've tested various meters and once they are adjusted to hit one test voltage they hit all test voltages.

Secondly the quoted accuracy of a meter is designed to allow a margin of error. When the meter is manufactured it should be adjusted to be close to the middle of the error band so that if it drifts in the future it will remain in spec. If the meter is initially adjusted right on the edge of the error band then any drift in the wrong direction will take it out of tolerance.

Doing this adjustment is expensive of course. First of all they should "burn in" the meters to let them stabilize, and then the meters should be adjusted carefully, not rapidly set up by a production line worker turning a screwdriver on the clock. I think they cut corners on this and that is why readings can be out on the edge.

It's my opinion that manufacturers are being given a free pass here and customers are not setting expectations high enough. The primary job of a meter is to measure and that is where the first critical evaluation of a meter lies. It's not enough to just scrape by, I expect products to exceed expectations. Others should too.

[alm]

It's my belief that meters in that class have a single reference for their DC volts scales and all measurements are ratio-ed against that. Once the basic adjustment is made all the rest is about linearity and stability. I've tested various meters and once they are adjusted to hit one test voltage they hit all test voltages.

There is the uncertainty/drift of the reference, which is constant for all ranges. In addition, there is the uncertainty in the divider, which is why the highest/lowest ranges tend to have the worst accuracy specs. Modern meters with closed case calibration tend to store two values (0V and near full scale) per range, to compensate for non-ideal divider resistors. Cheap meters tend to have divider networks with worse stability. This meter was described as having a bunch of trimmers, some of these might be for adjusting divider ratios.

Secondly the quoted accuracy of a meter is designed to allow a margin of error.

Calibration often involves adjustment if the value is not within 70% of the accuracy spec. Sometimes procedures only call for adjustment if a range is out of spec.

It's my opinion that manufacturers are being given a free pass here and customers are not setting expectations high enough. The primary job of a meter is to measure and that is where the first critical evaluation of a meter lies. It's not enough to just scrape by, I expect products to exceed expectations. Others should too.

Before condemning a manufacturer, I would first make sure that my measurement was valid. This involves comparing this to equipment with traceable calibration. It seems likely that the meter is almost out-of-spec, but it's by no means certain. You can never rely on a piece of equipment giving you lower uncertainty than the manufacturer's specs, unless you characterize the instrument yourself. This involves measurements over a range of environmental conditions and over a time period, not a single spot check.

[IanB]

Before condemning a manufacturer, I would first make sure that my measurement was valid. This involves comparing this to equipment with traceable calibration. It seems likely that the meter is almost out-of-spec, but it's by no means certain. You can never rely on a piece of equipment giving you lower uncertainty than the manufacturer's specs, unless you characterize the instrument yourself. This involves measurements over a range of environmental conditions and over a time period, not a single spot check.

I can accept this. I have measured my own voltage reference device at various times and temperatures to check for consistency and repeatability and have compared it with a calibrated HP3458A. I was very interested to verify the claims made for it. I am finding it to be very stable from day to day and week to week, and according to the HP3458A the voltage is indeed within the claimed limits.

The picture of the Fluke 17B is not mine though, and I cannot say whether the owner of that meter has done the same things I have.

[Achilles]

Well, I don't know, but I think the meter is just fine. It's not as accurate as other Fluke meters, but it's a lot cheaper. You may get a good one (low error) or a bad one (high error) within the specs. None of them equals another. If you need more precision you have to pay for it. At least they don't state it's 0.1% and ship it, because they won't offer any warranty or so anyway.  There may better cheapos for the price, but you may remeber:

-It's for Asian market. Maybe they want to sell "High Quality". Here we have Fluke (179 for example) and if you are on a Budet you take an Extech, Voltcraft or whatever. There you have a Fluke and take a Vichy when you need some spare money left.
-Next Point, it's a Fluke. So you may pay a lot just to have the name on it and show you got it. Same with cars....buy a Ferrari, but some parts are from Fiat...buy a Porsche and enjoy VW Technology.....pay more, have a status symbol.....
-Warranty, I don't know if they offer a warranty but I could Imagine that. A lot of cheap chinese suppliers don't care about that.

So for a 100USD Meter it's Ok if you want a Fluke. Otherwise, there are plenty of other meters around to pick from.

[kek]

My 17B was out a bit too. I downloaded the calibration manual and tweeked R18 a bit to get dc volts closer.
I have a couple of fluke 8050a's that I have adjusted to agree with the fluke 289 that I convinced my employer that we needed at work.
All my meters still agree with the 289.
Google 15b17b__cmeng0000.pdf to find a copy of the cal manual for the 17B.

Ken

[IanB]

All my meters still agree with the 289.

That's really the fascinating point. Even the cheapest meters seem to have good stability down to the 0.1% level or better. I have found meters I own to remain rock solid and consistent in their readings over one year, two years, as long as I have owned them. It's the original factory adjustment where they seemingly don't try very hard. Maybe they don't actually adjust them at the factory? Maybe they just check them and if they are reasonably within tolerance they let them go. Which is interesting. Maybe the quoted accuracy is not so much a technical limitation as a process limitation?

[vtl]

My 17B was out a bit too. I downloaded the calibration manual and tweeked R18 a bit to get dc volts closer.
I have a couple of fluke 8050a's that I have adjusted to agree with the fluke 289 that I convinced my employer that we needed at work.
All my meters still agree with the 289.
Google 15b17b__cmeng0000.pdf to find a copy of the cal manual for the 17B.

Ken

Thanks mate, was trying to look for that for ages.

All right sure enough its within spec (albeit 10mV within spec). But I don't think its unreasonable to expect better. People keep saying its because its made in China for the China market. Surely thats irrelavent? If you saw dave's $50 and $100 shootout none of them was that bad. This meter still costs around $100 and is not a $5 cheapo and if you look at the build quality is actually excellent.

[kek]

I've aquired quite a few meters at home and work since I've been watching Dave's blogs, high quality meters such as agilent 1242b, fluke 289, fluke 17B, fluke 8050a, fluke 79 II and some low cost ones such as the sparkfun VC830L and mastercraft 052-0060-2 (crappy tire), and also had a couple of meters from years ago: brunelle instruments 4060, tenma 72-2055. The meter that was most out was the 17B.

Now I just have to convince my wife to buy me a HP34410a for Christmas.

Ken

[Achilles]

Thanks mate, was trying to look for that for ages.

All right sure enough its within spec (albeit 10mV within spec). But I don't think its unreasonable to expect better. People keep saying its because its made in China for the China market. Surely thats irrelavent? ....

well, not that irrelevant. It is common to export higher grades and keep lower grades for inland consumption to keep a good reputation on international markets. Switzerland with cheese, Panama with banana, costa rica with coffee....... not so unusual.
There may be sure better catches for the money,sure, but maybe the Fluke is more secure...don't know. I wouldn't thing because it is a special brand it has to be great....
I think the 17B and it's chinese brothers were released there to get an low level entry in the asian market. With all the meters around there on the market it might be tough to keep a foot on ground. Well, and I don't think a Chinese worker/Factory would buy a triple price meter made in USA because everyone wants one . So get them a cheap entry with good build quality and make them want more (higher accuracy for example).

[Bored@Work]

well, not that irrelevant. It is common to export higher grades and keep lower grades for inland consumption to keep a good reputation on international markets.

And the other way is also common. This area produces a special alcoholic beverage. The good stuff is all sold locally, and the horse piss version that no one here would touch is exported.