Trying to choose a moderately decent meter... help!

[HKJ]

mV range is often high impedance.
xV range is usual 11Mohm (10Mohm+1Mohm)
xxV range is 10.1Mohm (10Mohm+100K),

Using any voltage higher than 0.xV to measure the impedance of the mV range will trigger the overload protection and give a considerable lower value (The mV range may have a 1Mohm resistor and then the overload protection).

In my reviews I am very careful to use a low voltage when checking impedance of the mV range. This spoils the precision due to noise, but I can see if it is 10Mohm or considerable higher.

[exe]

Using any voltage higher than 0.xV to measure the impedance of the mV range will trigger the overload protection and give a considerable lower value (The mV range may have a 1Mohm resistor and then the overload protection).

I'm pretty sure my measurements were affected by this, btw... I was too lazy to check this.

[David Hess]

Having a variable input resistance is typical for cheap meters because it simplifies autoranging.  This is not a problem for manual ranging meters and higher quality autoranging meters where the input resistance is constant.  Some meters disconnect the input attenuator entirely on the lowest voltage ranges which has advantages and disadvantages.  None of this has anything to do with input protection except insofar as the input resistance will change under overload conditions.

But if the specifications say "around 10M", then I expect the typical single ended autoranging divider which will vary between about 11M and 10M.  If the specifications unambiguously say "10M" like on the Amprobe 38XR-A, then anything else is a lie which is why I asked if anybody actually knows; I would be very interested in a new handheld meter which has a real 10M input.

My good meters have an input resistance of 10M within 0.1% no matter what range is selected.

If this seems like an overblown concern, then consider how much time can be wasted making precision measurements with an external divider when the meter's input resistance changes with range.

[HKJ]

Having a variable input resistance is typical for cheap meters because it simplifies autoranging.

Calling Fluke meters cheap is not entirely correct. I just checked Fluke 179 and 289 and they both goes between 10 and 11Mohm.

[exe]

consider how much time can be wasted making precision measurements with an external divider when the meter's input resistance changes with range.

Just use calibration on all ranges . Honestly, for me it's more "an implementation detail", rather than a serious consideration.

PS if probe loading is a problem, I think using 1Gohm DMM is a better solution.

[Lightages]

Price. Quality. Features. You can't get all for the same price. You need to pick what is more important to you. You have a budget that does not meet hat is available. The UT61E is probably the closest but I cannot recommend it for many reasons. If you want to spend more then I recommend the BM257S but it is a whole bunch more. But..... nobody ever complained about buying quality.

[Russ]

I am a newbee here. But I am really liking my Eevblog/Brymen 235 meter.

Russ

[HalFET]

Even some of the high-end the 7.5 digit meters change input impedance slightly when they range switch, but it's relatively speaking a lot smaller. Also you need an electrometer to spot it 

[David Hess]

Even some of the high-end the 7.5 digit meters change input impedance slightly when they range switch, but it's relatively speaking a lot smaller. Also you need an electrometer to spot it 

This only happens if the high impedance input buffer has excessive leakage.  The leakage current creates a variable error term which is dependent on the variable parallel output resistance of the divider.  I suspect this is a major contributor to the  "+ digits" parts of the error specifications on most multimeters.

Charge injection can also contribute to measurement errors of input resistance which is why I keep a couple of old multimeters around which do not suffer from this.

I am still looking for a modern handheld multimeter which has a constant 10 megohm input resistance.  I have found several which say they do but I am dubious.  Uni-Trend for instance does on the ones that I checked and I do not believe them at all.  B&K says their manual ranging models are 10 megohms which I believe but they are all low end models.  I guess I will have to ask Amprobe directly about their 37XR-A and 38XR-A.  To give Fluke credit, at least they say "about 10M" instead of just "10M".

[David Hess]

Calling Fluke meters cheap is not entirely correct. I just checked Fluke 179 and 289 and they both goes between 10 and 11Mohm.

Compared with high-end bench top DMMs with active input tracking amplifier, providing many GOhms of input impedance.

My complaint is about the variation in resistance and obtuse specifications detailing it rather than the magnitude as long as it is high enough.  It is absurd (except to the marketing department) to have high resolution meters with a variation in input resistance which causes additional errors greater than the basic accuracy or the 2nd to the least significant digit between ranges.  Too many times now I have either spent hours tracking down a phantom circuit problem (1) or helping someone else do so when it was the multimeter lying.

As far as Fluke, what am I to think about the Fluke 25, 27, 83, 85, 87, and 88?  One Fluke reference (2) warns about the variable input resistance of these meters but the manuals unambiguously say 10 megohms "nominal" without indicating that the input resistance is between 10M and 11.1111M and likely only "nominally" 10 megohms on the most sensitive range.  My multimeters which have an invariant input resistance of 10 megohms +/-0.1% are "nominally" 10 megohms.  That is marketing at its finest desiring to eat its cake and have it to look at.  What the user does not know will not hurt them, or at least will not hurt Fluke.  At least some manufacturers indicate greater than 10 megohms or list exact values. (3)

The difference between 10 and 11 megohms may not seem like much; after all, they are both high, right?  But what source resistance would create as much error as the multimeter's basic accuracy or even worse, 10 counts or more with a change of 1 megohm between ranges?  For 1% it is about 100 kilohms.  For 10 counts on a Fluke 87, it is about 5 kilohms!

So Fluke is cheap except of course in price.  They are not offering anything that their cheap competitors are not offering except their name and often less.

(1) It is usually a non-linearity problem which appears to be caused by excessive input bias current or poor common mode rejection.

(2) The manual for the Fluke 80k-40 high voltage probe gives detailed instructions for compensating for an input resistance other than 10 megohms.  So just plug-in the value for the range and get a correction factor.  Hmm, so what value do I use for "10M nominal"?  Is it the same "nominal" value for all Fluke meters?

(3) The Fluke 179 specifications say "> 10M" for all volt ranges.  The Fluke 289 specifications say "10 M" for all volt ranges but is that "nominally" meaning it could actually be 11.1111M or something else?  I do not know what to believe.

[Fungus]

The difference between 10 and 11 megohms may not seem like much; after all, they are both high, right?  But what source resistance would create as much error as the multimeter's basic accuracy or even worse, 10 counts or more with a change of 1 megohm between ranges?  For 1% it is about 100 kilohms.

The reading would be wrong anyway so what difference does it make?

Moral: Don't measure a 100k impedance voltage source with a "10 Meg" multimeter, nominal or otherwise. Reading voltages on high impedance sources needs special care/attention/tools.

[David Hess]

The difference between 10 and 11 megohms may not seem like much; after all, they are both high, right?  But what source resistance would create as much error as the multimeter's basic accuracy or even worse, 10 counts or more with a change of 1 megohm between ranges?  For 1% it is about 100 kilohms.

The reading would be wrong anyway so what difference does it make?

Moral: Don't measure a 100k impedance voltage source with a "10 Meg" multimeter, nominal or otherwise. Reading voltages on high impedance sources needs special care/attention/tools.

The problem is not the reading being wrong.  The problem is the reading being inconsistent between ranges beyond the accuracy specifications.

[Mr. Scram]

Price. Quality. Features. You can't get all for the same price. You need to pick what is more important to you. You have a budget that does not meet hat is available. The UT61E is probably the closest but I cannot recommend it for many reasons. If you want to spend more then I recommend the BM257S but it is a whole bunch more. But..... nobody ever complained about buying quality.

Exactly, you won't regret buying a good one. It doesn't have to be a silly top end meter, but a decent one like the BM257s will last you a long time. It has all the basic features you'd want and many more, is nicely built, has a bar graph and can even do datalogging if you someday wish to do so. More importantly, it's safe. You can be stupid and you will not hurt yourself, and likely not hurt the meter either. It'll only cost you your pride and a fuse.

It's better to have a few good tools than many tools you can't depend on. A good one will serve you 5, 10 or 15 years from now.

[Mr. Scram]

I have never once said "Geeze, I really wish this had a backlight".  Who probes in the dark?
Accuracy can be important, but very often precision is all you really need.

I'd hate to lose the backlight on any of my meters. Unless you have a lot of light on the bench, it can be convenient to have a backlight. If you're measuring things around the house or in the field, having a backlight is an absolute necessity.

[David Hess]

That brings up another point.  Our discussion of obscure issues like variation in input resistance may be interesting but is largely irrelevant for most applications.  So is the number of digits and counts and basic accuracy.  Sacrificing these things for toughness, quality, required features, and ease of use is a better trade off for most users.

As far as LCD backlights, I have never found myself in a position where one was required.

[HKJ]

As far as LCD backlights, I have never found myself in a position where one was required.

On my bench I have put up enough light that I do not need backlight, but I have sometimes been in location where backlight was nice.

[Fungus]

That brings up another point.  Our discussion of obscure issues like variation in input resistance may be interesting but is largely irrelevant for most applications.  So is the number of digits and counts and basic accuracy.  Sacrificing these things for toughness, quality, required features, and ease of use is a better trade off for most users.

Yep, 0.5% is plenty for just about everything. Anything beyond that is nice but won't have much impact in real circuits.

It's already way beyond the tolerance of most components used outside the test gear world.

PS: The Fluke 87 is really only a 6000 count meter at heart. The trick is to be 6000 counts +/- 1.

[Mr. Scram]

Yep, 0.5% is plenty for just about everything. Anything beyond that is nice but won't have much impact in real circuits.

It's already way beyond the tolerance of most components used outside the test gear world.

PS: The Fluke 87 is really only a 6000 count meter at heart. The trick is to be 6000 counts +/- 1.

And to still be that or close to it after 15 years.

[Fungus]

The Fluke 87 is really only a 6000 count meter at heart. The trick is to be 6000 counts +/- 1.

And to still be that or close to it after 15 years.

That, too.

eg. After Joe's latest tests the Fluke 17B+ started looking less overpriced than before. All the other meters were falling apart or grinding themselves to dust after a few thousand twists of the knob, even the fancy Agilents. The 17B+ was basically fine even after the full 50,000 twists.

[Helix70]

You would have to pry my Flukes from my cold dead hands.

[Russ]

I am a newbie to all this and currently enrolled in a Mechatronics program. Can I assume that it will be awhile before I outgrow my Eevblog/Brymen 235 DMM? How does it compare to Dave’s new 121GW model?

Thanks
Russ

[Fungus]

I am a newbie to all this. Can I assume that it will be awhile before I outgrow my Eevblog/Brymen 235 DMM?

Yes.