DMM6500 or 34465A

[TrickyNekro]

Hi!

I am thinking of going the extra mile and buying a proper bench DMM. I am already going through older forum posts but I would like to ask this question again now,
2 years after the last post I saw on the forum from googling around.

I do not really need the low Ohm range of the DMM6500 for what I do. Data logging is of greater interest.

I lean towards the 34465A due to availability in my country, better financing and support in my country.
Performance wise I lean towards the DMM6500 due to higher sampling rates but I do not know how well you
keep both precision and high sampling rates, it´s more of the impressions the DMM6500 left me from watching
reviews rather than concrete data, that´s why I also post here, to listen to experience.

I will not shy away from a third option that provides better performance in this price range, not only performance wise though
but also UI wise, I really want to be able to do some nice scripting and datalogging over ethernet.

What´s your opinion? I have not found a Youtube video yet that breaks down all the options pros and cons, would be really
happy to watch some!

Use is going to be a mix bag of hobbyist and professional use. Now I will be using it as a hobbyist / student, but I am looking
forward to use it one day professionally, target is low voltage devices, 48V tops, anything from light industrial actuators to IoT.

Thanks again for your replies in advance!
Best Regards,
Lefteris

[nctnico]

My general impression is that the DMM6500 has more features and could be considered better in many ways. OTOH it seems that the DMM6500 has some firmware issues but I don't know whether these have been ironed out yet. I own a Keysight 34461A myself but probably would have liked the DMM6500 better if it was available when I bought the 34461A.

[Kleinstein]

The higher speed mode of the DMM6500 use a different ADC with lower accuracy. The Bandwidth is limited by the input amplifier, so that the full sampling rate is of limited use. I don't know the speed of the input amplifier in the 34465, but is may be similar speed.

The 34465 is quite a bit lower noise, when it comes to just the ADC - at least at lower speed.
Normally the 34465 is quite a bit higher cost. Cost wise the DMM6500 is more comparable to the slower 34461.

The DMM6500 has the option to use a scanner card.

[daqq]

I have a DMM6500 at work, best multimeter buy ever. Great, responsive and intuitive user interface and what's also great is the option for expansion cards ( https://assets.testequity.com/te1/Documents/pdf/keithley/2000-SCAN-manual.pdf ) - this turned out to be VERY useful for us.

In terms of parameters it's OK for me.

Not saying that the 34465A is a bad instrument - I've worked with the 34461A, which at a quick glance is very similar to the 34465A. Good device, but I'd go for the DMM6500 any day.

[TrickyNekro]

The higher speed mode of the DMM6500 use a different ADC with lower accuracy. The Bandwidth is limited by the input amplifier, so that the full sampling rate is of limited use. I don't know the speed of the input amplifier in the 34465, but is may be similar speed.

Suspected as much, I will have to go through the specs and multiple reviews one by one.

The 34465 is quite a bit lower noise, when it comes to just the ADC - at least at lower speed.
Normally the 34465 is quite a bit higher cost. Cost wise the DMM6500 is more comparable to the slower 34461.

True, the difference is about 500€ which of course is something to consider as it´s almost 1/3 of the price of the 34465A.
But over 6 months payments interest - free, these extra 85€ are not that much of a pain.

Giving though 1100 - 1200€ just like that for the DMM6500 is going to be very painful. Plus I have to be buying from Germany probably.
There is supposedly a local distributor but their site left me with the impression that it was left a bit behind. Sure that´s only my impression
but you know... gut feelings...

But the ADC being less noisy is actually reason enough to prefer one over the other. I need to find the use cases where on is better than the
other really to be able to make a decision.

The DMM6500 has the option to use a scanner card.

I think that´s the DAQ6510 ( or is the DAQ6510 just a DMM6500 with this card? ). Which then goes about the same price as 34465A so meh pretty much the same.


Anyways, thanks for your reply!

[TrickyNekro]

Not saying that the 34465A is a bad instrument - I've worked with the 34461A, which at a quick glance is very similar to the 34465A. Good device, but I'd go for the DMM6500 any day.

Is something specific that makes you say that? This feature, that feature etc., I need to dig in a bit.

Thanks

[daqq]

Not saying that the 34465A is a bad instrument - I've worked with the 34461A, which at a quick glance is very similar to the 34465A. Good device, but I'd go for the DMM6500 any day.

Is something specific that makes you say that? This feature, that feature etc., I need to dig in a bit.

Thanks

General feel, how the graphing is done, user interface responsivity - somehow they made me like a touch screen on a measurement instrument, a thing unheard of! The numeric entry is usable... lots of UI stuff. But also the fact that it's damn fast and damn responsive.

And the scanner card option is really a great feature. Tad pricey, or you can build your own:
https://www.eevblog.com/forum/circuit-studio/example-project-relay-scan-card-for-k2000-dmm/

Plus the parameters are good.

I think that´s the DAQ6510 ( or is the DAQ6510 just a DMM6500 with this card? ). Which then goes about the same price as 34465A so meh pretty much the same.

Nope. The DMM6500 supports Keithley scanner cards (see 2000-SCAN and others).

[Someone]

I lean towards the 34465A due to availability in my country, better financing and support in my country.
Performance wise I lean towards the DMM6500 due to higher sampling rates but I do not know how well you
keep both precision and high sampling rates

The accuracy and precision at different sampling rates are relatively well documented in the data sheets, I think it gets more complex as the DMM6500 has a different "mode" for the fastest rates.

Don't forget that Keysight have models up and down from the 34465A, so there are many points to compare if you are pressed on budget. For example I have used the very similar 34461/65/70 meters and never found any killer app for the higher sampling rates, as high speed current and voltage could be handled by suitable scope probes. Equally for low current use I skip right past multimeters and use an SMU to eliminate burden. So whats best value is really in context of what other equipment you have, and the very specific measurement needs of your applications.

user interface responsivity - somehow they made me like a touch screen on a measurement instrument, a thing unheard of! The numeric entry is usable... lots of UI stuff. But also the fact that it's damn fast and damn responsive.

It took some time and many false starts, but the best touchscreen interfaces are now ahead of button/knob only. As you say, its the speed of the interactions that ends up critical for UIs, some button/knob interfaces fail there too!

[TrickyNekro]

Don't forget that Keysight have models up and down from the 34465A, so there are many points to compare if you are pressed on budget. For example I have used the very similar 34461/65/70 meters and never found any killer app for the higher sampling rates, as high speed current and voltage could be handled by suitable scope probes. Equally for low current use I skip right past multimeters and use an SMU to eliminate burden. So whats best value is really in context of what other equipment you have, and the very specific measurement needs of your applications.

Exactly because I have a scope, I am leaning towards the 34465A. Right now, I find the 34461A and the 34465A with a ~500€ difference, because though I can pay over 6 months, it doesn´t really dent me.
Or else I have to collect the money over the next 6 months to buy the DMM6500 and that´s at least 6 months without the instrument which could be a thing as then I would have to run between many places
to have access to measurement level DMM, plus the temptation of drinking that extra coffee or beer. 

The DMM6500 from reviews so far has me a bit more "wow"-ed. But I do not want to buy a 1k+ class instrument on "wow" factor. 

Still I wish to go the extra mile cause I know this will become part of my profession, so that´s only early investment, a worth one though.

I will go through old Dave videos and return with more specific questions. I really can´t remember if he made a comparison of the two. Sure
he has reviewed it, he and / or the "Signal path" guy ( name difficult to remember sorry! ).
But on the Signal path channel, I can not remember a direct comparison of the two.

Ah well...

[HighVoltage]

I have both and I like them both and really could not say that one is more recommended than the other.

I like the many direct physical knobs of  34465A but I also like the very intuitive touch screen of the DMM6500
It also depends on what you want to do with the DMM.

The 34465A is my daily "go to" instrument on the bench for most work.
It is very reliable and at least on my unit the calibration is rock solid even after a few years.

In order for you to decide, you might get a loaner unit of each and test them for a week or so for free.

[TrickyNekro]

Looking at this old Dave video, what I understood, at least from the video, is that the 34465A is a 34470A waiting for a LTZ reference to happen.
So if I would go to the Keysight route getting the 34465A over the 34461A is worth it.

Video here:

[TrickyNekro]

I think that´s the DAQ6510 ( or is the DAQ6510 just a DMM6500 with this card? ). Which then goes about the same price as 34465A so meh pretty much the same.

Nope. The DMM6500 supports Keithley scanner cards (see 2000-SCAN and others).

I see, I see. Found out the details about it also. To be honest though, I play around labview and know how to program a microcontroller or two. I could build such a card myself externally,
regardless of the DMM controlling it with even a simple Arduino or any other USB enabled microcontroller. That being said, I do not really need the multiple inputs from such a scan-card.

I would totally understand the need in a production environment with automated testing, where you do not want to play around.

Of course on the other hand, both seem to have nice labview drivers so given the need automation is not going to be an issue.

I will go through Dave´s DMM7510 review now. Try to find how hard the differences between the 6.5 and the 7.5 are.

Cheers!!!

[TrickyNekro]

Well I am going through the specs and it really makes it a hard choice.

The 34465A seems to be more accurate apart from the low Ohm range, which in all honesty I do not care all that much, I do not build shunts every day and even when I use them
there are other ways to calibrate them.
Generally with their "truevolt" trademark, they got better DC and AC stability. And a big difference also comes from the AutoCal feature, especially for the long term stability.

The DMM6500 has the interface, "wow-factor" low Ohm, higher diode testing voltage and so on going for it. If I was comparing with the 34461A then the DMM6500 would clearly
be better, much better as a feature and performance ratio.

Of course the 34465A costs quite bit more but that ACAL... that ACAL really makes a difference long term. Do I really need it? Yeah, I guess kinda, yes.

The DMM6500 has also better programmability and scripting, but it would be yet another scripting language I got to learn and then I will forget.
Since both companies support labview then I would much prefer to set up a labview program given the need for automation. Also from some random comments
I saw that people tended to be more happy with BenchVue and it is provided for free from what I understood. A confirmation on this would be great.

One great thing that the 34465A has, is the low current range. Since I work with microcontrollers and have also done some IoT before, current draw datalogging seems really
possible.

It is a really hard choice really. Between the DMM6500 and the 34461A, I would have gone with the DMM6500 hands down. Now it´s much more tempting to go with
the 34465A after I have done the comparison, but it´s also tempting to keep the money difference and equip my lab with a function gen, or something similar that is still
missing.

Tough call really, if the dealership comes through with a good enough offer, I will go for the 34465A. Else most probably I will go with the DMM6500.


If there is something I absolutely have to know, speak now or else be silent forever! 

Cheers!
Lefteris

[bdunham7]

I think  you might want to look at those specs again, perhaps post what you are looking at.  I don't see the 34465A as being more accurate.

One issue I have run into (with a different meter) is that while the DMM6500 uses a different current shunt for each of the lower ranges, the 34465A appears to use the same shunt for all 3 of the lowest ranges and recycles some others.  The DMM6500 has 7 different shunts, which is a big commitment to accuracy.  When you reuse a shunt and apply gain, you get some noise--it's unavoidable.  The 34465A probably has better digital trickery to smooth that noise, but their specs for the 3 lowest current ranges look suspect to me--I don't know why they include the comment "(typ)".  I'd ask.

Also, I find an accurate 10R range to be very useful.

These are just two observations.  I don't have either meter so I can't say much else.

[Kleinstein]

From the specs of the 34465 it looks like the 10 mA range is the same shunt as 100 mA, just with additional gain. This is visible from higher percentage error ralative to the range. The small ranges look more like they use a TIA instead of classical shunts. With just shunts one could hardly get that level of accuracy with so little burden. For stability a TIA still need some series resistance and thus the burden is not zero, but can be quite small.

The typical remark with the small ranges is likely mainly from leakage of the input protection, restricting it too room temperature.
They explain it at the front. A 80% confidence level would imply about an additional factor 2 in comparisom with the normal 95% or 99% level - though this may not be a normal distribution thing, more like a question of testing / rejecting.

I agree that the 34465 is better in digital handling the data to reduce noise. However this is more like a flaw on the Keithley side, that tends to add some LF noise with integration / averaging over some 0.5 - 50 seconds.

ACAL function can help to get very good long term stability (e.g. > 5 years) and allows then manufacturer to use cheaper parts in some places. It especially helps when comparing reading measured in different ranges (e.g. 1 V and 100 V) as ACAL bings that ratio essentially back to the 24 h specs, while without ACAL it would be more like the 1 year specs, unless directly after calibration with adjustment (or using measurend correction factors).
AFAIK CAL in the 34465 does no include all ranges ( espeically AC).

[HighVoltage]

Of course the 34465A costs quite bit more but that ACAL... that ACAL really makes a difference long term. Do I really need it? Yeah, I guess kinda, yes.

The ACAL on the 34465A works well but you only need it, if you have large temperature changes in your lab.
I don't think it has any influence on the long term stability.

[beanflying]

Well I am going through the specs and it really makes it a hard choice.

The 34465A seems to be more accurate apart from the low Ohm range, which in all honesty I do not care all that much, I do not build shunts every day and even when I use them
there are other ways to calibrate them.
Generally with their "truevolt" trademark, they got better DC and AC stability. And a big difference also comes from the AutoCal feature, especially for the long term stability.

snippity

If there is something I absolutely have to know, speak now or else be silent forever! 

Cheers!
Lefteris

If you seriously need low Ohms accuracy then a dedicated MilliOhm meter will run rings around the 34401A, 34461A (the two I have that are kept in Cal) and likely the 65A and the DMM6500 as well. So I would tend to cross this item off you list all together as a consideration in some ways and put the $4-500 saved toward a dedicated Milliohm meter. The Quadtech I scored some time ago is my go to and not my 6 1/2 or 7 1/2 digit meters (Advantest in the fleet too) for below a few Ohms.

Without rechecking the complete spec the 34465A is a hard sell for the extra $ unless you really need the high speed reading side of it the accuracy differences are fairly minimal.

Overall the question should be is how important are the last 10PPM to you now and into the next 5-10 years and if they are keep saving and buy a 3458A (got one of those in the repair que)

[TrickyNekro]

If you seriously need low Ohms accuracy then a dedicated MilliOhm meter will run rings around the 34401A, 34461A (the two I have that are kept in Cal) and likely the 65A and the DMM6500 as well. So I would tend to cross this item off you list all together as a consideration in some ways and put the $4-500 saved toward a dedicated Milliohm meter. The Quadtech I scored some time ago is my go to and not my 6 1/2 or 7 1/2 digit meters (Advantest in the fleet too) for below a few Ohms.

That´s the thing. I do not really. I much prefer a good DC current accuracy, to calibrate my shunts with their respective amplifier rather than the exact resistor accuracy. So I get I am a bit sold there.

Overall the question should be is how important are the last 10PPM to you now and into the next 5-10 years and if they are keep saving and buy a 3458A (got one of those in the repair que)

I will leave it for a day in greener pastures, when my kidneys are not at risk of being sold.

[TrickyNekro]

Of course the 34465A costs quite bit more but that ACAL... that ACAL really makes a difference long term. Do I really need it? Yeah, I guess kinda, yes.

The ACAL on the 34465A works well but you only need it, if you have large temperature changes in your lab.
I don't think it has any influence on the long term stability.

I will clarify this, but that´s ok even so. Summers here are brutal and you can´t be hugging the A/C all the time.

[daqq]

I would totally understand the need in a production environment with automated testing, where you do not want to play around.

Actually I had to log 4 temperatures from PT1000s of a device for a few hours. It was not part of an automated test jig, just an experiment.

Also, have you considered older meters? If you are concerned about user interface stuff, meters like the venerable 34401A, Keithley 2000 or similar can be found reasonably cheaply. In terms of parameters they aren't bad at all.

[TrickyNekro]

I think  you might want to look at those specs again, perhaps post what you are looking at.  I don't see the 34465A as being more accurate.

I was comparing from these two:
https://download.tek.com/datasheet/1KW-61315-0_DMM6500_Datasheet_052119.pdf
https://www.keysight.com/zz/en/assets/7018-03846/data-sheets/5991-1983.pdf

Granted the differences between the two are not earth shattering and maybe not worth the extra 500€.

One issue I have run into (with a different meter) is that while the DMM6500 uses a different current shunt for each of the lower ranges, the 34465A appears to use the same shunt for all 3 of the lowest ranges and recycles some others.  The DMM6500 has 7 different shunts, which is a big commitment to accuracy.  When you reuse a shunt and apply gain, you get some noise--it's unavoidable.  The 34465A probably has better digital trickery to smooth that noise, but their specs for the 3 lowest current ranges look suspect to me--I don't know why they include the comment "(typ)".  I'd ask.

Yes, that is something to "worry" about. Although that being said, this is the exact reason they have better burden voltages and I had it before that I had a problem with a device browning out / resetting
when exiting sleep mode.

So you can either take it as a bug or as a feature really... If they meet the spec, I wouldn´t really care, in the end of the day, even with the "caveats" these are top of the line companies, we are talking about.

[TrickyNekro]

I would totally understand the need in a production environment with automated testing, where you do not want to play around.

Actually I had to log 4 temperatures from PT1000s of a device for a few hours. It was not part of an automated test jig, just an experiment.

I am very sure that the day will come that I will say, "darn I should have bought the other". But that´s probably why one day I will try to buy both

Also, have you considered older meters? If you are concerned about user interface stuff, meters like the venerable 34401A, Keithley 2000 or similar can be found reasonably cheaply. In terms of parameters they aren't bad at all.

Ah, yes. But you know it is not the same. I know it is probably the illusion, and you are 100% right, but let me be happy, ok? 

[bdunham7]

I was comparing from these two:
https://download.tek.com/datasheet/1KW-61315-0_DMM6500_Datasheet_052119.pdf
https://www.keysight.com/zz/en/assets/7018-03846/data-sheets/5991-1983.pdf

Granted the differences between the two are not earth shattering and maybe not worth the extra 500€.

What I'm not seeing is how you come to the conclusion that the Keysight is better at all.  They're quite clearly head-to-head competitors, but the Keithley appears to be equal or better on everything except low and high frequency AC voltage.  I didn't read every spec--they both are probably more than good enough on most ranges--but you may need to pick the ones that matter.  I would be very wary of the asterisked lower current ranges.  Perhaps you were looking at some other specs that matter more to you?  If you look at features, they're very similar, but the Keysight does have the 1G resistance range while the Keithley has 1R and 10R and a better selection of diode ranges.  And of course the scan card option.  ACAL vs autozero is another issue that I think is mostly a non-issue in most cases, but I wouldn't regard ACAL as an advantage unless the meter is used in widely varying (outside the 18-28C window) temperatures.

Yes, that is something to "worry" about. Although that being said, this is the exact reason they have better burden voltages and I had it before that I had a problem with a device browning out / resetting
when exiting sleep mode.

So you can either take it as a bug or as a feature really... If they meet the spec, I wouldn´t really care, in the end of the day, even with the "caveats" these are top of the line companies, we are talking about.

Unless they have implemented some very special low-noise amplification in the lower current ranges, it's a bug-period.  You aren't going to be measuring individual picoamps because you are limited by the noise, accuracy and input bias current of the voltage section.  With a shunt for each current range, it is easy enough to get lower burden voltages by just selecting the next range up.  With the Keysight setup, you don't have a choice.  As I said I'm sure they are both excellent products, but I would have to ding Keysight here for a bit of marketeering and perhaps specsmanship.  I don't like the "(typ)" denotation because I think they are quoting a non-guaranteed spec.  If I needed thos ranges to be accurate for my work, I'd be pressing them to explain the details of that spec or finding someone that has reviewed and tested the meter in those ranges. I think their blurb on 'low input bias' is a sham with the bar graphs showing how they are better than the competition but they won't quote an input bias spec--and ignore the decades-old standard Fluke 8846A with its typical 15pA and guaranteed <30pA bias current.  Comparing your high-end 6.5 digit DMM to brands 'A, B and C' when none of the three are Fluke is a bit deceptive, IMO.  And if they are comparing their 'typical' measurement to a guaranteed spec, that's deceptive too.

Anyway, none of that should stop you from buying either meter if you like them and I'm not advocating for either.  There are other non-spec related issues as well, but since I don't have the meters I can't comment on the UI or things like that.

[TrickyNekro]

I was comparing from these two:
https://download.tek.com/datasheet/1KW-61315-0_DMM6500_Datasheet_052119.pdf
https://www.keysight.com/zz/en/assets/7018-03846/data-sheets/5991-1983.pdf

Granted the differences between the two are not earth shattering and maybe not worth the extra 500€.

What I'm not seeing is how you come to the conclusion that the Keysight is better at all.  They're quite clearly head-to-head competitors, but the Keithley appears to be equal or better on everything except low and high frequency AC voltage.  I didn't read every spec--they both are probably more than good enough on most ranges--but you may need to pick the ones that matter.  I would be very wary of the asterisked lower current ranges.  Perhaps you were looking at some other specs that matter more to you?  If you look at features, they're very similar, but the Keysight does have the 1G resistance range while the Keithley has 1R and 10R and a better selection of diode ranges.  And of course the scan card option.  ACAL vs autozero is another issue that I think is mostly a non-issue in most cases, but I wouldn't regard ACAL as an advantage unless the meter is used in widely varying (outside the 18-28C window) temperatures.

I got the feeling that yesterday night I confused the charts between them, or something. They are clearly head to head, in some cases one being slightly better than the other but nothing tragic.
In a day I could have a variation of 5 - 6 °C in the lab, especially in summer but it´s not the be all, end all of the game here.

The diode range is really a thing, but dirty old me says I will just hook my power supply in series in case I need the extra voltage.

I will try to find reviews on the lower current ranges. It is difficult for me to test them. I do not expect to count pA either to tell the truth. I need better accuracy on the 1uA - 100mA range than anything else,
with probably a single shunt that will not switch when a device wakes up.
Forcing a range is something, I have not thought really and that´s actually a good point to gravitate towards the Keithley.

Anyway, none of that should stop you from buying either meter if you like them and I'm not advocating for either.  There are other non-spec related issues as well, but since I don't have the meters I can't comment on the UI or things like that.

I really think it will come down to the dealership making me a good offer here. 500€+ difference is a big one. I could get me a second Rigol bench power supply or function gen for the same money.
Which is a lot if you think about it. I´ll see the next days really...

[Hexley]

... target is low voltage devices, 48V tops, anything from light industrial actuators to IoT.

If you plan to measure power supply current in a typical IoT device, it will be useful to use the 10A range of your 6.5 digit meter in order to get a very low burden voltage. The 34465A spec sheet alludes to this in the DC/AC current burden voltage spec, where they show that the burden voltage of the 1A range can be reduced from "<0.7 V" to "<0.05 V", with a footnote for the latter saying to use the 10A input range.

(The DMM6500A, though it only shows the burden voltage at 1A as "< 0.55 V" for DC and "<0.75 V" for AC, could be used in the same way, of course.)

The 10A range is for me the most-used current range for exactly this reason -- low burden voltage. So it is worth noting a small difference between the two brands -- the Keithley has the 10A input on the rear panel, while the Keysight has it on the front panel.

Since I frequently need to access the 10A terminal, I find its front-panel location in the 34465A quite convenient in this regard.

Just passing along this bit of real-world experience about a difference that might not otherwise leap off the data sheets that you are comparing. :-)

[TrickyNekro]

Since I frequently need to access the 10A terminal, I find its front-panel location in the 34465A quite convenient in this regard.

To tell the truth, this also went on my nerves with the DMM6500 and a lot of people are commenting about it. It is not an absolute deal breaker,
I would guess that the 1A / 3A shunt is low voltage burden enough.

Last experience I had, was with a BLE IoT I was developing was ~7μA sleep current, 20 - 40mA (depending on the mode) working current, which I guess
you can get away with a higher burden voltage. But I guess, if you have some kind of small motor, valve or anything on there that draws a bit more,
then it could potentially become a problem.

But in the end of the day is it not a deal breaker, I could always wire the back panel to the front externally. Though yeah, at 1.2k€ you don´t really call it optimal.

[Kleinstein]

For the higher current ranges, close to the limit of the fuses used, a large part of the buden voltage is often due to the fuse, not the actual shunt. The numbers in the spec sheet tend to be upper limites in this aspect. So the actual value may be a little lower.
Especially for the higher currents, the drop at the actual shunt is usually in the 100-200 mV or below range to avoid excessive heating. The rest is from the fuses and switches.

Measuring the supply current of a IoT thing is usually about the large dynamic range, but not much about absolute accuracy: nobody really cares if the average current is 100 µA or 110 µA, but the point is averaging current that may range from µA to 100 mA. If the burden is critical one may still want to use an external shunt resistor (and maybe an extra amplifier similar to the µCurrent), even though the absolute accurcy is relatively poor.In this case less protection can make the difference and a larger value for the shunt may be feasable than for high accuracy at the peak currents. For the short current peaks the circuit should have buffer caps.

[TrickyNekro]

Measuring the supply current of a IoT thing is usually about the large dynamic range, but not much about absolute accuracy: nobody really cares if the average current is 100 µA or 110 µA, but the point is averaging current that may range from µA to 100 mA. If the burden is critical one may still want to use an external shunt resistor (and maybe an extra amplifier similar to the µCurrent), even though the absolute accuracy is relatively poor.In this case less protection can make the difference and a larger value for the shunt may be feasible than for high accuracy at the peak currents. For the short current peaks the circuit should have buffer caps.

To tell the truth a 10% difference when building something that should last a couple years on primary cells is not exactly inconsequential, but I get your point.
Usually sleep currents are around 1uΑ - 10uA, so you need to be accurate there.

The thing is that that´s your sleep current. If you got a auto-power on feature, say you activate some sensors, do some readings etc. to see if your device that has an analog input
is somehow excited and you need to wake-up the rest of the device, some of these sensors when converting can actually consume in the lower 10s of mA (say 20 - 40mA) add to that
a couple LEDs when waking up, plus a BLE radio, you can go well in the 40 - 50mA where the shunt burden voltage is then really a problem.

I really can´t decide between the two to tell the truth right now. The funny thing is that how the instruments function, it makes me want to buy a 34465A now, as a "daily" bench drive
and a DMM7510 later, when and if I need the extra resolution. Now, of course this is a hunch, but the Keysight instrument gives me the feeling that it was designed for engineers but
the Keithley was designed for scientists.

And since I guess I am a bit of both, I will probably stick long term to the plan above. But I will have to address the "typicality" of the low current measurement of the 34465A first.

If someone can provide some examples, that would be great!

[bson]

I use the low μA range on my 34465A very frequently.  The burden voltage doesn't really matter since the meter goes between the power supply and onboard regulation.  Just need to make sure the supply is set high enough to accommodate the meter.  Either along the cable or using a 2-pin header.  In the latter case I'll also add a jumper to disconnect on-board regulation to measure the regulator quiescent current (and to null the meter).  Since its upstream of regulation and filtering/reservoir, it's pretty low frequency stuff and the digitizer can do pretty good 5-digit captures of current consumption for various operations, just to get a quantitative transient energy cost for various things.  (Also very useful when trying to tweak things to reduce power usage.)  It can even measure the current consumption of an MSP430 in various sleep modes and do some low-current SMU style transient captures.  (Though it just captures the data, plotting and analysis needs to be done on a computer in python or what have you.)  Of course, a load resistor can be used and the voltage measured, it's just handy to have an instrument that's calibrated and plug-and-go.

[TrickyNekro]

I use the low μA range on my 34465A very frequently.

I found out what the "typical" in their datasheets meant from the dealership. If I recall it correctly, that means that 80% of the devices meets the spec presented on the datasheet.
Now of course I do not expect that the other 20% will be out and about.

It´s probably the one conservative side of the 2σ being ruffly ~1.5σ, which of course it is not 100% but I can probably live with it.

Especially now that the dealership came through with a very sweet deal plus they are the official Keysight distributor and they also handle repair and calibration.
At this price point I have to consider after-sales.

So 34465A it is.
Still thanks everybody for their input. It helped me negotiate a bit, but also gave me a clear view on the trade-offs (which I guess they are not many).

Cheers!
Lefteris

[Bravo]

If you need speed without connecting it to PC, the DMM6500 is what you need. I use mine for monitoring uA  current draw in graph mode mostly, but also to see fast pulse current draw.  Check out the YouTube demos by Keithley, especially the one  testing the current draw by a Bluetooth device. Very easy to set up from the front panel.   

[Dr. Frank]

I use the low μA range on my 34465A very frequently.

I found out what the "typical" in their datasheets meant from the dealership. If I recall it correctly, that means that 80% of the devices meets the spec presented on the datasheet.
Now of course I do not expect that the other 20% will be out and about.

It´s probably the one conservative side of the 2σ being ruffly ~1.5σ, which of course it is not 100% but I can probably live with it.

Especially now that the dealership came through with a very sweet deal plus they are the official Keysight distributor and they also handle repair and calibration.
At this price point I have to consider after-sales.

So 34465A it is.
Still thanks everybody for their input. It helped me negotiate a bit, but also gave me a clear view on the trade-offs (which I guess they are not many).

Cheers!
Lefteris

Hello Lefteris,
as you have decided to purchase the 34465A - congratulations - make sure that after receipt of the instrument you apply for the free 2MB memory upgrade. That's very useful, and also really needed to make full use of the DIG option, which is already built in the firmware from # 3.0 onwards. In conjunction with the very good 1, 10 and 100µA DCI ranges, you will be able to monitor low currents on IoT applications.

https://www.keysight.com/de/de/cmp/promotions/free-memory-upgrade-for-bench-solutions.html

Maybe you also get the  BenchVue Software for your DMM free, w/o timely limitations, as it's currently promoted here in Germany.

https://www.keysight.com/de/de/lib/resources/installation-guides/benchvue-licenses-included-with-instruments-3046463.html

Please ask your KS salesman for details in Greece.

As you are SW experienced, I assume that you're able to write something better on your own.

Frank

[bateau020]

Maybe you also get the  BenchVue Software for your DMM free, w/o timely limitations, as it's currently promoted here in Germany.

https://www.keysight.com/de/de/lib/resources/installation-guides/benchvue-licenses-included-with-instruments-3046463.html

Any idea why all the links that say "Basic instructions for how to redeem and activate this type of license" or "See instructions to check for eligibility and activation instructions." turn into "Page Not Found" (at least for a small list of EU countries I tried)? Is it over, or is that just a badly maintained website?
(Not that I personally would want to use it, as it is Windows only --requires me to launch a VM--, and I only have 1 Keysight instrument, so I don't see the use for me. TC or sigrok or python will do for me.)

About the low current measurements and digitizing, some feedback with my experiences and disillusions with this instrument. A "real" bench DMM brings a learning curve with it. Just hoping that the following might help others.

I bought the 34465A thinking it would be great for creating those fancy power-draw-over-time graphs, at high resolution. Yes, 50k samples/s is not much compared to the 1M samples/s of the DMM6500, but at least I would have "real" low current resolution.

Totally wrong. Totally naive. For 3 reasons:

1) autorange is disabled when digitizing, at least on a 34465A. You will need to pick a range, and accept the limitations of that range. Don't know about the DMM6500.

2) device bandwidth. DCI BW -3db = 10kHz (any range). Yes, the DMM6500 does significantly better on 100mA range: 340kHz, but drops to 25kHz on 1A. So when you measure a device that peaks to 200mA (easily the case when you do WiFi), the difference between the 2 instruments is not that big anymore. On both, more than 50k samples/sec is even useless. When you can stay below 100mA (or accept overloads), DMM6500 seems a lot better though.

3) resolution. The higher the sampling rate, the lower the effective resolution. When doing 10k samples/sec on the 1A range, the 34465A will give you 5 effective digits, so a 100uA resolution. Not great. The DMM6500 would give you 10uA. BUT, here is something that I cannot confirm, I only got it from the specs --"Typical Digitize Signal Characteristics"--, so I may misunderstand: the DMM6500 will not, and not even close, be able to deliver you that resolution. At that sample frequency, it will only give you 8 effective bits out of the 16. Your scope will likely do just as good or even better! You will gain 2 bits when going to the 100mA range though. Yay. Still not 5 digits though. The DMM6500 will even drop to 5 bits on the 10A range, over 10k samples/sec.

So it all depends on your needs, but you will need to read the fine print. A digitizing bench DMM is not a high resolution scope.

[wizard69]

In the end you need to fit the meter to your expected usage.    These days expanded diode test range may be valuable.   Low OHMs capability might be seen as a big positive in tracing PCB shorts.

As for inherent accuracy this is one of those things that I often lead with the question is it really that important?   Very often it isn't.   These are 5.5 digit and more meters and frankly I'm not sure many people look a the last digit in the display to determine if they have the right value.   If they do, most likely they already know what specs are needed, the required calibration schedule and have resolved a host of other factors impacting the reading.    So unless you have a specific requirement to squeeze every bit of accuracy out of a meter of this type I really wouldn't worry about the specs much.

[bson]

You need greater precision than accuracy, and you need 5.5d accuracy to characterize a 16-bit ADC or DAC.  So in practice you want a 6.5d instrument for this.  You're not supposed to look at the last digit, it's just there for precision.  You care about the second least significant digit.

But of course, you may have a bit of output gain as well and a range that calls for a 7.5d instrument in many cases.