Keysight's new 34465A (6.5 digit) and 34470A (7.5 digit) bench multimeters

[G0HZU]

I doubt more than a tiny fraction of folk have a need for 6 1/2 or 7 1/2 or ... accuracy and indeed many of the purchasing decision are based on wanting the perceived latest & greatest versus needing it

Trying watching something simple, like the slow discharge of a battery, using a 4.5 digit meter. Good luck.
Having a 6.5 digit meter is like having a deep memory scope. You don't know how useful it can be until you get it.

Yep, I guess we'll all be doing slow discharge tests of batteries next to justify our 6 1/2 digit meters I work with battery powered designs all the time, haven't seen the need for measuring battery discharge with a 6 1/2 digit meter yet. An electronic load with logging to a laptop has been much more useful for my use when characterising a battery.

I have a 6 1/2 digit bench meter (got it free), use it occasionally, but certainly wouldn't rush out to buy one and I am designing/working with electronics every day. I agree that there are 'some' applications where it is needed, but I'll stick to my original claim that only a small fraction of users actually NEED one.

cheers,
george.

Agreed. I've worked in design labs for over 25 years and worked alongside hundreds of talented engineers across RF/HW/SW/DSP designing all manner of products in the defence industry and it's extremely rare to see a 6.5 digit DMM being used in our labs. We have loads of very expensive test
gear but the DMM isn't seen as anything special and is just seen as a dull and boring accessory and we  get by just fine with 3.5 to 5 digit H/H models.

In the test dept things are different and there are quite a few high end DMMs. some in ATE racks and some are on test benches where they have a definite status appeal.

[LabSpokane]

Anyone know who has the 34465A in stock in the US?

TestEquity and MicroLease both claim to have current stock.  Best price is MicroLease.  Fastest delivery for me is usually TestEquity.  I'm probably going to sell one of my 34461As and upgrade if anyone is interested. 

[EEVblog]

Haven't tried it but I wonder if the effect of temperature variations is more than that of discharge by the time you get to the 6th digit...

Generally not, no. The extra resolution lets you easily see rate of change when your load varies too, you can physically see ti counting quicker, it's a very useful and gives you a good intuitive feel for what's going on.

[EEVblog]

Agreed. I've worked in design labs for over 25 years and worked alongside hundreds of talented engineers across RF/HW/SW/DSP designing all manner of products in the defence industry and it's extremely rare to see a 6.5 digit DMM being used in our labs. We have loads of very expensive test
gear but the DMM isn't seen as anything special and is just seen as a dull and boring accessory and we  get by just fine with 3.5 to 5 digit H/H models.

I've found the opposite. Every good lab I've worked in has had a high resolution bench meter. We've always been big on production testing though were such bench meters are commonplace.
But again, it's one of those thing that you often don't know how useful it is until you have one to hand for a certain job. If a lab already has half a dozen decent meters, then it's hard to think you might need a good bench meter.
Many times though I've been without the high res bench meter and along comes a requirement and scream when it's not available.
Also, they have not traditionally been cheap.
If I was setting up a new lab here then I'd certainly spring the $1K for the 6.5 digit Agilent or less for the Rigol.

[Tom45]

Besides Test Equity and Microlease, I see that Newark is now listing quantities on hand for the new meters.

For the various Keysight bench meters, old and new quantities on hand are:

34461A: 43
34465A: 50
34470A: 10
3458A:    7

So they expect to sell 5 34465A to each 34470A.

When I contacted TEquipement about the 34465A, the response was essentially: huh? They later got back to me saying it would take 4 to 6 weeks to get one.

[Shim]

I returned my 61A which I paid $1075 and got the 65A for $1340. Both $CDN

I like the fact that the 65 can use thermocouples for temperature. I have 2. My Fluke 289 came with one and got another with the free promo test lead set when I bought it.

The 2nd display is great. I am doing audio product development and if using the dB scaling feature, you can display frequency or the actual voltage reading. The 65 also has full statistics when reading in dB, which the 61 did not (it only had min, max, and if think span, but no avg).

Been long day, haven't played with it more than a few minutes so far. More in the next few days.

[tszaboo]

I doubt more than a tiny fraction of folk have a need for 6 1/2 or 7 1/2 or ... accuracy and indeed many of the purchasing decision are based on wanting the perceived latest & greatest versus needing it

Trying watching something simple, like the slow discharge of a battery, using a 4.5 digit meter. Good luck.
Having a 6.5 digit meter is like having a deep memory scope. You don't know how useful it can be until you get it.

Yep, I guess we'll all be doing slow discharge tests of batteries next to justify our 6 1/2 digit meters I work with battery powered designs all the time, haven't seen the need for measuring battery discharge with a 6 1/2 digit meter yet. An electronic load with logging to a laptop has been much more useful for my use when characterising a battery.

I have a 6 1/2 digit bench meter (got it free), use it occasionally, but certainly wouldn't rush out to buy one and I am designing/working with electronics every day. I agree that there are 'some' applications where it is needed, but I'll stick to my original claim that only a small fraction of users actually NEED one.

cheers,
george.

Agreed. I've worked in design labs for over 25 years and worked alongside hundreds of talented engineers across RF/HW/SW/DSP designing all manner of products in the defence industry and it's extremely rare to see a 6.5 digit DMM being used in our labs. We have loads of very expensive test
gear but the DMM isn't seen as anything special and is just seen as a dull and boring accessory and we  get by just fine with 3.5 to 5 digit H/H models.

In the test dept things are different and there are quite a few high end DMMs. some in ATE racks and some are on test benches where they have a definite status appeal.

When you design something where the specification said 0.0x% or 0.00x% it is quite essential to have a meter like this, or even a 3458A. I like the one we have at work a lot. You can measure stuff you never even imagined it is even possible. Like you have a solid ground plane, but it is possible to measure the uV of voltage drops across it. If you ever worked with a 16 bits or more having one is essential.
Now, the 34470A seems like really a nice unit. I had a lot of trouble getting the 3458A to make trend charts, it requires a running network, PC, Ethernet-GPIB bridge a submarine and a spaceship to work. While I found it on the 34461 useless due to the inaccuracy of the meter, this is much better.
I'm also interested how did they make the good old artefact calibration in the meters.

[JohnnyBerg]

I remember back in the early 70's as we wanted to get telephone, and my father said: what do we need a telephone for?

Same with a X.5 digit meter: you don't know what you're missing, until you have one

[Howardlong]

I doubt more than a tiny fraction of folk have a need for 6 1/2 or 7 1/2 or ... accuracy and indeed many of the purchasing decision are based on wanting the perceived latest & greatest versus needing it

Trying watching something simple, like the slow discharge of a battery, using a 4.5 digit meter. Good luck.
Having a 6.5 digit meter is like having a deep memory scope. You don't know how useful it can be until you get it.

Initially I had this vision of a $3k DMM being tied up to a battery for weeks/months taking a reading once an hour or so, or am I misunderstanding?

Is it that you can see the self-discharge (or slow discharge) over a much shorter period, say a day or two, and extrapolate?

[Dr. Frank]

I'm also interested how did they make the good old artefact calibration in the meters.

You mean the AUTOCAL feature of the 3458A?

That is NOT implemented in the new '465A and'470A, quite obviously, although they also call it AUTOCAL.

If you have a look in the manual for these new meters, service / calibration section, you will note, that these instruments both have to be calibrated in the usual manner, i.e. by individual Cardinal Points for each mode and for each range.

They may only correct for some internal temperature drifts of amplifiers, what they call AUTOCAL here, but the A/D is far too bad (nonlinear) to realize a real / full artefact calibration, by only two references.

Frank

[tszaboo]

I'm also interested how did they make the good old artefact calibration in the meters.

You mean the AUTOCAL feature of the 3458A?

That is NOT implemented in the new '465A and'470A, quite obviously, although they also call it AUTOCAL.

If you have a look in the manual for these new meters, service / calibration section, you will note, that these instruments both have to calibrated in the usual manner, i.e. by individual Cardinal Points for each mode and for each range.

They may only correct for some internal temperature drifts of amplifiers, what they call AUTOCAL here, but the A/D is far too bad (nonlinear) to realize a real / full artefact calibration, by only two references.

Frank

I dont know, is it? For sure, they call it the same way. The manual said it takes 20-30 seconds to do this, it takes several minutes on the 3458A, because the 1000 NPLC on the 3458A, while the xx65 and 70 are 100NPLC. We need to see a service manual, because the user manual is missing the details. Maybe after the teardown it will be more clear. Maybe Dave should cover this if he makes a review of the meters or ask HP/Agilent/Keysight to clarify.

[Dr. Frank]

I dont know, is it? For sure, they call it the same way. The manual said it takes 20-30 seconds to do this, it takes several minutes on the 3458A, because the 1000 NPLC on the 3458A, while the xx65 and 70 are 100NPLC. We need to see a service manual, because the user manual is missing the details. Maybe after the teardown it will be more clear. Maybe Dave should cover this if he makes a review of the meters or ask HP/Agilent/Keysight to clarify.

This manual is already available, that is document 34460-90901.pdf.
This manual now contains special chapters for 34465A and 34470A, also describing their calibration routines, see snapshot from the DCV section.

Therefore, no artefact calibration!
See page 527,  a Fluke 5270A is required, 10V and 10k artefacts won't do the job.

The A/D is only 1 .. 1.5ppm linear, which may correct gain 10 errors in the order of 10..15 ppm only, impossible to achieve a full autocal calibration. (compare that to 0.02ppm linearity of the HP3458A) 


Then , the third indicator is in the specs.

All ranges diverge over time, i.e. the uncertainty gap between two distinct ranges increases over time.

On the 3458A, this gap is always constant, even after 1 year, as this instrument corrects all ratio uncertainties to the 24h specification, every time AUTOCAL is engaged.

Frank

[Dr. Frank]

Thinking a little bit further about the 10V DCV drift specifications (most stable range) I come to the conclusion, that the 34465A will probably have an LM399H reference inside, (20ppm/year drift), and the 34470A will probably have the LTZ1000A reference of the 3458A (8ppm/year drift).

As the operational ambient temperature is also specified for 55°C, they again have to run the references on about 90°C oven temperature.

Frank

[tszaboo]

I dont know, is it? For sure, they call it the same way. The manual said it takes 20-30 seconds to do this, it takes several minutes on the 3458A, because the 1000 NPLC on the 3458A, while the xx65 and 70 are 100NPLC. We need to see a service manual, because the user manual is missing the details. Maybe after the teardown it will be more clear. Maybe Dave should cover this if he makes a review of the meters or ask HP/Agilent/Keysight to clarify.

This manual is already available, that is document 34460-90901.pdf.
This manual now contains special chapters for 34465A and 34470A, also describing their calibration routines, see snapshot from the DCV section.

Therefore, no artefact calibration!
See page 527,  a Fluke 5270A is required, 10V and 10k artefacts won't do the job.

The A/D is only 1 .. 1.5ppm linear, which may correct gain 10 errors in the order of 10..15 ppm only, impossible to achieve a full autocal calibration. (compare that to 0.02ppm linearity of the HP3458A) 


Then , the third indicator is in the specs.

All ranges diverge over time, i.e. the uncertainty gap between two distinct ranges increases over time.

On the 3458A, this gap is always constant, even after 1 year, as this instrument corrects all ratio uncertainties to the 24h specification, every time AUTOCAL is engaged.

Frank

You are right. The calibration process still contains a lot more steps than what you would need for artefact. We should see some block diagrams, otherwise we have no idea what is going on inside the meter when we press the button.

[AndyC_772]

One thing I've found a 6.5 digit multimeter really useful is measuring the current of a microcontroller. Not because it was drawing a tiny amount, but because the difference between the burden voltages of adjacent ranges was enough that it triggered the micro's brown out logic. With the 6.5 digits being enough to give a decent level of precision on the higher range.

This is something I hope to now be able to do with reasonable accuracy. I design a lot of battery powered products using microcontrollers, and typically they spend a lot of time asleep, but wake at regular intervals to do something interesting before going back to sleep again. A fast, accurate, logging meter should be able to record current over a period of time, and even if it can't integrate total charge internally, it's simple enough to add up all the measurements in a spreadsheet.

My 34465A is ordered. Should be here tomorrow 

[6thimage]

Thinking a little bit further about the 10V DCV drift specifications (most stable range) I come to the conclusion, that the 34465A will probably have an LM399H reference inside, (20ppm/year drift), and the 34470A will probably have the LTZ1000A reference of the 3458A (8ppm/year drift).

As the operational ambient temperature is also specified for 55°C, they again have to run the references on about 90°C oven temperature.

Frank

I'm guessing you're right about the 34465A using the LM399. Looking through the parts on keysight, both the 65A and 70A now have fan shields (http://www.keysight.com/my/faces/partDetail.jspx?partNumber=5041-5261) presumably to limit airflow across the reference. It would be interesting to see how much of a difference the shield makes and whether there would be any benefit adding the shield to the 61A.

[6thimage]

I have had an update from Keysight about the 34461A's firmware:

We will be releasing revision 2.08 rather than 2.06 due to defects that were found and resolved during testing. I know this seems unending and I apologize. I am given release dates which I then pad in case of unforeseen issues. 2.08 is now scheduled for release March 16th.

I have attached a set of slides which summarize the changes/fixes for the 34461A that are contained in 2.08.

This is the same date (March 16th) as rosbuitre said, but they have increased the version number. I don't think there is anything new in the slides, but it does mention that the ethernet has HiSLIP and IPv6 support, which I hadn't realised before.

Link to the slides: https://www.dropbox.com/s/kynqv5v1opoda7i/34460A-34461A%202.08%20Enhancements.pptx?dl=0.

[rosbuitre]

I have had an update from Keysight about the 34461A's firmware:

We will be releasing revision 2.08 rather than 2.06 due to defects that were found and resolved during testing. I know this seems unending and I apologize. I am given release dates which I then pad in case of unforeseen issues. 2.08 is now scheduled for release March 16th.

I have attached a set of slides which summarize the changes/fixes for the 34461A that are contained in 2.08.

This is the same date (March 16th) as rosbuitre said, but they have increased the version number. I don't think there is anything new in the slides, but it does mention that the ethernet has HiSLIP and IPv6 support, which I hadn't realised before.

Link to the slides: https://www.dropbox.com/s/kynqv5v1opoda7i/34460A-34461A%202.08%20Enhancements.pptx?dl=0.

 

[6thimage]

I have had an update from Keysight about the 34461A's firmware:

We will be releasing revision 2.08 rather than 2.06 due to defects that were found and resolved during testing. I know this seems unending and I apologize. I am given release dates which I then pad in case of unforeseen issues. 2.08 is now scheduled for release March 16th.

I have attached a set of slides which summarize the changes/fixes for the 34461A that are contained in 2.08.

This is the same date (March 16th) as rosbuitre said, but they have increased the version number. I don't think there is anything new in the slides, but it does mention that the ethernet has HiSLIP and IPv6 support, which I hadn't realised before.

Link to the slides: https://www.dropbox.com/s/kynqv5v1opoda7i/34460A-34461A%202.08%20Enhancements.pptx?dl=0.

 

There's no need to be sarcastic - the original firmware update for the 34461A was due out, at the very latest, of December 2014 (I reported the histogram bug to them in January 2014). I was also told at the end of January that 2.06 was going to be released on March 1st. Then you said that bumping it up to 2.07 added a fortnight. So the fact that keysight have increased the version number again and not pushed the release back further is a small miracle.

[rosbuitre]

I have had an update from Keysight about the 34461A's firmware:

We will be releasing revision 2.08 rather than 2.06 due to defects that were found and resolved during testing. I know this seems unending and I apologize. I am given release dates which I then pad in case of unforeseen issues. 2.08 is now scheduled for release March 16th.

I have attached a set of slides which summarize the changes/fixes for the 34461A that are contained in 2.08.

This is the same date (March 16th) as rosbuitre said, but they have increased the version number. I don't think there is anything new in the slides, but it does mention that the ethernet has HiSLIP and IPv6 support, which I hadn't realised before.

Link to the slides: https://www.dropbox.com/s/kynqv5v1opoda7i/34460A-34461A%202.08%20Enhancements.pptx?dl=0.

 

There's no need to be sarcastic - the original firmware update for the 34461A was due out, at the very latest, of December 2014 (I reported the histogram bug to them in January 2014). I was also told at the end of January that 2.06 was going to be released on March 1st. Then you said that bumping it up to 2.07 added a fortnight. So the fact that keysight have increased the version number again and not pushed the release back further is a small miracle.

Hi
Not trying to be sarcastic, Am I getting the 34461A Friday and makes me happy improvements and corrections, only that

Regards

[6thimage]

I have had an update from Keysight about the 34461A's firmware:

We will be releasing revision 2.08 rather than 2.06 due to defects that were found and resolved during testing. I know this seems unending and I apologize. I am given release dates which I then pad in case of unforeseen issues. 2.08 is now scheduled for release March 16th.

I have attached a set of slides which summarize the changes/fixes for the 34461A that are contained in 2.08.

This is the same date (March 16th) as rosbuitre said, but they have increased the version number. I don't think there is anything new in the slides, but it does mention that the ethernet has HiSLIP and IPv6 support, which I hadn't realised before.

Link to the slides: https://www.dropbox.com/s/kynqv5v1opoda7i/34460A-34461A%202.08%20Enhancements.pptx?dl=0.

 

There's no need to be sarcastic - the original firmware update for the 34461A was due out, at the very latest, of December 2014 (I reported the histogram bug to them in January 2014). I was also told at the end of January that 2.06 was going to be released on March 1st. Then you said that bumping it up to 2.07 added a fortnight. So the fact that keysight have increased the version number again and not pushed the release back further is a small miracle.

Hi
Not trying to be sarcastic, Am I getting the 34461A Friday and makes me happy improvements and corrections, only that

Regards

My bad - sorry. Looking at your post now, I can see it the way you meant it - it just looked like slow clapping when I first saw it.

I don't think you will be disappointed with the 34461A. Just be warned that the update to the new firmware - if it is anything like the pre-beta version - will take a good few minutes to install over the 1.10 firmware. Mine sat on a white screen for what felt like an eternity and I was worried that it might have bricked it, then it burst back into life.

[rosbuitre]

Hi 6thimage
No problem, everything ok  , I have a few days to test it before the new firmware

Regards
Osvaldo

[jc101]

Slightly off topic, but I see that RS in the UK are offering the 34461A at a 10% discount at the moment (£638.10), looks like there is 1 in stock....

[AndyC_772]

You know how they say a watched kettle never boils? They should say that a hotly anticipated UPS truck never shows up... until about an hour ago. 

First impression of the 34465A is that it's more compact than my Keithley 2000, and the display is bright, clear, and has a wide viewing angle. Nice.

Auto-ranging, on the other hand, is so much slower it's not even funny. One of the things I really like about the K2000 is that I can apply a signal, and by the time I've looked up at the meter, it's already found the correct range and showed a measurement. The 34465A is much more like a handheld meter, really quite slow, and that's disappointing. Maybe there's a setting that'll speed it up?

There's also an annoying relay that clicks away if the meter is left in high-Z voltage mode. As the inputs charge and discharge, the meter switches range and back to accommodate the instantaneous voltage. Maybe I'll just have to leave it in 10M mode instead. It's a shame, but the K2000 doesn't have this quirk either.

On the plus side, within an hour I've had it hooked up to my LAN and controlled it via the web interface, and now - I think! - it's logging data to a USB stick. It's fun, in a nerdy sort of way, watching a battery charge 0.1mV at a time.

[LabSpokane]

Any one know if the software update for the '61A will include the ability to log direct to a USB stick?  That is one feature that is surprisingly missing.

I'm actually glad to see that Keithly has some competitive advantages still. That kind of pressure is good for all of us.