New Agilent 34461A 6.5 Digit Bench Multimeter

[ben_r_]

Stop it! Youre trying to make me feel bad for ordering the 34450A!

I'm returning the 34450A I just got (haven't even powered it) and ordering the 34461A. Hopefully my only loss is the shipping back and forth. I just hope Dave doesn't uncover something in his 34461A review I'll regret.

But seriously, if I had a Fluke 289, I wouldn't be able to justify to myself ordering a $1000 DMM. The Fluke 289 has stats, bar, trend, data logging, so you're right, it can do at least 95% of what any engineer would ever need. In my case my only other DMM is a cheap $50 handheld on which I don't trust to better than 10%. I was initially looking for something in the $500 range, and now the Agilent carrot on the stick led me to spending $1000.

My hope is that when they release the 34470A/34471A to replace the 34410A/34411A, that they will price it significantly higher not to tempt anyone here.

Yea I might be more tempted if I could even cancel my order, but once it was placed with TestEquipmentDepot for $787 out the door I was committed to it with no backing out and FOUR month delivery time. So no matter the temptation Im stuck until September anyway.

[EEVblog]

Don't you just hate it when they come out with a nicer and better model soon after you ordered an older model. I kind of wish I waited an extra 1.5 weeks.

That does suck!
Not even I knew it was coming 1.5 weeks ago.
Quick resell it on ebay before the masses find out!

[EEVblog]

Hmmm... Maybe. I guess more specifically where Im confused is that I have an Agilent 34450A on order and it does datalogging with a graphical OLED display and Im wondering whats lacking (if anything) in that functionality from what the 34461A offers and whether or not its just the same thing but with a color display.

The 34450A is 5.5 digits vs 6.5 digits for starters, and 3 times better accuracy or so. But it doesn't have dual display.
And no graphing capability, although I think it does some stats.
Also much slower update speed.
At $787 vs $945 for the 34460A, there isn't much in that price, and I think you get a lot more bang-per-buck there on the new model.

[EEVblog]

I would love to see a 6-1/2 digit DMM use the LTZ1000AH-- the goal would be lower 1/f noise (DC-10Hz), less drift between calibrations, and a 3-year calibration cycle (after the first year) vs. a 1-year calibration cycle.  This would lower the cost of ownership, and give more meaning to the last digit.

The new 34461A gives specs for 2 years cal intervals along with the traditional 24hr/90day/1 year specs.

[Hypernova]

If one has the facilities, what about swapping out the reference? If it's just another 399 based 6.5 the rigol DM3068  is much better value for money as it also have rough histogram/trend AND cap measurements. Every once in a while I open its product page and go "hmmm..." but never could pull the trigger. Dave's review is going to be a big factor.

[mikeselectricstuff]

Looking through the spec and user manual, it appears that although it can copy its 10K volatile readings memory to internal NVM or external USB, it doesn't appear to be able to do continuous logging to NVM or USB, which seems an odd limitation.

[bingo600]

Isn't there a fan in this unit , as opposed to the 34401A.

Dave ought to tell if it's as noisy as the 34410 (or just noisy) , before people starts returning their goods 

Seems like old 34401A users liked the silence ....

/Bingo

[EEVblog]

Isn't there a fan in this unit , as opposed to the 34401A.
Dave ought to tell if it's as noisy as the 34410 (or just noisy) , before people starts returning their goods 

I didn't really notice it, which means it can't be too bad...
Will have to pay attention when I next power it up.

[EEVblog]

Looking through the spec and user manual, it appears that although it can copy its 10K volatile readings memory to internal NVM or external USB, it doesn't appear to be able to do continuous logging to NVM or USB, which seems an odd limitation.

You could presumably simply manually get a single remote measurement, and do that as fast as possible?

[mikeselectricstuff]

Looking through the spec and user manual, it appears that although it can copy its 10K volatile readings memory to internal NVM or external USB, it doesn't appear to be able to do continuous logging to NVM or USB, which seems an odd limitation.

You could presumably simply manually get a single remote measurement, and do that as fast as possible?

It's not the speed, but the ability to do a long log (e.g. battery discharge over several days) standalone without connecting to a PC, and displaying the data so far on its own screen. It appears that this is possible with the trend function, but limited to the 10K sample memory (not looked deep enough to see what happens when it fills - presumably it does a fifo roll).
It would be very useful to be able to either have it write to NV memory during logging, or periodically empty the internal RAM to a CSV file in either internal filesystem or USB. The latter ought to be very easy to implement.

[EEVblog]

It's not the speed, but the ability to do a long log (e.g. battery discharge over several days) standalone without connecting to a PC, and displaying the data so far on its own screen.

Usually you don't need that much memory for that, hundred of points are enough, it's a just matter of setting the sample time if you know the estimated total sample time.
If you don't know the total time, then yeah, you need sample compression and/or level based compression or a crap load of memory to brute force it.

[robrenz]

Details of how 8846A Trend plot works here 

excerpt from that:
TrendPlot is a form of min/ max recording in which the meter automatically compresses the timescale each time the trend approaches the end of memory. As the meter starts to run out of memory, signal processors quickly go to work. They combine adjacent recording intervals into a new min and max. You still get to see the worst-case measurements and the overall trend. And because you choose when to stop the measurement, you automatically get the best time resolution with the available memory.

[quantumvolt]

This Agilent "Interactive Demo" http://wireless.agilent.com/html/dmm/ answers many of the questions here: Truevolt, Display, Trend, Fan, Statistics, Data Logging, Memory etc.

Except from USB/LAN and the primitive Display, this instrument has very little more than the 34401A. But of course - it is more sexy and has graphic display and colors. That makes the measurements much better for many people - including some here ...

Interesting: The RS232 port is gone and the GPIB is optional - I guess they are both outdated. Modernity is USB on the bench (as you see it in many handheld meters) and LAN/WEB when more than 3 feet distance is involved. If one has the cash to buy accessories I guess one can make a data logger / real-time math-statistics processing software with 1000 measurements pr. second until the end of the world ...

For a few dollars and an Arduino - the data stream from the 34401A is also real-time accessible for SD card writing, wireless, USB or Ethernet/Web and online PC processing. First test (I need to buy a DB-9 plug ... - the set-up monitors a voltage and sends the measurements real-time to a USB port or Ethernet shield or SD card writer).

[quantumvolt]

Details of how 8846A Trend plot works here 

excerpt from that:
TrendPlot is a form of min/ max recording in which the meter automatically compresses the timescale each time the trend approaches the end of memory. As the meter starts to run out of memory, signal processors quickly go to work. They combine adjacent recording intervals into a new min and max. You still get to see the worst-case measurements and the overall trend. And because you choose when to stop the measurement, you automatically get the best time resolution with the available memory.

As far as I can see from http://wireless.agilent.com/html/dmm/ the Agilent feature is just a data plot 'value against time' (last 60 seconds or all data available). The Fluke feature is signal processing and is part of the branch of mathematical statistics called 'Time Series Analysis' http://en.wikipedia.org/wiki/Category:Time_series_analysis.

[ftransform]

IDK, considering that I can have data being graphed in about 30 seconds on my PC, I just don't see the use.

I'm sure a program like labview or something can do it in real time too, once its setup. I'd rather pirate a program then fork over money to agilent.

I mean seriously, if you have intuilink running, you just move your chair 3 feet to the left, open up excel, click graph. And you have it on a nice big screen with alot of data points that you can easily run statistical analysis on on a good user interface instead of pressing buttons like its star trek. You can run a fourier transform, this that, easily and not have to deal with a 1 by 1 inch display or whatever. 24 inches baby

[Hypernova]

As long as the underlying reference (LM399, LTZ1000) don't improve, they are stuck.

[fpga]

Quick resell it on ebay before the masses find out!

Having ordered from a reputable, authorized distributor, I was able to return within 7 days for a full refund, minus the shipping charges.

[ben_r_]

Quick resell it on ebay before the masses find out!

Having ordered from a reputable, authorized distributor, I was able to return within 7 days for a full refund, minus the shipping charges.

Who did you order through?

[fpga]

Agilent has chosen to omit several features in the 34461A that were present in older and competitor's models. I'm curious as to whether they were omitted due to being of limited use or to leave as a carrot on a stick for future higher end models.

The first one is removing the dual measurement display, which given the new higher res color WQVGA display, they could have easily incorporated this feature. Apparently this feature on other models, including the 34450A, permits displaying frequency or DC offset while the main display shows AC V, or to display AC ripple V while the main display shows DC V. I can see use for these features.

The ability to display both V and I at the same time is nice in theory, but not sure if it is practical. First, you need to be able to attach multiple leads to your circuit, and second, most important, chances are that the V and the I would be measured using different ranges. I'm not sure about the Fluke and Keithley, but according to the manual, the 34450A performs the two measurements by switching back and forth. If this involves different ranges, the relays will switch back and forth, which I presume will impact the measurement accuracy and speed. Any experience with doing these measurements in practical use?

The second feature is Capacitance measurement. While it sounds nice in theory, it is of limited value in practical use. First, if you wanted to check if a decoupling cap on a board wasn't faulty, this won't work because you generally have many caps in parallel. If you wanted to verify the quality of a cap, this too wouldn't be of much use because high accuracy of capacitance is less important than quality. For the capacitor ESR, you need to use an LCR meter, which is based on different measurement technology. For checking just capacitance of loose caps lying around on a bench, you might as well use a cheap handheld DMM or a tweezer LCR meter.

So perhaps Agilent used common sense for omitting these features?

[fpga]

Who did you order through?

metrictest.com

If you put in a Web order, you get 10% discount. Sometimes they will match the 10% discount on their quotes. If it is a non-stocked item, like license key options, they sometimes quote with a 12% discount. All the distributors charge sales tax and the usual shipping charges.

[ben_r_]

Who did you order through?

metrictest.com

If you put in a Web order, you get 10% discount. Sometimes they will match the 10% discount on their quotes. If it is a non-stocked item, like license key options, they sometimes quote with a 12% discount. All the distributors charge sales tax and the usual shipping charges.

Ah. Yea not so good a deal for me as that 10% discount probably wouldnt even cover tax and shipping whereas with TestEquipmentDepot there is neither. But, good to know about anyway. Thanks.

[EEVblog]

The 8846A has 25ppm/year specs-- where's the beef?  People buy 6.5 digit DMMs because they want the "state of the art" in accuracy and stability-- this DMM uses the same 40-year old technology, with no effort made to improve it.  I can't figure out what market segment they are shooting for here--  This DMM doesn't even surpass the Rigol 6.5 DMM in specs. 

You expected it too for the price?

If they made this DMM with 10ppm/year basic DC accuracy (or better), and/or added self and external artifact calibration (or any other advanced feature) *then* I would stand up and take notice.  {...yawn...}  I'll wait for the review to form my final opinion on this DMM, but as it is so far it is a massive marketing "fail".   

Maybe the market just isn't there for a more expensive 10pmm 6.5 digit meter?

[EEVblog]

THAT would certainly be a good place for them to start.  There are many new op-amps and chopper amps that, in combination with an LTZ1000A could improve the accuracy of this DMM by an order of magnitude.  Then, there is the well-known technique of artifact calibration.  There have been massive improvements in resistor technology.  As far as I can tell-- NONE of these have been used in this "new" DMM.  {...sigh...}

Two significant claims they make are Noise and injected current and Input bias current.
Stabs at who "Brand X" are?

[mikeselectricstuff]

The 8846A has 25ppm/year specs-- where's the beef?  People buy 6.5 digit DMMs because they want the "state of the art" in accuracy and stability-- this DMM uses the same 40-year old technology, with no effort made to improve it.  I can't figure out what market segment they are shooting for here--  This DMM doesn't even surpass the Rigol 6.5 DMM in specs. 

Who actually needs that accuracy? Probably not enough to justify the work producing it. I'd guess that enough accuracy to use as a cal reference for most DMMs & other testgear  is the most that the vast majority of users will ever need.


You expected it too for the price?

If they made this DMM with 10ppm/year basic DC accuracy (or better), and/or added self and external artifact calibration (or any other advanced feature) *then* I would stand up and take notice.  {...yawn...}  I'll wait for the review to form my final opinion on this DMM, but as it is so far it is a massive marketing "fail".   

Maybe the market just isn't there for a more expensive 10pmm 6.5 digit meter?

[alm]

If you purely look at accuracy, then the latest significant improvement was probably the HP 3456a, introduced in the early eighties. Its revised 1 year specs are quite similar, if not slightly better, than modern 6.5 digit DMMs. It's also rated for 24h transfer standard accuracy. All improvements since then have been functions, features, connectivity and speed.