6.5 digit datalogging multimeter, opinions?

[bronson]

Thought dump here...  I need a multimeter to record tight voltage and resistance readings (load cell wheatstone bridges) and I'm looking for opinions on usability and maybe where to find it.

My requirements: high hundreds of readings per second at highish digits.  High tens of readings/sec is probably ok but more is definitely better.  6.5 digits should be fine, 5.5 is marginal, and my 4.5 digit Fluke is nowhere close.  I'm going to configure & process the data on a Linux computer in realtime so pretty sure I need USB or LAN (unless the deal is so smokin' that a $150 GPIB interface is worth it).  Flash drive won't do it.

I think these are the two finalists, with street prices:

Rigol DM3068      $800    (cheeeap but should do the job)
Agilent 34461a    $1100    (awfully new)

I'm leaning toward the Rigol because I'm spending my own money...  Is the Agilent worth another 30%?  In the past Agilent would nickel and dime me all to heck when trying to connect to a computer, will that happen here?

Been scanning ebay and SF Bay Area shops for used deals, no wins yet.  Getting frustrated enough to just to bite the bullet and go new.  How novel!

Others that I'm pretty sure would work...

Keithley 2001     $4380   (7.5 digits, lust)
Agilent 34460a    $945    (slow, no temp, no usb/lan?)
Agilent 34410a    $1346
Agilent 34411a    $2150
Agilent 34970a    $1600   (not a great multimeter, no usb or lan, $$$$ for options)
Fluke 8846a       $1600
Keithley 2002     $5800   (8.5 digits amazing, will have one day.  one day...)
Keithley 2000     $1100
Fluke 8845a       $1000

Is there anything obvious I'm missing?  Anywhere to find a decent used one other than ebay?  Thanks for any opinions!

    - Scott

[alm]

Note that reading rates are usually stated for DCV, and may be lower for resistance, especially high resistance values. A used HP 3457A should do the job, and even with GPIB adapter should be well under $800. Around $200 should definitely be possible in the SF Bay area.

I haven't used either of those meters. One thing I would check in advance is the programming documentation: in the past I wasn't impressed by Rigol documentation. If I remember correctly the DM3068 programming manual is a somewhat annoying mess of commands unique to the Rigol and commands compatible with the various Agilent DMMs separated into chapters for some reason. I would check for both if they use standard protocols over USB/LAN (eg. USBTMC or LXI) and whether these are supported with decent performance under Linux. I remember early USBTMC implementations in Rigol scopes to be somewhat nasty. Rigol software will suck, but you're not going to use anyhow it if you're running Linux.

This thread may be somewhat relevant for you: don't trust the DM3068 volatile memory spec and someone's complaining about the LAN interface.

[orin]

Thought dump here...  I need a multimeter to record tight voltage and resistance readings (load cell wheatstone bridges) and I'm looking for opinions on usability and maybe where to find it.

My requirements: high hundreds of readings per second at highish digits.  High tens of readings/sec is probably ok but more is definitely better.  6.5 digits should be fine, 5.5 is marginal, and my 4.5 digit Fluke is nowhere close.  I'm going to configure & process the data on a Linux computer in realtime so pretty sure I need USB or LAN (unless the deal is so smokin' that a $150 GPIB interface is worth it).  Flash drive won't do it.

I think these are the two finalists, with street prices:

Rigol DM3068      $800    (cheeeap but should do the job)
Agilent 34461a    $1100    (awfully new)

I'm leaning toward the Rigol because I'm spending my own money...  Is the Agilent worth another 30%?  In the past Agilent would nickel and dime me all to heck when trying to connect to a computer, will that happen here?

<snip>

All you need is the base 34461A to connect to a computer.  Unless you are replacing a 34401A, I see no real reason to use GPIB with it.  I access my '61A across a wireless link to a laptop which has its wireless and hardwired network adapters bridged.  The '61A is connected to the laptop via an ethernet switch (the switch is there so I can also connect to my Prologix ethernet adapter for GPIB devices).

The '61A even has a web interface to connect to it, but it does use java to to actually do anything.  I don't know where they download the java code from and running a network capture to find out is one of my least favorite activities.

Agilent have free capture software, but its for Windows.  I tried it briefly, but it's kind of basic.

FWIW, I attached a screen capture from the web interface of the 34461A monitoring my Fluke 731B.  Wish I knew what caused that glitch...  The drift is mostly due to the tempco of the 34461A.

Orin.

[bronson]

That's really funny that the 34461 uses a Java client.  Just when I thought client-side Java was finally dead...  Hope they don't rewrite it to use Flash.

It's great to hear the '61 comes ready for the network.  It only took HP two decades to realize almost all offices are wired for 10baseT, grr.  So, yes, your reassurances probably bump the '61 over the Rigol.

alm, the 3457A is a great suggestion.  Dang that's one ugly multimeter, and kinda slow too.  But, 6.5 digits for a few hundred bucks?  I can try it and not worry too much if it doesn't work out.  Definitely a finalist, even with a GPIB<->USB interface and having to mess with http://dawm.sourceforge.net/

Now to find one...

[grenert]

HP 34401A, cheap and local:
sfbay.craigslist.org/eby/eld/3988747551.html

[alm]

alm, the 3457A is a great suggestion.  Dang that's one ugly multimeter, and kinda slow too.

Yep, the display sucks. It's actually one of the fastest meters from that era (tens of readings/s in 6.5 digit mode and hundreds in 5.5 digit mode), but technology has advanced since then. The 34401A is (slightly) slower, for example. Still, I think it (or a used 34401A which is more expensive but supports RS-232) is probably the cheapest solution that satisfies your requirements. For the 34401A, check the maximum baud rate. In ASCII format a few hundred 5.5 digit readings might actually qualify as high bandwidth in RS-232 terms . Certainly more than a 9600 baud connection can handle. Both Dave and Martin did videos about the 3457a by the way. Dave did a teardown, Martin a review/trying to figure out how this thing works.

Definitely a finalist, even with a GPIB<->USB interface and having to mess with http://dawm.sourceforge.net/

Interfacing should be no harder than any other meter: you have a wire-level protocol provided by a library/driver (eg. the USB GPIB interface, USBTMC, LXI) over which you send plain text (I forgot if the 3457A supports SCPI, but they will be documented in the manual) commands. In the case of an USB GPIB adapter you would use the linux-gpib library (or interfaces on top of it) for 'real' GPIB interfaces and a serial port interface for Progolix interfaces. For USBTMC you might use the kernel usbtmc driver with whatever userland interface is available. Is there some turn-key solution for the Agilent or Rigol meters that works under Linux? Scraping a web interface isn't any nicer or easier.

[HKJ]

I would stay away from Rigol, it does not support a socket connection, where you can directly send commands. I do not want any driver for a LAN connection, I want a direct socket connection, this makes it independent of OS and OS upgrades.
I have been using the Fluke 8846A with LAN connection for some time and it works great and was easy to interface to.

[bronson]

Dave's teardown was good and the discussion thread was great.  Other than some very obsolete parts the 3457a looks fairly serviceable.  On the other hand, I sat thru 20 minutes of Martin's before giving up.  Does information density get better later on?

As for connectivity, Agilent are really finding their way into my heart.  http://www.home.agilent.com/agilent/editorial.jspx?cc=US&lc=eng&ckey=1189304&nid=-34952.0.00&id=1189304  The '61 has LXI, the '60 doesn't.  That page convinces me that I can connect to the '61 using all sorts of interfaces.  The code looks a little copy-n-pastey but that sure beats calling into big, brittle libraries.  At first glance it looks great.

As for the Rigol, they hide their programming guide behind an email collecting page and supply it in a .chm file.  SCPI doesn't look real pleasant and their code examples are really overwrought -- they seem targeted at people writing full GUIs, not at people who want to grab a quick sequence of bytes.  So, workable, yes, but will take a lot of effort.  (just confirmed by HKJ, nice)

grenert, that stupid thing has been listed for months.  I've called the guy twice, both times he claimed it's sold / not available and that the ad should be taken down then, bam, he relists it again.  No idea.  It sure doesn't seem to be selling.  Maybe it's a bait and switch tool?

[alm]

Dave's teardown was good and the discussion thread was great.  Other than some very obsolete parts the 3457a looks fairly serviceable.

Apart from the battery that you should probably replace after buying one (voltage is not a reliable indicator of remaining capacity for these batteries), I don't see any issue with obsolete parts. It will still be more serviceable than the current Agilent and Rigol DMMs with their custom parts and lack of schematics / good service documentation.

On the other hand, I sat thru 20 minutes of Martin's before giving up.  Does information density get better later on?

Not really, hence my comment about it being about him figuring out how the thing works. I remember cringing a few times when watching the video. It shows the meter in action though, unlike Dave who I think only took it apart.

My impression (not owning any Rigol equipment) is also that it will probably much more painless with the Agilent meter than the Rigol. The chance of running into weird issues like the meter getting confused after a certain sequence of commands, LAN connections being unreliable or the meter not living up to its specs is much higher with Rigol than with Agilent in my opinion.

[bingo600]

I got a 3457A with Opt-44491A (8 Chan relay board) , got it from*Bay  this spring for $260.

The 44491A , is a neat thing , as you can switch between 4-wire ohm and Volt without physical intervention.
... Can you do that wo. the Mux ? , i mean connect all 4 wires , and then just select DC (use the two) and select 4-wire and use all 4 ?
It might work .. My meter is out for Cal , so i can't test.

There is also a 44492A opt (10 x reed relay) , see *Bay #221280900890 , but 420$ & "as is"
I'm using linux-gpib with an Agilent USB-Adapter from *Bay , and toying around with the linux-gpib python bindings.
AlfBas (user here) has the same meter with Mux , and afaik. he's made some Windows prog. to do the Mux readings.

/Bingo

[Rigby]

That's really funny that the 34461 uses a Java client.  Just when I thought client-side Java was finally dead...  Hope they don't rewrite it to use Flash.

Java clients aren't what's bad, it's the in-browser Java stuff that's terrible.  Still, for cross-platform, client-side apps, .NET is probably the better solution, even for Linux and Mac.

I'm not trying to start a war, here.  I have years of experience with Java and .NET, professional enterprise-level experience, and this is just my opinion.

[WarSim]

That's really funny that the 34461 uses a Java client.  Just when I thought client-side Java was finally dead...  Hope they don't rewrite it to use Flash.

Java clients aren't what's bad, it's the in-browser Java stuff that's terrible.  Still, for cross-platform, client-side apps, .NET is probably the better solution, even for Linux and Mac.

I'm not trying to start a war, here.  I have years of experience with Java and .NET, professional enterprise-level experience, and this is just my opinion.

Not starting a fight, just asking. 
Aare you referring to Mono the third party partial port of .net?
I have read only bad about it, are you endorsing it?

[Tepe]

My meter is out for Cal , so i can't test.

Where are you getting it calibrated? I have a 3457A I have been wanting to get calibrated.

[saturation]

If you are in the USA, why buy when you can rent a 6.5 digit for $150/mo,  If you've a good credit rating and if your project is of short duration to make it cost effective.

Just example:

http://www.us-instrument.com/commerce/catalog/product.jsp?product_id=1426&czuid=1378996024896

Note, depending on who the renter is, quote rates between $90-150, be sure to include S&H and shipping insurance.  7.5 digit about $300, 8.5 digit about $500/mo.

[Rigby]

Not starting a fight, just asking. 
Aare you referring to Mono the third party partial port of .net?
I have read only bad about it, are you endorsing it?

Mono.  For what I've used it for, it's been excellent.  It doesn't include WPF or WWF (I think) but everything else is there.  Lately I've been working on an app that ties a LeapMotion controller to an AR Drone 2.0 quadcopter, allowing you to fly the copter with only gestures.  I just ran it on a Linux box with Mono a moment ago and it works identically to the Windows version.  I had to use different LeapMotion libraries (as they are not .NET under the covers) but after that small change, it worked fine.

[iceisfun]

I have good luck with mono too

Also I have a 34461A and its pretty awesome, with the firmware update the boot time is about 25 sec which was much longer before.

[WarSim]

Thanks I will put Mono investigation on my todo list. 

[Andreas]

Hello,

what do you really want to measure?

I think its neither resistance nor voltage.

In my opinion you want to measure the voltage ratio between the bridge supply and the bridge difference voltage.
So you will either need a DMM with ratio measurement and differential input,
or as alternative a simple dedicated ADC development board. (e.g. with the AD7190 ADC converter).

With best regards

Andreas

[bronson]

Andreas, that's part of the reason I want 6.5 digits.  That way I can measure the raw resistance change over different load profiles on different cells at different temperatures to get a baseline, then I can watch how my attempts at differential amplifiers affect the readings.

Since the '61a has >10Gohm isolation resistance, it should measure the difference in the wheatstone bridge no problem.  At 10V excitation, figure the cells will put out a few tenths of a millivolt differential...  Measuring that is easy.  I just need to figure out how to amplify it 1000Xish on the cheap -- linear with no noise -- and scale it for a microcontroller's A/D.

If I trusted my amps and these load cells more, you're right, I wouldn't need that resolution.

Besides, 6.5 digits sounds super handy.  Want!

[bronson]

Saturation, that's a good point about renting.  I figure $150/mo for four months is 60% of the '61a's purchase price, and six months ends up at 80%.  And, this sounds like a useful tool even after the project is over.  Tell me if I'm missing something...  (The $150 is a quote for '10a -- haven't found an available '61a for rent).

If I needed 7 or 8 digits I'd rent.

[orin]

Saturation, that's a good point about renting.  I figure $150/mo for four months is 60% of the '61a's purchase price, and six months ends up at 80%.  And, this sounds like a useful tool even after the project is over.  Tell me if I'm missing something...  (The $150 is a quote for '10a -- haven't found an available '61a for rent).

If I needed 7 or 8 digits I'd rent.

I'd think you'd be better off buying one and selling it after 4 to 6 months than renting one...  but I'd plan on keeping it.  I wouldn't part with mine for less than I paid for it.  (Metrictest had a 10% online discount that got eaten up with taxes and shipping so it was a little over $1100 when the dust settled.)  BTW, the warranty extension to 5 years is only $50 - I need to call Agilent to get it for mine.  Should have ordered it with the instrument.

[bingo600]

3457A on *Bay #360734525681  (Auction)

/Bingo

[bingo600]

My meter is out for Cal , so i can't test.

Where are you getting it calibrated? I have a 3457A I have been wanting to get calibrated.

This German company is offering a 3457A Cal. (ISO9000) incl. justage for 99€ + Vat.
I got an offer of 153€ incl VAT & Return shipping to DK , that's a very reasonable price.

kalibrierlabor.de

/Bingo

[Tepe]

This German company is offering a 3457A Cal. (ISO9000) incl. justage for 99€ + Vat.
I got an offer of 153€ incl VAT & Return shipping to DK , that's a very reasonable price.

kalibrierlabor.de

Thanks. That is definitely cheaper than what I had found so far.

[saturation]

If you find you'll use a 6.5 DMM beyond a projects time line, owning is a better option even if the project's timeline is only 1-2 months; a real issue is justifying the cost of new for a personal purchase.  A used DMM with a PC to do the data logging: a 3455a or 3456a typically sells for < $100, 3457a < $300 but you'll need some time to find one in good order, check it out and calibrate it.  They support GPIB at best.

For rental, its only a clear decision if you keep it for 1-2 months, beyond it it gets blurrier as the rental costs approximate purchase costs.   

Saturation, that's a good point about renting.  I figure $150/mo for four months is 60% of the '61a's purchase price, and six months ends up at 80%.  And, this sounds like a useful tool even after the project is over.  Tell me if I'm missing something...  (The $150 is a quote for '10a -- haven't found an available '61a for rent).

If I needed 7 or 8 digits I'd rent.