EEVblog #485 - Agilent TrueVolt 34461A Multimeter Teardown

[Rufus]

Inside the new Agilent TrueVolt 34461A 6.5digit bench multimeter. A replacement for the venerable 34401A.

Photos: http://www.flickr.com/photos/eevblog/sets/72157634225126018/



May as well start the thread.

I previously read a bit about multi slope III, don't know what IV changes.

Agilent's multi slope conversion is more like a slow delta sigma conversion.

You have an integrator with two comparators and inject the unknown current and a + or - reference current so the voltage on the integrator capacitor bounces between the comparator limits. Over time the ratio of time spent injecting + and - reference currents tells you what the unknown current is. Accuracy doesn't depend significantly on anything but the accuracy of the reference current. The comparators don't have to be accurate they just keep the integrator voltage in bounds.

For more accuracy in less time you measure the voltage on the integrator capacitor at the start and end of the conversion period. The current injection ratio gives you the most significant measurement bits and measuring the initial and residual integrator voltage can give you another 8, 9, 10? bits.

The integrating capacitor doesn't have to be stable or accurate, you can easily calibrate it with the reference current.  Dielectric absorption might still be an issue with this type of converter.

The manual recommends using the 10A input for currents over 1A. The 10A input has a fixed 10A full scale range.

Bits left of the board are interesting - wonder if there is another model to come or maybe they are alternate rather than additional parts, perhaps for the 60A? 

[duskglow]

I found the fact that it boots into a Linux kernel and then runs Windows CE on top of it to be really interesting.  No idea how that works.  Or is that booting Linux on one of the other ARM chips on the main board?

[Rufus]

I found the fact that it boots into a Linux kernel and then runs Windows CE on top of it to be really interesting.

I don't think it does. U-boot describes the boot image as "ARM Linux Kernel Image". Doesn't mean it actually is.

[duskglow]

Huh.  Guess so, at that.

[don.r]

Would the fan not cause some issues with the temp stability around the ref or does the cage and low tempco resistors take care of that?

[Dave]

Does anyone know why both the 10A and 3A current ranges are protected by 11A fuses? Does that fuse really protect the 3A input?

I also want to point out a factual mistake at 23:07: "... you can then switch it back down to ground and it's gonna discharge ...."
If you connect the input of an integrator to ground, no current will be flowing - it won't discharge. I think this calls for an annotation.

I was just hoping to see you connect it to your voltage standard. Guess we'll have to wait for the review. Looking forward to it.

[Monkeh]

Does anyone know why both the 10A and 3A current ranges are protected by 11A fuses? Does that fuse really protect the 3A input?

There's a cheap 3.15A fuse on the back to blow first. The 11A fuse internally is in case of a fault which that dirt cheap fuse can't handle.

Used to be seen on handhelds, too.

[c4757p]

Wow, Dave, what a fun little comment thread this video has. I think I've heard enough about current through capacitors for a month or two 

[EEVblog]

Wow, Dave, what a fun little comment thread this video has. I think I've heard enough about current through capacitors for a month or two 

Does current "flow through" a capacitor?
NERD FIGHT! 

I just contemplated doing a Fundamental Friday on this, but it's probably just one big huge can'o'worms, and I would almost certainly get some technical detail "wrong"...

[c4757p]

Oh god.  I can only imagine the results of digging up this one again.



If you value your sanity, don't go anywhere near that.

Fundamental Friday as a whole seems to have started a good few nerd fights. Never underestimate the ability of a pair of jackasses to argue about things that should be established facts...

By the way, at 12/page, this forum is about to reach critical emoticon mass. I suggest nobody use any for a few months or so.

[Dave]

Well, we might have gone a bit overboard on that. I in particular.
But it was an interesting discussion, was it not?

[Harvs]

I never read youTube comments but given the interest I took a look.  Can I get a refund on those 30sec of my life?

[ddavidebor]

Sure, i've read your post.

Now we have a debt of 30 sec each other, so, no debt.

Happy? XD

[duskglow]

Is that Kirchoff's law for debt?  All debt going into and out of a post equals zero?    Is this debt flow or displacement debt?

*ducks*    

[free_electron]

Does current "flow through" a capacitor?

Current flows in and out of a capacitor. But it does not flow 'through' (apart from some dielectric leakage)

[bexwhitt]

Is the current flow through a capacitor like the electron hole, sort of a thing what works in reality?

[FrankBuss]

Looks like Agilent uses the same boot sequence like in some of their oscilloscopes, with u-boot starting WindowsCE. With the scopes you can actually stop the u-boot booting process with enter on the RS232 input, and then you get some commands with "help" (you can't do anything when WindowsCE has started, then it's just a debug output). It is even possible to load and start a WindowsCE image from network (at least with the scopes), if someone wants to play with it

[uprightsquire]

Does current "flow through" a capacitor?

Current flows in and out of a capacitor. But it does not flow 'through' (apart from some dielectric leakage)

And then there's the fact that 'current' in a wire isn't really even 'flowing' through the wire anyway. Net electron velocity is on the order of 1.0 × 10-5 km/hr or something.

Maybe we can start arguing about that whole conventional current flow deal. 

[dr.diesel]

Maybe we can start arguing about that whole conventional current flow deal. 

Let's add what religion is best also! ha ha.

[bktemp]

I wonder how the digital rms measurement works.
I would have expected a fast high resolution ADC (16bit+, a few MS/s) feeding the data into the FPGA which calculates the rms value, but there are only the two low resolution ADCs.
Does somebody know, how it works? Are they using these low resolution ADCs with oversampling to get an accurate AC measurement or is there some other analog magic happening there to get the rms value?

[JackOfVA]

Wow, Dave, what a fun little comment thread this video has. I think I've heard enough about current through capacitors for a month or two 

Does current "flow through" a capacitor?
NERD FIGHT! 

I just contemplated doing a Fundamental Friday on this, but it's probably just one big huge can'o'worms, and I would almost certainly get some technical detail "wrong"...

As a matter of computational convenience, Maxwell considered "displacement current" to flow through a capacitor and we use that computational convenience today. Certainly a lot easier to think of it as "real" current than a mathematical construct.

[AlfBaz]

Maxwell considered...

Bah! What does he know, He couldn't even make decent tapes cassettes 

[rsjsouza]

I am pretty sure the mysterious battery manager TPS65073 device was used due to pre-existing device drivers on the embedded OS (WinCE).
These power management devices are very complex and probably Agilent did not want to spend much time in modifying this specific low-level device driver (TI has reference designs for their own SoC processors, where this device controls multiple power outputs)
The serial port is what one would normally get when booting any embedded OS on these embedded computing boards (BBone, RPi, etc.). As FrankBuss mentioned, however, the command prompt is disabled in WinCE unless you interrupt the u-boot sequence.

[RobB]

Looks like an impressive bit of gear. Just a couple of quarms with it though. the first is the extensive use of complex plastic parts. Surely that is not good for long term structural integrity. Take the power switch link for example, it clearly distorts when used. Try that again in 10-15 years and I'm tipping a failure.
Second I am loathed to trust any mission critical hardware managed by anything Micro$oft.  The communists like Rigol et al got it right on this one using an open sourced operating system.

[duskglow]

In their defense, Microsoft's OS really is not on the critical path on this one.  Since the data acquisition is done in hardware and the only real purpose for the OS to coordinate the *display* of that data, it's likely that any failure mode exhibited will be pretty obvious.  I would consider the odds of the WinCE software actually screwing up the display such that it displays the *wrong* value to be really small.  I would consider the odds of it crashing to be much higher.

I share your distrust for MS software but in this particular case - unless the mission critical aspect is 100% uptime, I'm not too worried about it.