EEVblog #485 - Agilent TrueVolt 34461A Multimeter Teardown

[cthree]

You base all of this on a sales pitch at an event some years ago? I actually fully evaluated WinCE along side VxWorks and we choose VxWorks because it was 'more engineery' than CE (did I mention the company failed?)

All of the statements poopooing Windows CE so far are based on ignorance and hyperbole. Please spare me the indignity of posting links to Wikipedia. You know how to get there without help.

[dr.diesel]

Then you must be willing to accept the USB and LAN functions would be forbidden in some corp environments.  I have worked in two huge corporations that forbid a Windows OS that could not run the corporate antivirus, absolutely no exceptions.

What's that got to do with designing test equipment? 

Do you want to use the USB or LAN connectivity in corp environments?  Meaning either connecting a USB flash or USB data cable to another PC or via LAN?

Many people use these data and software packages to pull info from the test equipment, which is usually done via USB or LAN.  This would be forbidden at several places I've worked because the equipment was incapable of loading antivirus.

[dr.diesel]

Then those corporations are stupid, plain and simple.  Also, this is very rarely true.  A corporation that can't set up a private T&M network segment should fire their IT department, shut down the company, and burn their building down.  Agilent developing a custom complex RT OS that will be guaranteed to be less secure and more buggy than WinCE just to satisfy some imaginary corporate rule would be stupid.

Ha ha, you've apparently never worked for large utilities, IT is worse than HR.

[Bored@Work]

You base all of this on a sales pitch at an event some years ago?

No, it was a paid-for, multi-day developer conference, only that Microsoft decided to make it more into a multi-day propaganda event.

We went there because we used CE, and thought we did something wrong, because the results sucked so much. After the conference we were sure it wasn't us who sucked, it was CE. We continued to use CE for some more time, shipped the products we had under development and then dropped the junk like a hot potato.

[mikeselectricstuff]

IMHO a simple multimeter display really doesn't need the price overhead of WinCE.
But if that is all the manager who's decision it was knows then it was probably the better choice.

Who thinks this is a "simple" multimeter?
It has USB interface, file systems, Ethernet, a huge graphical user interface display that does all sorts of stuff etc.

For a product like this, the need to use an OS is marginal - there are plenty of standalone graphics libraries, the UI is just buttons. USB functionality is very limited, and filesystem is not a huge deal either.
If you were just designing this product in isolation, then using any 3rd party OS could probably be avoided with similar development effort to learning and adapting one.
However for a company like Agilent, they need something that's useable on a wide range of architectures, including products that really need a complex OS.
If the OS was designed in a way that allowed it too be easily stripped down to only what was needed for a particular application it wouldn't be an issue -  what does the DMM OS actually _need_ to do at startup? Almost nothing - just some selftests.
I'm convinced that the bulk of boot time in OSs like this is nothing but lazy programming - inappropriate timeouts, trying to talk to hardware that isn't there, decompressing code that's never needed, not threading things like network initialisation properly etc.
It would certainly be possible to write an OS for this sort of application with negligible boot time.
The problem is how many people consider startup time a sufficient issue to be a major selling point? Unfortunately very few, so there is no incentive to invest in improving things.
This is a problem with the majority of embedded systems - many modern LCD TVs take longer to turn on than the old CRT ones.

BTW I hear on good authority that the priority for this product was getting something out the door that worked properly, and there is a new firmware version in the works which is likely to approximately halve the  startup time. 

However annoying boot time is, Software development time is always finite, and  given the choice between slow boot and flakiness in operation, the former is the least-worst option.

[John Coloccia]

Windows CE has been rock solid for 15 years.  The development environment is light years ahead of everyone else...Visual Studio is in it's own league and really has no competition.  Unix/Linux environments are much as they were 20 years ago...and 30 years ago...and 40 years ago.  It's like playing with tinker toys.  The threading model and libraries for Windows are more advanced and more usable than anything you'll find elsewhere.  Basically, it's a programmer's paradise if you approach it with an open mind.

Assuming you're going to use an OS at all, WinCE is the slam dunk, obvious first choice for a device like this, and by a wide margin.  Someone else mentioned cost.  I believe Core still costs <$5 per device.  It's not like I'm even a fan of Microsoft or Windows, but we're not talking about desktop environments.  It's about selecting the right tool for the right job.

[dr.diesel]

The problem is how many people consider startup time a sufficient issue to be a major selling point? Unfortunately very few, so there is no incentive to invest in improving things.

Good point.

Turn it on in the morning, off at night - if you remember.

[dr.diesel]

The threading model and libraries for Windows are more advanced and more usable than anything you'll find elsewhere.

Ha ha, tell that to the guys at top500.org.

[Stonent]

There's an air leak test device I saw before that pressurizes the system and does a leak down check and it runs an old version of Windows CE also, but it's pretty much completely non-interactive, no touch support. It just has a 10 digit keypad and some start and stop test buttons. They've got their test dialed in and doubt if they change it ever.

No need to reinvent the wheel. Once you get the kernel booting and something on the screen, you just get yourself a regular windows programmer to write the code.

[John Coloccia]

The threading model and libraries for Windows are more advanced and more usable than anything you'll find elsewhere.

Ha ha, tell that to the guys at top500.org.

Ha!  OK, maybe that statement needed to be qualified a bit 

[jzoeller]

The only thing keeping me from buying this meter is the lack of capacitance measurement.

[Christe4nM]

Well, IMO a decent LCR meter does capacitance more accurate anyway. So if you're after capacitance, why not go for both the 34461a and a U1700 series for example. Together the price point comes closer to the 34410a or 34411a, but now you have a good LCR meter too. Also there are good alternatives to the U1700's that are cheaper as well. Just browse the forum for examples.

[FrankBuss]

If the OS was designed in a way that allowed it too be easily stripped down to only what was needed for a particular application it wouldn't be an issue -  what does the DMM OS actually _need_ to do at startup? Almost nothing - just some selftests.
I'm convinced that the bulk of boot time in OSs like this is nothing but lazy programming - inappropriate timeouts, trying to talk to hardware that isn't there, decompressing code that's never needed, not threading things like network initialisation properly etc.
It would certainly be possible to write an OS for this sort of application with negligible boot time.

WindowCE has a good "platform builder" software, where you can select which components you want in the image. No need for USB or Internet Explorer? Just deselect it. Makes the image smaller and the boot time faster.

Linux has a similar system, Buildroot: http://buildroot.uclibc.org

I've worked with both and there is no best system, both have their pros and cons, depending on the product and on other factors, as written here before by others, like how fast you have to build program the firmware (writing a .NET application with USB support for thumb drives can be much easier than trying to implement your own USB host controller firmware without an OS), is there a BSP already available for it, do you have engineers who already know Linux/Windows programming etc.

[fpga]

A multimeter needs WinCE in the same way a guinea pig needs armor:

     , couldn't agree more.

How about a retro project with rotary switches and good old moving coil meters! If we want to go digital, we could use Nixie tubes! GUIs are for sissies!

Seriously, that could be a fun project. Make a retro front panel, complete with plywood, that controls a modern instrument over USB or LXI!

[fpga]

Then you must be willing to accept the USB and LAN functions would be forbidden in some corp environments.  I have worked in two huge corporations that forbid a Windows OS that could not run the corporate antivirus, absolutely no exceptions.

In these cases Corp IT doesn't care or understand the differences between the versions, any alteration would take an act of God to bypass.

I've experienced situations where a networked scope in a lab running Win98 had launched an inside attack on a corporate network. The IT department had to shut down all of the servers, check every desktop PC, and the last place they looked or suspected was a piece of test equipment in the lab. Nobody even thought about putting anti-virus software on a scope.

The WinCE based devices boot a fixed image from flash each time they are powered up. In the very unlikely event that they could get infected with a virus, all you would need to do is power cycle them to clear it.

[duskglow]

Running Windows CE on a piece of test equipment is defensible.  Running Windows 98?  Whoever made THAT decision needs to never work in the industry again...  That's like trying to run Commodore BASIC on a Cray.

[fpga]

Running Windows CE on a piece of test equipment is defensible.  Running Windows 98?  Whoever made THAT decision needs to never work in the industry again...  That's like trying to run Commodore BASIC on a Cray.

Most of the high end test equipment has a life of 10 to 20 years, particularly if it cost over $100k when new. Back in 2006 when I witnessed the event, that scope was still being used daily.

[duskglow]

My point is, that even when Windows 98 came out, it was a buggy piece of crap that no one in their right mind would have used for mission critical embedded applications.  Windows NT existed at around the same time, that would have been a much better choice, though probably way more expensive.

[FrankBuss]

How about a retro project with rotary switches and good old moving coil meters! If we want to go digital, we could use Nixie tubes! GUIs are for sissies!

Actually once I implemented a moving coil meter on a 7" TFT (for audio measurement), because the customer wanted the look and feel of the old instruments. Simulating a realistic needle movement for really old school VU meters was not easy, but marketing loved it

The WinCE based devices boot a fixed image from flash each time they are powered up. In the very unlikely event that they could get infected with a virus, all you would need to do is power cycle them to clear it.

Not in the Agilent DSO-X 3000 series scopes. They mount a partition at boot time where the scope software is installed. But the OS system programs and DLLs are still in the read-only image, so maybe unlikely that a virus infects the custom DLL and EXE files.

[dr.diesel]

The WinCE based devices boot a fixed image from flash each time they are powered up. In the very unlikely event that they could get infected with a virus, all you would need to do is power cycle them to clear it.

Very true, but big corporate IT morons are very static and don't listen to reason.     In cases like these, they'd never know anyhow, but it would be your ass if some one in a billion event happened and it caused trouble.  Out would come their 500 page IT policy and you'd be a gonner!

I've also worked were to receive LAN permission you had to submit a document, include the OS, what antivirus and list of all loaded patches! 

[fpga]

Very true, but big corporate IT morons are very static and don't listen to reason.     In cases like these, they'd never know anyhow, but it would be your ass if some one in a billion event happened and it caused trouble.  Out would come their 500 page IT policy and you'd be a gonner!

I've also worked were to receive LAN permission you had to submit a document, include the OS, what antivirus and list of all loaded patches!

I once worked for a European company that locked down the LAN to authorized MAC addresses in addition to VPN and domains. You would be automatically logged off after 15 minutes of non-use.

That same company required a 5 page approval form signed by your manager, the director, and the company CEO before they let you plug a desk lamp to a wall power outlet. You needed to make sure the lamp had all of the appropriate CE certificates, proof of proper training on its use, estimate its annual power usage, describe why it was necessary, steps you would take to minimize its use, responsibility of failure, responsibility of liability, and so on. Plugging in a phone charger was considered theft that would mandate disciplinary action and could result in dismissal. Luckily a coworker warned me before the manager could find out.

[duskglow]

Wow.  That would cost more money to take the time to fill out and approve the form than it would to power the lamp for a year.

[c4757p]

... lamp ... proof of proper training on its use

 

How do people get to be in charge of making these rules? By banging their heads really hard on a wall like this? Whoever lasts the longest wins?

Jesus Christ, I wonder if anybody ever thought to compare the money they save by making sure everything is done "properly" (and considering use of a phone charger "theft") with the money it costs to maintain all these layers of idiocy.

It's not even like Dilbert, it's like a parody of Dilbert. That's sad.

[fpga]

It's not even like Dilbert, it's like a parody of Dilbert. That's sad.

The thought of being dragged on the carpet in front of these people is even worse!

[mike_kawasaki]

IMHO a simple multimeter display really doesn't need the price overhead of WinCE.
But if that is all the manager who's decision it was knows then it was probably the better choice.

Who thinks this is a "simple" multimeter?
It has USB interface, file systems, Ethernet, a huge graphical user interface display that does all sorts of stuff etc.
It needs a complex high level O/S to operate this stuff.
Also, remember that the WinCE platform decision would have been made a long time ago, way before this meter was even though of.
You can't just go picking and choosing the latest flavour of the day O/S for your new product, that would be silly. Any smart company choses an O/S platform very wisely and then stick with it because they become heavily invested in using that standarised O/S platform across their products.
So anyone who's criticising the use of WinCE really needs to understand not only why the decision was made (which can be many many reasons), but also when, and over what range of products it's intended use was.
And when you do that, the choice may not be at all as silly as you think, or a bad one. In fact it may have been the best choice.

Dave, you are spot-on with your answer.

At Agilent we have standard platforms for different levels of test equipment (in laymen terms -- basic, mid-range and high-performance).  The 34460A uses the basic platform.  The platforms are chosen by a central Technology Leadership Organization in partnership with the Divisions.  The Technology Group standardizes on all the design choices to greatly simplify the product design engineer's job -- and so they can focus on the acquisition and display HW/SW design.

These decisions have become even more important as the Test Equipment value resides more in the software applications.  Standard platforms allow us to leverage the software design.  For instance, we just introduced the Real-time Spectrum Analyzer on our MXA model, just months after announcing it first on the PXA model.  In years past this would have taken at least a year to write the same code for a different platform.

I can't remember when the Basic Platform design was last updated, but it has been years since the decision to use WindowsCE.  As also mentioned in a few entries, Test Equipment is often used for >10 years, so the platform decisions require a balance between cost, design tools, functionality and vendor support.

As always, impressed with this group's knowledge and passion for Design Decisions!!

Mike