Very fast oscilloscopes

You shouldn't need to. MATLAB, or whatever else you'd like to let loose on your acquisitions, can run on the desktop PC, which need not run Windows either, incidentally. Networking should make this entirely transparent.

There's still some way to go, that much is true.

Jeroen Belleman

MatLab, LabView, all those are apps where users do not expect hard realtime performance. Just like they don't in MS-Office. From a scope I do expect hard realtime and so far all the Windows-based scopes I got to "enjoy" at clients have failed to deliver in that domain. To the point where we ended up schlepping an old boat anchor out of the basement so we had a real scope.

The idea is that Windows programmers are a dime a dozen, Windows itself is "cheap enough" (~$25-$100, depending on the version of Windows and the quantity, is nothing in an instrument with a five- or six-digit price tag!), and hence it's faster and cheaper to just use something "off the shelf" rather than rolling your own OS.

Heck, some software guys I know are currently lobbying to buy not just an OS, but an RTOS to run something on the order of complexity of a cordless DECT phone (i.e., LCD display, handful of buttons, two-way low-speed digital wireless, etc.); they've budgeted $50k for it. I'm not personally very enthusiastic about this, but I'd have to admit that if it saves them from having to hire one person even for six months to a year to write additional software, it will have paid for itself.

What I worry about is programmers who think that spending money can somehow magically fix all their bugs, when in actuality the bugs are largely due to the individual programmers and have very little to do with the tools they use... and getting more powerful tools can actually backfire, just giving them more ways to shoot themselves in the foot... or to blow off their entire leg rather that just a toe. :-)

---Joel

That Agilent scope appears to have the ability to let you write some Matlab scripts to perform, e.g., custom measurements and run them in "real time" as the scope is acquiring data, though -- that's pretty powerful, if you have an actual application for it. (Consider how annoying it would be if scopes didn't have built-in functions to measure rise times, frequencies, etc. and you had to transfer the data to your PC to do it... OK, I realize we all know how to count graticule lines and calculate this manually, but I think you get the point... even 25+ years ago when analog scopes were king many of the classics like the Tek 2465B were already digitally measuring these parameters for you...)

---Joel

You have to add the runtime license fees per unit into the equation. That cost is usually perpetual. The other cost in a portable application is the consequential energy overhead. Larger and more expensive battery, larger enclosure (which end users might not appreciate that much).

Key with any realtime OS is to select the right one. I had quite a positive impression with QNX, also WRT its footprint. But with phone type apps there may be better ones, where there's more pre-cooked modules tailored to that market.

I have yet to explore all the fancy trigger or math stuff in my DSO but it's amazing how fast a waveform shows up on the PC via streaming USB. Besides the normal screen shot view it also transfers Excel-ready data if that option is checked. I have not done it yet but I assume from there you are only a few VBA routines away from performing really fancy routines with the data. Excel can stomach running displays, I did that with a LabJack once.

The transfer is so fast that I sometimes use the laptop screen as the main display for this scope, especially when soldering fine-pitch SMT. That way I don't have to switch glasses but can just peek over the top of the 3x magnifiers.

I recently got this email from my friend Mark Kahrs, the sampling-scope-history expert:

==========================

Yes, the former Soviets did have a 7 GHz real time scope, and a 20 GHz transient digitizer. They both used the same technique to get the super high speed ? keep the deflection angle low.

When I managed a design project in the transient digitizer group at Tek, I went on a customer visit and met with engineers working for EG&G at the Nevada Test Site. The time was after Glasnost, near the end of the era of underground testing. A delegation of Soviet scientists were actually invited over to witness a test shot (They did the same, hosting a group of scientists from Livermore National Labs). EG&G was using the new 3 GHz transient digitizer from Tek, and thought its "leading edge" technology would impress their former foes. The Soviets were not impressed, stating that they had a 20 GHz digitizer. How did they do it? They way they surpassed many obstacles ? by not placing needless restrictions on the technology. A design requirement for the US made digitizers was that they must fit in a standard 19" rack. The Soviets took a pragmatic approach, and avoded any size restrictions. The CRT in their digitizer was just short of 10 m in length! They used traveling wave distributed deflections plates as our designs did, but the beam deflection angle was incredibly small, allowing very high BW. Likewise, the 7 GHz real time scope had a CRT over 3 meters long.

The funny part of this true story is that the EG&G team, confident that they had just showed what they thought was superior technology, asked if the Soviet scientists had any questions about it. They did. They asked "How do you get the walls so smooth in the tunnels?" The shot monitoring tunnels at NTS were drilled with diamond tipped tunnel boring machines, leaving a relatively smooth wall. The Soviets simply blasted their tunnels with dynamite.

=================

John

Great story John -- thanks for posting that!

It is rumored that the end of the Krasnojarsk over-the-horizon radar didn't come because of glasnost developments but more because of crumbling masonry and concrete.

On a sunny day (Fri, 30 Apr 2010 09:29:54 -0700) it happened Joerg wrote in :

Interesting article on Motorola making a profit again, because they moved from 6 OSses to one : Android.

When we started this turnaround, we had six mobile operating systems and 23 platforms, said Mr. Jha. We were able to reduce that to one and it allowed all of our creative energies on one platform.

"But Mr. Jha said Motorola's comeback went beyond Android. "There's a point of view that all that matters is the operating system," said Mr. Jha. "But consumers care about what features a phone has, what it can do and where they can get it. At Motorola, we have some better capabilities than others in those areas."

^^^ This quote I don't really agree with -- I think that for most customers you can pick any major OS you want (iPhone, Windows Mobile, Droid, etc.) and "there's an app for that"; customers really do still consider the operating system and the phone's form factor as well. Look at Windows Mobile 6: Up until a year or so ago, it had far more apps than the iPhone, it's always supported true multi-tasking, copy/paste, etc... but it has this kinda "klunky" look and feel to it, and customers preferred iPhones by a large margin.

What's going on with phones today reminds me a lot of the wild and crazy times with PCs back in the '80s -- lots and lots of choices and everyone competing fiercely on price and features to try to gain market share.

If I were Research In Motion, I'd be pretty worried that I'd be the next Palm right now...

In my experience you should choose a platform get really familiar with that and keep using it. Using third party libraries has its advantages but usually there are some weird quirks that need fixing or workarounds. Anyway, using a third party library is usually faster than implementing it yourself and fixing all the bugs instead of a few. With some luck a third party library has seen more testing in different environments.

The Agilent 5000/6000/7000 series are very good in this respect. They run on VXWorks, start in about 10 secs and you _never_ have to wait for anything to happen!

SD-24

VXWorks, start in

My Instek starts in about a couple of seconds :-)

I'd wager dollars to donuts that if your Instek even *has* an operating system as such at all, it's something they wrote themselves...

Some of the error messages sound like that. "DSO not connect" ain't exactly a Canadian accent :-)

They are quite silent about the innards. However, what sold me on this scope is that they laid the whole command set open, including stuff way down in there. You can talk to it via an emulated COM port on the USB link which makes the job rather easy. Also, they don't nickel and dime you, their own PC software for it is a free download. With certain other brands that's different, "Oh, you want fries with that? That'll be option xxx at $499".

?Everyone ran out of love? (KISS)?

They want you to upgrade when MS upgrades..

Just think of that cash machine! :)

-24

he

=3D=3D

I guess shipping would be a problem for these guys if they showed up on eBay.