Buying scope: Agilent or Tektronix

[TerraHertz]

It's not just the ADC. In fcat you can argue it's not the ADC at all...
Take the Agilent X series for example. They use the same ADC in their entire range from the $1K 70MHz X model through to the $12K 500MHz model, and likely the 1GHz model as well. So it seems that they can afford to put the same ADC in all the models.

See? A case of do the R&D once, use it for all models, then let the marketing dept decide the features vs pricing structure.

But they don't do the same thing with the front end. They have three different front end circuits, 200MHz, 500MHz, and 1GHz. I wonder why...

I'd guess the 1GHz version is 50 ohm? But for the high impedance ones, 200/500MHz, you keep arguing that because two different versions are made, that proves they have to do it that way.
Bear in mind that if they did try to maintain a pricing scale from 70/100/200/500 MHz while using the exact same input hardware for all, how long do you think they'd get away with it before customers became enraged?

You're the one with access to the guts of all these versions. Why don't you take photos of the input blocks for the 200, 500 and 1GHz versions, then think about what the relative manufacturing costs would be, roughly. Maybe I'm wrong, maybe the 500MHz version simply couldn't be made for a similar cost to the 200MHz one, no matter the volume production scaling. But I'd be surprised.

TerraHertz's comment that low end scopes use repetitive sampling is nearing a decade out of date!

Ha ha ha.. oh well, that's what I get for only being able to afford second hand gear.
It's true, I was not aware baseline scopes had gone to real time sampling right up to the full bandwidth.

Though still, in the case of high speed digital serial like USB, LPC, LVDS, SATA, etc, does anyone ever want to sample single shot waveforms? Thats what serial protocol analyzers are for, while scopes are for checking signal integrity, eye patterns and such. For which repetitive waveform sampling is fine.

[lewis]

Thus demonstrating their tendency to dishonest pricing structures.

My point is that they're choosing to make the lowest end one, solely to justify the higher price of the next model up the scale.

...they wouldn't, because it would reveal that they _could_ do 500MHz for less, thus pulling their whole pricing scale lower.

For the purpose of retaining a seeming justification for the medium to high prices. It's a pricing spectrum solely driven by marketing considerations, not actual engineering and manufacturing cost factors.

They _could_ just drop that pretense, and offer scopes of 500MHz (at least) as their low cost low end models. If they wanted to.

It's very naive to assume that corporations, even ones we as EEs instinctively want to support, never let their overriding imperative (maximizing profits) lead them into reasoning that is really quite evil on some levels.

...suggests the pricing structures are artificial.

Not to mention other advantages to society if more people could afford basic electronic test equipment.

Typical socialist banter, especially the last one which is an absolute classic. "Isn't it terrible that the nasty evil corporations, who only exist to chase dollar signs, exploit overseas workers and destroy the environment, keep their prices deliberately high to prevent Africa developing..." or some such bollocks. The things are expensive, deal with it. And if you don't like their attitude, don't buy. Simple.

...I've seen of how scope front ends work - programmed attenuator, protection, buffer, sampler, AtoD. This is such old, well established technology, and with stripline ICs and VLSI can be miniaturized so small and made so standard, that it _should_ be cheap.
And yet Tek, Agilent etc claim it isn't. We can't know their true costs, I'm just saying I don't believe they justify their low end scope pricing stuctures.

All of which is necessarily confined to a little block of a few cubic cm behind the BNC connectors, and a design which for basic scopes does not need to be changed for years, regardless of how much they choose to mess with the overall scope packaging and UI.

And you're pretending that there's been any actual R&D needed in the last 10 years for any kind of scope below 500MHz...It's ALL just industrial design, making the boxes look nice, changing around the probe interface connectors, improving the software for waveform display.

If it's so simple, make one! You'll sell shit loads if you can get the price to, say $1/MHz. After all you don't need to do any R&D, you can design it on the back of a fag packet in 5 minutes. You'll put those nasty corporations out of business overnight and stop those evil capitalists exploiting Chinese children once and for all.

[TerraHertz]

Typical socialist banter, especially the last one which is an absolute classic. "Isn't it terrible that the nasty evil corporations, who only exist to chase dollar signs, exploit overseas workers and destroy the environment, keep their prices deliberately high to prevent Africa developing..." or some such bollocks. The things are expensive, deal with it. And if you don't like their attitude, don't buy. Simple.

Except that if anything, I'm a Libertarian. Nonetheless...
"corporations, who exist only to chase <profit>" is true. That is by definition, charter and structure, exactly what corporations do. And cannot do anything but. The debate over whether it's a good idea to allow the existence of immortal, soulless abstract entities, legally granted the same rights (or better in some cases) as living persons, and who's implicit objectives are in conflict with fundamental human needs, is another whole topic.

"Exploit overseas workers" is true. Deal with it. Not to mention having transfered the jobs from local workers in the first place. All for profit maximization, and screw the national economy.

"Destroy the environment" is also in many cases true. Comes from that 'no ability or desire to do anything but maximize profits' thing.

Your "... Africa" bit is pure strawman bollocks. However with the 'keep from developing' part you're getting close to a truth. I didn't originally want to touch on this aspect but since you bring it up... google 'Agenda 21'. And don't bother trying to pretend it's some wacky conspiracy theory. It's real, it's a UN initiative, budget, power, official website, international conferences, the whole globalist works. I won't bother trying to explain it here, too complicated. A few primer links here: http://everist.org/archives/links/!_Agenda_21.txt
Then do some reading on the phrase 'disruptive technology', who was talking about it, and why. Enjoy the deep, deep rabbit hole.

Another linked theme is why western societies over the last few decades suddenly decided there was no social obligation to ensure the next generation had a decent education, without burdening them with crippling debt. But I suppose now your 'socialist hackles' are rising again, so little point trying to tell you it's all part of the same deliberate trend. Nope, despite that we have a highly technology-dependent civilization, any talk of needing to keep a useful fraction of the population technically capable (and that includes having access to useful test gear) is just socialist babble, right?

If it's so simple, make one! You'll sell shit loads if you can get the price to, say $1/MHz. After all you don't need to do any R&D, you can design it on the back of a fag packet in 5 minutes. You'll put those nasty corporations out of business overnight and stop those evil capitalists exploiting Chinese children once and for all.

Now you're being silly/abusive, and distorting what I said. Which wasn't that it's 'simple', just that the existing test equip companies have been doing this for so long _they_ could make lower end scopes cheaply if they wanted to.
Oh, and notice that there are quite a few small companies that are in fact doing small, cheap USB-based sampling heads, with scope software for PCs. I'm quite hoping these start to give companies like Agilent & Tek some serious competition, and force Tek/Agilent to quit their 'low end high prices' games.
Except of course, there's the patents problem...

And if you don't like their attitude, don't buy. Simple.

Exactly right. I don't. I buy older gear that does the job I need it for, *and* I can get full maintenance manuals with schematics.
If you want to know how politically extreme I can be, here's an example: I think it should be illegal to sell any product without the *full* technical documentation being freely publicly available. That includes all schematics, binaries, source code, etc. (Well, not so much 'illegal', but with arrangements that advantage entities that *do* make that information available. Something like a clear compliance branding, that only products where all the docs are available qualify to carry the mark. Thus letting the market encourage compliance.)

[EEVblog]

And you know the actual price of doing these front ends, how exactly?

No more than you do. Which is, not at all.

Actually, I do know more.
The head of scope development (or some such, forget the exact title) at Agilent told me that the 500MHz front end costs a lot more than the 200MHz front end. Which is why they were not able to offer the 500MHz front end in base model, and hence have full software bandwidth upgradability from 70MHz to 500MHz in the scope. That is what they were shooting for but they could not do it.

So your assertion:

I'm saying specifically that bandwidths in the 500Mhz range are something the large instrument makers learned how to do cheaply long ago, even in digital scopes. In fact _especially_ in digital scopes since the high bandwidth stuff is confined to the input sampling circuits,

is clearly wrong on two levels.
One being the actual price of a 500MHz front end.
And second being that the high bandwidth stuff is confined to the sampling circuit, which as everyone knows, it is not. There is a lot of input circuitry before the ADC that is not easy to design at high bandwidths.

Dave.

[TerraHertz]

Actually, I do know more.
The head of scope development (or some such, forget the exact title) at Agilent told me that the 500MHz front end costs a lot more than the 200MHz front end. Which is why they were not able to offer the 500MHz front end in base model, and hence have full software bandwidth upgradability from 70MHz to 500MHz in the scope. That is what they were shooting for but they could not do it.

Sigh. Well, I don't want to create bad feelings. So please take this as an abstract observation. You're running a popular electronics forum and product review youtube channel. Companies like Tek & Agilent will make an effort to be sure you get the 'right ideas'. This is only natural. So anyway, a guy tells you the 500MHz one costs a lot more than the 200MHz one (the way they chose to make it.) Do you see the slight problem with equating this to proof that it has to be that way, and there couldn't be anything behind this state of affairs other than purely honest efforts to build the best, cheapest products they could? (From people who do things like software crippling of bandwidth, requiring a payment to unlock.)

So your assertion:

I'm saying specifically that bandwidths in the 500Mhz range are something the large instrument makers learned how to do cheaply long ago, even in digital scopes. In fact _especially_ in digital scopes since the high bandwidth stuff is confined to the input sampling circuits,

is clearly wrong on two levels.
One being the actual price of a 500MHz front end.
And second being that the high bandwidth stuff is confined to the sampling circuit, which as everyone knows, it is not. There is a lot of input circuitry before the ADC that is not easy to design at high bandwidths.

Bearing in mind that the _design_ problems were solved long ago, and what we're taking about here is solely the manufacturing cost. But anyway...
I don't consider the first point disproved just because some guy at Agilent tells you otherwise. I'm sure being given loads of lovely gear to review gives you a warm glow towards them, but it doesn't logically prove they don't fib.

The second - oops that was just my poor phrasing. I meant the input circuits up to the sampler. Recall previously I made it clear I'm aware it's the whole chain, from the BNC through to the sampler. (And of course for real time sampling, right through to the memory buffer.)
The one big mistake I made was not realizing that real time sampling capability up to rated bandwidth has worked it's way down to low end digital scopes since the last time I used any (not quite a decade ago.) I suppose I should have expected that.
However, I'd be happy with scanned sampling up to 500MHz for a cheap price, since how many applications actually need real time sampling? Not that many.  I don't personally have any such need, and where I might need to see complex single shot sequences it's always going to require a logic or protocol analyzer not a scope.

Still moot I guess. There's no chance I'm ever going to be buying a recent model digital scope, especially given my other equipment buying priorities. Specific objective, struggling to acquire required gear in any form I can. Recent purchase - a gamma spectrometer, $1200. Next one, hopefully, something with as many GHz of fiber optic detector bandwidth as I can afford, and ideally spectral analysis too. That's going to be very ouchy, and won't happen for a while.

Just curious - that video of deemed-doomed scopes you mentioned, was that at some company we're talking about here? If so, you should extrapolate that mindset through other elements of the corporate philosophy, and see where it leads you. Not nice people.

Years ago I used to worship HP & Tek. It was a painful disillusionment to realize they changed, and became no different from any other extortionist corporation.

[marmad]

And you're pretending that there's been any actual R&D needed in the last 10 years for any kind of scope below 500MHz.  Are you serious? What R&D? They don't even have to develop the processor designs any more, since they are likely Intel PC reference designs, just adapted a little to fit the form factor. It's ALL just industrial design, making the boxes look nice, changing around the probe interface connectors, improving the software for waveform display.

I'm not a huge fan of tech equipment manufacturers - and I certainly don't expect them to be truthful about everything they publicly say - but it seems you're a bit behind in DSO design in more ways than just real time sampling to full BW. You're forgetting one of the major ongoing drawbacks of DSOs: their blind time. This is a simple mathematical fact - and the impetus, I believe, behind much recent R&D from Agilent.  DSOs can ALWAYS use faster throughput - and until the associated blind time is equivalent to or better than the typical vertical blanking interval of an analog scope, more R&D will likely be required.

[free_electron]

-You do realize that digital scopes don't work like audio digitizers, and don't try to always sample sequentially along a single shot waveform, right?

-The sample points are scanned along the waveform relative to the trigger point over many cycle times. Yes, it's nice to get the  conversion cycle time as short as possible, since the shorter that can be made, the fewer cycles of the source waveform are required to build up a reliable picture of the waveform.

-Did you imagine that a (say) 20GHz digital scope has an 80GHz AtoD?

wake up buddy. it's 2012... the time of time-interpolated sampling scopes is 20 years behind us... Yes 20 GHz analog bandwidth scope these days DO sample at 40 or 60 GHz ... and that is single shot. they use pipelined ADC's to do this work. Those are essentially charge shifters where every step in the pipe digitizes one or 2 bits and then the remainder charge is shifted on to the next step in the pipe.

Even the cheape rigols do 500 megasamples/s single shot these days. analog devices has plenty of ADC that run at those speeds.

The era of 'equivalent time' sampling is dead. It died together with the tektronix TDS5xx and TDS7xx series scopes.

Agilent hasn't done equivalent or time-interpolated sampling since they launched the infiniium machines almost 18 years ago. The ADc in those machines do run at full throttle. I have an old infiniium that has four a/d each clocked at 4 GHz. no problemo. Wanna do faster ? they interleave two adc to get 8 Gs/s.

And those machines , as well as the current models, are not simply a sampling head with a pc to do the post processing. in an Agilent scope the pc actually does almost no post processing... it's all done in hardware using massive ASIC's that can handle the raw data throughput.  That's why they tell you how many waveforms a second they can capture. They tell you what is their blind time. And it s pretty darn short.

[w2aew]

To be honest, at the high end of the performance range, both types of instruments exist. 

Very high bandwidth real-time digital scopes do indeed include extremely high speed ADCs.  Different manufacturers differ in their implementations.  Tek uses a single high speed S/H (sample and hold) that feeds a few muxed ADCs running at several GHz.  Agilent fans the signal out to hundreds of muxed S/H+ADCs.  Both schemes give you 100GS/s or more effective single-shot sample rates.

There are ALSO modern equivalent time sampling scope being produced today too.  They often sample at a low rate, like 100 to 200kHz or so.  The scope does a very precise job of knowing when each sample occurs with respect to the trigger.  This type of scope is for repetitive signals, or those that are synchronous with some repetitive trigger event.  Advantages of this type of instrument are that they are very accurate, since the samplers run slower, they are usually 14-16 bit converters vs. the 8-bit converters used in real time scopes.  Also, these equivalent time sampling scopes can have analog bandwidths of 50, 60, 70GHz or more.

Both techniques are still in use, each with their own applications.