NEW 100MHz Keysight Scope Release on 1st March 2017

[nctnico]

Literally the equivalent of pressing "single" after each trigger is all that would be needed.

That sounds exactly like "Normal" trigger mode (on my DS1054Z) to me. I'd be very surprised if other DSOs don't do it.

It is normal mode, but he is wanting a normal mode that uses the full memory length of the scope at the expense of the fast retriggering capability gained from double-buffering.

Again: most (all) other DSOs have double buffering without sacrificing half the memory depth so there is no issue needing single mode to get the full memory length. THAT is the big difference!

If you are going to double buffer, you have to have twice as much memory as is in a buffer - the clue is in the name.

Deep sigh... Don't you read/understand what I write? Other DSO makers just fit twice the amount of memory because they want to hold up to their claims regarding memory depth while also keeping the performance up. It is that simple because memory is cheap (especially compared to the price of a scope).

Actually Dave isn't far off with his 1/10th remark. If you enable 2 channels and 16 digital channels on a 4Mpts 3000T X series you'll probably have 1/8th of the memory depth for the analog channels (500kpts for analog and 500kpts for the digital channels). Unfortunately the datasheet isn't clear on how memory is divided it only says that  16 digital channels with the analog channels disabled gets you 2Mpts maximum (maximum=single shot mode and thus 1Mpts in continuous trigger mode). Assuming 1Mpts gets divided equally between analog and digital channels this leaves you with 500kpts. And we didn't even enable reference traces so maybe 1/10th of the memory depth of what it says on the badge is achievable.

[EEVblog]

Again: most (all) other DSOs have double buffering without sacrificing half the memory depth so there is no issue needing single mode to get the full memory length. THAT is the big difference!

You do realise you are arguing over semantics of how a specification is defined?
You are the only one complaining about this, so if if it's such a big deal to you then like I said before, post a big sticker over the front of the machine that says 2M instead of 4M and then you can sleep well at night.

[nctnico]

Again: most (all) other DSOs have double buffering without sacrificing half the memory depth so there is no issue needing single mode to get the full memory length. THAT is the big difference!

You do realise you are arguing over semantics of how a specification is defined?

That is one way of putting it but you can't deny that Keysight is very obscure about the memory depth in the various usage scenarios.

[pascal_sweden]

@nctnico: Do you remember that picture which you did send some time ago, where you suggested a solution to put a "pyjamas" with the picture of a dog, over the GW-Instek scope, for people who don't like the look&feeling of that oscilloscope? I was looking for that picture, and refer to it, as per Dave's comment about the sticker, to hide the specs. But unfortunately, I can not find that post from you anymore. Which is a real pity! Maybe you can help out here! Hope you recall the picture!

Maybe Wuerstchenhund can help out as well, as he was active in the same thread as well, if I recall correctly

[nfmax]

So would you be happier with a scope that delivered 1 megasample of memory, on all four channels, with reference waveforms on, digital channels on, double buffering on, segmentation on, and (say) 100 segments? Because that scope would have to have 3.2 gigasamples of memory installed. And even if you just wanted to make only a single-shot capture of a single channel, you would still get just that 1 megasample, and all the rest of that memory would have to be left idle, just in case you decided to turn the other stuff on. I think some people (though maybe not you) would complain about all this expensive memory that was unusable except in some contrived situation, when it could have been gainfully employed.

And remember that in the low-end Keysights (the Infinivision series) the acquisition memory is on the ASIC: it is not possible to increase it 'just by adding cheap RAM chips', only by doubling up the number of ASICs (3000A series and above)

[kcbrown]

Literally the equivalent of pressing "single" after each trigger is all that would be needed.

That sounds exactly like "Normal" trigger mode (on my DS1054Z) to me. I'd be very surprised if other DSOs don't do it.

It is normal mode, but he is wanting a normal mode that uses the full memory length of the scope at the expense of the fast retriggering capability gained from double-buffering.

Again: most (all) other DSOs have double buffering without sacrificing half the memory depth so there is no issue needing single mode to get the full memory length. THAT is the big difference!

Well, no.  They have double buffering without sacrificing the *advertised* memory depth.  But they sacrifice *actual* memory depth to achieve that.

There's another way to look at this:  all the other scopes fail to use the entirety of their memory in single shot mode -- they only use half of it, because they only record a single buffer's worth of samples when there's another buffer that they could use to augment the total.

Right?

[nctnico]

Literally the equivalent of pressing "single" after each trigger is all that would be needed.

That sounds exactly like "Normal" trigger mode (on my DS1054Z) to me. I'd be very surprised if other DSOs don't do it.

It is normal mode, but he is wanting a normal mode that uses the full memory length of the scope at the expense of the fast retriggering capability gained from double-buffering.

Again: most (all) other DSOs have double buffering without sacrificing half the memory depth so there is no issue needing single mode to get the full memory length. THAT is the big difference!

Well, no.  They have double buffering without sacrificing the *advertised* memory depth.  But they sacrifice *actual* memory depth to achieve that.

There's another way to look at this:  all the other scopes fail to use the entirety of their memory in single shot mode -- they only use half of it, because they only record a single buffer's worth of samples when there's another buffer that they could use to augment the total.

Right?

Not quite. Most other DSOs will use all the memory in segmented mode so the memory is used to it's fullest extend when needed the most. I guess my beef is that Keysight just isn't open about the amount of memory versus the operating mode and (worse) the fact the amount of memory available is actually way less then specified in many common usage scenarios. I wouldn't make such a problem out of it if Keysight had put a table in their datasheets which clearly indicates how the memory is used versus operating mode and what the limitations are. Now users/owners get a nasty surprise if they need the maximum memory depth long after they bought the scope.

[mikeselectricstuff]

Not to mention peak detect acquisition mode, which reduces memory by either 2 or 4 - don't recall what it is on the DSOX  but it was 4x on the old DSO6000

I'd agree that the x2 reduction in stated memory in the most commonly used acquisition modes isn't that obvious til you look into it, and the omission of a simple way for the user to select between maximum memory and maximum refresh rate is a pity.

[EEVblog]

Literally the equivalent of pressing "single" after each trigger is all that would be needed.

That sounds exactly like "Normal" trigger mode (on my DS1054Z) to me. I'd be very surprised if other DSOs don't do it.

It is normal mode, but he is wanting a normal mode that uses the full memory length of the scope at the expense of the fast retriggering capability gained from double-buffering.

Again: most (all) other DSOs have double buffering without sacrificing half the memory depth so there is no issue needing single mode to get the full memory length. THAT is the big difference!

Well, no.  They have double buffering without sacrificing the *advertised* memory depth.  But they sacrifice *actual* memory depth to achieve that.

There's another way to look at this:  all the other scopes fail to use the entirety of their memory in single shot mode -- they only use half of it, because they only record a single buffer's worth of samples when there's another buffer that they could use to augment the total.

Right?

Not quite. Most other DSOs will use all the memory in segmented mode so the memory is used to it's fullest extend when needed the most.

Most people for most situations do not use segmented memory mode.
Funny when the argument is turned on it's head you don't seem to have a problem with a scope using only half it's physical memory in the most commonly used mode (single shot), but when Keysight sacrife half the physical memory for another reason you think it's the end of the world?
 

I guess my beef is that Keysight just isn't open about the amount of memory versus the operating mode and (worse) the fact the amount of memory available is actually way less then specified in many common usage scenarios. I wouldn't make such a problem out of it if Keysight had put a table in their datasheets which clearly indicates how the memory is used versus operating mode and what the limitations are. Now users/owners get a nasty surprise if they need the maximum memory depth long after they bought the scope.

Is that not the same for most scope manufactures?
But I don't disagree, it could use a clear explanation in the manual in it's own section perhaps.
Actually, what I'd really like to see is the actual memory depth being used in the current mode displayed permanently on the screen like they do the sample rate. This is a very deliberate decision by Keysight to leave it out and it's always bugged me a bit. I guess their rationale is that people shouldn't have to think about memory depth, the scope "just works" and optimises everything. I don't think that's a bad argument. One of the things I like about the X-series is they "just work", and you don't have to fiddle with stuff like you do on many other scopes.

[EEVblog]

You do realise you are arguing over semantics of how a specification is defined?

That is one way of putting it but you can't deny that Keysight is very obscure about the memory depth in the various usage scenarios.
[/quote]

I don't deny that, see my last post, it's very deliberate on their part.
For something like the "hobby level" 1000 X-Series you could argue that's not a bad thing, it's one less thing beginners need to be concerned with.
I for one most times like not having to worry about the memory depth, it's one of the joys of using the X-Series.

[pascal_sweden]

They already were fuzzy about those things back in the HP54620C logic analyzer.
Nowhere in the datasheet or user manual can you find a table with the sample rate and memory size when all 16 digital channels are in use. Only way to find out, is try and check in the user interface.

[kcbrown]

Well, no.  They have double buffering without sacrificing the *advertised* memory depth.  But they sacrifice *actual* memory depth to achieve that.

There's another way to look at this:  all the other scopes fail to use the entirety of their memory in single shot mode -- they only use half of it, because they only record a single buffer's worth of samples when there's another buffer that they could use to augment the total.

Right?

Not quite. Most other DSOs will use all the memory in segmented mode so the memory is used to it's fullest extend when needed the most.

Well, I would expect it would use all the memory save for one segment's worth, so that it can maintain double buffering at the segment level.  But that's usually a fraction of the total memory so it's probably not worth mentioning.

I guess my beef is that Keysight just isn't open about the amount of memory versus the operating mode and (worse) the fact the amount of memory available is actually way less then specified in many common usage scenarios. I wouldn't make such a problem out of it if Keysight had put a table in their datasheets which clearly indicates how the memory is used versus operating mode and what the limitations are. Now users/owners get a nasty surprise if they need the maximum memory depth long after they bought the scope.

How many vendors provide that kind of information in the first place?  It sounds like the main problem is that Keysight advertises the actual amount of memory the scope provides, but doesn't go out of their way to say that the number of recorded samples will generally be half that when double buffering is active (which is not the case in single shot mode, and I wouldn't expect it to be the case in segmented memory mode either).

This also raises the question of whether or not Keysight's scope will use the full amount of available memory when in segmented memory mode.  One would expect it to since there would be no benefit to using half the memory for "double" buffering (one need only use a single segment's worth).

[memset]

They must be using the full 4Gsps capability of the ASIC. Nice.

Hope you're going to make detailed photos of the board for the teardown review

[JPortici]

Ditto. Also please dave, have an extensive look at the decoding/triggering menu to see which inputs can be sources for the various digital channels

[snoopy]

Hey Dave

Are you going to do a drop test on this scope like you did with the Tek MDO ?

cheers

[Someone]

Literally the equivalent of pressing "single" after each trigger is all that would be needed.

That sounds exactly like "Normal" trigger mode (on my DS1054Z) to me. I'd be very surprised if other DSOs don't do it.

It is normal mode, but he is wanting a normal mode that uses the full memory length of the scope at the expense of the fast retriggering capability gained from double-buffering.

Again: most (all) other DSOs have double buffering without sacrificing half the memory depth so there is no issue needing single mode to get the full memory length. THAT is the big difference!

Most other scopes require you to menu dive and drastically reduce the memory depth (or impair other functions) to improve realtime update rate, with the Keysight its all done automatically for you. Everything comes with plusses and minuses, and you prefer other scopes, we get it. Or we could take your inflammatory approach

OTHER MANUFACTURERS ONLY LET YOU USE HALF THE MEMORY DEPTH~!!!@@! THEY'RE CHEATING YOU OUT OF FUNCTIONALITY^^@!!@!!@

Literally the equivalent of pressing "single" after each trigger is all that would be needed.

That sounds exactly like "Normal" trigger mode (on my DS1054Z) to me. I'd be very surprised if other DSOs don't do it.

Normal: After this mode is selected, the oscilloscope enters the wait-for-trigger state and starts searching for trigger signals that meet the specified condition. If trigger signals that meets the specified condition are found, “T’D” is displayed in the running status bar and stable waveform is displayed. Otherwise, “WAIT” is displayed and the waveform of the last trigger is displayed.

Normal trigger simply waits and displays the last trace until a new trigger arrives, Auto trigger adds imaginary/speculative triggers to the acquisition system to keep the current signal on the screen whether its actually triggering or not. It doesn't have anything to do with the acquisition memory.

[nctnico]

Most other scopes require you to menu dive and drastically reduce the memory depth (or impair other functions) to improve realtime update rate, with the Keysight its all done automatically for you.

That is not the issue at all! The issue is that the actual memory depth in the most common usage scenarios is way less than what it says in the specifications from Keysight. The specification is in a grey area where it isn't a lie (from a legal standpoint) but it isn't the truth either. If Rigol or Siglent would pull a stunt like that the world would be too small but when it comes to Keysight a lot of people seem to see nothing but pink elephants and clear blue sky.  I start to feel like the boy in the story about the the emperor's new clothes https://en.wikipedia.org/wiki/The_Emperor's_New_Clothes

[Boson]

The specification is in a grey area where it isn't a lie (from a legal standpoint) but it isn't the truth either.

This industry standard for asian scope. Maybe Good Will Instek only exception? Asian scope for people who like put Tesla into autopilot and crawl to back seat. They never understand why slow & correct better, not fast & almost correct... or even very wrong. Possible it ok while people can choose and buy what like. Not ok when customer cannot choose and must crawl to back seat because no option. Hope Tek and other not give up and will offer front seat & manual control in future also to electronic top gun ace 

[saturation]

This might be something markedly innovative, the new Keysight 1000x design's performance/price.  They emphasize the words 'cheap oscilloscope'  and "Scrap the Toys, Get a Real Oscilloscope" quite a bit.



If the spec sheet is true, then the scope doesn't offer vastly more than competitors already in this niche [ except maybe the built in Bode plot and that their scopes live up to the spec sheet ], but if we have a true Keysight design  [ not an OEM rebadge or an slow and aging Tek TBS series ] aggressively priced at very cost conscious hobby-maker-entry level edu market, it will show that a major manufacturer recognizes the importance of this market not just for sales, but even as a loss leader, for brand recognition and as "taste test" units. 

IIRC, the original Agilent 1000 and Rigol 1052e entered the market at high $800 range, and are still in catalog for $520 and $330 respectively.   So given these clues it looks like Keysight could easily target $300 for the base model but it would be truly industry changing if in the $200, because then it would be position to take on the entire hobby market DSO as well, as far down as the likes of DSO nanos.



There isn't much to see, but looking at this one board, it reminds me of the highly integrated and modular GWInstek series using Zynq, so they can get performance and lower overall price and easily swap in out modules to get the other models made.

If it's $650 as per the Rupee document then I don't think this will in any way dissuade those to buy the 1054Z. Memory depth looks low compared to the Rigol to me...

And that would be $650 for the educational model (limited to 50 MHz without upgrade option, and probably limited to customers in the educational market). I would assume around $1000 for the 70 MHz "regular" model.

You will be pleasantly surprised

[nctnico]

The board is small but look what Tektronix made almost 20 years ago for their TDS200 series:

[Someone]

Literally the equivalent of pressing "single" after each trigger is all that would be needed.

That sounds exactly like "Normal" trigger mode (on my DS1054Z) to me. I'd be very surprised if other DSOs don't do it.

It is normal mode, but he is wanting a normal mode that uses the full memory length of the scope at the expense of the fast retriggering capability gained from double-buffering.

Again: most (all) other DSOs have double buffering without sacrificing half the memory depth so there is no issue needing single mode to get the full memory length. THAT is the big difference!

Most other scopes require you to menu dive and drastically reduce the memory depth (or impair other functions) to improve realtime update rate, with the Keysight its all done automatically for you. Everything comes with plusses and minuses, and you prefer other scopes, we get it.

That is not the issue at all! The issue is that the actual memory depth in the most common usage scenarios is way less than what it says in the specifications from Keysight. The specification is in a grey area where it isn't a lie (from a legal standpoint) but it isn't the truth either. If Rigol or Siglent would pull a stunt like that the world would be too small but when it comes to Keysight a lot of people seem to see nothing but pink elephants and clear blue sky.  I start to feel like the boy in the story about the the emperor's new clothes https://en.wikipedia.org/wiki/The_Emperor's_New_Clothes

These are counterpoints to your view, you don't like that the memory is less in many use cases but there are many products with many (headline or not) specifications that only apply in ideal or corner cases. All scope manufacturers are doing this in some way, its standard marketing.

You're happy to trade off realtime performance for increased memory depth, and there are products which allow you to do this. Its not a major failing of the Keysight products products that don't offer this control as it brings other benefits to trade off against which will be valued differently by different customers.

[EEVblog]

If the spec sheet is true, then the scope doesn't offer vastly more than competitors already in this niche [ except maybe the built in Bode plot and that their scopes live up to the spec sheet ], but if we have a true Keysight design  [ not an OEM rebadge or an slow and aging Tek TBS series ]

It is a Keysight design.

aggressively priced at very cost conscious hobby-maker-entry level edu market, it will show that a major manufacturer recognizes the importance of this market not just for sales, but even as a loss leader, for brand recognition and as "taste test" units. 

I was impressed by the price first time I heard it. Especially for a Keysight.
Just like in the previous X-Series you can bet they will allow themselves room to move on features and price further down the track.

[ebastler]

... So given these clues it looks like Keysight could easily target $300 for the base model but it would be truly industry changing if in the $200, because then it would be position to take on the entire hobby market DSO as well, as far down as the likes of DSO nanos.

 

Come on. How would they explain that pricing to their customers buying the high-end scopes?! They need to keep some consistency in their overall pricing structure, and I don't see them re-positioning Keysight as the cheapjack of oscilloscopes.

[scopeman]

Comments on memory (and math) and partial rant.

I do not know why all of the scopes released these days do not have a bare minimum of 2Mpts per channel of segmentable acquisition memory, quad trace (double memory for 2 and double again for 4) with a separate external trigger and also a trigger out jack (not just some barely useful mask pass/fail output).

My LeCroy 9354AL has all of these features and has had them since the 1990's. Not to mention independent cursor and zoom knobs.

True it does not have independent knobs for vertical gain and position (the two channel version does) but considering the math, trending and histogram capabilities it was light years ahead of the Tek and HP scopes at the time.

I still think if someone made a scope with those features (even 100MHz one) with the same zoom and cursor split screen (multiple grid as well) display method as the 93XX series they would have a really useful scope. Everyone seemed to copy the Tek window and cursor method which IMHO is a much poorer presentation. I really like the single movable up and down arrow cursors (in relative mode) that the 93XX used. Very clean display and you always know where you are. The zoom or math function window for a trace can be placed in a separate grid with the zoom area highlighted in the main trace just like the old analog scopes. Couldn't be simpler. On the quad trace machine you do have the distraction of switching the Vertical knobs from trace to trace with push buttons but at least you are not stuck with one knob to switch between cursors and zoom. or worse yet multiplex the vertical position controls with the cursor. What a PITA. It makes me wonder if the builders of scopes ever ask the people who actually are going to use their products before they make a blue million of them. Remember the first HP DSO's that were keypad entry only. And then the keypad and the single wheelie knob. They were awful.

When I drive a scope I want knobs and encoders are dirt cheap and I don't mind if my scope was a little wider to get them.

Yes there are times when I want a touch screen keypad for precise cursor location or zoom control but I use that as a supplement, not as the main way to drive the scope. If I wanted that I would buy a USB connected box like a PicoScope and drive it with Windows or Linux.  Yes I have one, and they are very convenient to toss in a briefcase but not my primary scope for sure!

Now mind you modern scopes have beautiful color LCD's and higher update rates and do a better job of interfacing with Windows (or run windows) but there is still something nice about powering up your scope and having it ready to work in 30 seconds after power on. My newer LeCroy's, although they are great instruments in their own right, need the time for Windows to boot to be useful. 

That can take a while. But that is the price you pay for the additional processing power. Maybe I should just leave the power on!

Now I will have to admit that I have not had the opportunity to drive any of the new Keysight (I almost typed Agilent there) scopes. Maybe I'll win one and do a review.

Sam
W3OHM

Independent LeCroy_Owners_Group Moderator on Yahoo! Groups

[snoopy]

The board is small but look what Tektronix made almost 20 years ago for their TDS200 series:

Amazing what they did with a Motorola 68000 processor. My Tektronix TDS784A only has a 68020 running at 16MHz but could capture over 350,000 waveforms per second !!

cheers