Let’s Talk About LeCroy Scopes, AKA… the “Wuerstchenhund Holds Court” Thread

[Wuerstchenhund]

LeCroy has feedback from just 40 buyers - so it must be quite new compared to e.g. Keysight's similar operation?

Yes, the ebay outlet is pretty new.

When will Tek join in too?

Tek already has 'Tektronixused', plus they were selling through another outfit before then. But seriously, who would buy Tek these days anyways unless those that are forced to? They pretty much have the most unattractive offerings of all big brands.

I wouldn't want to be a sales-rep trying to push new test-gear to customers at MSRP these days..

MSRP was always only something that only the lazy ones paid. Even the worst negotiator is usually able to get immediately some percentage off just by asking. The T&M business has always been about haggling 

[bson]

- passive probes PP024 are only 500MHz while scope is listed as 600MHz... oh well.

Active probes can be found at low, low prices.  I got an AP020 for $150, bought an AP022 factory tested from LeCroy for $145 (still waiting for it to arrive - there's one left at http://www.ebay.com/itm/Teledyne-LeCroy-AP022-Active-Probe-Kit-with-Performance-Report-/161686695014?hash=item25a5472c66:g:2dAAAOSwstxVPkIl ), and an AP034 differential probe.  The AP034 didn't come with any accessories and only handles +/-400mV at 1X, so I ended up locating a 10X attenuator (tip adapter) for it (which gives it a +/-8V range) that isn't here yet.  These ranges are top of the DC offset (up to 40V if memory serves).  The AP020 is a 1GHz 1.6pF single-ended probe, the AP022 is 2GHz 0.8pF.  The AP034 is nice in that it takes standard 2.54mm spaced square pins for tips, or they can be added to a board for TP's and the probe just "plugged in".  The single-ended probes are the more typical "pointed instrument" type.

The 500MHz 1:10 passive probes work great with my 53131A (opt 003, opt 010) frequency counter set to 10X attenuation.

[awallin]

any ideas on where to find list-prices on lecroy WaveRunner 8000-series scopes?

just got an e-mail they are selling ex-demo units at reduced prices...

[Wuerstchenhund]

any ideas on where to find list-prices on lecroy WaveRunner 8000-series scopes?

just got an e-mail they are selling ex-demo units at reduced prices...

I think RRP for the 500MHz variant starts at $14k up to $29k for the 4GHz variant. They don't show list prices for most of their gear but there's a quick quote function on the LeCroy website.

It seems they are selling ex-demo units for some 47% off which isn't bad for a scope that came out 5 months ago.

[AutomationGuy]

Hello LeCroy Experts,

I have a WaveSurfer 3024 and I like the scope becorse of its large functionality. There are some issues like the cursor control on a FFT. Its hard to zoom to a spike on the FFT. The general software stability increased latley. The current version 7.9.1.3 is the first stable version.
Now I found a Version 8.0.4.4 on LeCroys website and I downloaded it. Next day the version disappered. I am not so sure if I should install that version on my scope. Which version do you run on your Wavesurfer 3000 scopes?

Regards

AutomationGuy

[Wuerstchenhund]

I have a WaveSurfer 3024 and I like the scope becorse of its large functionality. There are some issues like the cursor control on a FFT. Its hard to zoom to a spike on the FFT.

Yes, cursors can be a bit fiddly. Some LeCroy scopes have a setting which allows you to change the acceleration profile for the encoders, but I can't remember if the WS3000 has it, too.

Rumors say that the WS3000 may get something like the SPECTRUM option on the larger scopes which would make FFT a bit easier to use, but that's not confirmed

The general software stability increased latley. The current version 7.9.1.3 is the first stable version.

Can't confirm that, we have now over 100 of these scopes out there in the field and they all have been pretty stable, aside from a few ones that catched a problem where it would sit in an indefinite boot cycle (but that was fixed a while ago).

If your scope has stability issues then it might well be a problem with your specific unit. I'd keep an eye on that.

Now I found a Version 8.0.4.4 on LeCroys website and I downloaded it. Next day the version disappered. I am not so sure if I should install that version on my scope. Which version do you run on your Wavesurfer 3000 scopes?

Pretty much a wide mix of versions including two scopes with some very early firmware (its users tend to spend  no time on firmware updates).

As to the version 8.0.4.4 you mention, I doubt that was for the WaveSurfer 3000, for which X-Stream Lite 8 is due to come out later this year, as far as I know.

The current version is 7.9.1.3.

[Keysight DanielBogdanoff]

This thread is intended to stimulate a discussion of LeCroy scopes...

It's aimed mostly against the Keysight DSOX3000T, which is a good scope, however there are a few points that speak for the WS3000 (taken from an earlier positing:

• The Wavesurfer comes with a larger screen with higher resolution (10.1" 1024x600 vs 8.4" 800x480 with the DSOX3kT)
• For scopes with bandwidths up to 1GHz I'd say in practical terms 4GSa/s and 10Mpts is probably much more useful than 5GSa/s and only 4Mpts memory
• FFT with the WaveSurfer is up to 1Mpts while the DSOX3kT uses only 64kpts which is pretty poor
• The WaveSurfer allows automatic and manual sample memory/sample rate management while the DSOX3kT is automatic only
• Unlike the DSOX3kT, which feature-wise is not that much better than the DSOX2k, the WaveSurfer 3000 has many features that can be found in LeCroy's high-end scopes, i.e. WaveScan and LabNotebook.
• The DSOX3kT has nothing comparable to WaveScan, which is a very versatile tool to find rare glitches and other issues and which works 'live' as well as on sampled data.
• The DSOX3kT also doesn't offer anything comparable to LabNotebook, which is a documentation tool and pretty neat if you have to document your measurements in some standardized format.
• The WaveSurfer 3000 uses the same probe interface (ProBus) all midrange and high-end scope from LeCroy use since the mid '90s, which means there's a wide range of suitable active probes out there, including a lot of second-hand ones which often sell for reasonable prices because they don't carry the "Tektronix" or "Agilent" label
• Integrated AWG: 25MHz 125MSa/s 14bit with 16kpts on the WaveSurfer, 20MHz 100MSa/s 10bit with 8kpts on the DSOX3kT (both not great, but still)
• LAN is standard on the WaveSurfer 3000 while it's a $400 option on the already expensive DSO3kT
• Not that important, but the WaveSurfer has four (2x front, 2x rear) USB host ports (Keysight two, one front one rear)
• Plus the WaveSurfer 3000 is noticably cheaper than the DSOX3kT

I don't want to derail the thread with a comparison shootout, but I feel obligated to chip in on a couple points for the Keysight scopes (bear with me)

• Keysight's capacitive touch screen vs WS3k resistive touch screen. Ask for a demo/loaner, you won't want a scope without it
• Serial decoding is done in hardware, so it's stinkin fast (also we have more supported protocols)
• Keysight FFT is hardware accelerated, can be signal gated, and has a peak search in the lister
• 3 year cal cycle vs 1 year cal cycle
• Keysight acquisition modes (normal, peak detect, average, high resolution) vs WS3k with "normal" mode only & "ERES" as a math channel
• Keysight DVM and hardware frequency counter & totalizer vs N/A
• Keysight 1 knob set per channel vs multiplexed channel knobs
• and of course waveform update rate and zone trigger... 

That's all, carry on.
 

[nctnico]

Hmm, how could I have missed the Wavesurfer 3000 has no peak detect 

[TAMHAN]

This is very odd now.  LeCroy already had Peak Detect in the 9354AM - see the video below:

[tautech]

I too find it very odd that a WS3000 does not have Peak Detect. 
One can only imagine that it's been left out of the incorporated features on purpose, if indeed it is missing. 
As a WS3000 is Siglent HW but apparently LeCroy couldn't trust Siglent to write the FW there only seems 2 explanations: Daniel's wrong or indeed LeCroy has left this basic feature out.

Both the Siglents that have evolved from about the same time as the WS3000 (Siglent SDS3000) do indeed have Peak Detect as do many of the Siglent range;

SDS2304X:

[David Hess]

The oldest DSO I am aware of with peak detection is the Tektronix 2230 which was first available in 1986 and implemented it with TTL using 74ALS574s octal D flip-flops for registers and 74LS684s 8-bit magnitude comparators.

Implementation of peak detection is not trivial since it has to occur at the full digitizer sample rate.  The 2232 which replaced the 2230 just 4 years later had 5 times the sample rate at 100 MS/s and implement peak detection in a custom ASIC which interfaced the digitizer to memory.  In theory discrete logic could still have been used but complex MSI functions like the 74LS684 were never made available in faster logic processes.  I hate to imagine how much power an ECL implementation would take but I am sure someone did it.

Peak detection became free or at least very inexpensive with programmable logic which was already used to either store the acquisition record in embedded SRAM or interface the digitizer to discrete memory but even so, low cost DSOs even now often lack this feature like many Rigol DSOs before the 1000Z series.  Of course if you are Tektronix, HP/Agilent/Keysight, LeCroy, or any other major OEM, you have been implementing this inside of an ASIC for a long time.

[JPortici]

haven't had to play with either of them but my two cents..

• Keysight's capacitive touch screen vs WS3k resistive touch screen. Ask for a demo/loaner, you won't want a scope without it

this actually depends on the quality/sensitivity of the touch screen. i'm sure we all remember older android phones and non-android samsungs (brr)

• Serial decoding is done in hardware, so it's stinkin fast (also we have more supported protocols)

that is why i always find myself leaning toward keysight. i'd gladly have that couple more  protocols.. but 4 Mpts memory.. just no.
i'd rather use a picoscope for the serial protocols other scopes lack, which is exactly what we do at work. no fancy scopes but a pico for our serial needs.

• Keysight acquisition modes (normal, peak detect, average, high resolution) vs WS3k with "normal" mode only & "ERES" as a math channel

i am sure that there is something simillar to peak detect, wavescan? not really the same thing but if you want to find aberration in a signal..

now. i judged on my experience and having no experience with the specific hardware on trial i'd like to ask to the judge, his honour, and defence lawyer: what's the deal with the loaner? can i small guy ask for a loaner to test for my home lab or is it only reserved for businesses? as i doubt we'll ever need new scopes unless our glorious tek breaks down (but i noticed a channel is probably in need of calibration. fingers crossed)  and even then unless we don't start doing more challenging hardware both of these will be overkill in bandwidth/sample rate.
but one of these will be in my lab. not today or tomorrow but soon.

oh, mr k: why did you go black a little while ago? is it because you want to have the coolest scopes in the market?   

[Wuerstchenhund]

I don't want to derail the thread with a comparison shootout,

So that's the commercials break then I guess?   

No seriously, you're welcome Daniel. 

but I feel obligated to chip in on a couple points for the Keysight scopes (bear with me)

Sure

• Keysight's capacitive touch screen vs WS3k resistive touch screen. Ask for a demo/loaner, you won't want a scope without it

I fully agree to try it on a loaner, because the difference is negligible. A capacitive touch screen is a big advantage on a tablet or a smartphone where modern operating systems use multitouch operations and gestures, however that isn't true for a scope where touch operation mostly consist of pointing at stuff and drawing a box.

At the moment, there's only one big brand scope which uses multi-touch and gestures, and that is the new LeCroy WaveRunner 8000. And this does have a capacitive touch screen.

Also, resistive touch screens have the advantage that they work well with gloves, which is a big advantage in environments where you're not supposed to touch the UUT with bare hands. There DSOX3kT's capacitive display is useless in these environments because it's capacitive screen doesn't work with gloves.

• Serial decoding is done in hardware, so it's stinkin fast (also we have more supported protocols)

You are right that the DSOX3k supports a larger number of protocols (the WS3000 does support the most widespread ones like UART/RS232, SPI, I2C, CAN, FlexRay while the DSOX3kT also supports I2S, MIL-1553, ARINC-429, LIN and SENT). Of course if you need one of these additional protocols and can't live with a cheap USB gizmo then the DSOX3000T is the only sensible option - and Keysight really charges a premium for that.

• Keysight FFT is hardware accelerated, can be signal gated, and has a peak search in the lister

First of all, the DSOX3000T's FFT only does a measly 64kpts while the WaveSurfer 3000 can process up to 1Mpts, that's 16 times the amount of data!

Also, you say "hardware accelerated", which usually means it's done through a dedicated ASIC. That isn't necessarily an advantage, though, because as we've seen especially in high-end scopes which through their high speed ADCs produce much more data than scopes like the DSOX3kT and WS3k, LeCroy's software-based X-Stream architecture handles large amounts of data a lot better than the "hardware accelerated" architectures from other manufacturers including Keysight.

• 3 year cal cycle vs 1 year cal cycle

I agree, it's an advantage if you need it. There's no technical reason why any modern scopes couldn't work on a 3yr cal cycle (they are all pretty stable these days) so I'd assume others including LeCroy will update to 3yrs as well, which means the days this is an advantage will very likely be numbered.

• Keysight acquisition modes (normal, peak detect, average, high resolution) vs WS3k with "normal" mode only & "ERES" as a math channel

As you say, on the DSOX (like pretty much any other scope) these are indeed acquisition modes while on LeCroy scopes they are math traces, but having them as acquisition mode is not an advantage, because it means that in any other mode than "normal" you lose all the actual acquisition data (they are destructive). On a LeCroy scope, where waveform-altering functionality is available as math trace, the original sampling data is retained. This has always been LeCroy's core design principle, and is the reason why LeCroy has been and still is the to-go brand for scientists when it comes to scope.

Also, the DSOX3kT, like pretty much any non-LeCroy scope (aside from some newer Siglents apparently, but god knows how good their implementation is), only offers the standard simple boxcar filter for its high resolution mode. ERES is a bit more sophisticated as it uses a linear phase FIR filter which avoids the various disadvantages of boxcar filtering (i.e. appearance of ringing) or the complete lack of controls over the filtering process.

Of course, things like ERES or using waveform-altering functionality as math traces while retaining the original sample data is much more processing and memory intensive, but the WS3000 seems to perform very well against simpler scopes like the DSOX3kT, despite having to process more data.

BTW, Averaging is available on the WS3k as well of course, again as a math trace so the original sample data is always retained.

Regarding Peak Detect (PD), you're right. The WS3k, like most LeCroy scopes, doesn't have Peak Detect acquisition mode.

This is very odd now.  LeCroy already had Peak Detect in the 9354AM

They pretty much dropped PD in 1998 (the 9384C was one of the few scopes which had PD, and also the last one).

I too find it very odd that a WS3000 does not have Peak Detect. 
One can only imagine that it's been left out of the incorporated features on purpose, if indeed it is missing.

It is on purpose. PD was a crutch to overcome the very small sample memories of older digital scopes, allowing to sample at full sample rate for an extended amount of time by storing only the minima and maxima of a sampled group and thereby extending the time length that can be acquired. As it is an acquisition mode, it is destructive (i.e. you lose the original sample data). You also lose timing information (you know in which sample period the data points were acquired, but you don't know where in that period, i.e. at the beginning or the end).

These days, scopes come with reasonably large sample memories, which means even in normal mode you can run the scope at full sample rate for longer timbases. Also, modern scopes tend to come with a much larger sample rate to BW ratio (the 750Mhz WS3074 samples at 4GSa/s, the 1Ghz DSOX3104T at 5GSa/s), which means there is lots of room for the sample rate to drop without losing any details. With its 10Mpts memory, even the 750MHz WS3 can aquire a 5ms period at sufficient sample rate (2GSa/s). And the lower the analog bandwidth the further the sample rate can be dropped without losing detail.

Modern mid-range and higher scopes also have sophisticated trigger and analysis tools so that the scope can capture the important event at high sample rate instead of having to capture a longer sequence via PD. Of course most entry-level scopes lack such functionality, and PD can then help there.

Bottom line is that PD is not offered because it's destructive and because LeCroy scopes like the WS3000 have suitable alternative ways (i.e. WaveScan) of capturing events at high time resolution.

I know that some people will probably disagree (I remember some discussions with nctnico about PD), but despite using mostly Keysight scopes at work which all have PD, I can't remember when the last time was when I used it (I vaguely remember one time in the early 2000's with some Tek scope). That of course doesn't mean much, and I'd be interested to hear about specific scenarios where people believe PD is still required - bearing in mind we're talking about a mid-range scope here, not some simple $400 Rigol box.

• Keysight DVM and hardware frequency counter & totalizer vs N/A

That's incorrect. The WS3000 does have hardware frequency counter, and it has the DVM (which was a free addition a few months after release, and a free upgrade for all existing scopes)

• Keysight 1 knob set per channel vs multiplexed channel knobs

Yes. If that's really an advantage it's up to your individual preference. My WavePro 7300A has individual controls, and I'd wish it was multiplexed because it allowed me to change channel settings without having to move my hand.

I'd rather have the space spent on a larger and higher resolution screen

• and of course waveform update rate and zone trigger... 

You're right with the waveform update rate of course (which is amazingly high for an entry-level scope). It's a nice marketing feature, but it's advantage in real-life is pretty limited, especially when compared to other scopes with decent trigger/analysis tools. Plus it comes at the price of very limited sample memory, which is a pretty big disadvantage.

But zone trigger, really? The WS3000 comes with WaveScan which can do zonal triggering and a lot more. The DSOX3kT doesn't even have anything similar, because that is only available for the Infiniium Series (InfiniiScan). And off course like pretty much everything at Keysight, it's a paid-for option . And having InfiniiScan on my work DSO91304A and WaveScan on my WavePro 7300A at home, I have to say that InfiniiScan can't even do half of what WaveScan can do. Plus WaveScan is free (comes with all LeCroy mid-range and high-end scopes, and when it came out it was a free upgrade for existing scopes). Go figure.

Since we're talking features, how about LabNotebook? It's a tool to create test reports and document testing directly on the scope. As with WaveScan, it comes standard with every LeCroy mid-range and high-end scope, and that includes the WS3000. What does the DSOX offer to make documenting test series easier? Exactly, pretty much nothing.


Lastly, lets have a quick look at pricing

The 100MHz 2ch DSOX3012T starts at $3500 while the 200Mhz 2ch WS3022 starts at $3.300 (it's price went up, actually, it used to start at $3k) - LeCroy gives you 2x the BW and 2.5x the memory for $300 less (or comparing similar BW, the 200MHz DSOX3022T is already at $4170 - that's $870 more which could be spent on probes and other tools instead)

Let's have a look at the 500MHz variants:
WS3054: $7500 vs Keysight DSOX3054T: $11704 - that's $4200 difference! You could even buy the 750Mhz WS3074 ($9200) and save $2500 which can buy you probes and other stuff.

Actually, the 500Mhz DSXO3054T is even more expensive than the 1Ghz WaveSurfer 10 which is $10k - double the BW, up to 16Mpts/Ch, and even more serial decode options than the DSOX3kT has for $1700 less.

Just for completeness, Keysight wants a whooping $14525 for the 1GHz DSOX3104T    Which puts it quite close to the WaveRunner 8000 - a sophisticated high end scope with 20/40GSa/s, up to 128Mpts/Ch, multi-touch, and large-capacitative touch display (the 500Mhz 4ch variant starts at $14k, the 1Ghz variant is probably around $16k).

I'm not sure that the separate vertical controls, PD, the excessive update rate and a few more serial decode options are worth the large price premium, even less so when it comes at the cost of tiny sample memory, smaller screen, and lack of other functionality. 

[Wuerstchenhund]

that is why i always find myself leaning toward keysight. i'd gladly have that couple more  protocols.. but 4 Mpts memory.. just no.

Well, if you look at the prices, at least if you're in for a higher BW scope (500Mhz and up) then same/less money gives you more protocols, more BW and more features from other brands.

Keysight is really squeezing it.

now. i judged on my experience and having no experience with the specific hardware on trial i'd like to ask to the judge, his honour, and defence lawyer: what's the deal with the loaner? can i small guy ask for a loaner to test for my home lab or is it only reserved for businesses?

In general it's available to anyone. If you're spending several grands on a scope then doing a test drive is not just adviseable, it's crucial IMHO. Of course some sales droids can get a bit iffy when a consumer calls, not a business, but if they are uncooperative just hang up and try again until you get a sales droid that is more flexible. At the end of the day, they have incentives to sell, and most don't care if it's for business or hobby if it's an easy sale.

Also, never forget to negotiate. No-one pays list price except the lazy. You can usually get options and probes thrown in as well.

oh, mr k

Who's Mr K?

[nctnico]

Even with long memory I use peak detect often at low sweep rates to make sure I don't miss a narrow pulse c.q. see a trace which has all the expected elements and nothing missing or malformed due to aliasing. It is true that timing information is lost but when looking at (for example) video signals it is nice to see the hsync and vsync pulses are all there. To me having no peak detect is a show stopper.

[Wuerstchenhund]

 

Even with long memory I use peak detect often at low sweep rates to make sure I don't miss a narrow pulse c.q. see a trace which has all the expected elements and nothing missing or malformed due to aliasing. It is true that timing information is lost but when looking at (for example) video signals it is nice to see the hsync and vsync pulses are all there. To me having no peak detect is a show stopper.

I understand. Well, on a LeCroy scope I would rather use WaveScan for that.

Your scenario is actually not too unsimilar to a pet project of mine, where one element relies on a set of (unevenly spaced) sync pulses. To find out if pulses are missing or out of spec I just throw WaveScan at it and let it run for a while, it then tells me any pulses were missing/out of spec, and if so presents me with a nice histogram showing when exactly that happened. If I wanted I could even set it up to do specific measurements on malformed pulses, or just let it do some screen shots everytime a deviation occurs, or do a range of other stuff.

It's pretty handy, and helped me to identify a problem where the sync generating unit producted malformed pulses in varying periods. It also helped me finding the source of a problem where the sync providing element occasionally threw out malformed pulses. With WaveScan and the statistics function I found out that the timing depended on the operating mode of that unit, i.e. power load, and that it was a flaw in the PSU which caused it.

Granted, on a entry-level scope which doesn't have any advanced functionality, PD is probably the best (only?) way to do that.

[JPortici]

Who's Mr K?

daniel, keysight.
a couple of months ago or so i saw new test equipment with black enclosures on keysight homepage, i remember it was mentioned even here at that time as a joke on them trying to look like  lecroy

https://www.eevblog.com/forum/testgear/opinion-keysight-new-design-theme/msg938112/#msg938112

don't need to take me seriously on that one

yes, what you say about loaner is true and what i fear is that they wouldn't care about loaning to privates, which is understandable to some degree.
it is also very stupid, on some degree. I already said i was a salesperson once and the worst thing you can do is to refuse the hint of a sale

[Wuerstchenhund]

Who's Mr K?

daniel, keysight.

Ah, I see

a couple of months ago or so i saw new test equipment with black enclosures on keysight homepage, i remember it was mentioned even here at that time as a joke on them trying to look like  lecroy

Yes, I remember. KS's new signal analyzers are in black, which is somewhat interesting.

It seems however scope will stay beige/white

yes, what you say about loaner is true and what i fear is that they wouldn't care about loaning to privates, which is understandable to some degree.
it is also very stupid, on some degree. I already said i was a salesperson once and the worst thing you can do is to refuse the hint of a sale

As I said it's only the occasional sales drone, most of them are happy to sell you a scope, or giving you a loaner.

[David Hess]

I too find it very odd that a WS3000 does not have Peak Detect. 
One can only imagine that it's been left out of the incorporated features on purpose, if indeed it is missing.

It is on purpose. PD was a crutch to overcome the very small sample memories of older digital scopes, allowing to sample at full sample rate for an extended amount of time by storing only the minima and maxima of a sampled group and thereby extending the time length that can be acquired. As it is an acquisition mode, it is destructive (i.e. you lose the original sample data). You also lose timing information (you know in which sample period the data points were acquired, but you don't know where in that period, i.e. at the beginning or the end).

...

I know that some people will probably disagree (I remember some discussions with nctnico about PD), but despite using mostly Keysight scopes at work which all have PD, I can't remember when the last time was when I used it (I vaguely remember one time in the early 2000's with some Tek scope). That of course doesn't mean much, and I'd be interested to hear about specific scenarios where people believe PD is still required - bearing in mind we're talking about a mid-range scope here, not some simple $400 Rigol box.

The connection between record length and peak detection reminds me of a recent post on the forum about the Rigol 1000Z linking record length and delayed sweep/acquisition.

The user in question was operating at a slow time/div but needed to magnify waveform details to such an extent at a point after the trigger point that the limited sample rate even with the Rigol's relatively long record length became a problem.  The question was whether the delay function could be used to acquire a full sample rate acquisition at the point of interest and I guess the answer was no; the Rigol's delay function *only* affects what is shown on the display and has nothing to do with the acquisition process despite what Rigol's documentation implies.  I managed to resist my inclination to respond that a gimpy Tektronix DSO with a 4K record length would have had no problem with this.

So both peak detection and delayed sweep/acquisition are crutches to overcome the limitations of a short record length.  I would also add DPO functionality to this list of crutches and I would rather have all three than a long record length if it means a faster update rate with lower blind time.

These days, scopes come with reasonably large sample memories, which means even in normal mode you can run the scope at full sample rate for longer timbases. Also, modern scopes tend to come with a much larger sample rate to BW ratio (the 750Mhz WS3074 samples at 4GSa/s, the 1Ghz DSOX3104T at 5GSa/s), which means there is lots of room for the sample rate to drop without losing any details. With its 10Mpts memory, even the 750MHz WS3 can aquire a 5ms period at sufficient sample rate (2GSa/s). And the lower the analog bandwidth the further the sample rate can be dropped without losing detail.

I am not sure if this is what you meant to say.  That modern DSOs typically come with a large maximum sample rate to bandwidth ratios makes no difference when the sample rate is limited by record length unless they have peak detection, delayed sweep/acquisition, or something similar going on between the digitizer and acquisition record.

[Wuerstchenhund]

The connection between record length and peak detection reminds me of a recent post on the forum about the Rigol 1000Z linking record length and delayed sweep/acquisition.

The user in question was operating at a slow time/div but needed to magnify waveform details to such an extent at a point after the trigger point that the limited sample rate even with the Rigol's relatively long record length became a problem.  The question was whether the delay function could be used to acquire a full sample rate acquisition at the point of interest and I guess the answer was no; the Rigol's delay function *only* affects what is shown on the display and has nothing to do with the acquisition process despite what Rigol's documentation implies.  I managed to resist my inclination to respond that a gimpy Tektronix DSO with a 4K record length would have had no problem with this.

Well, that is not surprising, considering that the Rigol DS1000z is a $400 scope with very limited functionality. I already said that for low end scopes PD is sometimes the only option. But we're not talking low-end scopes here.

So both peak detection and delayed sweep/acquisition are crutches to overcome the limitations of a short record length.

In essence, yes. PD may have some place in a low end scopes with limited trigger/analysis capabilities but again, this isn't what we're talking here.

I would also add DPO functionality to this list of crutches

Yes, it's a crutch, because Tektronix was simply unable to produce an architecture that delivers fast update rates in normal mode. DPO mode has several disadvantages, like no measurements, because like other acquisition modes its destructive (i.e. the original sampling data is gone).

DPO is one of many reason why pretty much no-one in its right mind buys Tektronix scopes these days, aside maybe from teh edu sector which now seems to be Teks main customers and which is full of people that live in their own world.

and I would rather have all three than a long record length if it means a faster update rate with lower blind time.

Great. But this tells me that you somehow missed the whole point I was making, which is that the lack of PD on the WS3000 is made up by other tools. Don't take this the wrong way but you sound like the type of person that if sat in front of a modern high end scope and asked to find and measure a glitch would resent to persistence mode and cursor readouts.

There's a reason why a modern mid-range or high-end scope has advanced toolsets, which is that you don't have to rely on crutches that pretty much only exist because of limitations in test gear 20 years ago.

These days, scopes come with reasonably large sample memories, which means even in normal mode you can run the scope at full sample rate for longer timbases. Also, modern scopes tend to come with a much larger sample rate to BW ratio (the 750Mhz WS3074 samples at 4GSa/s, the 1Ghz DSOX3104T at 5GSa/s), which means there is lots of room for the sample rate to drop without losing any details. With its 10Mpts memory, even the 750MHz WS3 can aquire a 5ms period at sufficient sample rate (2GSa/s). And the lower the analog bandwidth the further the sample rate can be dropped without losing detail.

I am not sure if this is what you meant to say.  That modern DSOs typically come with a large maximum sample rate to bandwidth ratios makes no difference when the sample rate is limited by record length unless they have peak detection, delayed sweep/acquisition, or something similar going on between the digitizer and acquisition record.

The point is that back then in 1996 a standard 100Mhz DSO like the HP 54622A came with 200MSa/s sample rate, while a modern day equivalent samples at 2Ghz or more. The large oversample ratio on modern scopes means it doesn't necessarily have to run at full sample rate to get all the details, meaning on a scope that lets you manually select the sample rate you can simply drop the sample rate to extend the acquisition period even more.

[nctnico]

Even with long memory I use peak detect often at low sweep rates to make sure I don't miss a narrow pulse c.q. see a trace which has all the expected elements and nothing missing or malformed due to aliasing. It is true that timing information is lost but when looking at (for example) video signals it is nice to see the hsync and vsync pulses are all there. To me having no peak detect is a show stopper.

I understand. Well, on a LeCroy scope I would rather use WaveScan for that.

Your scenario is actually not too unsimilar to a pet project of mine, where one element relies on a set of (unevenly spaced) sync pulses. To find out if pulses are missing or out of spec I just throw WaveScan at it and let it run for a while, it then tells me any pulses were missing/out of spec, and if so presents me with a nice histogram showing when exactly that happened. If I wanted I could even set it up to do specific measurements on malformed pulses, or just let it do some screen shots everytime a deviation occurs, or do a range of other stuff.caused it.

Granted, on a entry-level scope which doesn't have any advanced functionality, PD is probably the best (only?) way to do that.

You are missing my point slightly. For a first cursory look at a signal I'd like to see it's extremes at all timebase settings (even the slowest ones and roll mode) and for that peak detect is the only option. Sometimes I have to look at signals from systems which are slow but can have glitches. Even worse: I have no idea what to expect. So I set the scope to roll mode with peak detect on to get a feel for what a signal does (amplitude and if there are pulses at all) over a period of seconds to minutes. From there I can switch to triggering on glitches etc.

[Wuerstchenhund]

You are missing my point slightly. For a first cursory look at a signal I'd like to see it's extremes at all timebase settings (even the slowest ones and roll mode) and for that peak detect is the only option. Sometimes I have to look at signals from systems which are slow but can have glitches. Even worse: I have no idea what to expect. So I set the scope to roll mode with peak detect on to get a feel for what a signal does (amplitude and if there are pulses at all) over a period of seconds to minutes. From there I can switch to triggering on glitches etc.

I'm sorry and maybe I still miss your point but that sounds like a perfect scenario for WaveScan (which is *not* a trigger btw, it's more like a search tool/glitch finder, and it finds stuff that triggers won't). Instead of going through various time base settings I'd just enable WaveScan and let it search for deviations, and then just let it run for a while (5 seconds, 30s, a few minutes, ten days, whatever is appropriate). WaveScan will tell me exactly what went wrong at which point in time.

The problem with PD is that, while it can show small glitches on scopes that don't have enough memory to sustain a longer acquisition cycles at sufficient sample rates, it will show a somewhat malformed signal because of the loss of timing resolution. It also adds non-existing noise to the waveform, and overemphasizes infrequent amplitude deviations. It's OK for identifying glitches but pretty worthless to make a qualitative assessment of a signal. And PD won't show me anything that WaveScan can't find, without having me to stare at the screen to whole time waiting for an event to appear.

And while you used PD to find out that there are in fact glitches or dropouts in your signal and start to setup your triggers to catch them, WaveScan can already have done the work and provided me with screenshots of the glitches, or measurements.

One caveat though: WaveScan on the WaveSurfer Series is a simpler version of WaveScan on the high end scopes (WaveRunner, WavePro, WaveMaster, LabMaster) and lacks some of its capabilities. Not having a WS3000 myself (I only had one for a day) I can't exactly say what WaveScan on the WS3000 can and can't do. But I'm pretty sure, for your given scenario, it would easily find all dropouts and other problems with your signals.

[nctnico]

Wavescan sounds like a perfect tool for letting the scope find 'errors' in a signal. However every now and then I find myself in a situation where I hook up a scope to a system and I have no idea what to expect so the first thing I want is a slow recording (seconds per divisions) of some signals which may be interesting. Roll-mode or long time/div with peak detect are really crucial because that way I have both an overview of what happens on a long timescale and an indication if there is or isn't something happening on shorter timespans. From your description I don't see how wavescan can do the same but then again I have never seen Wavescan in action.

[AutomationGuy]

In WaveScan you can assign limits to any math function like rise time, peak width, duty cycle, RMS and so on ... on a trace and when the limits are -touched the trace will be stored. I am sure you can find any peak with WaveScan.
That way you can leave your scope running for days and review the stored anomalies next day. The high end scopes probably have  much more math functions.

Actualy I could find signals using LeCroy scopes in some cases where I couldn't find them using other scopes.

Here are some feature requests in case someone from LeCroy is reading this:
I would like to have WaveScan on FFT which would include math functions on a FFT trace.

Another nice feature on the WS3000 would be a SENT decoder.

A Hex number comparison trigger on the MSO would be nice as well. (Could be availble already. I was missing it in an earlier SW version)

One thing about the WS3000 which annoys me is the autosetup which probbaly breaks sampling for a few seconds each time its triggered. I could not yet find out when the autosetup is triggered. There is a button for autosetup on the scope but it has no use for me since the scope starts it automaticly.
The autosetup is realy helpfull and gives real good results but it interrupts sampling.

Direct replay from a trace to the build in function generator would be nice.

[tautech]

You are missing my point slightly. For a first cursory look at a signal I'd like to see it's extremes at all timebase settings (even the slowest ones and roll mode) and for that peak detect is the only option. Sometimes I have to look at signals from systems which are slow but can have glitches. Even worse: I have no idea what to expect. So I set the scope to roll mode with peak detect on to get a feel for what a signal does (amplitude and if there are pulses at all) over a period of seconds to minutes. From there I can switch to triggering on glitches etc.

I'm sorry and maybe I still miss your point but that sounds like a perfect scenario for WaveScan (which is *not* a trigger btw, it's more like a search tool/glitch finder, and it finds stuff that triggers won't). Instead of going through various time base settings I'd just enable WaveScan and let it search for deviations, and then just let it run for a while (5 seconds, 30s, a few minutes, ten days, whatever is appropriate). WaveScan will tell me exactly what went wrong at which point in time.

What features does Wavescan offer over a Mask test?
Sounds from what you describe it's much the same thing. 
Can it be used over non-repetitive waveforms or a continuous data stream?