Looks pretty good. Shame it wasn't available last month when I bought a keysight s series. Looks like it is better in a few areas
I like that there are some really high bandwidth options well over 8 bits now and I very much like that there are some sample memories that measure in gigapoints now. It's a nice direction to be going, now maybe in 10 years one will fall into my pricerange
... After briefly reading the spec sheet.
It's got a lot of good and some that is "bad" (which I'll post since that's more interesting)
It's full BW/samplerate in interlaced mode, and full sample memory only appiles in said interlaced mode (so 2.5 for 4 channels vs 5 in 2 channel mode).
The digital mem is 125Mpoints vs the 2.5/5Gpoints of the analog channels.
And the "analysis memory" is 0.5Gpoints?
Pricing and Availability
The WavePro HD is available now with pricing of $31,000 for the 2.5-GHz 254HD, $39,500 for the 4-GHz 404HD, $57,000 for the 6-GHz 604HD, and $71,000 for the 8-GHz 804HD. All are available in mixed-signal versions.
Press Release: Chestnut Ridge, N.Y., May 8, 2018
Really good oscilloscope from the banner specs. 8Ghz and 12 bit are really top performance . Also noise and Enob seem good
Does LeCroy (or anyone else) offer FFT analysis for 1G+ points?
LeCroy can do FFt on 500M points, none else should be able to manage so much samples.
For more points in an FFT, can you just use persistence mode? If that's what you mean by points, at least. If you mean bins.... I don't know of a scope that offers that, but it would take an immense amount of calculation to do...
The 20GS/s sample rate seems slow for 8Ghz BW. Within Nyquist, but barely.
If I'm not mistaken, LeCroy's position on the matter is that 2.5 oversamples is ok, as long as your filters are sharp enough and you use a correct interpolation.
The Keysight Infiniium scopes do FFT on 1Gpts (If you have the pricey option installed to unlock that much memory). Tho it can take it quite a while to crunch the numbers on something like that. Especially on the models that still use the older slow CPUs
The Keysight Infiniium scopes do FFT on 1Gpts (If you have the pricey option installed to unlock that much memory). Tho it can take it quite a while to crunch the numbers on something like that. Especially on the models that still use the older slow CPUs
What sort of time are we talking about (for the modern scopes)?
I haven't tested it but i would assume the time is measured in minutes.
In general FFT on the infiniium scopes becomes a slideshow once you get past 10 Mpts. Its not hardware accelerated in any way, the older CPUs don't have great floating point performance and the software is not very multi threaded in most tasks.
I rarely find massive memory depths useful. As soon as you start putting any automated measurements or math on these long waveforms things slow down to a crawl. Here and there they might be useful for capturing long events in detail or with protocol decoders to just snap a big chunk of time and move trough the data at your own pace.
I rarely find massive memory depths useful. As soon as you start putting any automated measurements or math on these long waveforms things slow down to a crawl. Here and there they might be useful for capturing long events in detail or with protocol decoders to just snap a big chunk of time and move trough the data at your own pace.
I have to agree with this. Deep memory is nice for multiple segmented captures. But for just a single transaction, not that important. As an example, a 64 byte page write to an I2C EEPROM is about 70 bytes (a little less), each with 10 clocks, giving 700 clocks. Capturing 10 samples per clock requires a whopping 7k pts. With 20k pts there's lots of leeway for imperfect trigger delays, inter-byte clock delays, clock stretching here and there, etc. 20000 points! Typically when working I'll stay at 50k pts. That's really plenty, while with such a short record any half decent scope is snappy. The exception of course is FFTs, but even then capturing at 2x the max frequency for something basic like audio (at 50kSPS) getting even 512k pts takes ten seconds - just to capture the data. Against this backdrop it doesn't matter how fast the FFT itself is - 10ms, 100ms, 1000ms... doesn't matter since the update time is dominated by the acquisition. And this is even worse when looking at anti-imaging filters with cutoffs at e.g. 100Hz to say monitor a voltage level. 512k pts at 200SPS is hopeless. (This where my HP 3577A VNA really shines as it can get down this low to sweep a filter in 10-20 seconds at SNRs that are totally unrealistic with an FFT based approach. Never mind it will also do log sweeps! Not to mention also gives me a phase trace, which for some reason is uncommon in scopes.)
I haven't tested it but i would assume the time is measured in minutes.
I've seen an FFT on the MSO S take > 20 minutes. Their support didn't think this was a problem.
For low frequency and audio stuff i have a HP 89410A (DC-10MHz Vector Signal Analyzer). Its workings are a bit strange in that it works like a spectrum analyzer but even the mixing is done in the digital domain while it still does some FFT in the end. It lets you have a RBW of 0.03Hz and includes a general purpose signal generator that can be used as a tracking generator. The high dynamic range lets you even measure the low THD of audio stuff.
But regarding FFT in infiniium scopes its still pretty nice. The scope is smart enough to set up all of its acquisition settings from the desired FFT settings. So if you want you can simply fullscreen the FFT window and just change the Start Stop RBW parameters on your FFT. It can do wide spans on few KHz RBW pretty fast. Makes it feel like you are suddenly using a real spectrum analyzer, but as long as you keep in mind the limitations of a scope signal chain. It doesn't have a massive dynamic range and it has some spurs that come from the scopes internal noise.
On most other scopes i found the FFT pretty crap (Never had the chance to use a high end LeCroy tho).
I haven't tested it but i would assume the time is measured in minutes.
I've seen an FFT on the MSO S take > 20 minutes. Their support didn't think this was a problem.
I've seen (and think i owned) a real SA that takes sweep about that time for particular (manual) setting (lowest RBW, full span) anyway FFT is a FFT regardless of any device, its not a real sweep SA, if you have immense reading you'll know the difference (things like leakage)... fwiw..
Price? 
If you have to ask, then you cannot afford it.
Really good oscilloscope from the banner specs. 8Ghz and 12 bit are really top performance . Also noise and Enob seem good
I wonder what its 12 bit settling time is. Higher resolution non-sampling oscilloscopes usually fail in this regard and the EOB of less than 8 bits is not promising.
Embedded Windows 10? What fun!
Embedded Windows 10? What fun!
where the heck did you see Windows Embedded...
?
With an I7 6th Gen and 16GB of RAM (32GB as an option), that doesn't sound like a winning couple...
And some of the options are to my knowledge not compatible with embedded Windows versions.
Test equipment that has a PC inside often uses the Embeded version of Windows. This is not to be confused with Windows CE, Windows IoT or Windows RT. Basically the embedded version is a more "no bullshit" version of the normal desktop version of windows where they let the OEM strip away anything they don't need. The licensing also works differently as you can't buy this version of windows on your own (Likely involves a contract and buying it in bulk). So anything that is compatible with the regular Win 10 versions should be compatible with this (As long as the OEM has not stripped out some essential module).
They had the embedded version since Windows XP, but i would imagine in Windows 10 it brings more drastic changes due to all the extra crap in it. Pretty sure things like Cortana, forceful updates, forceful reboots, ads in the start menu, nagging to log you into microsofts services etc... are all removed in the embeded version. It might actually be the best version of Win 10 to run on your PC.
Id still rather have Windows 7, but hey maybe they did a good job of cleaning the nonsense out of Win 10. Microsoft is already starting to wind down support for 7 so i guess they had to go to 10 to avoid having to transition to the new OS later down the scopes lifecycle.
I just finished a design job that used Win10 embedded. That choice that was well outside my part of the project, but I was able to see how it worked from the inside. Overall, it is pretty good. The end user has no idea and it allowed us to put more effort into custom hardware compared to most other options. The software developers liked it and the end result was very good - no complaints from anyone.
Embedded Windows 10? What fun!
where the heck did you see Windows Embedded...?
With an I7 6th Gen and 16GB of RAM (32GB as an option), that doesn't sound like a winning couple...
And some of the options are to my knowledge not compatible with embedded Windows versions.
The datasheet as shown below.
I just finished a design job that used Win10 embedded. That choice that was well outside my part of the project, but I was able to see how it worked from the inside. Overall, it is pretty good. The end user has no idea and it allowed us to put more effort into custom hardware compared to most other options. The software developers liked it and the end result was very good - no complaints from anyone.
If you are good, then the complaints come later. If you are Tektronix, then the complaints come immediately.
still can't see the word 'embedded' in there...
And honestly, with Lecroy, that doesn't matter... ALL x86 Windows Embedded scopes can run a 'full' Windows as well... (at least until now), so I wouldn't care too much about that... and while the scope is 'new' as such, their PCIe interface card was very probably not changed since the HDO series (why should they ?), there's shouldn't be any driver issues...