I have been doing a preliminary reading of the Lecroy MAUI interface manual, and they have a feature called Analog Compare which allows you to trigger on certain events occurring in the waveform
I get the impression that it can be setup to do peak detect type triggering. or am I not understanding something.
As nctnico says this isn't Peak Detect.
On a modern scope, the max(!) sample rate is usually more than high enough to avoid aliasing. However, the sample rate also depends on your available memory and your timebase setting, i.e. on the WS3000 which comes with 10Mpts per Ch, the memory is enough to capture up to 2.5ms at 4GSa/s. That means if the period you want to capture is longer than 2.5ms, the sample rate has to drop (i.e. to 2GSa/s for up to 5ms, or 1GSa/s for up to 10ms, and so on). With a drop in sample rate, the usable scope BW (which, according to Nyquist-Shannon, is < 2x f
sample) drops as well. For example, if you sample with 1GSa/s your usable BW will drop down to around 400MHz. Now if your scope is the 200MHz model then this might not be a problem as the real analog BW of the front end will very likely be lower than 400MHz, but if not then that means frequency components over 400Mhz will not be filtered out and get into the sampling system, which will cause aliasing (distortion of the waveform). Also, at lower sample rate, as other said you will generally miss very short events (i.e. the high frequency components) like glitches.
PD is a (one!) way around that. As nctnico says, PD means the scope samples at full sample rate but only takes the minimum and the max value over a sampling period that's appropriate for the long timebase setting. Because the sampling occurs at full sample rate, aliasing doesn't become a problem, and the acquisition system can still "see" high frequency components it would have missed in normal sampling mode with a much slower sample rate. PD does have its disadvantages, for example while it allows you to see high frequency events, because it only takes the max and min of a sampling period means the waveform on the screen might not be representative of the real signal. It will also look more noisy than it really is. Also, because PD is a sampling mode, the sample memory contains only the Extrema and not the real sample data, which means the data is pretty useless for any measurements (and you have to re-acquire in normal mode to get any measure-worthy data, which doesn't help in one-off events).
There's a very good paper on Peak Detect from HP:
http://www.hpl.hp.com/hpjournal/97apr/apr97a4.pdfIt's a bit dated (and scope BWs, sample rates, memory sizes and performance has progressed notably since then) but what it says remains still valid. PD was an important tool when scopes came with very low memory sizes (sometimes as low as 1000 points). Modern scopes which offer large memory have alleviated the problem to some extend, but if you want to look at high frequencies over a long period of time then even the large memories of modern scopes aren't enough to capture at sufficient sample rate.
Now, on a LeCroy scope, there is no PD. On a modern LeCroy X-Stream scope like the WS3000 you just use WaveScan and let it search for anomalies. WaveScan is a signal analysis tool where you can set certain criteria it should be looking for, and decide what it should do when it finds something (i.e. stop acquisition, take a screenshot, record it in an event tabel, perform specific measurements etc). You can run the scope at full sample rate and have WaveScan search for any anomaly in realtime (you can also use it to search post-acquisition if you want to). And because WaveScan maintains the full sample data and not just the Extrema you'll also maintain a representative waveform.
Keysight has a similar tool (although not as flexible) called
InfiniiScan for their Windows-based high end scopes (Infiniium Series).
The advantage of PD is that it requires very little processing power while WaveScan is pretty processing intensive, so it's simple to implement.
LeCroy stopped offering PD, after offering in I think only two scopes (9374 and 9384) in the late 90's when processing became fast enough to use alternatives like advanced digital triggers and math, mainly because PD is an acquisition mode which stores altered data (the sample memory only contains the extrema, not the real sample data). On a LeCroy scope, the sample memory always contains unaltered acquisition data) which is part of their design philosophy. Older LeCroy scopes (non-Windows/non-X-Stream) offered advanced triggering and math to capture issues, which on the older scopes sometimes was a bit complex to setup but when done properly worked really well.
As to WaveScan, there are some differences on its capabilities between LeCroy's mid-range (WaveSurfer, HDO Series) and high-end (WaveRunner, WavePro, WaveMaster, LabMaster) scopes., but even the variant in the WS3000 can already do quite a lot of pretty useful things (i.e. scanning for shifts in frequency, which in a repetitive signal is an indicator of the presence of some anomaly). It's a really versatile tool.
As I said, I haven't used PD in nearly two decades, and before that only on a few occasions, simply because there were always better (to me) alternatives available like triggers and math, even on HPAK scopes. And that includes stuff like waiting for a fast low d.c. pulse on a FPGA output (tggzzz's example), all without PD. It's probably down to what you're used to, i.e. if you used PD often back in the old days of low memory scopes then you might be more likely to still use it. But it's no longer the only option.