This seems to be a years old debate: Does the LogicPort have enough memory?
It does, for what it's designed to do.
It is hard to understand how on one hand "what kills the LogicPort is the short memory even with compression" but then hear on the other hand that the solution is to "capture a large chunk of data and look through it for a problem [rather than] having to setup exact triggering conditions which may or may not capture the problem". How do you capture a large chunk of data with a short memory? I'm confused , as is often the case
The way that you capture a large amount of data with a short memory is to store only the changes. I.e., if the current sample is identical to the previous sample, don't store a redundant copy. "But, but" I hear you asking. "How can that work?" It works by associating a counter with each of the sample memory slots. Instead of storing another copy of an identical sample, you just bump the count. That, along with the need to run at 500 MHz, is why the sample-size is constrained. You can't just "add more memory". They're already using specialized parts to achieve their current performance. And they don't come in bigger sizes (that's my impression, in any event).
Does this mean the problem is the memory length, or the difficulty of setting up the triggering (although I thought the triggers were generally agreed to be very good on the LogicPort), or maybe it's lack of familiarity with how the LA was intended to be used?
The triggers
are very good. But, a) they take time to set up, and b)
you have to know what you're looking for, in advance. When you've got 1M samples (or 6M, or 48M, etc.), you don't have to be selective. You can just capture a huge chunk, then browse it afterwards. As long as your post-processing functions are adequate (searching, markers, etc.), you can find all sorts of things without pre-configuring detailed triggers. In fact, with "error flagging" or "difference marking", you can designate a correct wave section and find all the variations without configuring
any complex triggers manually. And best of all, you can and will notice things (e.g. anomalies) you never knew were there, and
didn't even know you needed to set triggers for.
Net, net: it's hard to tell after reading the many posts regarding the LogicPort's memory size just how big of an issue this is, or how easy or hard it is to work around it, or how much memory would cure the problem...
Nico already explained it to you in a nutshell, that would be really hard to beat...
Its so much easier to just capture a large chunk of data and look through it for a problem then[sic] having to setup exact triggering conditions which may or may not capture the problem.
He nailed it.