Hi The Humble Oscilloscope. You're used to using it every day, and if you're doing, say, uh, troubleshooting a specific signal that you're trying to measure and seeing what's going on your circuit because it might be, uh, playing up, you're used to setting up your oscilloscope to capture a signal. You'll set your horizontal time base to what you want, You'll set your vertical time base to what you want, You'll set your trigger level to exactly what you want, and then you might single shot capture it. For example, you might do a run stop, and you might even set it to maximum memory depth here, for example, just in case you want to zoom in on the data and see extra detail.

And when you hit single like this, Bingo! You've captured your data fantastic. And you can zoom in, of course, to your heart's content because you've got a massive 200 meg of memory or whatever it is you got these days when I was a boy. Jeez, But one thing you can't do is zoom out. We're at five microseconds per division.

Take a look at this. Of course we can zoom in and we can see our data and everything so you can't see anything in there and you can't recapture. Let's just say, it's a one-off like event. You know, a rare trigger event you're trying to trigger on.

This isn't the best example. You've only got this one captured, but you want to see what happened off the screen over there or off the screen over here. You can't just move your horizontal like that because you've got no more data either side and put that back in the middle. And of course, you can't zoom out because there's no more capture data either side of that.

So it's capturing just within the window in there to whatever memory depth you currently set and you go well. Yeah, okay, fair enough. That's just normal basic scope operation. But what if I said not all scopes operate this way.

So let's go over this key side over here. Exactly the same signal, exactly the same horizontal vertical. I've fixed the memory depth. Yes, you can actually do this in a later start firmware upgrade.

You can actually fix the memory depth in digitizer mode on the scope so, but it doesn't matter whether it's in auto or not. I'm just you know, having the same thing. and if we single shot capture that, it's exactly the same, right? Five microseconds per division, we can zoom in on our data to see that, but we still can't see anything weird going on here. Once again, it'd be nice if we could actually zoom out on this capture data and see something.

Well, check it out. We can do this on the keysight and look at this. Looky. what we have here.

We actually see something. There's a little glitchy in there, isn't there. There you go. a little runt, pulse aha gotcha.

But if you didn't know, hmm, you might have never found this unless you have to recapture again. So we're at five microseconds per division that we captured it at, but our keysight allows us to go out 10 microseconds 20, 50, 100 and we will at 200 microseconds. We will eventually get to a point where, oh, it's at its limit. Hmm.

so why does the keysight allow us to zoom out and the cigarette doesn't? Well, before we try and answer that question, let's just take a look at how the keysight is operating here. Okay, we're at the five microseconds uh, per division that we were before. and if you run it, yeah, we can see the runt pole, so I know this isn't the best example. but you know, just run with me.

Let's assume you had some obscure, you know thing that you're trying to find and you only had one capture then being able to, actually, um, stop this thing and zoom out like that on the capture data could actually be handy. But anyway, let's zoom out and see how far it lets us go. Aha, it only lets us go to 160 microseconds per division. It, it actually goes really oddball.

Don't worry about the demo mode there, I'm just using the demo mode to actually generate this run pulse signal, but only in digitizer mode. Okay, with the fixed four Meg points, it knows that it can't actually display any more information outside of the screen, so it's actually not going to let us go any further than that. It's smart enough, intelligent enough to know that well. I'm not going to show you blank space either side.

What do you want to do that for? But it will do that if you take off digitizer mode and it's just doing auto memory. That's five microseconds per division and we stop it. Okay, it'll go out and bingo. It'll start showing us the black either side, so that's just an interesting difference.

The Keysight scope operates differently in digitizer mode than it does in auto memory depth mode, and you'll also notice there is actually a difference in the amount of data it captures it in run stop mode as opposed to single mode. Once again, we're at five microseconds per division. Nothing peculiarly about five microseconds. I just happen to be using that if we actually press run stop.

If we're at uh, in stop mode at 50 microseconds per division, Look, it starts showing the black so we're only capturing 50 100, 150, 200, 250, 300, 350, 400 microseconds worth of data. 400 microseconds before of data do the exact same thing, but press single so same time based settings, same everything. We're in auto memory mode, so zoom out like this and a hundred microseconds per division. So it's gone from 400 to 100, 200, 300, 400, 500, 600, 700, 800.

It's doubled the amount of data that we've captured in single shot mode than it does in run mode. So this is actually uh particular to the keysight's scopes and the Mega Zoom 4 Asic and the capture architecture that they're using inside this scope. Because all scopes are designed differently, they implement capture algorithms differently the way they use memory and or things like that. So scopes do vary in their operation.

So should you always use single shot capture mode instead of the run stop button? If you want to freeze something on the display just in case you just wander into a random lab, start using a scope and you don't know I've never used it before. You have no idea whether or not it has extra memory In single shot capture mode, which one should you use? Well, it's not as simple as that. Run stop mode actually exists for a reason. If you've got a very slow time base like we're in 200 milliseconds per division, here, you can see it right.

A really slow updating waveform. If you actually press, stop Here we go. If you see something and then you just oh oh, saw something and you press stop, then it actually just freezes it exactly in the middle of the sweep. Um, sweep.

as in, you know, old school analog scope, but you know, digitizing sweep. It stops it right then and there and displays the existing memory that it already had in the buffer. But a single shot capture mode? Like if where or part of the way through, you have to sit there and wait and it'll give you a full acquisition like that. So the single will wait for the next trigger point and then give you an entire capture in memory.

So there are different modes and they exist for a reason. Each one has advantage depending on what you're actually trying to do. That's why every scope on the market will have a run stop mode and a single button. Now, Keysight actually call this, uh, the ping pong memory buffer.

That's how they implement the memory inside their mega zoom for Asic, and I don't have particular details exactly how they do that. So why did it actually run out of puff at 50 microseconds per division there? Well, this has to do of course with the amount of memory that we actually have and our sample rate and our particular time base that we happen to be capturing now. and you'll notice that we're a sampling at five gig samples per second and we're in auto memory depth so we don't. The Keysight, uh, Asic, is quite smart in it'll use the maximum amount of memory of its full Meg sample memory maximum that it can based on whatever mode and time base and all.

You know everything else you're in. So let's actually wind this time base back a bit. 10 microseconds, 20 Watch this gig samples per second, 50. Aha.

50 microseconds per division? We've dropped. Well, bloody touch screens. we've dropped down to 2.5 gig samples per second. And at 100 1.25 gig, you'll notice that it's dropping because this thing's only got four meg of memory.

So it has to drop that sample rate because it can't get the maximum five gig sample per second sample rate at one millisecond per division. It just does not have the memory to do it. If this thing had 400 mega memory, yeah, we could actually probably still go to 100 millis one millisecond per division and still get five gig samples per second. So let's go to Dave Calc.

here. at five microseconds per division, we get five gig samples per second and we want to work out how much time between each sample. So you just invert your sample rate. Five gig samples per second.

That gives you 200 puff seconds, 200 picoseconds per sample. And if we multiply 200 picoseconds by our four meg, our maximum four meg memory depth of this scope that gives us 800 microseconds is that sounding familiar? And also 200 picoseconds times 2 meg of sample memory is 400 microseconds. Bingo. That's exactly how many divisions we were getting in this case in stop mode here.

if we go out at 50 microseconds per division, we had 400 microseconds. and as before, in single shot mode, when we go out, we've got eight divisions there times 100. That's our 800 microseconds. So that's why we can't get any data outside of there Because we've only got four sample, four meg sample memory, and this scope just happens to have two different modes.

Uh, where you have run stop and single have different memory depths in auto mode. But Dave, why does the Keysight scope have different memory depths in run stop and single mode? Well, this has to do with as I said, this ping pong buffer. which basically they split the memory in two. They've got two meg up here and two meg down here.

This is how they get their fast update rate. They, uh, fill up, Uh, the top two meg of memory. and then while they're filling up the bottom two meg, they're actually reading out the top two meg and displaying that on the screen. That's why Keysight is one of the fastest, you know.

Million waveform updates per second, blah blah blah. A lot of other scopes choose not to do that, so they're in inherently slower updating. There's other architectural things behind the fast updating as well, and the software is smart enough to know that well. if you, if you're choosing, you're the user.

If the user chooses to go single shot, then obviously they've set up all their triggering and their time base and everything's all set up nicely. And they're going to press single. They know exactly user knows exactly what they're doing. So aha, I don't need the ping pong memory anymore.

I'm just going to give all the memories so you'll get in this case, the trigger points in the middle of course, and you'll get 50 percent pre and 50 post trigger data. so it'll switch off, disable the ping pong memory and use the entire length. It's actually quite clever, so even on the key site that has this ability, it depends on the horizontal time base and the four meg of memory we've got in here. If we're on 10 milliseconds per division, we can stop that, but then we can't do anything like it is just limited to the screen length.

Because all of the four meg is being used on the screen, there's nothing left to go outside. Even if we go down right down to one nanosecond per division. As fast as this scope will go and we press stop, we get all that data. We can actually go all the way out.

10 20. Ah, we have to get to 50. uh, microseconds per division before we start coming to Guts Are There because once again, it's the five gig samples per second. But another scope like this signal just won't do it because it uses a different architecture and it's actually plainly visible on the screen.

Look, Even if we set this to 200 meg points right, we've got a huge, massive amount of memory here. Have a look down the bottom on the time base 20k points even though it's set to the maximum. You know this is not auto memory depth mode. This is manually set to 200 meg.

The scope simply will not use that 200 meg. It's only going to use 20 k points. So of course we're only going to get 20 000 points on that visible area if we zoom out. You're not going to get any extra data.

You've only got the 20 k points in there, even though we have 200 meg of memory. and this one can go right down to 500 per seconds per division, right? And but we only get 10 points down there. 10 not 10k points. Only 10 points whatsoever.

Even though we're in 200 meg memory depth, to get our 200 meg points, we have to go all the way up to 10 milliseconds per division there. But of course, if we stop that, it's only in the visible window. So why does Siegeland and other scope manufacturers do this? and Keysight do it differently? Well, your guess is as good as mine. Really, they just implemented their architecture differently.

I guess this hasn't gotten asic in it. It's got Fpga There's no reason why they couldn't even like change this in firmware or update the Fpga hardware or or something like that. because Fpga is a reconfigurable fill programmable gate array. There's no reason why they couldn't change it to actually do the same thing because you've got 200 meg of memory depth.

Sure, you're getting the fast updating, but Keysight can do it. They got the best of both worlds. They can have the fast updating and also have capture the data outside of the screen area. But yeah, only uh, the likes of Sigil and other manufacturers would have to get back to us and say no.

The reason we don't do this is because X and I'm sure there's uh, some reason. Or if they say oh yeah okay. that's kind of sort of useful because this is not a feature that you would actually rely upon all the time. It's nice if your particular scope has it, of course, but how you regularly use a scope is based on what's on the screen.

As I said, you set up your time, base your vertical, you set up your trigger level, trigger pattern, whatever you want to do, you set it up, and then you capture your data. And usually in most cases, if you want to see what's outside of that screen, then you recapture at a, uh, longer time basis. No big deal, That's how people have been using scopes for generations, so you know it like. this is just a little interesting tidbit.

So it's not a show stopper. like you wouldn't not buy a scope because it doesn't have this feature. Anyway, there's much about this. Uh, there's much debate about this over on the Eev blog forum, which is course test equipment central on the Internet.

There's just lots of test equipment nerds over there. I'm getting into the intricacies of all this and it's quite fascinating. but Keysight aren't the only ones that do it this way. Tektronix do it as well.

I don't know about in all models somebody on the forum said that they didn't but you know the same signal. run signal again. four microseconds because they do one two four sequence instead of one two five. If we stop that, we can zoom out Ten microseconds, twenty forty.

100 200 microseconds per division. Bingo. And uh, we're running 10 Meg points here. And if we run the numbers here, our 2.5 gig samples per second, 400 pico times 10 meg is 4 milliseconds.

So in theory, we should get four milliseconds worth of data if they're not doing some ping pongy type thing. But we don't. Five divisions times four hundred microseconds? That's two milliseconds instead of four milliseconds. So Tektronix must be doing the same ping pony type thing for the acquisition.

But unfortunately, their update rate is just atrocious. So um, yeah, they're famous for it. So if we go all the way down to 400 picoseconds per division and we press stop, you should be able to go right up. Yep, look at that.

Use our full 10 Meg points till we eventually get to our Black and Rhoden Schwartz. Use the Schwartz Luke. Uh, 2.5 gig samples per second, five microseconds, same signal, stop, change the horizontal time base. Nope.

Roden Schwartz don't do it either. We'll just do single shot capture on that and nope, but that's on auto Record length 150.15 k samples per second. Anyway, let's go to the full 20 meg samples per second. Stop: Go out if you fix it.

It does. So depends on what scope you've got. Some scopes won't do it on auto mode. The keysight does it on auto mode, but ones like the Roden Schwartz this is looks like can be an advantage to set in fixed memory size.

Does Gw instead do it? Let's go. Yep, it does. Look at that. Nice.

We're at 2 gig sample per second there. And where's the memory depth? Is that displayed there? I think it is. That's bummer, but we go into the acquire menu there. There you go.

It worked. It does that. No worries. In auto memory depth, it doesn't have anything else.

but it doesn't. How about Rigol's new bad boy with their uh, custom Asic? Once again, exactly the same five microseconds per division. we're in auto memory depth here. Stop? Nope.

Doesn't do it in auto. Let's fix it. You can't actually change that unless you're running, which is pretty dumb. so we're in a hundred meg.

Now let's stop it. And yep, it'll do it once again. Fixed memory depth. so it's a pretty handy reason to fix that memory depth.

And of course, you might trade off update rate of course for the ability to have to be able to zoom out like this. I don't know what do you call it the zoom out feature I guess zoom out on stop or something. But anyway, yeah, that's a reason to set your memory depth fixed. How about unity? Four microseconds per division? They got the one two four sequence Again, where's their stop button? And nope, not on auto memory depth.

But hey, we can. We can change that. Let's go to the full 70. Function is disabled.

What can I have? Seven Meg Please. Function is disabled. Do I have to be running it? Yeah, you've got to be running it before you can do that. And you saw just then when we switched to Um 7 to 70, the update rate actually changed on this sucker.

Of course that's the trade-off But on. once again, the fixed memory. No workers whatsoever, and you can see all the little run pulses in there. Neat.

And O1 here exactly the same One gig sample per second, one meg memory depth, stop and no problem whatsoever. And O1 doesn't seem to have an auto memory depth that's just fixed. Only now. I've also heard that Lacroix don't do this as well.

It works exactly the same as a cigarette. Probably not surprising considering that Siglin actually manufacture low end scopes for uh, Lacroix. So maybe there's some, you know, shared uh, technology there or something like that. But even in fixed memory depth, it doesn't actually have an auto mode.

You stop it and you can. You do not get any information outside the screen, so it does operate quite significantly different in fixed memory capture mode to probably the majority of other scopes out there. so it's really interesting. anyway.

I hope you found that video useful. If you did, please give it a big thumbs up. And as always, discuss down below, I'll actually link over to the existing Eev blog forum thread on this one, because that's where this particular feature, um, feature in quote marks is being discussed with much excitement, by the way. So anyway, yeah, it's just an interesting little aspect of using your scope.

so you really should get to know your scope in terms of, uh, you know what it's capable of and whether or not it does this in auto mode? What What differences between auto mode and manual memory mode and things like that, And it might just save your bacon one day knowing that you can just, uh, well, in this case, you can't But knowing that you can just zoom out? As I said in probably the majority of usage cases, it doesn't matter. You just uh, re re-trigger again at a slower time base, You know, you just go over here and you just go, oh well, stop it and then zoom in. I've got 200 mega memory, no worries, right? But you know there? It might occasionally be the case where you're working on some obscure glitch or something. You finally capture it like you see something in here and you go, oh geez, I wonder what happened out outside the screen here.

And if you can't see it and you can't readily recapture it at a slower time base, then yeah, that could potentially be a useful feature. So leave it in comments down below: Does your scope actually have this feature or not in auto mode or fixed memory mode? or both? It doesn't or your scope doesn't have auto memory mode. It's only got manual or vice versa, and it may not be specific to Uh manufacturers. There could very well be different models from the same manufacturer that operate differently, especially like older models, newer eye capture architecture, and all that sort of stuff.

So yeah, it's a little fascinating. uh, quirk of scopes. I anyway, I hope you learned something there and you found it useful. Catch you next time.