Hi. We got a follow-up video on the Jim Williams pulse generator and if you remember last time, we tried to measure the performance of Silvan's little Uh board here based on the Jim Williams circuit that he sent me for the mailbag. and well, the $10,000 1 GHz Agilant 3000x series just wasn't quite up to the task of measuring the rise time of this thing. So I thought we need to maybe Step It Up by say an order of magnitude.

So unfortunately out it goes Oh well. Turns out that I had something lying around the lab here Gathering dust and uh, we can step it up by an order of magnitude. Here it is Agilant Infinium 13 GHz 40 gig sample per second DSA 91304 a isoscope worth approximately Oh, about $40,000 Australian No worries I Think we might be able to finally measure the rise time of this thing. Let's give it a try.

Now if we take a look at the horizontal down here, it uh, goes all the way down to that's we're in the Picoc second region Now 50 20, 10, 5 Pico seconds per division. that's Pico folks, not nano, but that's what you get with a 13 G gz 40 gig sample per second scope. So you might think oh, like we expect like 300 odd Peak a second Rise time. So you might think this thing would balls it in, just plug it in and measure it.

Turns out it ain't that easy. and by virtue of their incredibly high bandwidth, these Scopes are all 50 ohm input impedance which is great which is what we need, but Taada the Trap Check it out. Plusus: 5 Vols maximum Cat 1 you used to, you know, 300 volts uh maximum like on a regular uh scope but these things incredibly easy to blow the ass out of the front end. and as much as I'd like to smell what $140,000 worth of Magic Smoke smells like eh? I Don't think we're going to do it today.

So what we need is an attenuator because this little pulse gen outputs uh, you know, too higher voltage to measure Direct ly on this oscilloscope. So what we need here is a wide bandwidth attenuator and uh, thanks to Charles Trio Smart Cow I was able to get one. It's not something I ordinarily have in the lab and uh, it's a Mecca um, 65021 F4 and that's 20 DB and it's uh, it's got a 4 GHz uh rated bandwidth. It may go higher than that, but that's all it's actually uh rated to.

So really, even though we got this $140,000 13 GHz oscilloscope, we're limited by our attenuator here. Crazy. But anyway, this should be more than good enough to uh, measure the performance of uh, the Jim Willians pulse generator. and of course we had to fit various adapters cuz it's got an N connector on here so we had to fit n to BN C and then a a BNC uh sex adapter there and you know, crazy, but that's the shortest path I can possibly get directly from the BNC output of the pulse generator.

So let's give it a go. So there's our pulse generator hooked up through our attenuator and let's check out the signal TDA And if we have a look at the rise time here, we're talking about 254 Picos seconds. uh, average and a fall time of 420 PCS average. So that's uh, you know that that will be somehow limited, uh, by our A by our attenuator of course.

Um, because we're not essentially limited by the Uh bandwidth of the oscilloscope here. But there you go. That's a really nice clean pulse. Maybe if we turn some average on on, uh, we'll get a cleaner away for and there we go.

With 16 averages on there, we're getting about 253 PCS rise time and 420 PS full time on average. but it's quite a few people uh, pointed out in various uh comments that, um, because this waveform doesn't have time to settle up here, it is just a pulse. It it falls as quickly as it rises. It's uh, really, uh, no good for measuring the bandwidth.

You really need it to rise and then flatten off and then fall for. um, those formulas we used last time to be valid the uh, you know, .3 or 0.4 on the Uh bandwidth. So On The Rise time. So um, really? uh, we're going to have to modify the circuit and see if we can flatten that out a bit.

And also, as a few people pointed out, one way to do this is to solder on a length of coax across. uh, the main uh, two peaka farad capacitor I think it was actually? well, I replace that with a bit of coax. So I've just wired that in parallel I've got some Rg174 building uh 35 CM worth. Of course the longer it uh goes, the more you can, um, stretch that pulse out.

But anyway, let's have a look. what? 35 CM solded across the existing Uh couple of pea farad capacitor in there can do? Let's give it a try And bingo that's actually worked. A treat. Check that out.

It's uh, extended the pulse by. uh, what are we? 2 NS per division? So it's extended it to 4 NS that pulse and you can see the ring in up the top there and it eventually, uh, flattens out and then drops back down. So let's go in and measure the rise time of this baby. We still got our average in on in there and what are we get? What do we know about 288 P seconds rise time? Beautiful.

And of course we can zoom all the way in and go down to 5 Picos seconds. Uh, per division just because we can. But that's beautiful. So still got the averaging on there.

And uh, that is a little. That's a nice useful addition too. extend the uh length of that pulse. So let's get the official figure here.

We've got to have the full pulse on the screen. there is, so it knows the Uh level where at it measures 10 to 90% And 296 odd peer seconds. So let's call it uh, pretty close to an even 300 P seconds. Now if you're wondering what happens when I remove the capacitor and just have the coax there, look at this.

We don't get a nice. there's no more ringing at the top there, but we don't no longer get a nice stepped waveform. We sort of get this and then this little kink in here and then slowly rising up. so it looks to be better to at least have the uh, the old uh, the the original cap in there, the original uh Jim Williams circuit.

and of course this is the Uh Jim Williams recommended add-on I can't remember offhand if he uh, recommended leaving the original cap in there, whether or not he said to replace the capacitor or to with coax or to just add the coax in parallel. But yeah, that is not a uh nice. uh, you know, rise time input at all. It's um, you know it's pretty darn horrible.

and you know the rise time of course is going to be way out of whack. You know, 800 P seconds or 1.4 nond four times. Still, cuz that has all that down there. it's you know it's uh yeah, it's no good as a pulse generator for uh, measuring bandwidth and we'll step it up just a little bit by putting on what.

Well, the real says 100 m of coax. but I don't think there's 100 m still left on there. So let's call it. you know, maybe 75 M or something like that.

it's Phillips branded I I Googled that part number. Nothing comes up. Um, so you know it's once again, it's thin. uh, coax like the RG uh, 174.

But anyway, let's give this a try and see what we get. And here we go. This is with no uh, original capacitor in there. so it's just like the 75 M of coax or whatever.

We're talking 184 p a seconds according to this. but that increases if we get the whole pulse on there, the whole pulse, we're talking 50 NS per division 150 like we're talking 170 nond pulse or thereabouts now. and uh, but we get a a nice clean Edge there. but then it sort of.

We get a little little bit of ringing and then it starts. still starts to rise up there. So I really don't like that. and then we've almost got a linear fallback at the top of the waveform there.

So yeah, I think I might put the original cap back and see what we get. No, we get pretty much exactly the same with that linear ramp going down at the pulse. it's the same uh, pulse length of course and but we get this massive overshoot here at the first Edge the rising Edge But anyway, if we let the scope, uh, call that in terms of uh, rise time, we're talking 180 p a seconds there. So yeah, I don't know, you know you could, you could take it at any point you deem uh, you know to be uh, worthy of being the 90% point and as it turns out, I just killed my Two Pea Farad capacitor and the end cap just fell off.

Um so I've put in a five pea farad the nearest one I had to hand quickly and uh, it's slightly increased amplitude I believe but uh, there we go. We're back to our original Uh cap with no um, uh, coax on there, just the five Peak AER capacitor and we're talking 295 odd peak of seconds rise time and 598. oh say even p a seconds fall time and I've solded my coax uh back on the Uh 35 cm of coax cuz I really like that. That seems, uh, like an ideal thing to use.

bit of overshoot there, but we can get the rise time nicely and we are talking 270 Picos rise time and uh 1 nond uh fall time. So I'm going to use that as the reference I'm going to leave the circuit alone now, not going to play with it anymore. Um, and now we'll do some bandwidth measurements on our other Scopes Uh, based on that reference uh, rise time of 270 PCS So let's go back to our Uh Agilant 3000 Series the 3054 uh X Series Oscilloscope 500 MHz I'm using the exact same attenuator exactly the same as we uh, just did. Wanted to keep it all the same, didn't want to put it directly in there.

We'll do that next I Just wanted to see what we get, keeping the conditions the same. Let's check out the rise time and as you can see, the wave shape is I believe almost identical to what we'll getting on our $140,000 13 GHz Agilant. But of course our rise time is greater and it's basically uh, jumping between 610 and 630. It's mostly been about 630 P seconds there.

So I'm going to take that as the value and if we punch that into our calculator 0.4 / by uh 630 PCS what do we get? We get Tada 635 megahertz. So we were definitely overestimating uh before because we didn't Um, because we only had the pulse which didn't settle out, but now we've got the pulse which settles out there. The scope is able to take the average of the top, get the N get the true 90% and give us the rise time so 635 MHz But that doesn't take into account the uh absolute rise time of our unit which we uh saw was 270 Pak a second. so we haven't taken that into the calculation yet.

but even so, that value 635. If you go back and look at my Uh Marone Uh 2023 video which I'll uh, have to link in here. Um, we measured the bandwidth of this thing. the minus 3db bandwidth of this at guess what? It was actually 637.

mahz so you tell me, is that just a coincidence or is this actually good enough to you know measure to a reasonable degree of accuracy. The minus 3db bandwidth using that formula maximumly flat formula in quote marks of 0.4 On The Rise time it gave it gave us within a couple of megahertz the exact value because there's error in this of course. and there was a slight error in the Uh measurement of our Marone 2023 uh, minus 3db bandwidth as well. But you know it, it's pretty darn close.

but we haven't taken into account that 270 Peak a second. so I'm not sure if we have to. but hey, it is basically spot on and if we take a look at it here, this is directly connected straight into the scope. There it is.

so it's really, it's really bang on. As good as we can get it. No coax in, uh no, no coaxial connection, just direct B and C Connection in there and the amplitude is a little bit over it's off the screen. It'll give us 590 P seconds there when it's off.

But really, you need to get the whole thing on the screen for that to be an accurate measurement. and it is jumping between that 630 and that 610 Peak a seconds like we had before. So if anything, it's slightly lower and hence, um, overestimating our Um bandwidth again like we did, uh, last time. So um, in this case, we might have to take into account the Uh 270 PCS absolute rise time.

but still? I Think this probably demonstrates that you can, um, probably just measure the uh uh bandwidth of a 500 MHz scope using this basic Jim Williams pulse generator with the Coax Uh mod on there as well I Got 35 cm of coax with the original cap in there and that that works. That works! a treat and a few people have asked to see the trusty old Ryo to see what it gives. So there's the fabulous boot screen. All riall.

Now come Shi with this boot screen and we'll give it a go. Um, I'm using a 50 ohm uh termin 50 ohm in inline Terminator because of course the Ry gold doesn't have Uh 50ohm termination on it. so let's have a look. I've got the 75 M of Coax set up here and of course, we' got our downward slope on the waveform here.

So once again, where do you take the 90% uh point from I'm not sure where the Ry Go's taking it. it's telling us the rise time is less than 2.4 NS But really, you know it's not that accurate. Less than four depends where you want to take it. If we want to, um, say uh, zoom into the position here and take that as the value.

Um, then we're talking. you know, 2 nond. So what I'm going to do now is measure the uh real bandwidth, the actual minus 3db bandwidth of this uh scope using my Marone Uh 2023 RF signal generator. So what I've done here? I'm feeding in 1 MHz sine wave at 1vt Peak to Peak So we want to uh increase the input frequency um until it drops to 0.707 and that'll give us our real bandwidth.

So let's give it a go shall we? Okay let's wind the wick up and you can see that lovely effect of there is due to the averaging I got 16 averages turned on. we're at 50 mahz now so let's go up increase the time base so we want 0.70 or 7 170 m volts. so let's keep going. We at 123 MHz now it's 707.

Once again this is 50 ohm terminat of course and it's jumping around. It's hard but let's call that um at is that's let's call that as uh the minus 3db point and we're talking 136 mahz So that is our the real bandwidth of our Uh 100 MHz modified of course um 100 MHz uh Rial scope. Beautiful, That's our reference point. So clearly the two NCS that we're uh reading here for the rise time.

That's 200 mahz. that's way overestimating. So that's with our 75 M of coax. So let's uh, well, you know if we get four Nan seconds.

Once again, it's not very accurate down there. so you know there's jumps up to 2.4 Maybe you know it's not that great. And to get our figure of 136 MHz we need a smid and under uh to measure a smidge and under 3 NCS rise time. So it's clearly not going to give us that doesn't have the Uh resolution to do that I'm afraid.

So let's um, that's with the 75 M coax. Let's give the 35 CM coax a try. and here it is 35 cm of coax and if it looks familiar, it's because it's like the standard pulse Again, like the without the coax that we saw. And the reason for this is because this bandwidth of this scope is lower.

so it just you know, it hasn't got time for the thing to settle out. So um, we really need to add maybe just a little more coax to it. Uh, so we can do that. We're getting 2.48 NS So you know it's not the uh value of three that we require.

We can't really, uh, turn that up a tad cuz it can't measure the rise time there. But yeah, it's uh, 2.5 it's 2.6 Um, it's not going to cut the mustard. Need more coax? and I've doubled that to 70 cm of coax and as you can see, it does flatten out before it drops back down now. But uh, we're still getting.

You know, the resolution's not bad on the rise time uh measurement here. it's um, but we're still only getting just over 2 nond which is up near 200 mahz. so clearly overestimating once again. And how about 105 cm of coax? I Keep adding on these 30 35 CM uh segments and uh, nope.

No good. at least with the automated Uh rise time measurement just over 2 nond. Still, now what I'm going to do is manually set up my Cur cursor measurements here to the 10 and 90% Mark I've used a my Uh variable vertical uh attenuator to set it for five divisions Peak to Peak to that to the basically the top of the waveform. they're ignoring the overshoot there and then I've set my cursors to the 10% and 90% Mark And if we zoom in here, we can get let's have a look.

let's move that to the where it just crosses that vertical graticule division there and go across where 2 nond per Division and where it crosses that one. It looks like it's only one division, so it's still 2 NCS Pretty much exactly what it was measuring automatically, so it looks like we're entirely unsuccessful here of using this uh Jim Williams pulse generator even with the coax mods to extend the PSE to Uh accurately measure the Uh, or at least give us an reasonably accurate Park measurement of the minus 3db bandwidth. we're way overestimating here at like, you know, in the order of 200 MHz Um, instead of the 136 which we Uh measured is the true analog uh, minus 3db bandwidth and that's using the formula of 0.4 Of course, it may or may not, uh, actually apply to this Uh scope because it may not have a maximally flat response. Who knows, uh, what's going on here, but yeah, um, you know it relies on a whole host of factors.

You got to get the correct waveform uh type, with no over as little overshoot as possible, You got to extend the Uh pulse as much as possible. and uh, you've got to, You know, um, make sure it's a suitable pulse length for the bandwidth of the oscilloscope you're using, and then you've got to know what the correct factor is. Uh, what the that correct formula is based on, uh, your particular scope scope and what method it's uh implementing to do that. So yeah, it's it's not looking that great.

It seemed to work for the Um Agilant 3,500 mahz scope, but can't get this thing to work on the Ryo Really? I mean even taking our Uh Gan formula of 0.35 On The Rise time, we're still overestimating by quite, you know, tens of megahertz at minimum. Quite a significant error. So anyway, uh, more food for thought. We might, uh, do some, uh, more tests on this later.

But anyway, I uh, hope you enjoyed that. That's the Uh Jim Williams uh pulse generator with coax uh modification on it and it does certainly seems to extend the pulse, so if you got one of these, definitely give that extra Jim Williams mod with the coax a try and see what you can get. So even though it may not be fan, Fantastic for measuring the bandwidth of a scope with those Uh formulas. Unless you get the waveform absolutely, uh, perfect.

Um, it could still be used for, you know, measuring uh, the rise time of Uh systems and uh, things like that. the effect on the uh, the slew rate having just having a fast rise time pulse uh can be quite useful for uh, uh, quite a few different applications. so if you enjoyed the video, please give it a thumbs up. And if you wanted to discuss it, jump on over to the Eev blog.

Forum Catch you next time.