Hi if you saw my previous video. I Got a package from uh Lei Hung and he sent me a passive uh probe uh, but I thought we'd uh, take the uh board he gave me in the little uh RF uh box a little shielder box and have a play around with an LC pie Fielder um why I don't know why not I just wanted to uh, basically show off um the performance of such a field and how to use a spectrum analyzer basically and with a tracking generator to get the response of such a filter like this. So um I'm on a website uh here? Uh, this is Calculator Edge.com and they've got an online filter for a uh low pass um LC Butterworth uh Pi filter and if you have a look here, you can see it's Piie shaped if you just ignore L2 C3 and all the rest of those through there. If you just look at C1 C C2 and L1 you can see it's actually a Pi shape like that and that's why it's called a Pi filter.

It's also called a capacitor input filter because it's got a capacitor on the input. Um, but yeah, I think it's more commonly known as a Pi filter due to the shape. So I thought well what if you wanted to say do a Um LC filter Pi filter like this. It's a single stage, only do a single stage one cuz the board that uh Le Chi gave me only supports a single stage on there.

So let's say you wanted to do say a 20 mahz uh filter. That would be a 3db down of course. cut off frequency of Uh filters is usually always measured at the minus 3db point. and let's do 50 ohms because our Spectrum analyzer tracking gen output and Spectrum analyzer input is all 50 ohms so well, that's the Uh impedance of the load.

So the number of components we want is three. So we've got a three component design here and let's calculate what values there are standard formulas for this. It's going to vary depending on your uh PCB uh, car parasitics and stuff like that, but we get a value of 795 Nano Henry's let's convert it to micro Henries, which you're probably more familiar with Um8 micro, Henri's and 159 paa farads or Puff as it's called in the trade. Um, now I don't have that exact value I've only got say a one micr Henry inductor in my standing kit, so let's call that say 15 MHz Let's try that again.

Recalculate it. there we go. Bang one micro Henry pretty close and 212 paa farads. Let's change that to a standard you know E12 type range.

You can get a 220 paa farad capacitor I'll have those in my kit. So um, let's build up a Pi LC low pass filter with 220 paa farad capacitors, both of them and a one micro Henry inductor and see what happens. We should be able to measure a minus 3db cut off point of round about 15 MHz give or take based on the component tolerances and some parasitics on the PCB as well. Let's give it a try.

So here's our little low pass pie filter here. We've got our one Micro Henry inductor there. We got a couple of Uh links here because there were extra components on the board there and we've got our Uh 220 Paa Farad ceramic cap there and there so you can see the Piie shape in that. This is the ground along here and this is the Um this is the input over sorry input over here and output over here.

Let's give it a go. So what we're going to use here is our Ry Gold DSA 815 9 khz to 1.5 GHz Spectrum analyzer with with tracking generator option and this is absolutely vital. You need one of these tracking generator options to measure the Uh frequency response, the frequency performance of filters and other circuits like this. So if you're going to buy a spectrum analyzer like this, this is like a 13,200 $1,300 base model without the tracking Gen.

But spend like an extra 300 bucks and get that tracking generator option because if you don't get that, you can't do any of this or it's incredibly difficult. So the tracking Gen makes is well worth the money. Trust me. Now what this Uh tracking generator does is that when the Uh Spectrum analyzer um takes the sample at a particular frequency because instead of an oscilloscope which is time domain, this is frequency domain It at a particular frequency, which it's measuring that signal out.

It outputs that exact same frequency on the tracking generator output as it sweeps across so effectively. This is an RF Sweep generator. It starts at the low frequency and goes all the way up and it depends on the displayed Uh frequency. You know if we're over the full range like we are here 1.5 GHz Um, 750 MHz Center Basically from zero Uh to 1.5 GHz then it's going to sweep over that entire range on the tracking generator output.

And it's that sweep Frequency sweep which allows you to get the frequency response of your particular circuit under test. Anyway, what we're going to do is turn on our tracking Gen here. I Don't know why that button. You can't just press it twice and it automatically switches it on I Don't know.

that just would have been nice. Anyway, we switched on our track tracking Gen here. our level's going to be zero Dbm Fine, we don't want to power sweep off the power sweep. Um, we just want a constant Um output voltage level from this thing.

So we'll go into Uh frequency. Well, we'll go into Span here and we'll set full span. So we're doing our full 1.5 GHz span here. So our Center frequency down here is 750 MHz We're going over the full 1.5 gig range or there about.

So uh, let's go into amplitude and by the way, I've um, got the just the track engine uh joined straight through to Via a uh 50 ohm coax straight to the RF input and of course the tracking gen output is 50 ohms and the RF input is 50 ohms impedance as well. and that's what we calculated for our PI filter here. So we're going to go into amplitude here and we'll just autoscale that and look at that. isn't that funny.

What do we expect? We actually expected. Well, here's the reference level over here: Zer Dbm. We expected that to be flat across the whole range because we've just got a coax in there. We've got no other circuit.

So In theory, you know this coax can you know it should be able to do uh 1.5 gig I Mean, not all coxas can, but you wouldn't expect to see all that garbage in there. You'd expect a flat line. Why aren't we getting it? And the answer's really easy. Go check the manual for this thing and if you take a look at the Uh tracking generator option output, you can see it's rated for plus minus 3db output level over the entire 1.5 gig uh frequency range.

So it's not a great tracking gen at all, and that's not terribly surprising. It's not easy or cheap to make a completely flat and linear tracking gen from you know, a DC to Daylight uh basically 1.5 GHz I know, that's not daylight to the RF go guys out there. Heck, you know anything under 1 GHz is DC as far as they're concerned. But anyway, it's not that easy or cheap.

So what you're buying here is a spectrum analyzer and you're not really buying that good a Tracking Gen, You're only paying a couple hundred bucks for it. Its performance is not going to be completely rul of flat over the entire frequency range and this is the actual performance of it. Believe it or not, because um, the Spectrum analyzer uh, input of course is going to be pretty good. This is a, you know, a reasonably decent uh Spectrum analyzer.

So if we had another a better a ruler flat track engine, we would see a ruler flat line. But because the track engine is not that great, this is what we get. Now how do you measure I Mean you know the amplitudes here I mean it's spec in the manual is up plusus 3db so it's well within that I mean you know we're only talking sort of a peak around 1.3 DB there and minus 1.7 or thereabouts DB over the entire range. How do we fix that? Well, riol know that uh, this is a problem.

So you go into tracking Gen here and you use What's called the normalize function and that basically compensates for the poor performance of the tracking Gen so it can. What it can do is it can. store this wave form and then subtract it from your final signal to give you a you know a proper response and it has this capability built in. It's easier and cheaper to add this capability in software than it is to Uh design and build a ruler flat.

Uh. Tracking Gen on this thing. So what we want to do? Uh, you use your coax directly, input and output. So before you take any measurements, you compensate for the track Eng Gen.

So you just store it and you turn normalization on. Oh, it didn't like that. Press it again. There we go.

It's updating the reference trace and bang. There's our rule of flat response. so now we can disconnect this and we can plug our circuit in series with it and test it and it will automatically compensate for that stored reference. It'll normalize it.

So what we want to do here is go into frequency, set our set of frequency to say 10 MHz or thereabouts, and the frequency span. Okay, let's just do 20 to start off with and and we'll take a little filter here and we'll plug it in series here with the input Tada There it is. Okay, so we got a filter in here and let's try and measure the response of this thing and Bingo! look at what we get. Look at.

here's our filter response dropping off there. this garbage here. Um I'm going to say that's um from setting up the Uh reference, um, uh level. We did it over the full uh span range.

We could uh, redo the uh normalization across a smaller range and we should be able to get rid of that. but I Just want to show you the roll. so let's just assume that that's flat. Okay, so let's change our um our frequency span here.

it's currently set to 20 MHz so this is 0 to 20 mahz here with a center frequency of 10 MHz And let's expand that and we can see where it gets minus 3db down. Actually, we'll try and measure that with one of the markers. So how we do that is, we just hit marker over here and we've got different. We've got various markers and various modes, but here's the little marker number one and it tells us.

um, and we can adjust. Use the knob over here to move that back and forth So we want Minus 3db and we're expecting around about 15 mahz for that and what do you know? Pretty Dar Close look at that. Ah, Magic 15.06 MHz So we've accurately measured the uh minus 3db uh rolloff point of our Um LC Pi filter here. So let's have a look.

let's expand. Let's go into span here and expand this range. I'll use my knob again and you can see the changing value I could type it in I Can just type in a value on the keypad here if I wanted to. but let's just use the let's just increase it to 30 MHz and you can see it's going off here.

So let's rescale that. let's just Auto Scale that. There we go and let's expand let's span again. Let's go up in frequency.

You can see it's rolling off nicely, beautiful, and it's starting to level out. Actually, there we go. We're up to 100 megahertz span now. so this is 100 MHz and you'll notice it's actually going back up.

Look at that and if we keep increasing our spam, we're at almost 300. MHz Now look at. this isn't this interesting. You know we're at 600 MHz span there and you can see the fall off of our filter.

But it hasn't just continued to fall off like a brick wall. It's actually recovered. it's reversed. and is, you know we're talking -7 DB The -5 DB up here.

I Mean it was down to around minus 27 DB there. but it's gone back up and it has a response like that. And of course, if we go over the full 1.5 gig range of this spectrum analyzer, it rolls off and it recovers like that. Why is it doing that? So what's causing it? Let's take a look at our circuit here.

We've only got basically, uh, four components on this entire thing. I Know you're thinking, well, there's only three. No, there's actually four. Um, there's the inductor.

Of course, we've got our two capacitors plus the circuit board as well. Remember, circuit board is a component. Like anything else, it has a dialectric constant act as inductance, capacitance, all sorts of stuff like that when you start getting up into high frequencies. But this is F Fr4, it's going to be more than good enough.

especially at the uh, um, 70 mahz that we're uh, seeing, um, this thing, uh, change on us. So we we're using um, uh, NP ceramic capacitors here. they should be more than good enough. uh for this sort of frequency range.

So you go aha, the inductor. So let's take a look at the inductor I used and this is what I used I had it in my kit here. it's an NLC 45 3232, uh one, uh, micro Henry but it's a basically a um, a power inductor. It's not designed for RF stuff.

So what we're seeing is the self. the Lish self-resonant frequency of this inductor. It is not a good inductor at high frequencies. It doesn't work as it's supposed to.

and that's exactly what we've seen on the Spectrum analyzer here. so we can see it here there. I've got my Uh marker there. You can see the little one going around and it's uh, reversing there at about 65 MHz So that's going to be around the uh self-resonant frequency of this inductor.

I.E Where it, uh, rolls off and stops becoming the inductor. you expect it to be. and then you get in to all sorts of parasitics cuz it's wire wound. There's a, you know, there's a whole bunch of wires in there, wound around and you get into winding, capacitance and all sorts of stuff and it's not working as the inductor.

it was down at these lower frequencies below 7 70 odd megahertz there. Now let's try a slightly different one. the NLC 3252 once again, one micro Henry Let's see if that changes that uh, self-resonant frequency point and you can really see the wire wound nature of this one. The other one was fully encapsulated so we couldn't see it.

but this one here. you can clearly see the inductive winding inside this thing and exactly how many turns it's got. And if I just disconnect the filter here so we've got nothing going in, you can Bingo see the noise floor of the unit down here. Okay, I've solded the new inductor in there.

Let's plug it in and see what we get. We're exactly the same as before. Ah, what? No, we're still on around about 70 MH hurts there Bummer. I Was hoping we'd uh, see at least a bit of a difference I managed to uh, uh, steal this one off a old VGA video card and I measured that at uh uh, 500 uh Nano Henry So half a micro Henry So it's you know it.

it's going to, uh, adjust our frequency somewhat. All right, let's give this one a try here and let's have a look. Ah, bummer. Still around that.

70 MHz Mark All right, let's try one of these. SMD Ferite beads once again salvaged from aboard I Have no idea of its value, but uh, let's see what it does. Hey, it's uh, not better. The Uh frequency has changed though.

we're now about 125 MHz before it starts bending back there. but uh, certainly its performance is uh, uh, pretty terrible up at the high end up here. I mean you know, only talking. Uh, well, it's you know, peeking around -7 DB Awful.

But really, all that experimentation is for N because we are not going to escape this characteristic. Notch Response: You can see how it goes down and then recovers. That's called a notch response in our low pass filter there. so it's It's not working nearly as good as a low pass Fielder up at the high past a certain frequency point.

So when you're designing one of these fields, you will design it for that particular frequency which you're trying to attenuate uh, at best there. And because after that, it's just going to recover and it's going to have that Notch type response. And this is actually characteristic of one of these LC pie filters. You can't escape it.

The only way to escape uh that Notch response is to change to a different Uh type of filter either a Uh T type uh, filter or a multi-stage filter or something like that to really attenuate this high-end stuff because it all has to do with the parasitics of the inductor and the parasitics of the board and everything else in there. It's just totally characteristic of a very simple LC pie filter. Nothing you can do about it. Hey, it was fun trying a few different values anyway and I'll just show you another thing with the uh noise floor here: I've got the uh full uh frequency range here 1.5 GHz So we've gone into span.

we've gone full span there and we're uh in the amplitude. we're Auto scaling that and you can see it's you know, around about - 65 uh DB I've got no um, normalization uh on here and you can see that it's uh, you know round about that figure. but there's an RF preamp you can switch on and if you go into RF preamp here and switch that on, you'll notice no floor bang goes much much lower. And if you want to know if the Uh tracking turn, switching on the tracking Gen makes a difference to the noise floor with the preamp on very little.

Watch this there we go. It's just slight. You can see a slight difference there by switching on the track Eng Gen So that would be the internal cross talk. So I hope you like that.

That was just a bit of a uh play small play around with an LC uh pie filter and measuring its response on a uh, low-end uh Spectrum analyzer with a tracking generator output. There are other ways to do it, of course, with your siggen and your oscilloscope and stuff like that. Um, and there are ways to do it without the a tracking gen option in Spectrum analyzer, but it is, actually, uh, quite hard. But with that tracking gen option, you can see how easy it is.

and perhaps we can follow this up later with different things. but it's worth playing around with. Something like this was good fun. So I hope you liked it.

If you want to, uh, discuss it, jump on over to the EV blog forum and if you do like it, give it a big thumbs up. Catch you next time.