So let's take a detailed look at the main board here and we're only going to be concerned with the top side here because if you have a look at the bottom side here, there's just nothing of interest there. It's just all our passives bypassing and regulation maybe things like that. so nothing special at all. Now, it might look daunting at first with all these distributed and element fielder's and everything else, but as we've seen before, you can see that's pretty much a modular block approach, and I've done a handy little overlay here that will attempt to hopefully explain all the different functional blocks and the signal flow on the board.

So let's get to it. So let's start by taking a look at the RF input in the top left corner. This section here, of course contains the 50 ohm input impedance, but that little SOT 23:6 package you'll see are four of these here. These are actually our single-pole double-throw switches.

so they can actually switch in the 50 ohm load and various other stuff. So we'll go to a higher res photo for this and then zoom in on the RF section here and we can see that the input is AC coupled there through C10 and then that goes into U1 which is a 955 see as all the other ones here. these SOT 23:6 parts. They're some form of single-pole double-throw switch which I can't find the datasheet for if I can I'll link it in down below, but you can see that one side of the switch there I believe pin one.

There switches in C9 and R1 which is the 50 ohm load there. so it's not a permanent 50 ohm load input, and you'll notice that there's actually four diodes unpopulated there. so there's a distinct lack of input protection here. So unless there's something inside that little wimpy U1 switch there, there's you know, not much here at all.

There's basically nothing. On the other side, there is a tiny little diode d7 there, but jeez, it's wimpy. And if we scroll down here, we've got a couple of more of these switches here. and there's some diodes for diodes there.

So I'm not exactly sure what's doing there, but that looks like some power supply clamping protection there. At least they start to have something now, and if we have a look at the pin one of you - there, you can see that there's a control impedance trace coming in to that. Obviously this is an imp. good path where they can switch in some sort of, you know, system test signal.

something like that. I Don't know where that comes from, what there would be, maybe part of the self test or or calibration or something like that. So yeah, that's coming from somewhere, but normally that wouldn't be part of the measurement system. just allows them to switch stuff in and a bit further down here.

You can see that VR one there. It's got 20 written on it and that's a 20 DB attenuator there and you can see that's basically switching in C 16 that that straight are controlled impedance line there. So it's basically it's selecting either straight through or a 20 DB attenuator here. Next up we go down into a hey CH MC Once again here tight.

They're everywhere. They've got the entire solution for this thing, the H MC 307 and this is the digital attenuator. So when you go into the spectrum analyzer and you set the input attenuation, you can send in 1 DB steps up to 31 DB over and above the 20 DB input attenuator and that's exactly what this chip does. So the software is limited by the capabilities of this chip, but yet nice device DC to 4 gig.

Do you see the daylight and I Really like the way the designers have laid out this chip. Look at this, there's the input pin and then there's the two ground pins right there. so you can see all that via stitching to separate the input and the output so there's no coupling there and then the pin below. that is the output.

So from a layout point of view, it allows you to lay it out with a minimum amount of coupling. Nice. But we're not done with our input section yet. If we scroll down a little bit more, we'll see the signal flow down into our next section.

Which is good. Of course, the preamp This thing has I Believe it's a 10 DB preamp gain on it, so once again, selectable. So we expect to see the digital switches there. and that's exactly what we get.

so it can never bypass the preamp or switch in the preamp. But of course, in this case, you'll notice that the switches are bigger, They're a different package and we can't actually get the datasheet. Surprise surprise. It's another he type part.

a single-pole double-throw It's a non-reflective switch up. DC - not quite daylight this time. 3.5 gig. It's a non-reflective switch.

You can see the internal diagram there. It's actually got internal 50 ohm termination resistors in there, but basically it's just a switch. It allows us so they use a combination of two of them. You can switch in your preamp or switch it out easy.

Now that's all bread-and-butter stuff. But look at all these other blocks in here and this is the complex operation of a spectrum analyzer. Not all spectrum analyzers operate the same, but they use very similar techniques. So what we're going to do is take a look at a basic block diagram here.

So we've looked at basically just one block here. the RF input attenuator in near the signal input there, and that includes the switching in the preamp and everything else. Now we expect to see a low-pass filter in here, and that's what we'll see in a second, and then that goes into a mixer, which then uses a local oscillator. Mixes the two signals together, generates a higher frequency called the intermediate frequency, and then we expect to see a gain stage.

Here there's that gray amplifier block there attenuator. We won't see this in this one, but it doesn't matter that I F thing goes, you know what? IR filter will definitely see that and then goes into a log AB and envelope to take that video filter and display. But that's not quite how this one works. We need to look at another block diagram for that and as it says here, most spectrum analyzers used to or for mixing steps to reach the final intermediate frequency that we can then in this case or do all digital processing and actually display that because this is an all digital I of system instead of a traditional analog spectrum analyzer.

Anyway, so there's gonna see several steps here. By the way, these diagrams come from the keysight application. node A in 150 I'll link it in down below. Highly recommend.

It's one of the best reads on how spectrum analyzers work and everything else, so we expect to see in more. In this case, what we're gonna see is two local oscillators. The first one goes in the first mixer and then the second one that goes into the second mixer. Here if we take a look at the first mixer on the left-hand side, there, that's the green circle with the X there.

We need this because we need to generate a higher frequency than our frequency range of interest. In this case, our spectrum analyzer can go up to 3.2 G so we have to generate an intermediate frequency higher than that because if we don't do that, then there will be dead bands within the measurement window that just won't work. So we have to actually mix that with a high mix our input frequency with a higher frequency to generate and intermediate frequency above our maximum 3.2 gig input range. And if we go back to our original block diagram here, what we expect after our input stuff is a low-pass filter and then a mixer with a local oscillator feeding into that mixer.

Do we get that? Well, let's take a look. Yes, of course we do. You can see the preamp there on the left that we looked at before. it then feeds into a down into that low-pass filter which is again a distributed element filter there with the various URLs and C's and then that goes into a mixer I see there which then accepts the signal from above it there from that nice-looking bowtie distributed element low-pass filter and that will come from the local oscillator as we'll see, but it's a bit more complex.

It's not like the local oscillator feeds straight in. we're doing some tricks with our local oscillator in this particular case, but anyway, the output from the mixer then goes into that amplifier gain stage as we saw on the block diagram. and if we take a look at a high res photo of the mixer and that amplify our AF amplifier, our stage. Once again, we've got two headlight parts yet again: the HM C4 Double 8 mixer there on the left and the HM C Seven One Six amplifier.

Let's take a look at the data sheets and this mixer can go from four to seven gig, which is exactly what we want. It's above our operation or our frequency range of our amplifier. and if we have a look at the specs here, then our way: intermediate frequency range DC to 2.5 gig and then now o AF amplifier chip the HM C Seven One Six. It's exactly what you expect.

It's A in this case, it's an 18 DB gain amplifier, but it's got the bandwidth of 3.1 to 3.9 gig. so it's designed to operate within that range which is above basically our 3.2 gig maximum operational frequency range. and that's where our frequency is going to sit somewhere about 3.2 gig. The exact value we don't actually know unless we do more investigation or some measurements.

But before we follow that intermediate frequency out, we want to see our local oscillator. Because I Said before, it wasn't as simple as just the local oscillator feeding into the mixer as it shows on the block diagrams for spectrum analyzers. So if we zoom in here, we can find our first local oscillator, our main voltage controlled oscillator, and this one uses a Zed comm part. Therefore, the VCO the voltage controlled oscillator and which is the big metal can there and another here tight PLL there to form our local oscillator.

Now this is made by a company called Z Communications and they make a ton of different variants of these with different ranges and things like that and this one is going to cover the frequency range that we need. If you have a look at the tuning voltage here, it goes from 1,800 to 4200 megahertz or 1.8 to 4.2 gig and pretty much exactly the range we need here. and this is our sweep generator we saw in the block diagram on the bottom left There, the red sweep generator feeds into the local oscillator and then feeds into the mixer. But as I said, there's a few more steps after our local oscillator before we get to the mixer in this particular analyzer.

But as part of that local oscillator, we've got a key light HMC 7:03 Fractional synthesizer which forms part of the ultimate PLL local oscillator loop and we can see that here. If we take a look at the demo board, you can actually get for this chip. It shows that there's a VCO integrated as part of the system here in this case ahead I'd HMC 508 But in the case of the signal and spectrum analyzer here, we're using a VCO from ZD Communications and if you believe the sales blurb here, check it out. This platform has the best phase, noise and spurious performance in the industry.

Yes, thank you very much. but once again, you know decent choice is being made here to enable a pretty decent performance at a low price point. Well done segment, but even with all that magic, the output of the first main local oscillator here is not high enough in frequency. so it goes into a frequency doubler there.

and this is the I Four to two to four gig input. So doubles that anywhere from four up to eight gig. but once again, the exact bandwidth our frequency range we're talking about here we don't exactly know unless we did further investigations or measurement and the frequency double are being used again. A here dot h NC 189 here two to four gig input as I said.

so four to eight gig output. It's designed for exactly this job and this particular part isn't obsolete. Unlike if you were very keen you would have noticed a plus and over the data sheets for a couple of chips before we would have seen that they're actually obsolete. So yeah, why they're still using them I Don't know.

Maybe there's nothing better at the price point, but we're not done yet. No siree. Bob The output of the frequency Ee here for our local oscillator goes into two single pole double throw switches which then can select one of three bandpass filters in this case our then this particular physical arrangement and the distributed element filter is called an inter digital bandpass filter and so three different frequencies. You can actually see that they're different geometries there which actually selects the bandwidth and the response.

And then there's three single pole double throw switches on the other side. so the software can select one of three bandpass filters on our local oscillator. and these switches are different to what we've seen before. These are obvious Wa2 - 63 bla bla bla bla bla.

and these are about high isolation, absorptive single pole double throw switches with integrated seamless drivers and all sorts of weird and wonderful stuff. And we don't care about the quiescent current really. and 500 - 6, 500 mega - 6 gig bandwidth. Pretty decent and we're almost there.

I've mentioned this before. you can see the output of that wasn't that selectable bandpass filter there. Then I goes through just a little bit more stuff there and goes through another bowtie low-pass filter. It's called a bowtie low-pass filter because it looks like a bowtie.

that's where it gets its name from. and then that finally goes into the mixer. So that block diagram we saw before and you see for all speech analyzes, the local oscillator goes straight into the mixer. Well, yes, you've seen It's a bit more complicated than that for various are performance reasons.

But if you're keen I you would have noticed something in between there the output from the inter digital filter after the switching and probably some little buffering there or something then goes into this odd-looking arrangement here on the board which is coupling the signal to go over. If you follow the trace on the other side its coupling over to go up to the tracking generator local oscillator SMA connector and that jumps on over to the tracking generator module we saw before. So finally out of our mixer and then through our I F gain stage which we have looked at we expect to find an IR filter and well you look at the output of the amp and the 18 DB I have amplifier down here. Bingo! It goes into another bandpass filter, another interdigital type, once again different geometry in there to give you a different range and response of the thing and then that's followed by another rub and cute-looking bow-tie low-pass filter as well.

once again just to take the upper edge off for something. and if you curious about how these into digital bandpass filters actually work when you can clearly see that both like the input signal comes in and then it basically goes down to ground with a trace sticking up and then the other than the trace on the right hand side next to that goes up to ground at the top side and then the next one goes down the ground. So how does this actually work? Well, it's because we're at high frequencies here. These work at, you know, several hundred megahertz, up to, you know, several gigahertz or something like that.

They're basically are coupled resonators. They're also known as inter digitated coupled resonators. So yet they resonate between the two and then it propagates along and resonates. And that's why you might see different spacing in there is to give a different pass band characteristic for this thing.

Anyway, you have to get into real complex RF microstrip type theory. To you know, figure out exactly how this works and there's a time into it and I'm sure you could google it if you're really interested. But yeah, even though it goes down the ground there, it gets through. But we said here before that this particular spectrum analyzer arrangement uses two mixing techniques and so we need to find that second mix.

Our end the second local oscillator as well. And if we pan across here, bingo, the output of our filter there goes into another mixer. the four double-a four, double eight. Exactly as we had before, but just like on the block diagram here, you'll notice that the output of the second mixer is a much lower frequency.

It's with way under way within the pass band of our spectrum analyzer in this block diagram 322 megahertz. But in the case of this particular one here, it's actually at 810 megahertz. And the reason we know that is because hey, look we can look at the filters on the output of the mixer and we can see that there are saw filters or surface acoustic wave filters and we can have look at the data sheet for this particular one. they're available in all different frequencies.

This one happens to be an 810 mega Hertz saw filter so we know that's the output frequency of the second mixer, but this isn't low enough frequency for now us to do digital Ifr sampling on. So what we want to do is feed it into another third mixer just like what's showing here to actually down convert it to a frequency that we a baseband frequency that we can actually sample with like a Joe Bloggs you know, 16-bit analog to digital converter and we can see that here, the output of the soft feel that goes into this little white block here which is a mini circuits. Yes, we finally get a mini circuits win in the design Here, it's not all he died many circuits, one of the biggest art providers of these sorts of mixers and so this will go in and we can take a look at the data sheet for this mini circuits mixer as well. But there's nothing terribly exciting to see here.

it's just a you know, basically five megahertz to one gig mixer designed for this sort of application down conversion to a baseband signal. But wait, we're not finished with the mixer. Every mixers gotta have a local oscillator input. Where's that coming from? Well, it is coming from the second local oscillator, but we need a much lower frequency, so you'll notice that the second local oscillator here as like feeding the second mixer across to the left.

There, it also goes up and that same signal feeds a is divided by four and then that gets fed into the third mixer which does the down conversion. So we've got our final RF frequency bandwidth here, and this goes into our curiously a single pole, four throws switch. and that's what the IC is it. So I'm not exactly sure what it's selecting there, you know, and there's some sort of different filtering options that it's doing there I'm not exactly sure what Anyway, that then goes over into another single pole for throw switch here, which has only half the stuff populated, so that's quite unusual.

Why did they leave that out Now as a user by the name of Vigo Zoo if I'm pronouncing that correctly on the EEV blog forum postulated for this one, it it certainly looks like another bandpass filter in there with inductors and the caps in there, and that would be one of going into presumably one of the channels of U85 on the left hand side there, the single pole, fourth row switch, and presumably there would be a software option for this to have another additional bandpass filter on the final if' before it goes into the sampler. So maybe there's even a secret menu option for it if you could hack the firmware or whatever. Or maybe you know they had an early version of firmware, they decided they didn't want it. I Don't know, it could still be there.

Who knows. could be interesting, but yeah, I don't know if you can find it. you might be able to hack in your own bandpass filter in there for some additional functionality. And the good thing about an experiment or hack like that is that you're not really.

You know damaging anything. You're populated in existing footprints in there with an existing digital switch that's only affected if you enable a software option in the firmware to actually flick that switch and in, you know, put that filter in series with the final if' there so you know you can play around if your heart's content without really risking damaging anything. So that's it. We're finally through our complete block diagram here, but this envelope detect that we don't have that because as I said before, this spectrum analyzer uses what's called an all-digital If' filter.

So it does everything. after the If' stage, the intermediate frequency stage, it just samples that directly with a high resolution, a high sample rate analog to digital converter, and then does everything in software as we see in this Keysight application. That note here here is how the Keysight X-series signal analyzers do, and all digital. AF They've got an ADC in there with a gain and the alias filter, everything else but it goes in there.

then a custom IC which in this case would be that Spartan 6 Fpga we saw is doing a Hilbert transform and thence doing some filtering and then it can do the video bandwidth in there and does logs and powers and all sorts of and the detector. all sorts of stuff all with inside that'll be happening inside that Spartan 6 FPGA no doubt. and then that goes into the pro and probably to be doing the FFT in there as well. and then that just goes out to the display applications processor which we saw earlier.

So now we have to go full circle right back to the main PCB under that block where we found our main reference oscillator before and what are we fine? surprise, surprise, and a DC driver designed specifically for AI F baseband process in in this case it's the National Semiconductor numbers Texas Instruments Rubbish. L Mh-65 1:7 is designed exactly for this for a 16 bit ADC and there's the block diagram down the bottom, so no surprises to find what's down below this. I'll give you one guess and congratulations you in a brass. Razoo It's an analog to digital converter.

it's the analog Devices Ad 9235 actually 12-bit Surprise Surprise No is 16-bit rubbish. I guess for Cichlid Moe 12-bit will do the job just fine. And yeah, it's designed for ultrasound equipment or low-cost digital oscilloscopes. There we go, We know winner chicken dinner and you'll notice that we've got the - forty part there.

Which it means forty Meg samples per second. This parts available from 20 up to 65 make samples per second, so 40 make samples per second. We know that our F baseband frequency has to be somewhere below 20 because you know all that Nyquist stuff. Really annoying.

Yeah, so it's got to be at most half of that sample rate. So I hope you enjoyed that sort of building block walkthrough of a spectrum analyzer. In this case, the sequent SSA 3000 I Did do this video a couple of years back, but it was embedded in the teardown and it was a new style of edit I Wanted to try where I took my high-res photos and actually then just in. My editor actually did the voice commentary with my mic here and do that over the top and then you know, zoom in and pan and do all that sort of stuff.

so that was I was quite proud of that and it was kind of like, you know, just tucked away in this tear down. So I thought I just take that out and move it on over to a separate video and I've got a new monitor set up here I Just wanted to try edit some video so I hope you found that useful. And if you like that style of tear down video, please mint let me know. It does take a bit more work than just my usual wow thing where I just literally like opened up and I stand behind the camera I've got my little poker in there and I just like poke at stuff and then you know zoom.

Basically the zooming I do is zoom in on the camera or changing the macro lens or whatever and just you know, waffling on it. Just press a record and figure out something to say. I do the same thing here. It's not like I have a script or anything.

it's um, I still do sort of like the off-the-cuff commentary and stuff like that, but it's done at the editing stage which is a different style of video to what I normally do. So anyway, if you liked it, please give it a big thumbs up. Let me know down below. Um I won't do this kind of style of video for all tear downs because sometimes that's not appropriate.

Sometimes it's just easier and quicker to do it just standing behind the camera and just do it off the cuff just poking. you know, straight at the thing. This is still off the cuff, but it requires a lot more editing magic and I'm not really the world's best at it a bit. Hey, it worked I think Anyway, catch you next time.