Hi, welcome to Tear Down Tuesday It's not every day that Tetronics release a new oscilloscope, especially these days, but they have. It's the new Mdo 3000 mixed domain oscilloscope a uh, a sort of a lower-end followup to their Uh Mdo 4000, which we've reviewed and tear down before and I'll put the links down below if you haven't seen that one. So I expect this one to be a similar high quality design and construction inside. It's got the built-in RF Spectrum analyzer which is going to be fantastic to take a look at and this is the 1 Gz uh version so it will actually contain different Hardware to the lower end 100 to 500 MHz version at least on the front end anyway.

But oh yeah, so of course we could Power It Up and review it. But hey, you know what we say here on the Eev blog. Don't turn it on, take it apart, and here it is it's It is small and lightweight, and pretty compact. Well, at least you know, small and lightweight for the size.

I mean the screen is absolutely enormous on this thing and uh, well. I see a couple of screws on the back here, so let's whip it open. It looks like, you know, uh, traditional tectronics? uh type type design here. just a few screws on the back.

the whole back panel will lift off I can see the boards, the big fan in there I can see the boards uh, through or the rear board through here. so uh, it looks like it just whips straight off and there's a metal shield on top so there'll be more uh items in there. They looks like uh yeah, this back panel will just lift off. We don't have to undo any nuts on the Bnc's there, Maybe on the inside? well, almost certainly uh, on the inside before we can get that metal can off.

And of course this is going to be shielded really well. The Spectrum analyzer built in just like the Uh 40001 will have its own metal uh, diecast uh, metal can inside there so there will be some RF porn inside there. No doubt it's a 3 GHz bandwidth. As I said, this is the 1 Gz model and it is physically different to the 100 to 500 MHz model.

If you want to upgrade, if you buy the 100 MHz model and you want to upgrade to 500 MHz you can with just a software license. but if you want to go to 1 GHz like this model is, you have to physically send it back and they swap the board. I'm fairly confident it'll only be the front end that actually changes the amplifier on the front end. I Don't think any of the processing or anything like that actually uh, changes inside.

So really, if we are able to see the analog front end, that'll be the only difference between this and the lower-end model. The Uh RS Spectrum analyzer the 3 GHz bandwidth one is the same on all models now, of course. Uh, Tektronics are famous for their Uh hybrids. They're a huge company, of course, one of the big Scope one of the big three scope manufacturers and they of course roll their own uh hybrids for all sorts of stuff their own custom Asic chips and things like that.

so they're notorious for uh, you know if they fail like in the old Tectronic Scopes are getting Replacements Um, very difficult indeed. Or although, of course these modern ones, they're so so complex that you know, uh, you probably wouldn't uh, repair these suckers like you know in 20 years time, you're probably not going to be able to repair one of these things if it goes. HeyWire Let's lift it off because this is my own unit. Um, they did donate it to the Eev blog lab here, so it is now the highest bandwidth scope in my lab at 1 GHz So we're going to have lots of fun playing with this thing over time and uh, so yeah, I really don't want to damage this even if it was theirs I wouldn't uh, want to damage it Anyway, it did get stuck in Custo sorry for the late notice on this thing.

um, but it did get stuck here and I had to pay Customs fee to get it cleared and get it into the country. but I was supposed to have this before the launch deadline. It launched on the 25th of February and I'm sure you can actually buy it right about now. Prices start from I think don't quote me about $3,300 for the base model unit.

Looks like there is another second Shield under there cuz I Still don't see the uh nuts for these BNC connectors and I am taking nut highres photos of this as I go along. So if you want to see those, jump on over to Eev blog.com where they will be uh Linked In the link will be down below for that. Yes, I've got my new Sony uh NEX 5T camera with my macro lens. very nice Tada and H look at that.

we're in like Flynn and the main board exposed. Fantastic! And the power supply out the back? look at that. All very neat, but you'll notice. The first thing of course is that there's no shielding from the switch mode power supply to the main logic down here.

go figure. Well, not a huge deal in the logic cuz this is all the digital logic stuff. Uh, you know around here that's near this uh switch mode. All the RF all the stuff that matters.

the RF on the front end of course. uh RF The RF Spectrum analyzer will have its own metal cam, but all the analog uh front end for the four channels will be all in its own individual individually shielded thing, but still a scope of this quality. I would have expected shielding around the can like that cuz that's power of the course in other Scopes that are cheaper. So yeah, not that impressed now.

I Do really like this. uh rear connector board like this PCI card Edge connector which then flips over onto the main board I Really like that. So we did. didn't have to undo any of the connectors on the back to get it open.

This is a really very well-designed case so you can see the thought that's gone into this. Small touches like the fan over here, they've gone. Oh well. You know ordinarily you would power that from the power supply board over here, but oh, we don't want to run a cable all the way over there.

Oh, look at this. let's make it nice and neat and just the right length and mount it on that rear panel connector board. there, you know. Absolutely brilliant.

The Main's wiring. very neat, very tidy down in there. Look, the cable runs have been kept to an absolute minimum. They cable tied here.

They've just got exactly the right length cable to go in there. Beautiful now, even though the power supply is, uh, unshielded. but I'm sure Tech have done their homework on that. So I'm just nitpicking with the shielding.

It's just not best practice that's all. But they're obviously getting away with it and well, okay I Trust Tech to know what they're doing. But the actual power supply looks like very high quality. They've got an Emerson uh Transformer down here.

you know, top quality brand. It's all you know TUV um compliance on it and all sorts of stuff. looks really really welld design. Of course Tetronics wouldn't have done this.

they would have. uh Sho this out. But yeah. I you know, look at that beautiful Sastic holding everything in place.

All your protections are there all you got your um uh, boots, even even over the uh Earth connector there. we've got a rubber boot over that Beautiful. We got some input protection down in there. It's you know they're doing everything right on this power supply.

High quality, but hey, that's what you'd expect when you pay high price for a tectronics. And for those who' love to know the capacitors as we all do. And you should because they should be high quality in a tectronics. Yes, it is a Rubicon brand.

No problems whatsoever. and that one by the way was 105 C rated and this one down in here even though you can't see it, trust me, is a nipon chemicon. So once again, top quality and 105 C rated. No problems.

Someone really has gone to town with the Sastic gun. Look at all this. it's all gunked down just well. You know it's beautiful.

Of course it looks like we might have some uh small PCB fuses down there. Perhaps they got silicon in the end on uh, the end on through hole mounting like that. Fantastic. And the pot is sealed nice and the fan once again also.

uh, good quality brand name SEO Dany No problems at all. So our main processor board is, uh, significantly less complex than the Mdo 4000. It's big brother product and that's what you'd expect. Why? Well, it's obvious they're trying to meet a much lower price point with this thing, so they're going to sacrifice uh, some horsepower and capability uh compared to the Mdo 4000.

And if you compare it which I'll link in the photos and the video down below for the 4,000 Well, this has bugger all. compared to the 4,000 we've only got three main heat sunk devices here to share across all four channels. Now could be more stuff on the other side, but certainly not. Uh, big heat sunk devices like this whereas the Mdo 4000 had I think a separate ADC uh per Channel heat sunk and all that sort of stuff.

But look, three main devices. The sample memory here. Almost certainly the sample memory. Why? Well, there's four of them there and they're coupled in with uh, presumably match length uh, traces.

To get the timing right, we'll take a close in look at that. Um, so there's one device per two channels like that. So of course, your uh, sample rate or half have if you turn on uh, both channels presumably. And then they've got the uh third main Heaton device over here.

that could be a uh, you know, the main acquisition uh ASC or you know, something like that triggering all sorts of other Jazz Um, but that's tied between all four channels. So really, they have cut cost here presumably not the same um Asic as they use in the MD 4000 I presume that they've uh, spun new ones for this Mdo 3000 and here's one of the main As6 as well. a totally custom uh job presumably uh, you know Tech part number you probably can't get any info on that uh to save your life, no doubt. and they've got some more memory coupled into that here closely.

And if you follow the traces out like that, you'll notice that it is coupled into these channels over here. So there we go. You can follow that bus all the way around over into this Main basic and presumably on some of the other layers. There will be another uh, you know, parallel bus going from here over to that one as well.

So this is obviously some sort of main Uh acquisition and uh processing Asic probably does all the uh DPO uh way. you know, intensity, grading, and things like that. perhaps part of the display Asic perhaps I Still don't know where the display is plugged into this thing, so that's the thing with these when you doing tear Downs Like this, you can tell a lot of the functionality due to the proximity of stuff. like if the display connector was here and then coupled in.

well, you know that this is doing the main display processing as well. So anyway, this thing does have a couple hundred, uh, waveform updates per second. so they're obviously doing that in Hardware instead of the main processor down here which will take a look at that handles all the gooey and the controls and uh, and the communications and all that sort of stuff. And if we have a closer look at those memories, I've checked the data sheets for these: I SSI brand.

Of course these are 8 Meg by 32 or a total a 256 Meg synchronous. Dam So there you go. that's couple. There's two of those, maybe another couple on the back side of the board.

We don't know until we flip it over coupled into this main Asing So presumably that would all be for the DPO uh, intensity grading and display, uh, processing and stuff like that? That would be my guess. Anyway, until I flip the board and find exactly where the processor the display is connected into. We know won't know, but that's a good first guess. I think and back over here to the Uh Asic which shares two channels.

We've got ourselves uh, some Micron memory here that's the Micron symbol and um, you have to decode this part number here. but thankfully Micron on their website have a part number decoder for these small BGA they can't print the full number for those playing along at home. it is actually an Mt 42 H64 M16 HR and that is a 1 Gbit or 64 megabit by 16 memory. So where is the rest of it used for? well clearly used for the Uh logic analyzer would be uh, my guess I see the dedicated logic analyzer circuitry but I don't see uh logic analyzer memory unless we flip the board out and there's our main clock down in there looks like 100 MHz don't know the exact brand.

It looks like it's got decoupling on like a little surface mount hybrid module there that's quite interesting. hooked into a Analog Devices ADF 4360 and that's a um, no surprise. a synthesizer and voltage controlled Uh oscillator. So that gives a three that chips capable of a 350 MHz to 1.8 GHz output.

So clearly that's not Uh providing directly the main sample clock for the ADC this thing uh samples at 5 gig samples per second. so there has to be some additional mod ication happening somewhere else and no surprises for seeing. Tied into that is a Mcrel. um H1 U I Had a look at the data sheet for that one, found it.

No surprises. It's a Peele buffer so that's actually uh providing the drive here we go. Here's the input coming from the output of the Vco there coming through here into the buffer and then check this out. There's the output there coming across, they're tap clearly tapping off that signal there and there, through to the uh bottom layer.

So an internal layer through those veers and then they've got this massive LC filter Network Huge number of stages here and then popping out the other end there to drive something else. Fascinating. So you've got to wonder if that's some sort of uh LC delay line or something like that. H Now, this is rather interesting.

Check out this lattice device here. It's a Um Lf3 series Fpga and it's not a particularly, you know big beast. and it's not heat sunk, so it's not doing any real serious work. but uh, it's only a 33,000 lookup table.

uh, Fpga with one megabit of memory built in. They've got some uh ROMs you know, some flash memory uh coupled into that. So you got to wondering. You got to wonder what this thing is doing? Is it running some sort of internal well? Must be running some sort of internal processor if it's got some flash memory hooked up to it like that.

So Softcore processor. But take a look at all of these test pads out here. This is not for an unpopulated connect I Believe that's a, uh, you know, some sort of production, uh, test pad or something like that. So maybe could this be a real, uh, expensive, and uh, dedicated way to do production testing? Perhaps.

I Don't know. Maybe it's doing something else. You can clearly see the tracers coupled up here. Look, they're heading all the way up there, coupled into um, one of the input uh Asic over here as I'll call them and presumably there would be a similar, uh, yeah, yeah, no, no, there could be a similar bus heading on over to the other one.

You would think so. Anyway, so yeah, what is that puppy doing? On second thought, it's actually very likely to be the serial, uh decoding feature of this thing. cuz this is like a, you know, Advertiser 6 in1 uh instruments. and that's one of the main instruments is the serial decoding and stuff like that.

So maybe they're doing that in Hardware in this dedicated Fpga and it just looks like we have some uh, power supply Down stuff down there, presumably for that uh, lattice local regulation core voltages for that lettuce fpga and we have ourselves another oscillator. It's one of the espresso series can't quite make out the part number of that, but yeah, presumably just another uh clock multiplier. And we got a couple JTAG headers here in here. conveniently fully populated with the proper header connector.

so go and plug straight in and uh, hack and uh, debug and play around to your heart's content I guess and I have absolutely no idea what's going on here. Look at this. It's almost as if it's some sort of big touch sensor. Um, you know, some capacitive touch sensor.

They got some, you know, digital tracers running underneath there, so that's not good practice to be begin with. but like, why? why is that? look, you know my finger is, you know that's sort of like the size of my finger I wonder what it does be interesting to power this thing up and just touch your finger on there? Maybe because look, there's a couple of V's on there, so it's not. you know? So they're electrically connected in two separate halves. It's not some weird ass spark gap uh, you know, input protection thing cuz there's no solder M removed of it or anything like that with the sharp Corners it it's a capacitive touchpad.

it's all I can figure and check it out. I Found a little bodge in there. look at that curiously on just one channel of this multi- channel uh, logic analyzer input here. here's our the logic analyzer actual connectors on the other side of the board down in there, presumably surface mount on the other side and no surprises for guessing these are your input comparators.

Yes, very fast ad CMP 562 is at 1 nond comparatives 500 p a seconds rise, time differential pickle outputs A it's all happening there. and just above those is a voltage regulator. but not just any voltage regulator. a fast pulse response voltage regulator specifically for driving these uh input comparators.

No doubt because that's the real disadvantage with having extremely fast, well devices of any digital nature like this. They take huge gulps of current when there are switching. So really, you need a fast, transient response voltage regulator up there local just to power those devices. and that's a Um Ldo 725.

Nothing much happening around here. This ICL 3 Uh 321 looks important, but it's just a Max uh, sort of 232 equivalent Uh part for RS 232 interfacing that brings us over to our main processor or main applications processor as you'd uh, really call it because this is no surprises for finding a freescale. Uh IMX 6 Series This is a COR Arm Cortex A9 Processor 1 GHz this Uh series of chips. You can actually get quad core versions, but this is only this.

What's called the Solo chip. Uh, hence the s in the Uh, the 6s there in the part number. Um, so this only has one core in it, but this thing is quite a beast. You operating at 1 GHz capable of 1080p encoding and decoding in fantastic stuff.

So as you can see, we've only coupled in one Uh DDR Uh 2 memory. Here you can see there's an un unpopulated footprint on the bottom. there could be another one on the bottom side there, of course. And of course you notice the Uh Serpentine traces.

That's to match. Uh, that's to match all the length of the differential pairs going to the DDR memory there. Now, whether or not this thing is driving this display directly, or whether uh, the display is actually handled by this main tectronics Asic over here because if you're getting a 100, you know, a couple hundred, waveform updatings, updates per second. It's going to be hard to do that through even a real beefy applications processor like this.

But hey, they could be doing it. Um, so yeah, we don't know. We have to find out where the LCD is actually physically connected into this interface here. But yeah, there you go.

I Mean this is going to be handling all of your M You know, your main front panel, your user interface functions, your operating system, your main display stuff, of course, all your menus, and everything else. It's going to be handling all that. whether or not it's actually the real live data. Real live waveform data? Uh, directly to the screen.

I Mean they could be mapping it through this AC cup here. You know that that would be a better guess than trying to funnel everything into this poor little applications processor, even though it is pretty beefy. Um, yeah, it's more likely to be handled up here the display parts of it. That's what they do in the Agilant ones, for example.

Or the Uh Asic handles the direct mapping of the waveform to the screen, and then the main processor just sends the data to that processor, which then you know, adds the menus and cursors and everything else around wrapping around or overlaying that waveform data so you can get the fast waveform updating and of course, all your menus and everything else. Ah, they can be as slow as a wet week and you don't care and there's nothing else terribly exciting on here. We've got a couple of uh, push buttons down here CP Power on off Very interesting Global reset built-in switches like this PCB Mount they've gone to a bit of trouble to uh Mount those things in uh for design development and servicing and uh, stuff like that I don't know. Yeah, we've got another um, uh, big uh multi Pro multi- Channel switch mode uh Supply up here doing some stuff I'm not sure why they just why they did that on the board and they got this full main power supply board all out here doing stuff.

but anyway, big multi- channel. uh, big multi-way connector up here connecting the main power supply board on. They got some miscellaneous stuff down here that's just handling the USB port the front panel USB port down the bottom so nothing terribly exciting there. So that's the main overview of this board.

Not a huge amount on there, but as you expect, all you know Tectronics go to town in terms of developing custom As6 and you know, really, engineering this thing to the hilt to get the uh cost down. and well, we main three main heat sunk devices over here. obviously. uh, two of them are drive two of them are identical and handle in two channels each.

We've got a third one which handles all four in some way. We've got this mysterious Fpga over here doing something with some uh ROMs hooked up to it. We've got a main tectronics uh, process? Uh, well. Asic over here, not heat sunk.

Curiously, we got an applications processor which handles all the real time OS and we've got our logic analyzer stuff down here. and well, there's not a huge amount more. Okay, let's lift this main processor board out. Taada.

Ha. We're in and here's the main board and it's pretty much what you'd expect. Not much on the bottom apart from some additional memories. Just to get the density, it makes sense to put them on the back just from a PCB Uh routing Point of View flip them on the other side of the other chip on the top.

Um, it just, you know, works out fairly well. So here are your three main heat Sankar As6 one of these per sharing Channel We got some extra memory over here, so we've now got a total of two gig bits memory per Channel There you go. And uh, of course our uh main uh Asic in here coupling uh, all two channels together. There's nothing else on the back of that Fpga over there.

Of course there's our logic analyzer input connector that's actually the front panel connector there. We've got some more high-speed comparators on the front end there, but apart from that, not much else. Here's our main Tectronics Asic down here. Yes, we've got more memory coupled around that, so tons of it.

Got two Bard high speed board to board inter connects and they're going to be actually bringing the data over from the front ends. Um, presumably this is where the systems engineering in this gets interesting because you notice the Adcs. Well, the the front ends sorry are not on this board, they're on the other board and they got to get across. the data's got to get across.

Well, this high-speed connector here, cuz this one over here is coupling in, um, some other stuff over here to the main tectronics. uh, processor Asic over here. So yeah, go figure it. We do have an extra memory up here on our um, um, Applications processor over there, but apart from that, not much else.

By the way. I Just checked the photos from the Mdo 4000 tear down and the main Tectronics Asic here. Uh, on the 4000 was the Tech 05b so this is presumably a uh, new revision of that part. Now here's where the surprising part comes in.

Just like the Mdo 4000, Yes, they have a separate Uh input board here which contains the analog front ends. These are the 1 GHz front ends four Channel Scope of course. Um, and this is our R F front end over here. but that's all there is to it.

I'll show you a photo from the Tectronics M4000 and it was much bigger like this and that contained all the big RF can all up there for all of the spectrum analyzer stuff. but look, there's nothing else inside this thing. All we've got is this tiny RF front end over here. No wonder they're keeping the cost down on this thing.

This thing is nothing like the Mdo 4000 and not only in the Spectrum analyzer part of things, but also there's no Adcs on this board directly, at least not on this side. The once again here's a photo from the Mdo 4000 and it had above each one of these ADC sorry each one of these analog front ends here. Um, it. It had a separate ADC per Channel with a big heat sink on it protruding up through the top side of the main processor board.

The heat SN was so big it had to go up through the through cutouts in the main board. This has none of that. Where are the adcs? These are the analog front ends 1 GHz bandwidth. But hey, are they just taking the analog data through this connector here? Anyway, if you send your scope back to Tectronics to upgrade it from the 100 500 MHz version to the 1 GHz version I've got here.

this is the board they will swap. Note out the processor board would still all remain exactly the same, but uh, they just SWA this board over not a huge amount of COs in this thing I was expecting basically a a similar analog or RF uh section to the Mdo 4000. but it's nothing like it. just this tiny can over here I Doubt there's anything on the flip side of this.

it's just going to be the front panel, the keyboard, and stuff like that We will go further to, uh, try and find out, but they've really completely re-engineered this by the looks of it and built it down to cost because the MD 4000 was. you know, a ridiculously expensive scope. Uh, you know, incredibly. Uh, you know, Versatile, but very, very expensive.

They've done some serious re-engineering to get the cost down for this Mdo 3000. That's why they can afford to include the RF Spectrum analyzer onto. You know, the base model 100 MHz dual Channel scope. That's it.

And they software limit the RF bandwidth. So this is going to be a 3 GHz RF Spectrum analyzer regardless of whether or not you buy the 100 the cheapest 100 mahz model. But granted, as I said, right at the start, the 100 MHz model duel channel is uh, $3,300 or there about starts from that price and that only includes a software limited 100 MHz front end. but the full 3 GHz front end is there.

But obviously it's not your traditional Spectrum analyzer. There's nothing in there now. I'll take this board out as well, of course. but um, there is not going to be any analog to digital converter on this board because when you start talking, you know 5 gig samples.

uh, per second ADC right? You need some serious heat sinking. That's what that board over there is, right? These are still going to be your Adcs sharing two channels there. But the difference? the MD 4000 had them directly coupled on this board. So the only thing that can be happening here is they' got the differential analog signal from the buffer here coming out.

it's probably this Trace here. Look, you can see it's snaking its way up there. They've got some length matching on that, going through something there and going into this connector. So they got the analog signal running through this high-speed connector over here.

Nothing inherently wrong with that, you're just driving at a little bit more distance. These connectors are, really, you know, expensive, and really well engineered for that sort of thing. And of course, they have alternate grounds and pins and sorts of uh uh, stuff going on in there to, to, um, ensure that there's no cross talk and stuff like that so you know it's not bad. But yeah, they're running the analog connectors over that board to board analog signals for all four channels over that thing.

Yeah, look, they're obviously matching the length. You can see this one over here. This pair goes all the way over there and they go. Oh, we have to.

This one is physically closer so we have to match the length of that pair because you can't just have different length pairs. So the propagation layer screwed. So that's why they have to whoop switch it back there. So the length of that Trace would be precisely matched to the length of that one and this one here as well.

You can see it. Ah here it is running in here for this channel three I guess it is and snaking up there all around there, down there and back up and into there Woo! And the other fascinating thing is look, we found our display connector. There it is there. There's a ribbon cable going off to the display and all it is is going through a buffer over here coming from this main board to board interconnect which goes up to the main processor board.

Wow, they haven't even they''re going via this power supply board here. Why? So coming back to our main processor board, here here is our board to board interconnect. This is the only thing that the front panel uh, display and keyboard has to go through and unfortunately I can't see the traces on either side of those. So they're obviously running in the middle layers of this board somewhere now where they're actually running to.

My guess would be that they're running over to this Tech Asic over here and then that as I said is, the main applications processor is mapping uh, any required user interface data over this main Tech Asic which is then driving the display exactly how it works in the Uh Agilant. uh X Series Scopes But that's only guess I do stand to be corrected on that. It could very well be coupled directly into the applications processor. But as I said, that's not going to handle a couple of hundred thousand uh, waveform.

All your waveform update processing is going to be done in this main. Tech Asic All your DPO stuff and things like that is going to definitely going to be done in here. and whether or not they're actually transferring that and just updating a slower rate to the screen, they could certainly well be. They could well certainly be actually doing that because the there is a there is a big difference between the waveform update rate I.E the processing waveform update rate of this thing I think is like 250,000 waveforms updates per second.

There's a big difference between that that doesn't necessarily mean that, uh, that information is being updated on the screen. Of course, at 250,000 waveform updates per second, it's not. It's being done in memory and then a slower version of that, A slower C of that information is being transferred to the display. so it could certainly all.

the Dplo is certainly happening in here, but it could be funneled and all the processing for the display is done via the applications processor. Actually, that might make more sense cuz this does have a big Uh display processing Uh engine in it and Driver as well. And it does look like inside this thing, they've got one huge PCB on the front. Here you can see this matte black uh PCB material here with the pins for the Uh intelligent probes which actually plug onto this thing.

I Think there's one big matte black board covering this whole front panel and you can see that it extends up here to these plug-in application modules as well with the contacts directly on the board. And by the way, for those wondering, the Spectrum analyzer over here, this is literally going to be just the front end Uh amplifier and attenuators and everything else for the Spectrum analyzer. After that, it's purely digital and I believe I Read somewhere that uh it's doing because this has a very wide capture bandwidth. by the way.

1 GHz Wide capture bandwidth. Absolutely fantastic. If you uh, need to uh resolve U and capture and debug different signals over an entire you know you might have something operating at 400 MHz and you have something else operating at 800. This is a way to really capture the whole bandwidth at once and see time correlation between the two signals.

Now speaking of which, the time correlation for this this this Mdo 3000 is not the same as the Mdo 4000. They've deliberately cut this the performance in this thing down completely so that you cannot do simultaneous Uh time sampling and correlation between the Spectrum analyzer and between the analog inputs as well. And that's an incredible shame. And one of the big disadvantages of this thing is that you know one of the huge advantages of the MD 4000 was that you could do that fantastic magical time correlation between your analog signals and your RF signals and your digital signals as well.

hence the mixed domain name. So this thing almost doesn't deserve the mixed domain name because if you can't do time correlation and Sample at the same time between the RF Channel and the analog inputs, well, what's the point? It's just a tacked on Spectrum analyzer in it in a box. Anyway, enough ranting about that. Uh, yeah, this is just an RF front end.

That's it. I Read somewhere that it's doing uh, it's sampling this. The 3 GHz bandwidth has being sampled at 10 gig samples per second. so I reckon they're tight.

Well, obviously the analog Uh channels are being sampled at 5 gig samples per second. They obviously tieing at least two of those, maybe even four of those together to sample the RF front end. And hence why you can't do time correlation between the Uh and Sample at the same time between the RF Channel and the analog channels. Because well, to when it's sampling the RF, it's working as a spectrom metalizer sampling that RF front end.

It needs all the power of those Adc's which are over here on this board to actually do that and then it's got no resources left over for the analog channels at the same time. So you swe you either use this as a spectrum analyzer or as a regular time domain oscilloscope. Aha, what I've done is I've undone four uh um, machine screws down in here or three actually and looks like we might be able to lift this out independently of the rest of the shazzy. So let me carefully lift this sucker out.

Yeah yeah, she's coming out. Oh look at that. That is a beautiful design. Aha, tons more stuff on the back.

so I'm actually quite impressed by the design of this thing, the modularity and the way that you can, uh, service it and or assemble the thing. Fantastic! I Was afraid that I'd have to take out this entire shazzy. I actually started to take out the main self Tapper screws here which go into the front panel plastic. I Thought all that would have to lift out but it doesn't I just lifted out this RF board here.

the RF and analog front end board. Fantastic. Obviously because it is designed, they put some thought in this designed for that upgradeability because when you as I said if you send your scope back and you want decide, you want to pay the extra and get the 1 GHz version. Well they have have to make the upgrade of this board really easy and they've done so.

Fantastic! and you can see I think I was right about the front panel connector. Maybe you can't see it due to the exposure of the camera, but there is one big matte black board going right on the front panel. I may not actually get that far in the tear down cuz there's nothing interesting on there and that's it. Check it out! Look, they haven't even bothered to Shield the the front side put a can on the front side of this R front end.

it is so Inc incredibly simple compared to the Mdo 4000 which was more traditional uh, you know Spectrum analyzer sort of. uh Construction and design as well. We'll take a look, try and see what that little puppy in there is. Obviously, look at all the V stitching around here.

they're obviously getting a super low well in fact all the way around here, obviously getting super low inductance. uh, ground path there. but yeah, they're not bothering to Shield the top like they did. There you go.

They got the can on the front side there so we'll try and open that up and get a look at the RF front end. but that is all there is to it. Unbelievable. Looks like some sort of uh, little processor over here just controlling the front end now.

I am sorry, but I will not be able to show you the 1 GHz analog front end because these solder connections here here and over There are the um. metal shielding cans over the 1 gig front end. So this is the back side of it here, so the obviously got a chip. There we go.

It's like some sort of heat sunk Qfn package something like that. so they got a bit of heat sinking on the bottom side of there. So all of your good stuff for the RF front end is on the other side of the board. eh, sorry about that.

Some hatte C4 Logic here by old School 77400 can't Pete it? You got to have that somewhere. There's one of those per Channel and further on up each channel three six pin so 23s. Not sure what they're doing. probably regulation by the huge traces uh, going in and out like that with the huge Fers and the decoupling.

So they're probably uh, local voltage regulation for the front end and we have ourselves a Maxim Max 9601 and that is no surprises. A high-speed pickle comparator for the triggering, obviously the triggering. uh uh. front end.

So there you go. Um, that's the proper analog triggering section. 500 p a seconds. you know, really, really kickass uh.

comparator there and I don't know what that one there is doing? Uh 441 L AA I don't know you could try and uh, figure that out. don't even know that symbol off hand. But anyway, one of those per channel. So we're basically looking at a duplication of all that circuitry all there across all four channels.

and aha, backside of that. RS Spectrum analyzer. No real surprises. it's another ADF 4360 Exactly the same uh part we saw on the main uh logic board that is the synthesizer and Vco to generate the sampling clock required.

and curiosity is going to get the better of me. So I'm going to lift up that label, lift up the skirt, see what that thing is? Of course it's a dead giveaway. It's some sort of processor cuz anyone that has a label like this, this means it's been flashed and has code in it. Yeah, there you go little uh, free scale.

Hs8 8bit uh processor free scale of course. Well, they use a free scale main application processor, so hey, might as well use the same manufacturer for your little 8bit micro. Now these tracers are interesting. Look at these.

all these uh, highspeed differential pairs all going over to this connector on this far side of the board. I Thought that' all be popping out that middle one, but they not. There's a ton going over to here. Now here's where we look at the systems engineering and see where all this stuff is.

Flowing To figure out what's going on, We we know we've got these highspeed differential pairs, one coming from each Channel probably even from the RF Uh Channel over here. So there's a whole bunch of them going to this connector over here. And where's that one connected? Well, If we have a look at our main board here, it's this connector over here. What is that connector near Aha, that Main Tectronics Asic over here.

here's the other one. So as I said, all of the analog channels would be going through this connector through to Adc1 and Adc2 here, which handle all four channels. but all but then they' got a whole bunch of other pairs coming over here coming out of if you have a look where they're coming out of that chip. we had a look at down in there, which is your highspeed comparator for your trigger.

So these are actually the triggering outputs from each. Channel Going through that connector and that connector pops up here. There it is. It's mounted on the back side of that there, coupled into this main tectronics.

ASC over here. So this is handling all of the triggering. but of course, if that's handling all of your triggering, then what is this third device over here? I You know these are obviously your main um, Adc's and Acquisition As6. As I said, one per two channels there? Well, what's that third common device doing between all four channels? If it's not doing the triggering? Check out this.

Someone was thinking when they laid out this board. look at this. They have put, uh, notes to whoever was designing the case or the system. You know, the whole putting the whole thing together.

The person who laid out the board went, oh, okay. it's important we don't short out the positive tab of this battery to anything cuz it's a through hole part. If we accidentally short that pin out to some metal or if it's very, very close, that could you know in production or something like that with tolerances short out to one of the metal cases or something like that, then well, it's going to ruin your day. So they've put a radius around there where we must cover.

or whoever designs the next part of the system must cover that pin there and look even. May cover. Hey, if you want to Guild the Lily let's May cover that part and they've followed that up with it on the bottom just in case you missed it. Look at that.

Brilliant. And of course we can't let this go without having a peak in the RF section. I've undone three screws there. Yes, we have our gasket.

look at that completely separated. You can see the signal path how there's the RF connector there. It's flowing up through here, flowing up around there and out, well there somewhere. It's popping out on the top side.

so let's take that aside and flip our board over and see what we have. Hey, not much there we go. No solder mask on there. That's pretty typical, but there's hardly anything in that front end at all.

and as you can see, it follows the path directly of what that was covered with our metal diecast top on there. And that's very typical of these. Spectrum analyzers. Con construction, all controlled impedance of course.

All you know, very critically. Uh. designed. We've got some length matching around there, and uh, you know, basically when you're talking in the gigahertz range, everything on your PCB becomes an A component.

I've talked at this before extensively on previous videos and how you can actually distributed element filters manufactured out of your PCB bandwidth. uh, you know, low Pass High Pass bandwidth filters manufactured magically into the tracers of your PCB Doesn't seem to be a huge amount of that going on here or any of that going on here. Really all I see is our there's our Um input, there's our AC coupling and then we've got some filter in there on the input we'll have a look at uh, what that is, that's our input amp and then uh yeah, we got some more in series termination here. Some more AC coupling going on.

We'll have a look at that, but some more filtering happening around there. Little Transformer couple little Balon sorry uh going there and well, not a huge amount else. Now if you're concerned about those solder joints I Know they look ugly, but that's what happens with a high when a high thermal Mass component meets leadfree solder. It just.

it looks pretty horrid like that. But no, there's nothing actually inherently wrong with that. And smack on the input here, no surprises for finding that we've got an attenuated chip. In this case, it's a Hitte brand Hitte Microwave Corporation HMC 624 Lp4.

It's a digital attenuator. I.E By digital, it means that it's got a digital serial interface serial or parallel uh interface controlling the attenuation in there. So yeah, we basically just got our AC coupled input little bit of filtering happening there and go straight to the attenuator which gives your input all your input attenuation ranges and as always, I will link in all the data sheets down below so you can go and check these out for yourself and follow through. And what do you need after an attenuator? Well, you need your amplifier of course.

And yeah, once again, we've got a Hite Corporation Hey, if you're going to get all your stuff, you may as well get it from the one company, especially the company who specializes in these sort of chipsets for these sort of RF front ends. and this is is a DC to 6 GHz gain Block It's the HMC 311 and for all you RF fans, it is a hetero Junction bipolar transistor amplifier. a Hbt with 142 DB of gain. Fantastic! And you might notice if you look very, very carefully It might be a bit puzzling that there's no other pins connected except the input and the output pin.

How does that work? How does a chip like this work? If there is no Power Pin at all? In fact, this is true. Take a look at the data sheet here. there is only an input and an output pin. There's no power pin.

How does it work? Well, Easy read further down in the data sheet and you'll see that the power is actually coupled in via the output here via this inductor. Here there we go. so it's coupled into the output poers the ship internally. This is very common for these sorts of RF amplifiers and you'll find these in all sorts of things.

Like you might be familiar with the your TV M head amplifier. for example, it'll be a similar thing. You feed in your voltage down at your you know at your TV Outlet you actually feed it up the coax and it's powered from the output. Exactly the same thing happened in here.

Oops, and I forgot before and after that little uh, eight pin packages here labeled 360 I don't know what they are. There's no equivalent, you know, HMC 360 from HTE or something like that. But basically based on the operation of this thing, there we go. There's our input attenuator, our digital attenuator from our RF front end.

So our RF comes in goes through our digital attenuator and obviously you can see that there's a bypass Trace here. of course, controlled impedance Trace Going over and bypassing our uh amplif fire our game block there. So obviously these, uh, little RF switches that, uh, digital line there would switch between. well, switch to bypass, uh, completely our RF amplifier there.

And then we've got our signal coming out of our filter here, going up to what looks like a little Transformer and that will be a Balon of course cuz at this stage, we're still operating single-ended It is not a differential, uh, pair. it is a single-ended reference to the ground. You can see all the via stitching around here, a beautiful low impedance, and you'll see that out of the other side of the Balon here. Bingo It comes out differential so we're no longer referenced to ground at this point.

but curiously, then it just pops out, does some extra. It looks like we got some load match in there and looks like no, there was supposed to be like some filtering happening there, but I I Don't know. And then we've got some extra delay introduced by this Trace around here. and then they're basic, basically just joining up again single-ended over at this side.

so not sure why they've actually done that. and this is where it ends. and well, curiously, well, you've got to think, well, a spectrum analyzer. It's got to have that uh, Vco, got to have that local oscillator.

and here it is. It's a once again her Tite HMC 429 and 4.45 to 5 GHz Vco and that's all it does. it just basically voltage input. It can generate a Uh frequency out within that range.

So at this point I started to get rather confused actually because well, we've got ourselves that Vco there and anyone who knows the basic operation of a traditional Spectrum analyzer which you've seen in many of my previous videos. Once you see well they have a Vco like that and the Vco feeds into a mixer along with the RF signal goes into the other side of the mixer and bingo out the other side pops you inter immediate frequency if signal which then goes on to further pass down in the Uh Spectrum analyzer and well we've got a Vco here. So I thought well okay there's going to be your traditional mixer in this thing, but well it just didn't make sense. I mean where is the mixer in those two little uh White four pin white things there and and a couple of Uh tracers with some termination? It just didn't make sense.

So I contacted uh fellow blogger Alan Waly. If you haven't subscribed to his YouTube channel, you certainly should who actually is an applications engineer at Tectronics and he wasn't quite sure either exactly how this new Uh product actually worked, but he had a couple of ideas and well, it started to make some more sense. Now there's one thing which we both knew was that uh, the Tectronics Mdo technology actually samples the RF the entire RF Spectrum in in this case, uh, at 10 GHz So obviously it combines the uh, traditional ad sees in the oscilloscope and uses those to do the sampling directly on the RS Spectrum. But hey, we had this this Vco in here right acting as a local oscillator.

And when you've got a local oscillator in a spectrum analyzer, it implies that it feeds into a mixer and your RF comes in and you get that intermediate frequency out. So you know. We didn't quite know what was going on, but Alan knew that there was a technique um, where you dither the where you inject a dithered RF signal into the RF signal you're trying to measure and then sample that and that would increase your performance uh of your RF front end and Bingo! he pulled up a um, rather obscure I think um, application note or white paper from Tectronics that explains exactly how the Mdo RF uh, Spectrum technology works. And this was the other thing that was puzzling me: I didn't quite understand how they were getting the performance out of this spectrum analyzer when they were using just the regular ADC in the oscilloscope.

Sure, they're sampling at a high sample rate. In this case, they're combining all the Adcs together, presumably to give that uh 10 gig sample per second. uh sample rate for our 3 GHz RF bandwidth. So there sampling the RF directly doing that.

But hey, what's the difference between just a regular Uh oscilloscope with an Fft function and this Tectronics Mdo technology? You know, like yeah, you can do some Box Car averaging like your uh familiar with with the highres mode and stuff like that, but we didn't have the extra bandwidth there. You can do some other averaging and things like that, but it just didn't make sense. I mean an 8bit converter at Best is only going to get you know around about that Uh 50 DB dynamic range. And well, this spectrum analyzer has.

you know it's got like 100 DB in the order of 100 DB dynamic range. So how do they get that increased performance using an 8bit converter? And here's the white paper I was referring to Now I will link this in down below. So if you want to read all the Gory details yourself, you certainly can. And well, here it is.

Learn how the Uh Mdo series Scopes including the 3,000 This is for the 4,000, but it will apply to the 3,000 with uh uh, you know, the subtraction of something we'll show in a minute. Uh, they're able to leverage existing oscilloscope acquisition technology to achieve spectral Fidelity on par with entrylevel Spectrum analyzers. So the design techniques used in the Mdo series allow them to achieve that Fidelity far in excess of that provided by the typical Fft feature found in other oscilloscopes. And that is the key point.

The key point of difference, even though they're using exactly the same 8bit ADC Now let's take a look at the basic block diagram here and it's Um, this is for the 4,000 So the 3,000 In fact, they're actually missing something out of here, which we'll show in a minute as well. even on the Uh 4,000 Which what? What we're doing? Basically, the RF input can is just here, which is what we've seen. Well, at this point, that's all it is. Input attenuator, some gain, some you know, uh, some low pass filtering and stuff like that.

but it's basically just an RF front-end amplifier and it's going into what's called a block down converter because um, in the Mdo 4000 series. Well, they're trying to get up to 6 GHz bandwidth. but we don't need that in the Mdo 3000 bandwidth because we're only going to three gigahertz here. So we're just operating within that one block.

so we don't need a block uh, frequency down converter. so that is gone out of the Mdo 3000 and then, well, yeah, a trigger signal comes out of that and it just goes into an analog digital converter. In this case, it'll be all of the A Adcs that are regularly used on the analog channels. They tie them all together so they can get 10 gig samples per second.

so massively High sample rate. And as it says, uh, further on here we go: a bandwidth in excess of 5 GHz So the actual ad itself has an input bandwidth up to 5 gig. the ADC hybrid or whatever ASC developed by Tectronics That massive 5 gig bandwidth so can they can easily get a 3 gig input Spectrum analyzer. They can sample the RF signal directly so you know none of your traditional um, uh, Spectrum analyzer type stuff.

It's basically just operating like a regular scope. That's pretty much all all it is except you, you know you got got some nicer RF sort of front end with your 50 ohm load and your attenuator and your gain and you know, some filtering and that sort of stuff. But apart from that, it's just like another analog Channel really. And as such, you can forget about all the rest because well, yeah, it does some digital down conversion and stuff like that and some discrete fuor transform which they don't show in here.

but that's all just your digital processing inside the thing. It's basically RF straight into your analog to digital converter except for something that. but how does it do that? Exactly? How does it get like a 100 DB of dynamic uh range instead of 50 DB you'd expect with your regular 8bit analog to digital converter? Well, it's not using averaging, it's basically using two different, well, three different things effectively. But processing gain is one of the main ones which is all done in the digital uh domain.

I Won't bore you with the formulas and stuff like that, but it's basically saying what I've been saying is that a regular 8bit converter only gives you 50 DB Um, Noise floor? That's you know, pretty much it. But whereafter in a spectrum analyzer like a you know, 100 DB for a you know, a half decent entry level Spectrum analyzer. So that's what we're getting. still with that 8bit ADC But what they can do is they can do some digital down conversion and some uh discret Furia transforms in the software after the ADC So this is all software processing and that gives you what's called process gain which effectively lowers your noise flow.

It's over a smaller bandwidth and it can calculate and do all that in digital. So even with an 8bit ADC you can get a noise floor down at, you know, minus 100 DB Fantastic magic. And there you go. They give you an example of a 10 MHz span for example, and a resolution bandwidth of 10 mahz uh with the 10 gig sample per second.

Uh, Mdo 4000 and Mdo 3000 samples at the same rate in this respect. Uh, they are the same instrument working the same way so that improves the signal to noise ratio by roughly 57 DB Massive. So they get 107 DB noise floor out of an 8bit converter? Unbelievable. And they're go on to say that a traditional Uh Spectrum analyzer like modern ones.

They're also doing digital ADC sampling. you know very few of them are you know, really true analog anymore, but they're very low sample rate so if they're sampling at like 20 meg samples per second, it requ Rees a 12.5 bit ADC to get the same signal to noise ratio as they can use an 8bit ADC in the Mdo series. Scopes And of course, hey, you're already paying for that 8bit ADC inside your oscilloscope. so why not use it? It's very clever, but there's one more trick up there sleeve which confused Us in the tear down because there was that uh, local oscillator in there.

that Vco chip. what the hell was that doing in there? Well it was adding uh dither in into the RF signal before the ADC to increase the spous free dynamic range and they explain exactly how that works here. I won't bore you with all the details, but basically um, it does this in the Mdo 4000 as well, but they don't show it on the blog diagram. but in the Mdo 3000 it's much simpler because there's no block down conversion.

There's no IF frequency or mix or anything like that. the RF just goes straight into the ADC Well, but they add in or they mix in a dither signal like that. and basically a dither signal is just a random noise. And of course, the Vco.

The output frequency changes based on the input voltage here, so you put a a random uh, you know signal on there. you're going to adjust your frequency or dither your frequency that is added onto there and that reduces your spirous free dynamic range by spreading it over your entire range of your A or a wider range of your ADC. But yeah, you can read the white paper if you want to know how it works. So there's no block down converter in the Mdo 3000.

it's just the RF straight in. and then they add the dither signal on here, straight in the ADC. Now, the dither signal has to be just below somewhere just below the Uh Nyquest sampling frequ Nyquest uh frequency. Of course, we've got 10 gig samples per second here.

So uh, five. It has to be somewhere under 5 gig. And of course, as we saw in the data sheet, this HMC 429 Vco is designed for just under 5 gig. Perfect chip to add some dithering into there.

so that's all they got. So that's the only cost that they're adding to this thing. Is this Vco? just some RF attenuating and some switching and some amplification? That's it. Bang straight into the existing analog to digital converter in that that you're using for the analog channels and Bobs your uncle.

So the result of all this uh, you know, RF and digital processing magic is that they're able to leverage the existing analog, the digital converter already in the scope, and to add some just some lowcost RF input parts that, as I said, only cost like you know, 2530 bucks tops and bingo, you got yourself a full 3 gig bandwidth Spectrum Analyzer that uh has a massive capture bandwidth over the whole effectively the whole range or 1 gig, uh, capture bandwidth I Think in the case of the MD 3000, it's absolutely incredible that you can get this performance. Of course, it is not as good a performance as a proper you know World design Spectrum analyzer with a um a better ADC in here of course, but hey, you know it, it matches the performance going to be similar to entry level Spectrum analyzers on the market and hey, that's good enough when you include in just some basic Spectrum analyzer functionality into a scope. It's brilliant. Unless they got this thing patented I Expect all the other manufacturers are going to start doing this with, you know, in the not too distant future.

But I Suspect: As with a lot of things, it's probably not as simplistic as a look. You can't just whack it in there and do a couple of routines and Bobs your uncle, you got yourself a spectrum analyzer. It's probably a bit of secret source and some you know tongue angle, you know Gray beard tweaking that goes on inside this thing to actually get it to work really well, but still. hey, you can get that spectr analizer for a low cost.

Ah, it's brilliant. all right. I haven't put it fully back together yet, but I'm curious to know what happens when you power it up a without that Communications with the case open like this and uh, without the comm's board plugged into it and also I want to see what that mysterious I think finger touch switch does so I don't know. let's give it a go.

Only one way to find out. And here we go. Wh? we have some lights? No yes, it's booting up I didn't have to touch the uh power switch which is a soft power switch by the way. T Come on, two little thumbs, see how long it takes to boot up quite a lot? I think it takes like 30 seconds or something I Think it's on par with the Agilant Haven't actually timed it yet and it's getting there.

It's getting there. It's getting there. It's getting there. Not yet.

Oh dear. there we go. We're on power on self Test failed. If the error persists after power cycling, refer to the oscilloscope to qualified repair personnel.

There you go, you can proceed. If it's possible, Proceed Menu off. However, this Ccope may not perform. There we go.

Menu off. Warning: The oscilloscope is not compensated Tada But it has booted up and if we push our RF button over here, there we go. our RF is sampling. So let's um I found a bug in this.

By the way, check it out. If you press the RF button again to turn it off, you would think it would go back to your digital channels. But watch this, it doesn't. It just freezes dumb.

Oh well, it hasn't actually locked up. but uh. anyway, just press that and you're back. So RF takes you into that and then going back.

You know you can just press that and you're back in H. Just a silly little thing now. I'm curious to know what happens when you touch that magic finger sensor. Only one way to find out.

Nothing Bummer. Whoa. Look at that warning. Internal temperature is approaching a point where the oscilloscope could be damaged.

If you continue, check that the fans are operating and there's sufficient clearance. blah blah blah yeah. Tech Scopes They run hot. All right.

Well, let's do a quick, uncalibrated temperature of this thing with my Flur E8 Look at that. I mean I don't want to leave this thing here for too long. but uh yeah. I mean well, there silver heat s so.

I may not have the emissivity set correct and all that sort of stuff. but we're we're talking. You know, 50. at least those heatings.

There's a real hot spot up in there 50 on the other side of the board there. so I'm not sure what's uh, what's going on there. but anyway, this thing it I think it did actually automatically switch off. so uh yeah, it's definitely got some sort of internal uh, temperature measurement and it does not like being operated with the case off, that's for sure.

I mean it was barely up for like a minute before 2 minutes before that warning came on. and yep, there we go. I Just pressed stop and it switched itself off. So yeah, well, at least it won't damage itself, hopefully.

And of course, system airflow wise. that's why they've got the fan in this position and that's why they've got it blowing inwards. The air is sucked in from the outside here from this, uh, vent on the outside, sucked in directly over Wham right on top of the three main heat s here across the board and then out the other side up there. and no, the fan is not very loud in this thing.

For those who are concerned about that thing, it looks like it does have variable speed on the fan cuz you can hear it. sort of were