I have to say that my initial reaction to the membrane switches was negative - but I say now, that I have used it, a lot of effort went into the material selection and the membrane buttons have a great tactile feel. Definitely designed for the industrial market - completely different league than consumer grade membrane buttons.
Alan is biased though, of course he likes it! ;-) j/k If there is someone else that is going to pick this thing apart, Alan would, even IF he is on the payroll! Looks like a darn good bit of kit TBH. Though metal inserts into the plastic on the backside would have [should be] a part of the design. Penny pinching on a Tek product should be sacrilege.
Ooo this having a VESA mount is handy. You could get a swivel desk mount and mount it at the perfect level on the bench and not take up bench space. Handy if you've got a compact bench.
Just wish the power plug had a back access, even if they had to have two jacks, you lose the ability to hide the cables off the side, it ruins the aesthetic and is inconvenient pragmatically because you can't put it up against anything on that side you have to have the area for the cable clear. The VESA mount is nice though since it's light enough it'll probably fit fairly well on some floating arm stands.
@@JanBruunAndersen on a scope in this price range even as it's "entry level" it's a design consideration that should have been addressed before a final product was released. This might be 'entry level' but they start at 1,800 dollars and what I've seen so far doesn't really give me great feelings about the amount of true effort they put in towards good usability engineering. One thing they did was used self tapping screws for the main screws for the case, and that's just not a good sign in my book. No one should have to be taking a Dremel to one of these things because they decided not to address something so basic as sensible cable management,
Immediate thinking about an pen driven interface. Start loving a full sticked TM5000 with modules and 50 knobs 40y ago ... now why not something radical different.
I’ll watch your video next and give a comment and thumb up :) you both give great and comprehensive reviews on this stuff so you both have earned all the views you can get.
Nice video. Also, the OS is running on the DDR4. The memory controller is on the (PS). Ultraram is used for the programmable logic (PL). OS is on programmable system (PS). eMMC is used for bootloader, PL bit file, OS … and is attached to the PS.
The 2 metal posts could be useful for the case to feel a bit more solid, when you pick it up. If feels really cheap and nasty when you buy a new bit of kit and the back is just all squishie
Thought on the BNC back Shield, is that the bumps are providing 2x functions, 1: reducing RF capacitance (it could have been creating an undesirable filtering effect if it was closer, and changing the characteristic impedance of the connection) from any protrusions of the BNC center pin, and 2: mechanical stiffening against the back plastic. Edit: Thought of another possibilityUL compliance/clearance for transients from voltage sources… just remembered of a student using DC coupling on a 240VAC transformer primary.
@@reinerfranke5436 Isn't it 1M || 13 pF for this scope? Also i came to around 400pF and 200pF with the bump. I assumed 1 cm² area and 2 mm distance without and 4mm with bump. Not sure this is even the correct approach here. Any RF experts here?
Looking very carefully, U11, CR3 and Y3 appear to be covered in a clear coating. U11 is certainly shinier and blacker than surrounding un-coated ICs. Possibly coated to prevent high humidity or airborne contaminants from upsetting the oscillator frequency (?) or flattening the battery.
At 31 min. They wouldn't be using the internal memory for the OS. Only ~1.5MiB. Likely using the external memory for the OS and the sample memory. The internal RAM might be used for FIFO buffers between the ADC and the external memory for storing the samples. The Zynq typically also runs form executable code in RAM. So it would load (copy) the binary for the processors from the FLASH into RAM and execute it from there. Possible otherwise, but I would say unlikely.
Not to mention that the PL (FPGA fabric) and the PS both use the same interface to the memory controller, and thus share the address space. They could split it up however they like!
@@RobertHancock1 Yes, but that is On Chip Memory (OCM) which is different to all of the memories shown in the videos. The memories mentioned here (except DDR, that is shared) are all PL based.
@@RobertHancock1 What kind of "on chip memory" do you mean? Boot ROM, caches or distributed memory blocks in the FPGA array? However, large sample buffers, framebuffer and the operating system must be outside the FPGA in the DDR4 chips. There is not too much space for the samples...
Soldered on coin cell battery is always a no no for me. Had to wait 3 months last year to get my hands on a BR-1225/HCN for a serial concentrator at work. Checking today I would be on a 6 week lead time. It's just not necessary those coin cell holders are not letting a battery come loose ever.
@@NavinF Unfortunately I did because the fecking solder tabs have different PCB layouts. You don't think I didn't look into that? I guess I could have botched it but I decided just to stick it back in the rack while I waited for the right part to become available. Panasonic have the same battery with more than one solder tab layout for crying out loud. I cannot for the life of me fathom why you would use a solder down coin cell over a battery holder and bog standard 3V coin cell. The battery will run down and then you need to breakout a soldering iron to replace it.
@@jonathanbuzzard6648 Ah I assumed there would be a de facto footprint and you could just bend the tabs to account for small differences. I agree that soldered batteries are asinine, especially in the sluggish test equipment industry. A mid end scope from 10 years ago is still considered a mid end scope today. No way the battery can last as long as the scope.
@@larzblast I probably could have bodged a button cell holder. I did consider getting a carrier PCB made. However in the end it's only a serial concentrator for emergency out of band management of all the network switches/routers etc. on an HPC system. That the RTC lost the time if the power was lost was not that important on a UPS and generator backed system. Yes it moaned once a day for three months in the logs that battery was flat but I could live with that. The issue is why the hell put a soldered on battery that is unlikely to last the lifespan of the device. It turns a job that would just require a screwdriver to something that requires extensive disassembly and then breaking out the desoldering tools. If it's going in a rocket then perhaps it is justified, in a fricking oscilloscope or serial concentrator no way is it *EVER* justified. The designer(s) needs taking out the back and given a jolly good thrashing with a clue stick.
I had no idea anyone else had one. When I got mine in April they were so rare that not even w2aew Alan Wolke who works at Tek had even seen the battery compartment add-on that I got given.
@@EEVblog Damn! I figured you probably just didn't know, or maybe there was something special about this one (I commented before finishing, and you know what they say, the quickest way to get an answer to a question is not to ask it, but to be wrong on the internet), but after watching them, nope, same. Obviously you each add your own unique styles and info, but yeah. Was there an embargo date or something for video release? I thought it was extremely odd (and kinda amusing to be honest) to see two videos doing a teardown on the same thing, one claiming to be a "world exclusive", literally right next to each other in my sub box. How could you have known? :)
LOVE these teardown videos! And I love this oscilloscope. For people like me who love having things in order and hate sacrificing desk space, this new scope format is amazing! Time to order one! :)
If they had the upper right 3 buttons real, I could probably forgive the rest being membrane. Those 3 buttons are used too much to deal with membrane switches. There's a reason they are presented as extra large.
Probably they are not using the whole bandwidth because of clock source jitter (probably a higher end model with cleaner clock sources can manage it). Clocks at that speeds needs precision to prevent noise produced by clock sample mismatch inside the ADC's. High speed ADC's are composite ADC's that split the phase in 4-8-16 parts to achieve that bandwidth, so any mismatch can cause phase misalignment distortion. Those gsps adc never provide good dc performance, so probably the input ASICs can provide some means of integrated ADC's to cover frequencies from DC to HF. This means also that signal is composed and reconstructed from spectrum inside the FPGA, to show you an equalized and properly timed signal (with lots of gains in signal quality). Believe it or not, the all-integrated signal processing needs less shielding, and JES204 interface needs less wiggly lines and less fancy pcb's, at cost of no-user perceivable latency. I would like a more "innovative" user interface (knobs at both sides with red tips like old tek), and real buttons. The rest is kind of OK. Old scopes are heavy and bulky
@@reinerfranke5436 that's sad. Probably that frequency has poor coverage. High sample rate adc's have pretty thin bandwidth. You safely can assume 20% of total range at both sides of fs/2. Composition at this rates is a tough task, and needs calibration at certain time intervals to avoid spurious free dynamic range degradation. And if different sources are used, different calibration procedures are needed. Clock sneaking into the main signal means that they don't have a way to filter it. A way to have less noticeable ebob degradation is to use more bits. 12 bit high speed adc are now off the shelf parts, and you can easily can get from dc to hf. (I.E ads4229) hope the can polish this scope design approach. I really like this part/cost reduction and focus in integrate design. This make the whole unit more reliable and can leverage flexible and high quality processing. It would be fun to see an oscilloscope using Xilinx Ultrascale RFsoc's
@@bigalejoshileno The NSC ADC is interleaved internal, 1% interleave channel mismatch give -40dB spur and some newer ADC are calbrated to better than the internal device mismatch, but it need finally a test source, so a factory test with the application, if calbration channel randomization could hide the spur below noise floor but it will still be part of the noise budget, similar to the jitter.
I don't think the PL has enough for 10Mega samples for four channels. That would be 40 Megabytes minimum. The largest Ultrascale, i.t.o. memory (XCVU190), has a mere 16MB. For buffering (FIFO) PL memory is probably being used with a DMA to DDR.
That power input connector is the same as used on a lot of reciept printers for 24VDC power, so readily available if it breaks. EDIT: might be incorrect there as I believe the reciept printer connector is a 3 pin (i'm not at home so can't check my epson TM88), but the 4 pin was/is still used on a lot of things so still easily available. I had a cheapo IDE NAS that used that connector for +12/5V and I believe my SCSI JAZ drive is the same connector. I knew I'd seen it before somewhere.
Why haven't Tek chosen to power the oscilloscope using USB-C which is becoming standard on laptops, notebooks, tablets and phones, and just mandated by EU? It is powered with 24V and rated 60W maximum.
@Christopher Grant I dunno, so far all my USB-C connectors and cables are outlasting the others. I've only had 1 cable go bad thusfar. I've probably got 10 of em. Higher current is a huge win. Zero mechanical failures and zero dirt/dust/grime issues so far.. knock on wood right ! High end Pro audio equipment is using them. These are not consumer grade products. However, I don't think power & data should be mixed. I've seen noise issues arising from mixing data & power. Everything evolves, maybe noise will be solved one day
Safety earth bonding is a requirement. It must be guaranteed through the connector or you have to make the customer earth bond the scope some other way which can be a challenge. If you use a standard interface like USB-C the customer may use any USB-C power supply which may or may not be safety earth bonded and thus pose a safety risk.
Looking at the rear panel, there would be ample space for a decent capacity battery pack inside the enclosure. It goes to show that Tek do not have a lot of experience in compact (phone, tablet-, notebook) design, with the amount of air inside.
4channel version is around 2400euro on tektronix site. Way too expensive to get no powered probes, auto detect. Horrible interface and membrane buttons. How about Wi-Fi, usb mouse and keyboard, remote control? I don’t know, it just looks like an expensive toy and not a proper instrument.
Wifi would be reason to ignore the product. When I measure somethink I don't want to have unnecessary radio source on the desk. There's gig LAN. Wire is wire, you know...
@@EEVblog Yes, but only on certain boards. Analog front ends and acquisition boards are component-level repaired, but encoder boards, displays, certain power supplies, etc. are discretely replaced. The worst part of the component repair is supply chain issues. Many parts have to come by ship from China and new products generally take quite a while to have parts stocked adequately. Happy to answer any other questions you have.
@@jalalhobbs Haha, yes, but probably not like what you’re imagining. Two person crew mostly makes ready-to-eat meals that are refrigerated in company stores across the buildings. I can only speak for service/mfg in building 19. I’d love to see what the cafeterias looked like before the corporate buyouts.
That’s one of the lowest-end parts in the Zynq Ultrascale+ line. Tek are probably paying something like $50 per unit or less depending on the volume. Part of the external RAM will be being used for the ARM system and part of it for the programmable logic fabric.
If we still did component level repairs on the regular, I’d love a couple of these on VESA mount arms above our workstations. But for as often as we’d need one anymore, our old Agilent will do the business just fine.
The DDR memory is probably mostly the OS code, you'll not get a Linux kernel for arm64 in under 16MByte alone, and at-least another 16MByte for even a small system... they may well be using most of the DDR for their software.
My only question is why hasn’t this been a thing for years? So many scopes you’ve torn down are empty space. Dunno why this hasn’t been a thing up till now.
Maybe the 'thunky car door' marketing ? The first depth reduction, losing the CRT, made useful extra bench space. Losing another couple of inches adds a certain amount if utility but also reduces the physical stability and apparent quality. Doubtless we'l now see a race to the slimmest, laptop-like.
10:14 If someone is more experience in RF stuff I have a question about the BNC/Frontend shielding. Could the bump in the metal shielding be there so the GND/Shield is not so close to the BNC center pin? Would this influence the impedance?
The bumps on the input shield probably just contact against the back cover and press down on it to keep it in place... And a Tag-Connect TC-2050 footprint (J2)!
perhaps those pins are to make the panel a bit firmer when you're pressing buttons etc. especially if it's being used as a portable scope, it would make it feel a bit more rigid if you were pressing the buttons with thumb force
No, they only serve to align the pcb, as they facilitate the assembly of the pcb given the amount of protrusions that have to come out of the front panel.
Nice tear down. A perfect example of modern product design where the emphasis has shifted to ease of manufacture without too much in the way of a compromise. Tektronix continue to lead in build quality and design and still manage to incorporate a hint of their trademark blue colour in the handle. Whenever I see the Tektronix blue I immediately think of those wonderful 500 series ‘scopes from way back.
Just as a general note. Kudos to Tek for making the handle on the stand classic Tektronix blue. Really brings back memories. Wish we'd see more metal cases though.
19:17 Those Micron memory are likely shared between acquisition memory and application memory for driving the user interface. That ISSI chip is an eMMC for the Zynq to boot from.
Sounds like it uses the ultraram for sample memory, that external ram is obviously for the applicatiin processor, OS and framebuffer and secondary sample storage/functions
Not having had it in the hands, I think the metal pins are guiding pins for the battery holder back, so that it will ensure the mating power connector is aligned before it actually mate the electrical pins. This surely can be a precaution in case some I'll advised user forgets to take the batteries out before mounting the battery holder. But for that, of course, the back of the case cannot be snug fit around the pins and it must have some clearance around the pins, and the pins must protrude some bit from the back, (or if not from the back, protrude from some recess that may exist around the pins). Well... My guess is as good as anyone's.
Interesting to note the use of what appear to be National Semiconductor chips. I thought they were acquired by and absorbed into TI. Unless they've brought the brand back into use or something.
Seems like a lot of the time they don't bother to redo the branding on the chips after acquisitions like that. Still lots of NXP chips with Freescale logos on them for example.
They really could have made the battery option a lot thinner - just a few lipos and a BMS board. It would be similar to the second battery option on some Thinkpads.
Bump on a back shield around BNC pin is most likely for signal integrity or creepage reasons, instead of mechanical only. Must cover/may cover usually means epoxy coating on top, so at assembly they would know where epoxy should go/not go.
Could the raised bumps on the back of the low profile shield be there to keep a distance from the back of the connector pins......to avoid any coupling?
I don't know this for sure as i havent measured one but I would bet that the outer set of threaded holes on the back are also VESA compatible. If the holes in the middle are 100x100 I would guess the outer holes are the 300x200 compatible mount
Yes. Only look-a-like. I've owned and used the ato1104 for something like 4 years now and both love it and hate it. I hate the touch screen, but otherwise love the scope. I'm going to get sto1004 learnelectronics just reviewed. Did the sto1004 basically replace the sto1104c?
4:30 These pins are maybe for further design decisions. Maybe they will offer a version with internal battery other back cover. These pins are for alignment of the battery I think.
I am worried about the lack of a vesa metal bracket it does not weigh a lot but i would place this on a monitor arm and grab the side of it to move it around the workbench and over years they might pop out
the batteries look like film batteries I wonder if they're using a fairly stock battery that could be sourced from third party vendors. Like how infotainment lithium Sony batteries are so generic. If that's the case that would be a very good move you can have your high capacity originals and third party backups.
Alignment pins strike me as like a pc motherboard using the center standoff as a pin - getting the board lined up is pretty important because ALL the knobs/bncs would looks wrong if off by just a fraction of a mm. Rigid pins with a tight fit on the board ensures it can only go in one way and one precise alignment. Board edge clips can break or bend.
🤑The Series 2, now starting at $1,800. I'm over here trying to figure out what scope under $400 to get? I'm looking at the ds1202z-e and sds1202x-e. Sure would be nice if there were a comprehensive head-to-head of these two budget boxes.
Congratulations champions, you have just made an oscilloscope 24 times more expensive with less analysis capabilities. Great job yours. You should post on fashion channels and not on electronic ones
I would have to think the bumpouts on the shield around the back of the BNC is for clearance for voltage withstanding, or to reduce capacitance to the center pin or both.
The metal pins = when you have it mounted on the smaller VESA mount, the weight of the unit plus the wires connected to it, plus and yanking that might happen... The metal pins support the back plate from flexing under load.
Very well executed packaging. Not a fan of the separate battery back; just include it (maybe minus batteries). I’ll have to see those membrane switches in person; I have seen many worn through on machines. Hopefully the switch panel is replaceable and not expensive.
No, they only serve to align the pcb, as they facilitate the assembly of the pcb given the amount of protrusions that have to come out of the front panel.
31:33 That FPGA is configured with 64-bit wide DDR4 memory. Tek could have used a DDR4 SODIMM slot here instead of soldered chips to allow further upgrades. For example if you want to further expanding your memory depth you can just buy a Tek SODIMM module (the SPD EEPROM can contain memory timing configurations along with a license key.)
i really like the design... and the tought and detail thats gone into the BNC connectors is great.... ...but...IMO...the software lets it down a bit... the DDR memory is probably also being used for the OS RAM.. the eMMC would contain just the OS/Software itself...
External memory is too slow - most likely, it is used by the OS and the graphics accelerator. I'm assuming the memory for signal sampling is on the FPGA (12 meg)
@@HitoPrl DDR4 is very bursty in nature and has both latency and refresh issues for high speed sampling (look at the number of clock cycles needed for activation, RAS and CAS, plus the number of cycles needed for a page refresh). Very probably it’s program memory, the internal SRAM/BRAM on the ZYNQ is on the FPGA side of things, and is too small for a Linux build with GUI anyway.
They might be using distributed RAM (in LUTs) in addition to the dedicated internal SRAM to get additional fast internal sample memory. Or/also, they may have two separate DDR4 channels with one being used by the A53 cores and another being used for sample memory.
@@EEVblog the mistake is in thinking that there’s no connection between the FPGA and HPS parts of the chip. Data can be captured to BRAM, and as each block is completed it can be queued to be written to HPS RAM. The HPS can reserve blocks of RAM for FPGA use, and the latency of traversing the FPGA=>HPS and then HPS=>DDR4 interfaces would be hidden from the capture process. 10M sounds like an entirely arbitrary size though, and with little effort it could be increased or decreased, subject to how much of that 2GB of DDR4 is being used by the HPS for code/data.
the os is probably using a significant amount of the 2 gb of ram because for linux with a gui you need a minimum of 300mb or more so would expect at least 500mb of it to be used for that
Nothing runs "from eMMC" directly. eMMC works like a disk, content must be loaded to RAM to be executed. The DRAM4 is more than likely used for the application processors, not just for sample storage (samples may even be stored/buffered in FPGA internal RAM first). There isn't enough internal RAM to run all the modern software with all the graphics and its abstraction layers.
Wow. Scopes are just getting thinner and thinner. What is the big black connector(?) seen at 38:29. I think I might have missed where it was mentioned. Is that part of the logic analyzer part of the unit? You mentioned that feature when pointing to that general area of the board.
I have to say that my initial reaction to the membrane switches was negative - but I say now, that I have used it, a lot of effort went into the material selection and the membrane buttons have a great tactile feel. Definitely designed for the industrial market - completely different league than consumer grade membrane buttons.
membrane buttons on my 1983 yamaha dx7 still work great
@@cappnuwu4021 the DX7 was built to last also.......nice one!
@@cappnuwu4021 Lots and lots of DX7's dont work anymore due to the membrane switches!
@@dietervolke7811 Let's remember to check in on how Tektronix 2 series are doing in 2060 Haha!
Alan is biased though, of course he likes it! ;-) j/k If there is someone else that is going to pick this thing apart, Alan would, even IF he is on the payroll! Looks like a darn good bit of kit TBH. Though metal inserts into the plastic on the backside would have [should be] a part of the design. Penny pinching on a Tek product should be sacrilege.
Ooo this having a VESA mount is handy. You could get a swivel desk mount and mount it at the perfect level on the bench and not take up bench space. Handy if you've got a compact bench.
Just wish the power plug had a back access, even if they had to have two jacks, you lose the ability to hide the cables off the side, it ruins the aesthetic and is inconvenient pragmatically because you can't put it up against anything on that side you have to have the area for the cable clear. The VESA mount is nice though since it's light enough it'll probably fit fairly well on some floating arm stands.
@@JWH3 - seems to me like it would be a Dremmel job to cut a new hole for the power connector and move it.
Me i have a TDS640 Tel Scope !
@@JanBruunAndersen on a scope in this price range even as it's "entry level" it's a design consideration that should have been addressed before a final product was released. This might be 'entry level' but they start at 1,800 dollars and what I've seen so far doesn't really give me great feelings about the amount of true effort they put in towards good usability engineering.
One thing they did was used self tapping screws for the main screws for the case, and that's just not a good sign in my book.
No one should have to be taking a Dremel to one of these things because they decided not to address something so basic as sensible cable management,
What are the odds of both of us choosing a similar thumbnails independently?!
LOL, probably inevitable! Didn't know you had one too.
Spider-Man pointing meme
Immediate thinking about an pen driven interface. Start loving a full sticked TM5000 with modules and 50 knobs 40y ago ... now why not something radical different.
@@EEVblog "Didn't know you had one too.", surprise mf LOL
I’ll watch your video next and give a comment and thumb up :) you both give great and comprehensive reviews on this stuff so you both have earned all the views you can get.
Looking forward to the Siglent version of this for $500.
haha yeah
Or Micsig, they will release a 500 MHz model too according to their Chinese website
Nice video. Also, the OS is running on the DDR4. The memory controller is on the (PS). Ultraram is used for the programmable logic (PL). OS is on programmable system (PS). eMMC is used for bootloader, PL bit file, OS … and is attached to the PS.
The 2 metal posts could be useful for the case to feel a bit more solid, when you pick it up. If feels really cheap and nasty when you buy a new bit of kit and the back is just all squishie
@the fltsimbuff yes, just make it feel a bit more solid. 👍
@@Radio_Flakes could be cost cutting, but it could be swapped to plastic inserts to reduce cost. Im glad I dont have to build moulds. 🤣 👍👍
@@Radio_Flakes I bet thats why I stayed away from them and stuck to PCBs.
Thought on the BNC back Shield, is that the bumps are providing 2x functions, 1: reducing RF capacitance (it could have been creating an undesirable filtering effect if it was closer, and changing the characteristic impedance of the connection) from any protrusions of the BNC center pin, and 2: mechanical stiffening against the back plastic.
Edit: Thought of another possibilityUL compliance/clearance for transients from voltage sources… just remembered of a student using DC coupling on a 240VAC transformer primary.
My thoughts exactly - but you got there long before me!
1M || 15p some 50f stray is irrelevant
@@reinerfranke5436 Isn't it 1M || 13 pF for this scope? Also i came to around 400pF and 200pF with the bump. I assumed 1 cm² area and 2 mm distance without and 4mm with bump. Not sure this is even the correct approach here. Any RF experts here?
You are correct , look at the area around the centre pin to the plane ✈️ on the pcb
Yes, on 2nd thought I think it's CAT compliance clearance
The "MUST COVER" / "MAY COVER" is likely instructions for conformal coating, which has not been applied on this prototype device.
Does it follow the RFC2119 outlines? Ha.
Looking very carefully, U11, CR3 and Y3 appear to be covered in a clear coating. U11 is certainly shinier and blacker than surrounding un-coated ICs.
Possibly coated to prevent high humidity or airborne contaminants from upsetting the oscillator frequency (?) or flattening the battery.
At 31 min. They wouldn't be using the internal memory for the OS. Only ~1.5MiB. Likely using the external memory for the OS and the sample memory. The internal RAM might be used for FIFO buffers between the ADC and the external memory for storing the samples.
The Zynq typically also runs form executable code in RAM. So it would load (copy) the binary for the processors from the FLASH into RAM and execute it from there. Possible otherwise, but I would say unlikely.
Not to mention that the PL (FPGA fabric) and the PS both use the same interface to the memory controller, and thus share the address space. They could split it up however they like!
Usually the on chip memory is just used for initially loading the boot loader from flash before the external DDR is initialized.
@@RobertHancock1 Yes, but that is On Chip Memory (OCM) which is different to all of the memories shown in the videos. The memories mentioned here (except DDR, that is shared) are all PL based.
0:31:00 - You can refer to timestamps like this if need be :)
Format is: hours:minutes:seconds (hh:mm:ss) with a space after it.
@@RobertHancock1 What kind of "on chip memory" do you mean? Boot ROM, caches or distributed memory blocks in the FPGA array? However, large sample buffers, framebuffer and the operating system must be outside the FPGA in the DDR4 chips. There is not too much space for the samples...
Soldered on coin cell battery is always a no no for me. Had to wait 3 months last year to get my hands on a BR-1225/HCN for a serial concentrator at work. Checking today I would be on a 6 week lead time. It's just not necessary those coin cell holders are not letting a battery come loose ever.
Ehh you don’t have to replace it with the same part. Any 3V coin cell with solder tabs would have done the job.
@@NavinF Unfortunately I did because the fecking solder tabs have different PCB layouts. You don't think I didn't look into that? I guess I could have botched it but I decided just to stick it back in the rack while I waited for the right part to become available. Panasonic have the same battery with more than one solder tab layout for crying out loud. I cannot for the life of me fathom why you would use a solder down coin cell over a battery holder and bog standard 3V coin cell. The battery will run down and then you need to breakout a soldering iron to replace it.
@@jonathanbuzzard6648 Ah I assumed there would be a de facto footprint and you could just bend the tabs to account for small differences. I agree that soldered batteries are asinine, especially in the sluggish test equipment industry. A mid end scope from 10 years ago is still considered a mid end scope today. No way the battery can last as long as the scope.
With a bit of creativity, I'm sure you could bodge in a button cell holder.
@@larzblast I probably could have bodged a button cell holder. I did consider getting a carrier PCB made. However in the end it's only a serial concentrator for emergency out of band management of all the network switches/routers etc. on an HPC system. That the RTC lost the time if the power was lost was not that important on a UPS and generator backed system. Yes it moaned once a day for three months in the logs that battery was flat but I could live with that. The issue is why the hell put a soldered on battery that is unlikely to last the lifespan of the device. It turns a job that would just require a screwdriver to something that requires extensive disassembly and then breaking out the desoldering tools. If it's going in a rocket then perhaps it is justified, in a fricking oscilloscope or serial concentrator no way is it *EVER* justified. The designer(s) needs taking out the back and given a jolly good thrashing with a clue stick.
Surprised the lithium memory retention battery is not user changeable. I wonder if they will start using "super caps" instead of batteries some day.
The screws under the VESA mount screws on the label are for structural support integrating it to the front
Yes, you want some rigidity in the centre like that.
Most of the DDR4 will be dedicated to the ARM CPU. There's no way you can run the OS from the internal FPGA memory.
I am not sure if the "World Exclusive" thing is a joke, considering The Signal Path did a teardown too that released like 7 minutes after yours did.
I had no idea anyone else had one. When I got mine in April they were so rare that not even w2aew Alan Wolke who works at Tek had even seen the battery compartment add-on that I got given.
@@EEVblog Damn! I figured you probably just didn't know, or maybe there was something special about this one (I commented before finishing, and you know what they say, the quickest way to get an answer to a question is not to ask it, but to be wrong on the internet), but after watching them, nope, same. Obviously you each add your own unique styles and info, but yeah.
Was there an embargo date or something for video release?
I thought it was extremely odd (and kinda amusing to be honest) to see two videos doing a teardown on the same thing, one claiming to be a "world exclusive", literally right next to each other in my sub box. How could you have known? :)
@@chemputer There was definitely an embargo date since Dave had it since April.
LOVE these teardown videos! And I love this oscilloscope. For people like me who love having things in order and hate sacrificing desk space, this new scope format is amazing! Time to order one! :)
with all do respect, Its nice that the power input is on the side, It lets you wall mount it eye level. handy for smaller labs or saving desk space.
Yes, but I also would have liked to have seen a rear option.
If they had the upper right 3 buttons real, I could probably forgive the rest being membrane. Those 3 buttons are used too much to deal with membrane switches. There's a reason they are presented as extra large.
I had the same thought. Once i configure my scope settings those are the three buttons i use 90 percent of time.
The battery pack has double the PCBs as the main product... Somebody goofed IMHO :D
Probably they are not using the whole bandwidth because of clock source jitter (probably a higher end model with cleaner clock sources can manage it). Clocks at that speeds needs precision to prevent noise produced by clock sample mismatch inside the ADC's. High speed ADC's are composite ADC's that split the phase in 4-8-16 parts to achieve that bandwidth, so any mismatch can cause phase misalignment distortion. Those gsps adc never provide good dc performance, so probably the input ASICs can provide some means of integrated ADC's to cover frequencies from DC to HF. This means also that signal is composed and reconstructed from spectrum inside the FPGA, to show you an equalized and properly timed signal (with lots of gains in signal quality). Believe it or not, the all-integrated signal processing needs less shielding, and JES204 interface needs less wiggly lines and less fancy pcb's, at cost of no-user perceivable latency. I would like a more "innovative" user interface (knobs at both sides with red tips like old tek), and real buttons. The rest is kind of OK. Old scopes are heavy and bulky
Quite possible.
SignalsPaths Multicarrier test show a spur at 500MHz which is related to to CLK, not low but must be part of the 7.3 ENOB
@@reinerfranke5436 that's sad. Probably that frequency has poor coverage. High sample rate adc's have pretty thin bandwidth. You safely can assume 20% of total range at both sides of fs/2. Composition at this rates is a tough task, and needs calibration at certain time intervals to avoid spurious free dynamic range degradation. And if different sources are used, different calibration procedures are needed. Clock sneaking into the main signal means that they don't have a way to filter it.
A way to have less noticeable ebob degradation is to use more bits. 12 bit high speed adc are now off the shelf parts, and you can easily can get from dc to hf. (I.E ads4229) hope the can polish this scope design approach. I really like this part/cost reduction and focus in integrate design. This make the whole unit more reliable and can leverage flexible and high quality processing. It would be fun to see an oscilloscope using Xilinx Ultrascale RFsoc's
@@bigalejoshileno The NSC ADC is interleaved internal, 1% interleave channel mismatch give -40dB spur and some newer ADC are calbrated to better than the internal device mismatch, but it need finally a test source, so a factory test with the application, if calbration channel randomization could hide the spur below noise floor but it will still be part of the noise budget, similar to the jitter.
I think the main FPGA's internal memory is used for fast sampling and the external memory is used to run the program and maintain internal samples.
I don't think the PL has enough for 10Mega samples for four channels. That would be 40 Megabytes minimum. The largest Ultrascale, i.t.o. memory (XCVU190), has a mere 16MB.
For buffering (FIFO) PL memory is probably being used with a DMA to DDR.
That power input connector is the same as used on a lot of reciept printers for 24VDC power, so readily available if it breaks.
EDIT: might be incorrect there as I believe the reciept printer connector is a 3 pin (i'm not at home so can't check my epson TM88), but the 4 pin was/is still used on a lot of things so still easily available. I had a cheapo IDE NAS that used that connector for +12/5V and I believe my SCSI JAZ drive is the same connector. I knew I'd seen it before somewhere.
Receipt printers are 3 pin. A lot of POS terminals use the 4 pin though
This would be a great use for USB-C PD
Why haven't Tek chosen to power the oscilloscope using USB-C which is becoming standard on laptops, notebooks, tablets and phones, and just mandated by EU?
It is powered with 24V and rated 60W maximum.
not nearly as rugged as the chosen connector.
@Christopher Grant I dunno, so far all my USB-C connectors and cables are outlasting the others. I've only had 1 cable go bad thusfar. I've probably got 10 of em. Higher current is a huge win. Zero mechanical failures and zero dirt/dust/grime issues so far.. knock on wood right !
High end Pro audio equipment is using them. These are not consumer grade products. However, I don't think power & data should be mixed. I've seen noise issues arising from mixing data & power. Everything evolves, maybe noise will be solved one day
@Christopher Grant The newer stuff is USB-C
And why still have the idiotic CHASSIS GROUND??? Why not have each input isolated, floating, and differential??????????????????????
Safety earth bonding is a requirement. It must be guaranteed through the connector or you have to make the customer earth bond the scope some other way which can be a challenge. If you use a standard interface like USB-C the customer may use any USB-C power supply which may or may not be safety earth bonded and thus pose a safety risk.
Looking at the rear panel, there would be ample space for a decent capacity battery pack inside the enclosure. It goes to show that Tek do not have a lot of experience in compact (phone, tablet-, notebook) design, with the amount of air inside.
The extra space around the PCB is part of the thermal / air-flow design - keeps the audible noise down to barely audible.
@@w2aew I agree on the silent part. I upgrade to noctua whenever possible. So much better !!
You need that air for cooling.
4channel version is around 2400euro on tektronix site. Way too expensive to get no powered probes, auto detect. Horrible interface and membrane buttons. How about Wi-Fi, usb mouse and keyboard, remote control? I don’t know, it just looks like an expensive toy and not a proper instrument.
It does support mouse/keyboard, as well as remote control and programmatic control.
@@w2aew and graphic tablet TOO please
Wifi would be reason to ignore the product. When I measure somethink I don't want to have unnecessary radio source on the desk. There's gig LAN. Wire is wire, you know...
I’m a scope board repair tech for Tek in Oregon and I’m stoked to hear your thoughts.
I'm also a scope repair tech for Tek in Oregon and I'm also stoked to hear your thoughts.
Do you guys do actual component level repair on scopes?
@@EEVblog Yes, but only on certain boards. Analog front ends and acquisition boards are component-level repaired, but encoder boards, displays, certain power supplies, etc. are discretely replaced.
The worst part of the component repair is supply chain issues. Many parts have to come by ship from China and new products generally take quite a while to have parts stocked adequately. Happy to answer any other questions you have.
Do they still have a cafeteria?
@@jalalhobbs Haha, yes, but probably not like what you’re imagining. Two person crew mostly makes ready-to-eat meals that are refrigerated in company stores across the buildings. I can only speak for service/mfg in building 19. I’d love to see what the cafeterias looked like before the corporate buyouts.
That vesa mount is such a good idea!
That’s one of the lowest-end parts in the Zynq Ultrascale+ line. Tek are probably paying something like $50 per unit or less depending on the volume. Part of the external RAM will be being used for the ARM system and part of it for the programmable logic fabric.
If we still did component level repairs on the regular, I’d love a couple of these on VESA mount arms above our workstations. But for as often as we’d need one anymore, our old Agilent will do the business just fine.
The DDR memory is probably mostly the OS code, you'll not get a Linux kernel for arm64 in under 16MByte alone, and at-least another 16MByte for even a small system... they may well be using most of the DDR for their software.
Tekronix: "Please Dave, don't take it apart, just turn it on" Dave: "I will do both"
My only question is why hasn’t this been a thing for years? So many scopes you’ve torn down are empty space. Dunno why this hasn’t been a thing up till now.
Maybe the 'thunky car door' marketing ? The first depth reduction, losing the CRT, made useful extra bench space. Losing another couple of inches adds a certain amount if utility but also reduces the physical stability and apparent quality. Doubtless we'l now see a race to the slimmest, laptop-like.
10:14 If someone is more experience in RF stuff I have a question about the BNC/Frontend shielding.
Could the bump in the metal shielding be there so the GND/Shield is not so close to the BNC center pin? Would this influence the impedance?
My thought as well.
Or for isolation distance. 1000volts (beacuse students) on that bnc and you need some space.
@@ulwur "Because students" should really be a formal error submission code. A more specific version of "Because reasons" and "because users".
Might they be mechanical only, needed to keep clip on shield in place by being in contact with the back shell?
I think this is almost definitely the case, yes.
The bumps on the input shield probably just contact against the back cover and press down on it to keep it in place... And a Tag-Connect TC-2050 footprint (J2)!
Nice to see the jtag using the tag connect cable
The Signal Path was 45 minutes slower to release his teardown :)
perhaps those pins are to make the panel a bit firmer when you're pressing buttons etc. especially if it's being used as a portable scope, it would make it feel a bit more rigid if you were pressing the buttons with thumb force
No, they only serve to align the pcb, as they facilitate the assembly of the pcb given the amount of protrusions that have to come out of the front panel.
Nice tear down. A perfect example of modern product design where the emphasis has shifted to ease of manufacture without too much in the way of a compromise. Tektronix continue to lead in build quality and design and still manage to incorporate a hint of their trademark blue colour in the handle. Whenever I see the Tektronix blue I immediately think of those wonderful 500 series ‘scopes from way back.
Just as a general note. Kudos to Tek for making the handle on the stand classic Tektronix blue. Really brings back memories. Wish we'd see more metal cases though.
I guess the qualification build explains the massive number of fingerprints on the screen 😀
Probably the bumps in the shield do have a function, some capacitance thing.
Possibly capacitance to center pin, maybe.
PT - Plastic Tapper
19:17 Those Micron memory are likely shared between acquisition memory and application memory for driving the user interface. That ISSI chip is an eMMC for the Zynq to boot from.
I'll just add one into cart.. Oh wait, are those MEMBRANE switches?!
*closes tab*
Sounds like it uses the ultraram for sample memory, that external ram is obviously for the applicatiin processor, OS and framebuffer and secondary sample storage/functions
Not having had it in the hands, I think the metal pins are guiding pins for the battery holder back, so that it will ensure the mating power connector is aligned before it actually mate the electrical pins.
This surely can be a precaution in case some I'll advised user forgets to take the batteries out before mounting the battery holder.
But for that, of course, the back of the case cannot be snug fit around the pins and it must have some clearance around the pins, and the pins must protrude some bit from the back, (or if not from the back, protrude from some recess that may exist around the pins).
Well... My guess is as good as anyone's.
I fell in love with this VESA mount
The power comming out the side is so yo can charge it while the battery is plugged in.
Interesting to note the use of what appear to be National Semiconductor chips. I thought they were acquired by and absorbed into TI. Unless they've brought the brand back into use or something.
Seems like a lot of the time they don't bother to redo the branding on the chips after acquisitions like that. Still lots of NXP chips with Freescale logos on them for example.
They really could have made the battery option a lot thinner - just a few lipos and a BMS board. It would be similar to the second battery option on some Thinkpads.
Bump on a back shield around BNC pin is most likely for signal integrity or creepage reasons, instead of mechanical only. Must cover/may cover usually means epoxy coating on top, so at assembly they would know where epoxy should go/not go.
Could the raised bumps on the back of the low profile shield be there to keep a distance from the back of the connector pins......to avoid any coupling?
I think it's for CAT voltage clearance
I don't know this for sure as i havent measured one but I would bet that the outer set of threaded holes on the back are also VESA compatible.
If the holes in the middle are 100x100 I would guess the outer holes are the 300x200 compatible mount
290x155
Loved all the expressive hand motions reflecting in the screen during the intro
I jus tautomatically do that behind the camera.
@@EEVblog Relatable, am Italian
8bit scope with membrane switches for 2000+usd in 2022.
PT and M3 may refer to plastic thread (self tap) and threaded screws.
Finally, a Micsig-look-like Tablet Scope ?
Yes. Only look-a-like. I've owned and used the ato1104 for something like 4 years now and both love it and hate it. I hate the touch screen, but otherwise love the scope. I'm going to get sto1004 learnelectronics just reviewed. Did the sto1004 basically replace the sto1104c?
4:30 These pins are maybe for further design decisions. Maybe they will offer a version with internal battery other back cover. These pins are for alignment of the battery I think.
They might be for connection to a back spraying which was found unnecessary. Or simply didn't get done on this example, if it's preproduction.
I am worried about the lack of a vesa metal bracket
it does not weigh a lot but i would place this on a monitor arm and grab the side of it to move it around the workbench and over years they might pop out
Real concern!
An Aussie should always remember his knobs!
For the time base, it is likely they are using the FPGA to generate the time reference for the PLL chip.
Sweet! I would have appreciated half the price though...
I'm just waiting for one of these pre-production models sent in to EEVBlog for teardown to have silk screen that says, "Hi Dave!"
i am buying one as soon as it is available
... Perhaps "must cover" and "may cover", refer to conformal coating areas when in mass production.
Those mystery pins are triggers for the internal patent protection explosives. Better run!
the 2G memory is for the Linux OS, possibly including a ramdisk root filesystem ...
I used that exact bulkhead power connector on a system I designed. Easily sourced last I ordered. Who knows with supply chain, though.
the batteries look like film batteries I wonder if they're using a fairly stock battery that could be sourced from third party vendors. Like how infotainment lithium Sony batteries are so generic. If that's the case that would be a very good move you can have your high capacity originals and third party backups.
A thing of beauty is a joy forever
Alignment pins strike me as like a pc motherboard using the center standoff as a pin - getting the board lined up is pretty important because ALL the knobs/bncs would looks wrong if off by just a fraction of a mm. Rigid pins with a tight fit on the board ensures it can only go in one way and one precise alignment. Board edge clips can break or bend.
🤑The Series 2, now starting at $1,800. I'm over here trying to figure out what scope under $400 to get? I'm looking at the ds1202z-e and sds1202x-e. Sure would be nice if there were a comprehensive head-to-head of these two budget boxes.
NO "in video" COMMERCIALS. How do you do it? Easy to over look for us viewers (that don't pay extra). Good ON you. Cheers.
I simply don't enable them.
Congratulations champions, you have just made an oscilloscope 24 times more expensive with less analysis capabilities. Great job yours. You should post on fashion channels and not on electronic ones
Always love to see scope stuff!
I doubt those pins are for alignment. I suspect they are supplying mechanical support to the central area of the back panel.
They're almost certainly running their OS (likely Linux) from the DDR4. That's not just sample memory.
I would have to think the bumpouts on the shield around the back of the BNC is for clearance for voltage withstanding, or to reduce capacitance to the center pin or both.
What an alluring form factor. Love it.
Ultraram isn't available for application memory. It's in the PL side.
The metal pins = when you have it mounted on the smaller VESA mount, the weight of the unit plus the wires connected to it, plus and yanking that might happen... The metal pins support the back plate from flexing under load.
That thing looks pretty dope.
TEKBAT01 batteries are being sold for $330USD each! .. jeeeeez, you can get a low end scope for that!
Very well executed packaging. Not a fan of the separate battery back; just include it (maybe minus batteries). I’ll have to see those membrane switches in person; I have seen many worn through on machines. Hopefully the switch panel is replaceable and not expensive.
I suspect those metal pins are to make sure the case doesn't get crushed.
I wonder of those two "alignment pins" actually transfer mechanical load, especially from the battery pack?
No, they only serve to align the pcb, as they facilitate the assembly of the pcb given the amount of protrusions that have to come out of the front panel.
31:33 That FPGA is configured with 64-bit wide DDR4 memory. Tek could have used a DDR4 SODIMM slot here instead of soldered chips to allow further upgrades. For example if you want to further expanding your memory depth you can just buy a Tek SODIMM module (the SPD EEPROM can contain memory timing configurations along with a license key.)
I’d be paranoid of knocking it off a desk or bench.
i really like the design... and the tought and detail thats gone into the BNC connectors is great....
...but...IMO...the software lets it down a bit...
the DDR memory is probably also being used for the OS RAM.. the eMMC would contain just the OS/Software itself...
And here I am still with a tektronix 465
External memory is too slow - most likely, it is used by the OS and the graphics accelerator. I'm assuming the memory for signal sampling is on the FPGA (12 meg)
It's DDR4, a single chip should have plenty of bandwidth to handle even 2.5 GSPS.
@@HitoPrl DDR4 is very bursty in nature and has both latency and refresh issues for high speed sampling (look at the number of clock cycles needed for activation, RAS and CAS, plus the number of cycles needed for a page refresh). Very probably it’s program memory, the internal SRAM/BRAM on the ZYNQ is on the FPGA side of things, and is too small for a Linux build with GUI anyway.
There is not enough internal SRAM in the FPGA, it's has 10M per channel + the digital channels. Many scopes use external DDR4 sample memory.
They might be using distributed RAM (in LUTs) in addition to the dedicated internal SRAM to get additional fast internal sample memory. Or/also, they may have two separate DDR4 channels with one being used by the A53 cores and another being used for sample memory.
@@EEVblog the mistake is in thinking that there’s no connection between the FPGA and HPS parts of the chip. Data can be captured to BRAM, and as each block is completed it can be queued to be written to HPS RAM. The HPS can reserve blocks of RAM for FPGA use, and the latency of traversing the FPGA=>HPS and then HPS=>DDR4 interfaces would be hidden from the capture process. 10M sounds like an entirely arbitrary size though, and with little effort it could be increased or decreased, subject to how much of that 2GB of DDR4 is being used by the HPS for code/data.
Winner!!!
Wow new tektronix osciloscope
the os is probably using a significant amount of the 2 gb of ram because for linux with a gui you need a minimum of 300mb or more so would expect at least 500mb of it to be used for that
if some parts are underused according to their max specs does it mean its prime for a hack to get more performance than factory shipped?
Siglent now has 12 bit oscilloscopes. Hurry!
Nothing runs "from eMMC" directly. eMMC works like a disk, content must be loaded to RAM to be executed. The DRAM4 is more than likely used for the application processors, not just for sample storage (samples may even be stored/buffered in FPGA internal RAM first). There isn't enough internal RAM to run all the modern software with all the graphics and its abstraction layers.
Wow. Scopes are just getting thinner and thinner. What is the big black connector(?) seen at 38:29. I think I might have missed where it was mentioned. Is that part of the logic analyzer part of the unit? You mentioned that feature when pointing to that general area of the board.
The raised bumps are there to lower the chance of sparks.
I'd say that those unknown pole things would be for crush protection as they expect people to carry this around in backpacks and carry-on luggage
Wow! That’s one dope scope!