Amazing to see the tight integration of everything. microsd instead of nand chips, integrated inductors inside chips, one big heatsink, standard usb power benefitting from scale instead of expensive in-chassis DC-DC, standard android OS... It's built down to a cost, but I'd argue that a lot of these choices are preferred over more custom solutions. E.g. industry-standard isolated USB-C PSU isn't going to easily fail as it already has millions of units in the field.
An added bonus is maybe you can back up everything in case the card goes bad or gets corrupt. I wonder if there's any other flash on the board that would need to be in sync.
That USB-C port for power is the only thing I'm a bit worried about. It looks like it could pop out with an unintended move, or it could bend and break. I'm not overly happy with the various SBC boards I have which are USB-C powered. Other than that, this looks great. It just may be my next scope.
Re. populating all the passives - bear in mind that small passives and larger chips are typically placed on different machines, so may also be a factor in simplifying their production process
"small sparrow" is a reference of a Chinese saying that goes like “麻雀虽小五脏俱全”, meaning "The sparrow may be small, but It is fully-equipped". It's really a nice touch!
The reason they likely populated the bypass caps and termination resistors is likely to prevent the traces turning into antennae and injecting noise into the unused inputs. Basically the same as tying unused logic ic inputs to either power of ground.
Also it might have been EMI tested and certified with the additional ram on (as a prototype/Engineering sample or to have the option in the future), I’m almost certain that by just removing the ram there will be no noticeable adverse effect on compliance, removing bypass caps, even if they are nominally the removed memory ones I would not be so shure, thus to avoid re-testing.
A "Dumb measurment" for correct components specs via bed-of-nails station with "one size fits all Jig". The function test itself will be made in Software which goes deeper in the "options & variants" available. The minuscal ammount safed not populting these caps and resistors, compared to a more complex or extra different test-jig, is not worth it over a loooong time :)
Not the bypass caps, they have no impact on that. And if they are dumb enough to drive unused lines at high frequency and then have to terminate them, that's just as stupid.
MicroSD also means not having to do any in-circuit programming on the final board - could save on production time & test fixturing, as copying to onboard emmc could take significant time. They probably also have a "special" version for production test
The Rockchip RK3399 supports a eMMC chip boot. But as @mikeelectricstuff said, saves having to load a whole software image over USB 2.0 OTG using some of Rockchip's dodgy looking Windows tooling.
They could easily order preprogramed eMMC from manufacturer, most (if not all) manufacturers offer that option, same apply for microcontroller manufacturers. And you do not need ecen to order some insane amount I was supposed how low is minimal quantity for preprogrammed microcontrollers by Microchip, so probably second reason is Mike state is right reason why they chose that path.
@12:00 will be the same mould, it will have an interchangeable insert to block that section off for that model. The moulds are actually fairly complex with under cuts and side wedges along with small features.
SD card is interesting - could simply be that it's cheaper than an emmc card, and surprising there doesn't appear to be an emmc footprint. Suppose it makes it easy for them to up-size it easily in future, but it also seems to scream "hack me"
We have had massive issues with availability on eMMC (and hence prices have skyrocketed) the past 3 years - that could have been a factor in the design decision process, to go with an SDcard solution.
@@bobshowrocks that’s fair. I’ll admit that I wasn’t paying much attention to that last night. eMMC is definitely more feasible in that size than in the few GB range.
20:24 - the LA inputs look like they are differential, so presumably use an external converter to do threshold selection etc. Might be a risk of static damage though with just 33Rs on the inputs
the reason why the cheap passive are populated for the unpopulated parts is because they will go through a separate PnP process, and for tooling, its cheaper to just run the base board. And in the second process applying the larger more expensive parts, that's when they delegate what board goes to what model.
Bingo, I think we have a winner, especially if they plan to come out in the future with an up-specked version, it would make sense to just do the DFM and CAM work for the larger board and just tell the PNP not to populate the expensive component that are not needed especially if they do parallel lines and not batch
Worth your time to dig into the Xilinx IO pin datasheet. Reading between the lines there is a LOT of analog capability in the pins in addition to a ~1GHz 4-bit shift-register, frequency multiplier, and 1-2ns variable delay line. The analog capability is a side effect of the 100+ differential, semi-differential, and low voltage swing logic formats the FPGA supports. Ends up that a high speed rail-to-rail comparator on every IO pin is the simplest way to support every logic standard. You can either do fully differential inputs, or feed in a threshold voltage for a whole bank of pins. This comparator is VERY useful and I'm surprised it's not officially documented in the datasheet. TLDR, logic analyzer wired directly to the FPGA will work great but be a bit more ESD sensitive than I'd like.
@@dreamcat4 it's 100mm. But you could 3D print something to fit a 120mm fan, or use an 80mm/92mm one with a 3D printed shroud and preserve the VISA mount.
@@SeanHoulihane They come in many flavours, different static pressure is one of it. But, the original small fan is enough to cool the DSO, I would be surprised if a big 120 mm pc fan wouldn’t.
The logic analyzer has LVDS input, the probe does the level conversion. Also, the reason for lower bandwidth may be the cheaper FPGA - it runs at 1.2 GSa/s.
Those SD cards are not great for long term reliability but just used read only it'll probably do the job and provides an easy way to hack the thing and make backups and whatever. As for the cooling I'd just try to screw in an external large quiet fan to that VESA mount flushing the whole heatsink with air.
"long term reliability".. ? Hey I wonder..If the TERRIBLE reliability rating for the otherwise wonderful (and expensive!) TC Helicon voicelive3 vocal processor might be due to IT'S os being on one???
@@petegaslondon But unlike an on-board NAND chip the SD card can be easily replaced with no desoldering and you don't have to find a pin compatible chip, so you can just buy one off the shelf.
That DID cross my mind - sadly the Voicelive is marketed to not-so-tech-savvy Musicians .. So I'd like to investigate if this is the recurrent problem ! @@volactic8495
I suspect so. Rigol have basically lead every revolution in the entry level scope market. First the DS1052E, then the 4CH DS1054Z , and now the touch screen 12bit compact VESA mount DHO800.
@@EEVblogIf you think about it, they are continuously undercutting Tektronix from the start. Probably most of the old giants of the day, but their line-up follows the Tektronix like-up much more closely.
It's really nice that they populated all those extra small components. Those are always the hardest to find if you want to hack such a device. I tried to add Bluetooth to a slightly older car radio. It has an audio input selector that is controlled by I2C and has traces going to an unpopulated RCA jack on the back used for a cd-changer. It has some kind of a bus system on the board that checks if devices are present and communicates with them. If you unplug the Cd part, there is no CD option in the input selection anymore. With that functionality I can't just solder wires to those rear solder joints but I would also have to populate the components used for the cd-changer bus and somehow trick the CPU into thinking that there is an active cd-changer connected. I now just soldered the Bluetooth module to the front audio-in pins and can change the input by flipping a switch to turn on/off the BT receiver. I could even use that thing as a simple Bluetooth receiver if it is turned off with the audio-in jack as an audio-out.
Seems crazy that Rigol are offering the logic analyser and arbitrary function generator features on this smaller model when it was not on the DHO1000 or DHO4000 series. I'm interested in the DHO914S which includes these features with the 125MHz analogue bandwidth (maybe hackable to 250MHz), so would like to see reviews of how these features perform given that past reviews of Rigol have shown their offerings to be quite buggy and lacking refinement.
The discretes being fully populated would depend upon the MFG method. For very high speed /throughput MFG you have one reel for each form factor of components that is pick and place. So for example you would have a 0603 reel and it has all resistor, caps, inductors on it. This allows the pick and place to work in a linear fashion across the board. MUCH faster and if the parts cost fractions of a penny the savings for this high speed is likely higher than having different master reels made. Discretes in volume in China are basically the cost of material +5 to 7%. Master reels (where these discretes are placed in the appropriate position from raw reels to master reels) cost virtually nothing and are often just a contract line item. And of course populating the discretes terminates otherwise open lines. This is an exceptional design. Hats of to the Rigol engineers.
They used to do this for through hole parts which were physically much bigger so you couldn't as easily line 10s of different components up next to each other on a machine. But surface mount components are so much smaller and tapes are what 8mm wide at minimum? You can fit 50-60 maybe more different reels across each side of a high volume PnP machine... Do they really bother to populate custom reels of different surface mount components in a particular order just to make the machines job easier? Doesn't make sense to me...
Yes they indeed do. If you watch a video on very high volume production consumables (like cell phones) it is phenomenally fast. Reel changes and repositioning adds up very quickly. Generally the Production Engineer will evaluate the cost benefit and order the reel stacking. During the pre-release design review stage (or earlier) you bring in the PE and make sure the Design For Manufacturability needs are met. Part of this review are cycle, throughput and test times. A good PE prints profit really.
Oh that niiice - I'd recommend maybe, if you ARE gonna go inside,(void your warranty!) A blob of epoxy over the USB-C? Oh I WANT that Logic analyser one its niiiice - with the generator and in black, is like oohhhh!! (never had anything much more advanced as my poor ol Tektronix rack valve jobby, that i had before the nasty Eviction - kit-built a few little pieces of Chinesium test stuff, but ooh I LIKE this thing) . . PS Maybe the termination res's are applied so the open transmission lines dont resonate/ring? OR that they can rework their OWN boards easily - OR that you really CAN do it yourself :)
I got the 924S and it heats quite hard to the point the BNC on the front are hot. Same with both BNC at the back (OUT and function generator). I have put a 100mm fan on the back and it cools it down well. I bought it for the PLA, output to monitor and the fact you can use a mouse and the monitor without a computer. With age is harder to see these small displays
The Artix-7 isn't very expensive as FPGAs go - if anything the Zynq may be more expensive as it's a combined SoC-FPGA as opposed to a standalone FPGA. Strange they would use one here when they already have the Rockchip as an application processor.
They did indeed save money on the FPGA, however it has nothing to do with it being a zynq (which is either an artix or a kintex with arm cores bolted on) but by dramatically cutting down the logic portion size (from 100T of the previous to 015T) however I really do not expect Dave to actually verify his speculations anymore
Maybe Xilinx had some kind of special offer. But for that kind of cost down it would certainly be interesting to save the Rockchip and use the Zynq arm core instead. Would certainly mean some investment on the software development side.
@@DirkEibach my fairly educated guess (as a heavy zynq 7020 user) is that, the chip does not have nearly enough grunt to do both acquisition and graphics at the same time in logic, and no dedicated GPU, for that you would have to move to Ultrascale (which has a gpu) but then the cost go up substantially. Also the ARM CPU in the zynq is really weak in modern terms, so overall designing in the rockchip was cheaper than either doing the engineering of squeezing in everything on a single FPGA or going with the zynq ultrascale
@@sanjikaneki6226 as far as I know it does not have a particular meaning, it is just part of the part number of Xilinx most series 7 FPGAs that distinguish the device size (SoCs and Spartans don’t have that, however there is no large feature disparity)
I took a look at the dump file from the card, no partition at all, but there is random raw streams of date through the card. DRM+Logs+cache? There also no OS on the card
Huh? Just zoom in, it’s plainly visible that they’re controlled-impedance differential pairs. (And the pair of series resistors is also plainly visible.) There’s a separate guard trace to the shield.
The only things that really concerns me about this are the USB C power connector and the fan. If any of them break you could have a bad time. Otherwise it looks like a great deal.
@@absurdengineeringReplacing a USB C connector is easy. What concerns me is that if you need to replace it, there's a good chance there's damage to the PCB already. But that's not the main reason i dislike it. The main reason i dislike it is the need for an external power brick. With the provided USB C power back you need some more desk space (more than the DS1054Z had in the back), or you need to buy a slimmer third party one (that would add cost, storage complexity and general inconvenience). Standard power cables are really convenient since I already have a box with 20 of them. I love USB C for devices that 100% require small form factor - laptops, smart phones, headphones, etc, but I'm not sure I like it for T&M. Or maybe i'm wrong and i just need to give it some time in order to see who it would fit.
ปีที่แล้ว
@@Nik930714 I prefer USB-C on most of my gear. I can just use one powerbank for everything. It's great to have an oscilloscope isolated from the ground.
If I would buy one of these, first thing I would do after the warranty expires is to clone that SD card. It will almost surely fail/get corrupted within the first 5 years. At least it's a decent brand, but still a guaranteed failure point. At least it's easily replacable, so I like this idea against the soldered-on flash memory, which is only a bit more reliable than an SD card, but takes much more effort to dump its contents and clone it.
I've had so many devices that boot from an SD or microSD die within 4 years -- lots and lots of music equipment like an MPC. I always screw the warranty and go in and copy the filesystem/dump. I try to put it back together without looking like its been tampered. Its saved my machines at least 10 times or more from corruption. I just buy another SD/micro.
Someone hacked it already, it has a very weird filesystem setup. But all the necessary android files are there, which means you could turn a DHO804 into a DHO924S by swapping the SD card.
@@EEVblog I still think I would have preferred a case that was even 10mm thicker and used a standard fan that I could get from RS/Digikey/Mouser/Farnell etc. How easy is going to be to get a replacement fan in 10 years time? That said I think I would have preferred spending an extra $10 and having a bigger case and heatsink with no fan 🙂
I've been thinking of buying a scope for a long time, but was held back because the DS1054Z looked a bit old for my taste. If I land a new job within a month I'll treat myself to the DHO800. All because some Australian with a huge-ass knife told me to.
Very interesting. I'm not that interested in buying one to hack, so the 900 series seems more attractive to me. The top model wouldn't be too much more than what I paid for my old Siglent, but it comes with all the bells and whistles, whereas I had to pay extra to get the arb option on my Siglent. Really tempted to add it to my Christmas present list.
The SD card is also an easy way to factory-program the thing: no messing around with JTAG and such, just stick in a preprogrammed SD card containing all the firmware. That doesn’t explain why it won’t read on Linux though. What does dmesg say when you plug it in?
JTAG isn’t just for programming. JTAG literally stands for “joint TEST action group”: JTAG’s original purpose was to enable circuit testing without needing to add as many test points, complex test jigs, etc. It’s commonly used for production testing.
I bet it would read with the right software. It's probably just not formatted with a standard partition table. Something like Balena Etcher would probably read it just fine. Making sense of what you read, that would be the difficult bit.
SD card would have a squashfs image on it, no FAT or Ext3/4 file system on it, just a bootloader in the first LBA block, that is loaded and executed, that then will extract the file system, mount and decompress, and run it. Would suggest to clone that card to another, as it likely does write logs and temp files to the mounted file system, and that will kill the SD card in short order with all the writes going to the same blocks, as that card probably does have wear levelling, but it does not work well with a non ExFAT file system. So around 50k writes before you get issues.
For temp files it would make more sense to use tmpfs, especially for a device like this. If the SD card contains a squashfs image, it's easy - you can't write to a squashfs image. It's read-only.
@@invisi1407 Yes but there will have to be some storage there, as either an image or as a block that is mounted, at least ot store waveforms for comparasion, and I would not put that the final image still has some logging enabled, or some swap, as the main RAM is very likely always running near full. Wonder if there is an exposed USB header that you can use to add in an external flash storage device somewhere on the main board, without needing to use the front panel USB port.
Not much of a bargain in Europe... the price is 499 euro here. Also, it has got interleaved sampling: it's 1.25Gsa/s, but only when one channel is used, otherwise 625 Msa/s for 2 channels, and only 312.5 Msa/s for 3 or more channels. I'd advice to consider adding other 300 euro and get the DHO924s (priced 799 euro), which is the 250MHz fully featured version, includes the AWG and the Digital Analyzer, and it's got non-interleaved 1.25Gsa/s sampling....
I would have really liked to see a metal plate inside for the mount, im not worried about its weight im worried about grabbing it and overcoming the monitor mount friction several times a day as you tilt it around across the bench
I'm finally jumping into automotive component bench testing and someone suggested that I save my money for a few and purchase a DH0800 as my first scope. I believe I'll take his advice and get this unit. I'll probably never use but one third of what this device is capable of accomplishing. Overkill? Yes please and thank you!
The first console that you opened up was probably the Android shell which is a derivative of Linux shell. You should be able to list files and folders with `ls`, `cd` into directories and maybe even list some hardware specs (e.g. system RAM) or look at how they are using the SD card and stuff. I think the shell could be using the Busybox toolbox so the basics should be there.
Strange to use a micro SD card like that. I've seen that done for a prototype or development revision, but then for later builds they swap it out for an emmc module to save cost. Gluing it in also makes it seem like they don't intend users to swap them out as a means for an upgrade, but maybe for a technician to upgrade?
It's possible they want to be able to quickly/frequently update firmware in the factory during assembly. With SD cards, you can do that with a cheap duplicator machine, but it's harder with an emmc.
Some eMMC modules have very high failure rates. I have a bunch of network devices here with dead eMMC modules, which I can’t fix because the ROM has a very limited eMMC driver. And eMMCs are on small daughter cards, so they were designed to be swapped! If these devices, had micro SD, they’d be simple to repair. Never solder in flash memory if you can, because it wears out too quickly.
I think that caps and resistors are installed because they separate the boards for different models after that "small parts placing machine" and it's cheaper in the production
Wow, a most exciting 12bit oscilloscope. Something hard to not purchase. However, the 12bit hobby scopes are getting more and more in demand. Let's see what the competition is up to? 😎 Does the DHO800 have a CAN, SPI, I2C Decoder? Thank you for the new must have list.
Thank you Dave! Will buy a DHO804 next month hoping the folks on the eevblog are done hacking it by then. Looking forward to the full review, but i'm already sold.
More likely the passives are placed, well the bypass caps anyway, so that the PDN behavior is the same. You'd have to (well, ideally you would, but we all know how often we do it ourselves in practice...) evaluate the supply quality for every assembly configuration. Easier just to leave it alone, and let the low impedance sink a little more current in the case it's unused. The termination resistors hanging around though, beats me =)
To see how far we have come with regard to scope market watch Daves first video from 14 years ago where he reveiws Rigols DS1052 that he got for the crazy low price of 642 AUD 😂
I'm trying to understand if I can supply the scope directly from 3 Li-ion cells (11.1V discharging to 9V) or 4 Li-ion cells (14.8>12V). I suppose the scope has an internal dc-dc converter to provide all the necessary voltages that can handle some supply voltage changes. Based on your experience and the results of the teardown, please let me know what you think.
Love a new small scope with more features! However... as a retired engineer who tinkers around, I'm pretty happy with the 100MHz 4 channel 8 bit Rigol DS1104. I'm not sure I can envision a situation where a 12 bit scope would be noticeable improvement. Maybe when dealing with very small signals?
well i suppose if you think of 8-bits as being a max resolution of 1024 peak to peak. (but only when the scale takes advantage of the entire range)... that is like a 1080p hd sort of resolution. so then 12-bits is analogous to going from hd to 4k i suppose? does that sound about right?
@@dreamcat4 No question that it provides more resolution, but I can't think of any instance where I was dealing with a signal of a couple of volts where I needed more resolution. If I was working with a microvolt signal, then possibly. There's also the issue of signal to noise ratio. If I've got anything other than a very clean signal, then the extra resolution is overshadowed by the noise.
First thing I do with most electronics I got (NAS, routers, etc) that have a fan is to replace the fans with Noctua ultra silent industrial ones. They make the typical sizes and come in all voltages. But I would never dare opening up a scope, because I don't know how the Noctua fan might interfere with the EMI, and if the built-in fan is specially optimized for that and that's why it is so loud. Bear in mind, an ultra silent, decent static pressure Noctua fan of that size can cost upwards of 25 Euro/USD, so that's a decent chunk of the scope's price. Plus this fan looks like that would be almost impossible to be replaced.
Wow. The price value is amazing. Also love the form factor. The USB C power that is cool and innovative. Nice. Could use battery banks easily, standard PSUs, but the provided PSU also is nice, high quality and small. Cool. The USB Type C socket looks a bit anemic, could use some more solder, or some glue there. Also would be nice to see some screw in USB Type C. There are few standards, and they are really and nice way to secure Type C for more permanent and secure installation.
Could any of the unpopulated spots be due to last minute changes from the "chip shortage"? Maybe the reistors and caps, that are populated, are there to stop the unused traces from becoming resonant antennas from stray RF?
The visa mount that's an option but yeah would have been nice for the plastic to have some cable mgt, I like the fact they left everything populated so easy to hack, its funny allot of people do populate caps and resistors ect but not the big cost maybe its just easier to have the same cfg on the pick and place and just leave out the more expensive components as i think that would be another pick and place machine for the big BGA's
Maybe they keep the best boards with no issues for the Higher model. And boards with problems or not as to spec like snr are sold as the lower model that's why some parts are still populated/ installed?
Rigol needs to address the flimsy connectors. The logic analyzer upgrade is a ripoff if the power connector breaks off. They obviously weren't paying attention to the 1990s through the 2010s when laptop power connectors constantly fell off their motherboards.
It is literally the cheapest they can buy with high speed serial transceivers which are probably necessary for the ADC converter (which I would bet is using the JESD204 protocol, which is an LVDS high speed interface)
Absolutely this. It has half of the multi-gig transceivers of the Artix 7 100T but has an extra 12-bit ADC. Plus, the PS has a bunch of "freebie" peripherals that can potentially reduce the BOM costs significantly across the models that they use it in (why buy an FPGA that you're not filling up AND an ETH PHY, when you can use an SOC/FPGA with enough fabric and 2 ETH PHYs for cheaper).
SD cards can be used in either SD or SPI mode and each others interface might not read the correct data. Or the cards are hare programmed to only use one mode. You could get an SD to SPI adapter used on Arduinos to try to read that card. They might use them as a cheap eMMC alternative which is something I once wanted to try on a QNAP NAS with broken eMMC flash. Sadly I wasn't able to remove the old chip so I just sold that NAS. It was a nice and modern model I got for really cheap. I'm a bit sad I didn't keep it to try it again. I was also able to control it via a soldered on serial port which was kind of nice.
You have to use the USB-C connectors with the magnetic attachments so it would easily disconnect if the cable is pulled. I use those cables on USB-C power connection on the my Apple MacBook AIR. I wonder if accidentally cutting power would corrupt the SD card though.
The cost of the discrete components is so low that it is logical to leave them in circuit compared to the cost of recoding the populator bots and R&D the effect of floating traces turning into mini aerials.
I think it is to do with cost of manufacture why they put the passive components on the logic analyser and not the chip. When you get a PCB manufacturer to do boards like that they charge extra fro mounting chips of that size and pin configuration. It suggests to me they place them by hand after the machine has placed all the other SM parts. The large ribbon connector socket is also in that category of manual placement. The idea is you do a large manufacturing run of the basic boards then you can use that board for all the different models and populate then according the the demand for each model. It's more costly to split it up into about ten different manufacturing runs.
5:30 Make sure to get a full, bit-for-bit image of that SD card. I've seen too many of those fail, especially in daily use specialist hardware. Edit: I wonder if someone's going to reuse the older 8-bit architectures to introduce sub-$200 'scopes? I think sub-$100 is still wishful thinking on something this niche, but given the R&D is likely long paid off... 🤷♂
I have to ask. Is the 802 fully populated too with the added exceptIon of the the BNC connectors and relays for channels 3 and 4? Are the front panels the same except for the front bezel and ch3, ch4 switches? I bet you see the lights flash behind the plastic on startup.
Might be they're fitting a lot of unnecessary parts for initial batches to give them a lot of boards they can easily convert to the higher tier models while they finish their development? We've done that before, and then later in production cycles removed the parts not needed for that SKU.
Front panel mould can still be the same, just have parts of the mould set that are swapped out to have the logic analyser subsystem in there. Back one they simply use a sticker to cover the holes, cheaper to make 3 stickers with silkscreened labels and part numbers. Termination resistors and bypass capacitors are populated, probably to minimise reflections on the lines, and also the caps are easier to leave in place, so as to not need to change the solder paste stencil, and the glue dot stencil, making it easier to change board numbers on line, just place the extra chips, which are the major cost, while the passives probably added an extra 5c to the BOM cost.
Amazing to see the tight integration of everything. microsd instead of nand chips, integrated inductors inside chips, one big heatsink, standard usb power benefitting from scale instead of expensive in-chassis DC-DC, standard android OS... It's built down to a cost, but I'd argue that a lot of these choices are preferred over more custom solutions. E.g. industry-standard isolated USB-C PSU isn't going to easily fail as it already has millions of units in the field.
An added bonus is maybe you can back up everything in case the card goes bad or gets corrupt. I wonder if there's any other flash on the board that would need to be in sync.
@@Okurka.
The Liteon also manufacture in China. "Generic Chinese garbage" is often de-branded designs from the same factories.
I prefer a larger unit because it looks more substantial and pro in my lab. It's like having a wall of Marshall amps--it makes you look cool!
That USB-C port for power is the only thing I'm a bit worried about. It looks like it could pop out with an unintended move, or it could bend and break. I'm not overly happy with the various SBC boards I have which are USB-C powered.
Other than that, this looks great. It just may be my next scope.
@@tohaasonJust got my 804 the other day. So far impression is good. The USB C cannot really wiggle since it is held snuggly by the back cover.
Re. populating all the passives - bear in mind that small passives and larger chips are typically placed on different machines, so may also be a factor in simplifying their production process
Especially for the DIP header, I would do the same and leave the passives on.
"small sparrow" is a reference of a Chinese saying that goes like “麻雀虽小五脏俱全”, meaning "The sparrow may be small, but It is fully-equipped".
It's really a nice touch!
Mao Zedong would have hated this product.
The reason they likely populated the bypass caps and termination resistors is likely to prevent the traces turning into antennae and injecting noise into the unused inputs. Basically the same as tying unused logic ic inputs to either power of ground.
Precisely, all those caps provide high frequency shunts to ground, which dampen any oscillation that might be induced on those unused trace stubs.
Oh you beat me to it! :)
(I kinda said that, up-thread)
Also it might have been EMI tested and certified with the additional ram on (as a prototype/Engineering sample or to have the option in the future), I’m almost certain that by just removing the ram there will be no noticeable adverse effect on compliance, removing bypass caps, even if they are nominally the removed memory ones I would not be so shure, thus to avoid re-testing.
A "Dumb measurment" for correct components specs via bed-of-nails station with "one size fits all Jig".
The function test itself will be made in Software which goes deeper in the "options & variants" available.
The minuscal ammount safed not populting these caps and resistors, compared to a more complex or extra different test-jig, is not worth it over a loooong time :)
Not the bypass caps, they have no impact on that. And if they are dumb enough to drive unused lines at high frequency and then have to terminate them, that's just as stupid.
MicroSD also means not having to do any in-circuit programming on the final board - could save on production time & test fixturing, as copying to onboard emmc could take significant time. They probably also have a "special" version for production test
The Rockchip RK3399 supports a eMMC chip boot. But as @mikeelectricstuff said, saves having to load a whole software image over USB 2.0 OTG using some of Rockchip's dodgy looking Windows tooling.
They could easily order preprogramed eMMC from manufacturer, most (if not all) manufacturers offer that option, same apply for microcontroller manufacturers. And you do not need ecen to order some insane amount I was supposed how low is minimal quantity for preprogrammed microcontrollers by Microchip, so probably second reason is Mike state is right reason why they chose that path.
@12:00 will be the same mould, it will have an interchangeable insert to block that section off for that model. The moulds are actually fairly complex with under cuts and side wedges along with small features.
SD card is interesting - could simply be that it's cheaper than an emmc card, and surprising there doesn't appear to be an emmc footprint. Suppose it makes it easy for them to up-size it easily in future, but it also seems to scream "hack me"
You simply cannot get small eMMC chips, but small microSD cards are still widely available.
@@Finder245 the card in the video is 32GB, hardly what I call small
We have had massive issues with availability on eMMC (and hence prices have skyrocketed) the past 3 years - that could have been a factor in the design decision process, to go with an SDcard solution.
@@bobshowrocks that’s fair. I’ll admit that I wasn’t paying much attention to that last night. eMMC is definitely more feasible in that size than in the few GB range.
20:24 - the LA inputs look like they are differential, so presumably use an external converter to do threshold selection etc. Might be a risk of static damage though with just 33Rs on the inputs
I'll have to check what the DIY probes have.
the reason why the cheap passive are populated for the unpopulated parts is because they will go through a separate PnP process, and for tooling, its cheaper to just run the base board. And in the second process applying the larger more expensive parts, that's when they delegate what board goes to what model.
Bingo, I think we have a winner, especially if they plan to come out in the future with an up-specked version, it would make sense to just do the DFM and CAM work for the larger board and just tell the PNP not to populate the expensive component that are not needed especially if they do parallel lines and not batch
Worth your time to dig into the Xilinx IO pin datasheet. Reading between the lines there is a LOT of analog capability in the pins in addition to a ~1GHz 4-bit shift-register, frequency multiplier, and 1-2ns variable delay line. The analog capability is a side effect of the 100+ differential, semi-differential, and low voltage swing logic formats the FPGA supports. Ends up that a high speed rail-to-rail comparator on every IO pin is the simplest way to support every logic standard. You can either do fully differential inputs, or feed in a threshold voltage for a whole bank of pins. This comparator is VERY useful and I'm surprised it's not officially documented in the datasheet.
TLDR, logic analyzer wired directly to the FPGA will work great but be a bit more ESD sensitive than I'd like.
Which Logic Analyzer is good nowadays?
I would be temped to fasten a 90* usb-c extender. That could take physical stress off the board and allow one to lay the scope flat on the bench.
Dave - we need a DHO800 vs Siglent SDS800X HD comparison !
21:33 that looks more like connector footprint than an IC - 2 locating holes in the middle, so maybe a sub-board for ARB
You could mount a 120 mm pc cooling fan to the vesa mount for ultra quiet operation.
thats great idea... just wondering if those hole spacing is either 100mm or 150mm ?
@@dreamcat4 it's 100mm. But you could 3D print something to fit a 120mm fan, or use an 80mm/92mm one with a 3D printed shroud and preserve the VISA mount.
Are those fans designed for the right static pressure? External fan and ducting seems like the solution thought.
@@SeanHoulihane They come in many flavours, different static pressure is one of it. But, the original small fan is enough to cool the DSO, I would be surprised if a big 120 mm pc fan wouldn’t.
The logic analyzer has LVDS input, the probe does the level conversion. Also, the reason for lower bandwidth may be the cheaper FPGA - it runs at 1.2 GSa/s.
Those SD cards are not great for long term reliability but just used read only it'll probably do the job and provides an easy way to hack the thing and make backups and whatever.
As for the cooling I'd just try to screw in an external large quiet fan to that VESA mount flushing the whole heatsink with air.
"long term reliability".. ? Hey I wonder..If the TERRIBLE reliability rating for the otherwise wonderful (and expensive!) TC Helicon voicelive3 vocal processor might be due to IT'S os being on one???
@@petegaslondon I have no idea about the device but could be.
@@petegaslondon But unlike an on-board NAND chip the SD card can be easily replaced with no desoldering and you don't have to find a pin compatible chip, so you can just buy one off the shelf.
That DID cross my mind - sadly the Voicelive is marketed to not-so-tech-savvy Musicians .. So I'd like to investigate if this is the recurrent problem ! @@volactic8495
@@petegaslondonno one cares
I've been wanting to get a scope for awhile and finally a decent brand with decent specs at a decent price. It's gonna sell like mad.
I suspect so. Rigol have basically lead every revolution in the entry level scope market. First the DS1052E, then the 4CH DS1054Z , and now the touch screen 12bit compact VESA mount DHO800.
@@EEVblogIf you think about it, they are continuously undercutting Tektronix from the start. Probably most of the old giants of the day, but their line-up follows the Tektronix like-up much more closely.
Hope that price don't jump up
Anyone know when it’s available in the UK?
@@Okurka. The DHO814is the 100MHz bandwidth version that costs more than the $389 DHO804, which is the 70MHz bandwidth version
This is the new low-spec king. The size is absolutely fantastic. A huge upgrade from the DS1000Z.
This is unchellenged and will be for years probably
I still think Siglent SDS1104X-E is a better option. It's got a logic analyzer option and 2x1GS/s ADCs for just $100 more.
@@miloradowicz but the logic analyzer is expensive as hell
No CAN decoding, so it's useless.
@@Z0DI4C Probably will be a software addon or maybe just a software "fix"
It's really nice that they populated all those extra small components. Those are always the hardest to find if you want to hack such a device.
I tried to add Bluetooth to a slightly older car radio. It has an audio input selector that is controlled by I2C and has traces going to an unpopulated RCA jack on the back used for a cd-changer. It has some kind of a bus system on the board that checks if devices are present and communicates with them. If you unplug the Cd part, there is no CD option in the input selection anymore. With that functionality I can't just solder wires to those rear solder joints but I would also have to populate the components used for the cd-changer bus and somehow trick the CPU into thinking that there is an active cd-changer connected. I now just soldered the Bluetooth module to the front audio-in pins and can change the input by flipping a switch to turn on/off the BT receiver. I could even use that thing as a simple Bluetooth receiver if it is turned off with the audio-in jack as an audio-out.
Seems crazy that Rigol are offering the logic analyser and arbitrary function generator features on this smaller model when it was not on the DHO1000 or DHO4000 series.
I'm interested in the DHO914S which includes these features with the 125MHz analogue bandwidth (maybe hackable to 250MHz), so would like to see reviews of how these features perform given that past reviews of Rigol have shown their offerings to be quite buggy and lacking refinement.
You can get right angle USB-C cables for working flat
The discretes being fully populated would depend upon the MFG method. For very high speed /throughput MFG you have one reel for each form factor of components that is pick and place. So for example you would have a 0603 reel and it has all resistor, caps, inductors on it. This allows the pick and place to work in a linear fashion across the board. MUCH faster and if the parts cost fractions of a penny the savings for this high speed is likely higher than having different master reels made. Discretes in volume in China are basically the cost of material +5 to 7%. Master reels (where these discretes are placed in the appropriate position from raw reels to master reels) cost virtually nothing and are often just a contract line item. And of course populating the discretes terminates otherwise open lines. This is an exceptional design. Hats of to the Rigol engineers.
They used to do this for through hole parts which were physically much bigger so you couldn't as easily line 10s of different components up next to each other on a machine.
But surface mount components are so much smaller and tapes are what 8mm wide at minimum? You can fit 50-60 maybe more different reels across each side of a high volume PnP machine... Do they really bother to populate custom reels of different surface mount components in a particular order just to make the machines job easier? Doesn't make sense to me...
Yes they indeed do. If you watch a video on very high volume production consumables (like cell phones) it is phenomenally fast. Reel changes and repositioning adds up very quickly. Generally the Production Engineer will evaluate the cost benefit and order the reel stacking. During the pre-release design review stage (or earlier) you bring in the PE and make sure the Design For Manufacturability needs are met. Part of this review are cycle, throughput and test times. A good PE prints profit really.
Oh that niiice - I'd recommend maybe, if you ARE gonna go inside,(void your warranty!) A blob of epoxy over the USB-C?
Oh I WANT that Logic analyser one its niiiice - with the generator and in black, is like oohhhh!!
(never had anything much more advanced as my poor ol Tektronix rack valve jobby, that i had before the nasty Eviction - kit-built a few little pieces of Chinesium test stuff, but ooh I LIKE this thing) . . PS Maybe the termination res's are applied so the open transmission lines dont resonate/ring? OR that they can rework their OWN boards easily - OR that you really CAN do it yourself :)
I got the 924S and it heats quite hard to the point the BNC on the front are hot. Same with both BNC at the back (OUT and function generator). I have put a 100mm fan on the back and it cools it down well. I bought it for the PLA, output to monitor and the fact you can use a mouse and the monitor without a computer. With age is harder to see these small displays
As you noted in EEVBLOG 1566 one other difference is that there appears to be no provision for an RTC backup battery.
The Artix-7 isn't very expensive as FPGAs go - if anything the Zynq may be more expensive as it's a combined SoC-FPGA as opposed to a standalone FPGA. Strange they would use one here when they already have the Rockchip as an application processor.
They did indeed save money on the FPGA, however it has nothing to do with it being a zynq (which is either an artix or a kintex with arm cores bolted on) but by dramatically cutting down the logic portion size (from 100T of the previous to 015T) however I really do not expect Dave to actually verify his speculations anymore
Maybe Xilinx had some kind of special offer. But for that kind of cost down it would certainly be interesting to save the Rockchip and use the Zynq arm core instead. Would certainly mean some investment on the software development side.
@@DirkEibach my fairly educated guess (as a heavy zynq 7020 user) is that, the chip does not have nearly enough grunt to do both acquisition and graphics at the same time in logic, and no dedicated GPU, for that you would have to move to Ultrascale (which has a gpu) but then the cost go up substantially.
Also the ARM CPU in the zynq is really weak in modern terms, so overall designing in the rockchip was cheaper than either doing the engineering of squeezing in everything on a single FPGA or going with the zynq ultrascale
@@pelor92 what does that T rating mean ?
@@sanjikaneki6226 as far as I know it does not have a particular meaning, it is just part of the part number of Xilinx most series 7 FPGAs that distinguish the device size (SoCs and Spartans don’t have that, however there is no large feature disparity)
Amazing. Great video as always!
Dave, let clarify prices to not confuse people. 814 does not cost 389... it cost 499. 804 cost 389 and 802 is 329
Your title is misleading a bit.
@@xlr8r171499 and Amazon today
actually in the EU the dho802 is a bit over 400€ these days. to confuse people even more ;)
Still love my DS1054Z introduced to me by EEVblog. I wish I knew how to use it properly!
I'd love an Agilent to look at, but it would just sit on the shelf without being understood either!
Oh wow it seems pretty decent and at that price point seems like a good purchase for hobbyists
@@Okurka.$500 on Amazon now
I took a look at the dump file from the card, no partition at all, but there is random raw streams of date through the card. DRM+Logs+cache? There also no OS on the card
It doesn't look like the paths from the frontend to the ADC are differential, just single-ended referenced against the ground and shield.
Huh? Just zoom in, it’s plainly visible that they’re controlled-impedance differential pairs. (And the pair of series resistors is also plainly visible.) There’s a separate guard trace to the shield.
The only things that really concerns me about this are the USB C power connector and the fan. If any of them break you could have a bad time. Otherwise it looks like a great deal.
It’s not terrible to replace USB-C connectors as long as the traces stay put. I say USB-C power should be a standard for T&M that can use it.
@@absurdengineeringReplacing a USB C connector is easy. What concerns me is that if you need to replace it, there's a good chance there's damage to the PCB already. But that's not the main reason i dislike it. The main reason i dislike it is the need for an external power brick. With the provided USB C power back you need some more desk space (more than the DS1054Z had in the back), or you need to buy a slimmer third party one (that would add cost, storage complexity and general inconvenience). Standard power cables are really convenient since I already have a box with 20 of them.
I love USB C for devices that 100% require small form factor - laptops, smart phones, headphones, etc, but I'm not sure I like it for T&M. Or maybe i'm wrong and i just need to give it some time in order to see who it would fit.
@@Nik930714 I prefer USB-C on most of my gear. I can just use one powerbank for everything. It's great to have an oscilloscope isolated from the ground.
The USB C is through hole. That is, thank goodness, a step up from those useless surface mount ones.
Attach a 90° USB adapter to the back cover, JB weld or hot glue it in place. Bob's your Uncle.
If I would buy one of these, first thing I would do after the warranty expires is to clone that SD card. It will almost surely fail/get corrupted within the first 5 years. At least it's a decent brand, but still a guaranteed failure point. At least it's easily replacable, so I like this idea against the soldered-on flash memory, which is only a bit more reliable than an SD card, but takes much more effort to dump its contents and clone it.
I've had so many devices that boot from an SD or microSD die within 4 years -- lots and lots of music equipment like an MPC. I always screw the warranty and go in and copy the filesystem/dump. I try to put it back together without looking like its been tampered. Its saved my machines at least 10 times or more from corruption. I just buy another SD/micro.
I would expect the SD card to have either an EXT4 or F2FS file system. But if Linux can't identify or mount it, it really might be something special.
Yeah someone mentioned trying a faster card for improved boot time - I'm assuming soemthing like DD could make a binary copy?
Someone hacked it already, it has a very weird filesystem setup. But all the necessary android files are there, which means you could turn a DHO804 into a DHO924S by swapping the SD card.
@@petegaslondon Yes, dd will make an image of it no matter what file system and partition type is used.
It MIGHT also be due to the sector size used for the filesystem. Linux on x86/amd64 doesn't support larger sector sizes than 4096 bytes for ext4.
@@0x8badf00d The term you're looking for is block size, and this limitation could be removed very soon with the current work on folios.
Gparted will be able to read the format on that SD card.
Already done, dump is on the forum. It contains the OS.
Great teardown. Thank for sharing.
One issue I see is the fan is going to be difficult to replace when it inevitably fails. You are not going to be able to get that off the shelf.
or take out the fan and srew a bigger one on the outside ... I did that also on my 1054
It might be a pretty standard graphics card fan.
I found many off the shelf that look identical, they are used on GPU cards.
@@EEVblog I still think I would have preferred a case that was even 10mm thicker and used a standard fan that I could get from RS/Digikey/Mouser/Farnell etc. How easy is going to be to get a replacement fan in 10 years time? That said I think I would have preferred spending an extra $10 and having a bigger case and heatsink with no fan 🙂
I've been thinking of buying a scope for a long time, but was held back because the DS1054Z looked a bit old for my taste. If I land a new job within a month I'll treat myself to the DHO800. All because some Australian with a huge-ass knife told me to.
Very interesting. I'm not that interested in buying one to hack, so the 900 series seems more attractive to me. The top model wouldn't be too much more than what I paid for my old Siglent, but it comes with all the bells and whistles, whereas I had to pay extra to get the arb option on my Siglent.
Really tempted to add it to my Christmas present list.
The SD card is also an easy way to factory-program the thing: no messing around with JTAG and such, just stick in a preprogrammed SD card containing all the firmware.
That doesn’t explain why it won’t read on Linux though. What does dmesg say when you plug it in?
Maybe the content inside is not a valid filesystem. Maybe the bootloader inside the processor know how to handle. A raw dump will clear the things.
JTAG isn’t just for programming. JTAG literally stands for “joint TEST action group”: JTAG’s original purpose was to enable circuit testing without needing to add as many test points, complex test jigs, etc. It’s commonly used for production testing.
@@tookitogo Sure, but you can do both: fully test all boards using JTAG and then later personalise them as needed with little effort.
I bet it would read with the right software. It's probably just not formatted with a standard partition table. Something like Balena Etcher would probably read it just fine. Making sense of what you read, that would be the difficult bit.
SD card would have a squashfs image on it, no FAT or Ext3/4 file system on it, just a bootloader in the first LBA block, that is loaded and executed, that then will extract the file system, mount and decompress, and run it. Would suggest to clone that card to another, as it likely does write logs and temp files to the mounted file system, and that will kill the SD card in short order with all the writes going to the same blocks, as that card probably does have wear levelling, but it does not work well with a non ExFAT file system. So around 50k writes before you get issues.
For temp files it would make more sense to use tmpfs, especially for a device like this. If the SD card contains a squashfs image, it's easy - you can't write to a squashfs image. It's read-only.
@@invisi1407 Yes but there will have to be some storage there, as either an image or as a block that is mounted, at least ot store waveforms for comparasion, and I would not put that the final image still has some logging enabled, or some swap, as the main RAM is very likely always running near full. Wonder if there is an exposed USB header that you can use to add in an external flash storage device somewhere on the main board, without needing to use the front panel USB port.
Surely they are total amateurs and destroy their SD card with logs. 😂
Not much of a bargain in Europe... the price is 499 euro here. Also, it has got interleaved sampling: it's 1.25Gsa/s, but only when one channel is used, otherwise 625 Msa/s for 2 channels, and only 312.5 Msa/s for 3 or more channels. I'd advice to consider adding other 300 euro and get the DHO924s (priced 799 euro), which is the 250MHz fully featured version, includes the AWG and the Digital Analyzer, and it's got non-interleaved 1.25Gsa/s sampling....
@@you2ber252 You can edit Your own YT comments anytime.
@@norbert.kiszka Ok. Corrected
I would have really liked to see a metal plate inside for the mount, im not worried about its weight im worried about grabbing it and overcoming the monitor mount friction several times a day as you tilt it around across the bench
3:00 earth-isolation is actually a nice feature^^ It's actually not so easy to earth-isolate a device without additional equipment like a trafo
I'm finally jumping into automotive component bench testing and someone suggested that I save my money for a few and purchase a DH0800 as my first scope. I believe I'll take his advice and get this unit. I'll probably never use but one third of what this device is capable of accomplishing. Overkill? Yes please and thank you!
The first console that you opened up was probably the Android shell which is a derivative of Linux shell. You should be able to list files and folders with `ls`, `cd` into directories and maybe even list some hardware specs (e.g. system RAM) or look at how they are using the SD card and stuff. I think the shell could be using the Busybox toolbox so the basics should be there.
Strange to use a micro SD card like that. I've seen that done for a prototype or development revision, but then for later builds they swap it out for an emmc module to save cost. Gluing it in also makes it seem like they don't intend users to swap them out as a means for an upgrade, but maybe for a technician to upgrade?
It's possible they want to be able to quickly/frequently update firmware in the factory during assembly. With SD cards, you can do that with a cheap duplicator machine, but it's harder with an emmc.
Maybe uodate firnware easily
Some eMMC modules have very high failure rates. I have a bunch of network devices here with dead eMMC modules, which I can’t fix because the ROM has a very limited eMMC driver. And eMMCs are on small daughter cards, so they were designed to be swapped! If these devices, had micro SD, they’d be simple to repair. Never solder in flash memory if you can, because it wears out too quickly.
What model of LITE-ON psu was show on 2:55 ?
I think that caps and resistors are installed because they separate the boards for different models after that "small parts placing machine" and it's cheaper in the production
I was intrigued by the possibility of using this with a battery pack. But 36W, oof... Anyway, sweet scope!
this new scope have a slow data logger (saving data on usb stick) and Bode diagram features, like Siglent ones ?
We need other video, maybe a comparison with the ds1050z?
Wow, a most exciting 12bit oscilloscope. Something hard to not purchase. However, the 12bit hobby scopes are getting more and more in demand. Let's see what the competition is up to? 😎 Does the DHO800 have a CAN, SPI, I2C Decoder? Thank you for the new must have list.
Hopefully there will be a few secondhand DS1054Zs hitting the market soon. :D
Thank you Dave! Will buy a DHO804 next month hoping the folks on the eevblog are done hacking it by then. Looking forward to the full review, but i'm already sold.
Please do a full review!
Paint net is great for displaying images and scrolling using Ctrl + mouse wheel to move left-right and Ctrl + mouse to zoom in-out.
More likely the passives are placed, well the bypass caps anyway, so that the PDN behavior is the same. You'd have to (well, ideally you would, but we all know how often we do it ourselves in practice...) evaluate the supply quality for every assembly configuration. Easier just to leave it alone, and let the low impedance sink a little more current in the case it's unused.
The termination resistors hanging around though, beats me =)
That "Joshua" login suggestion was awesome! Just watched Wargames the other day!
To see how far we have come with regard to scope market watch Daves first video from 14 years ago where he reveiws Rigols DS1052 that he got for the crazy low price of 642 AUD 😂
I'm trying to understand if I can supply the scope directly from 3 Li-ion cells (11.1V discharging to 9V) or 4 Li-ion cells (14.8>12V).
I suppose the scope has an internal dc-dc converter to provide all the necessary voltages that can handle some supply voltage changes.
Based on your experience and the results of the teardown, please let me know what you think.
If the power supply is isolated... does that mean you don't get a ground loop and can put your probe earth on something like a CRT ground safely?
They should have included a angled USB-C plug and a fixation point. That is the least one could expect! That would have really cost nothing!
Love a new small scope with more features! However... as a retired engineer who tinkers around, I'm pretty happy with the 100MHz 4 channel 8 bit Rigol DS1104. I'm not sure I can envision a situation where a 12 bit scope would be noticeable improvement. Maybe when dealing with very small signals?
If you wanted to measure noise in an audio circuit or something, I could see it being useful.
from 8 to 12 bit is a huge gain!
well i suppose if you think of 8-bits as being a max resolution of 1024 peak to peak. (but only when the scale takes advantage of the entire range)... that is like a 1080p hd sort of resolution. so then 12-bits is analogous to going from hd to 4k i suppose? does that sound about right?
@@dreamcat4 No question that it provides more resolution, but I can't think of any instance where I was dealing with a signal of a couple of volts where I needed more resolution. If I was working with a microvolt signal, then possibly. There's also the issue of signal to noise ratio. If I've got anything other than a very clean signal, then the extra resolution is overshadowed by the noise.
@@dreamcat4in what world do 8 bit have 1024 counts?
That's a cool heat sink, I like it.
Yeah, worth the effort.
Pun intended?
CMXT RAM chips, that's interesting. I guess they now source more chips from Chinese suppliers, and this is another point in price cutting.
First thing I do with most electronics I got (NAS, routers, etc) that have a fan is to replace the fans with Noctua ultra silent industrial ones. They make the typical sizes and come in all voltages. But I would never dare opening up a scope, because I don't know how the Noctua fan might interfere with the EMI, and if the built-in fan is specially optimized for that and that's why it is so loud. Bear in mind, an ultra silent, decent static pressure Noctua fan of that size can cost upwards of 25 Euro/USD, so that's a decent chunk of the scope's price.
Plus this fan looks like that would be almost impossible to be replaced.
@your
Are you twelve?
Wow. The price value is amazing. Also love the form factor. The USB C power that is cool and innovative. Nice. Could use battery banks easily, standard PSUs, but the provided PSU also is nice, high quality and small. Cool. The USB Type C socket looks a bit anemic, could use some more solder, or some glue there. Also would be nice to see some screw in USB Type C. There are few standards, and they are really and nice way to secure Type C for more permanent and secure installation.
SD could have partitions similar to a router: boot loader, kernel, squashfs, jffs2.
25:12 "Joshua" I love this homage to the movie "War Games" !
Could any of the unpopulated spots be due to last minute changes from the "chip shortage"? Maybe the reistors and caps, that are populated, are there to stop the unused traces from becoming resonant antennas from stray RF?
16:23 - perhaps signal integrity changes if you have those floating?
The visa mount that's an option but yeah would have been nice for the plastic to have some cable mgt, I like the fact they left everything populated so easy to hack, its funny allot of people do populate caps and resistors ect but not the big cost maybe its just easier to have the same cfg on the pick and place and just leave out the more expensive components as i think that would be another pick and place machine for the big BGA's
Maybe they keep the best boards with no issues for the Higher model. And boards with problems or not as to spec like snr are sold as the lower model that's why some parts are still populated/ installed?
Could you mount a 92mm or 120mm fan on the back?
92 mm sure has to sit there.
With enough hot-snot, you can mount anything to anything.
@@ghydda I was thinking of a 3d printed adapter/mount.
Rigol needs to address the flimsy connectors. The logic analyzer upgrade is a ripoff if the power connector breaks off. They obviously weren't paying attention to the 1990s through the 2010s when laptop power connectors constantly fell off their motherboards.
Love the Wargames ref! 😂
We're living in a new golden era of test equipment.
Not sure how much they have saved on the FPGA, looks like the zync XCZ7015. Pretty heftly with dual ARM cores included. Thanks for the tear down.
It is literally the cheapest they can buy with high speed serial transceivers which are probably necessary for the ADC converter (which I would bet is using the JESD204 protocol, which is an LVDS high speed interface)
Absolutely this. It has half of the multi-gig transceivers of the Artix 7 100T but has an extra 12-bit ADC. Plus, the PS has a bunch of "freebie" peripherals that can potentially reduce the BOM costs significantly across the models that they use it in (why buy an FPGA that you're not filling up AND an ETH PHY, when you can use an SOC/FPGA with enough fabric and 2 ETH PHYs for cheaper).
My Father's brother was named William, so "Bill's my uncle"... just saying... :P Great video as always! Good on Ya!
SD cards can be used in either SD or SPI mode and each others interface might not read the correct data. Or the cards are hare programmed to only use one mode. You could get an SD to SPI adapter used on Arduinos to try to read that card. They might use them as a cheap eMMC alternative which is something I once wanted to try on a QNAP NAS with broken eMMC flash. Sadly I wasn't able to remove the old chip so I just sold that NAS. It was a nice and modern model I got for really cheap. I'm a bit sad I didn't keep it to try it again. I was also able to control it via a soldered on serial port which was kind of nice.
Warranty void stickers aren’t valid in the US, EU, or AU
You have to use the USB-C connectors with the magnetic attachments so it would easily disconnect if the cable is pulled. I use those cables on USB-C power connection on the my Apple MacBook AIR. I wonder if accidentally cutting power would corrupt the SD card though.
Are those warranty void stickers not illegal in Australia?
When we can see a new design series or fundamental Friday again? We miss those days dave
My old Rigol DS5202CA has been in service for 15 years. I affectionately call it "the old coffin"...
Does it have a FFT function to get frequency spectrum plots?
The cost of the discrete components is so low that it is logical to leave them in circuit compared to the cost of recoding the populator bots and R&D the effect of floating traces turning into mini aerials.
I think it is to do with cost of manufacture why they put the passive components on the logic analyser and not the chip. When you get a PCB manufacturer to do boards like that they charge extra fro mounting chips of that size and pin configuration. It suggests to me they place them by hand after the machine has placed all the other SM parts. The large ribbon connector socket is also in that category of manual placement. The idea is you do a large manufacturing run of the basic boards then you can use that board for all the different models and populate then according the the demand for each model. It's more costly to split it up into about ten different manufacturing runs.
With this sd card solution it's up to the "standard" of the frnsi industrial garbage
There is much more to this "standard" for example noisy but unreliable fan.
Yeah, when you see SD card held in place with a blob of Araldite you think Hmmm.
Why do they include an earth lead? Isn’t it better for the scope to be isolated?
That depends on your situation. There is no one right answer to that.
Siglent also released SDS1000 HD series. It's a lot similar to Rigol's DHO900 seris in terms of price and capability.
5:30 Make sure to get a full, bit-for-bit image of that SD card. I've seen too many of those fail, especially in daily use specialist hardware. Edit: I wonder if someone's going to reuse the older 8-bit architectures to introduce sub-$200 'scopes? I think sub-$100 is still wishful thinking on something this niche, but given the R&D is likely long paid off... 🤷♂
I have to ask. Is the 802 fully populated too with the added exceptIon of the the BNC connectors and relays for channels 3 and 4? Are the front panels the same except for the front bezel and ch3, ch4 switches? I bet you see the lights flash behind the plastic on startup.
Might be they're fitting a lot of unnecessary parts for initial batches to give them a lot of boards they can easily convert to the higher tier models while they finish their development? We've done that before, and then later in production cycles removed the parts not needed for that SKU.
here in EU Rigol DHO924 is 700 and the DHO924S is 800 Euro's
Front panel mould can still be the same, just have parts of the mould set that are swapped out to have the logic analyser subsystem in there. Back one they simply use a sticker to cover the holes, cheaper to make 3 stickers with silkscreened labels and part numbers.
Termination resistors and bypass capacitors are populated, probably to minimise reflections on the lines, and also the caps are easier to leave in place, so as to not need to change the solder paste stencil, and the glue dot stencil, making it easier to change board numbers on line, just place the extra chips, which are the major cost, while the passives probably added an extra 5c to the BOM cost.
我有一个dho804，不得不说，我在国内买2299元，在美国也差不多是这个价格，可是美国的平均工资比国内高很多，买这个示波器就相对容易很多啊！😋
Ooh last time I was this early the door was still locked