EEVblog 1545 - R&S MXO4 TEARDOWN. The World's Fastest Oscilloscope!
Вставка
- Опубліковано 21 лип 2024
- The world's fastest oscilloscope, the Rohde & Schwarz MXO4 teardown!
Wow, this is an impressive beast in more ways than one.
www.rohde-schwarz.com/us/prod...
Teardown photos: www.eevblog.com/2023/05/25/ee...
Unboxing and testing:
• EEVblog 1529 - R&S MXO...
ADC: www.ti.com/lit/ds/slvsdr3c/sl...
FPGA: www.digikey.com/en/products/d...
Forum: www.eevblog.com/forum/blog/ee...
00:00 - Rohde & Schwarz MXO4 The world's fastest oscilloscope
00:55 - The handle doesn't cut the mustard these days
02:02 - The production flap
03:54 - ALL scopes should use this power connection!
04:32 - This is gorgeous, and it just gets better.
06:38 - Look at it! Look at it!
08:14 - RF goodness
08:35 - Dual 100MHz Arb generator
08:54 - Best probe interface solution ever
09:55 - The world's best low noise fan?
10:53 - Main PCB
13:17 - Inside the front end RF can
19:48 - The MXO4-EP ASIC
21:30 - External trigger system
22:43 - $3300 12bit ADC
23:29 - The $3400 Xilinx Zynq FPGA and applications processor.
26:08 - I'm now uploading videos natively on Twitter
If you find my videos useful you may consider supporting the EEVblog on Patreon: / eevblog
Web Site: www.eevblog.com
Other channels:
EEVblog2: / eevblog2
EEVdiscover: / eevdiscover
T-Shirts: teespring.com/stores/eevblog
#ElectronicsCreators #Oscilloscope #Teardown - Наука та технологія
ADC chip alone is worth more than my entire scope.
ADC chip alone is worth more than my last car!
That is the reason for the flimsy handle axle. With all the expensive thechnology they had to cut the cost at least somewhere. Not sure how much they managed to save though.
I used an adc 16bit for current measuring shunt and it costs like 4$ for the chip breakout board alone almost as much as my mcu. Are ADCs generally expensive?
@@omniyambot9876 Ones that we use for our projects are usually pretty cheap, but they are over 1000 times slower.
Bah! Half a probe breaks my budget 😢
It is the first time i see Dave using a wriststrap for a teardown. He really must love this scope.
Spot on Jeroen! I find this ShOcKiNg too.
Considering the price not really a surprise. And we don't even know who is the owner of the scope.
You know it's a good ADC when the specsheet at 23:30 mentions "Electronic Warfare" right there
Watched at 1440p on a capable monitor, stunning video quality. It's like I am holding that Pcb!
Don't sell that camera Dave.
R&S is German, so the "RuS" etched on that front-end amp is "Rohde und Schwarz".
Ah, thanks.
So that is not a special edition of the chip designed specifically for the russian market...
"Whoever laid out this board takes pride in their work and they know what they're doing."
That Rohde & Schwarz engineer should be putting that sentence with a link to this point in the video into their resume (or perforrmance review)
😆
Oh trust me, they are all watching this video :)
i hope one day i would be able to layout a board like this
I'm gonna take a screencap, print it out, and hang it on my wall as inspiration.
…and on the silkscreen of the next board. 😊
@@pauldenisowski say hi to them. hiii
As a PC enthusiast, love seeing the NB eLoop there! Until the Noctua A12X25 came out, the eLoop and Nidec GT were top dog. I don't really know electronics, but enjoy these teardowns.
You‘re absolutely right about the fan noise. If you have ever been the last person in the lab and switch off all the gear in the evening and everything is quiet, you notice how loud some test equipment is.
You've trained me .... I can't call something "A thing of beauty!" without following with ... "A joy forever!" I'm going to have to launch a class-action suit :-)
Thanks Dave! Nothing I like more than a good teardown video :)
agreed 😎
Do you know how many people contributed to the pcb design on this?
@@orbita1 Not sure I can share that information :)
@@pauldenisowski I am a pcb designer myself but mostly for an academic research group and I always wondered how many people work on projects this bit and polishes. It's really impressive.
I wonder what the schematic of this 'monster' looks like and how many people worked on this, the service manual and so on.... all must be huge!
yeah the 3000 series has loads of issues, like frontends dying randomly, knobs falling off, that kind of thing. My uni had to send all 10 of them we have in the lab back to get the fixes installed.
The most action is happening in the software, like VHDL and also in the ASIC. So not so an impressive schematics I suppose. Old skool top line Tek scopes scopes would have much larger schematics.
Upvoting for the 12V power connection.
i got to build the PCBA, safety, performance and calibration testers for the production line for these units.
some my job is building PCBA testing fixtures and platforms.
and let me tell u, these where the biggest challenge jet to build.
the amount of testing and level of performance of every detail is amazing.
like all USB outs get a over-current event, the trigger time and recovery time are both tested and log'd.
and ges who has a few of these pre-production models just floating around different work spaces.
we do.
I can't remember the last time you did a teardown on an oscilloscope with hardly a complaint. I'm tempted to take a mortgage out to get one!
This is definitely the teardown of the year! Wanted to see the PSU board btw.
My bets would be on a bought in top quality module from Delta or TDK-Lambda.
Can't believe I forgot to do that!
I love that usb board, removes the need to solder if someone is too heavy handed or gums up the USB ports. Love it.
Watching with open mouth, drooling...
Dave, thanks for the tear down. Wouldn’t dare attempt with the MXO4 on my bench. Not enough can be said about the fan and lack of noise from this scope. It really is a thing of beauty.
Hello from Romania! Nice teardown Dave! :)
I’d rarely use it, but dang do I want one
That's what she said.
I don't think I've heard Dave say "look at it!" as much as this. Great bit of engineering here.
Except possibly on a 3rd date? No disrespect :) Cheers BP
Love the advancement in Toroidal propeller designs for boats and multirotor drones.
As far as I’m aware they are not magic. Fan rotation speed will still have the biggest impact. A 120mm fan at
that 20k pricetag seems really reasonable if you look at about 10k for the ADCs and FPGA alone...
Built my latest custom watercooled PC with those exact fans, brilliant noise/airflow ratio
1:40 Man I just love the way you handle a 20.000 USD instrument. No regard for cosmetics... I could never
I want to see LockPickingLawyer in this mental state during showing of best designed lock ever.
Nice to be able to see inside such a machine.
"Dave nuts over an oscilloscope for 27 minutes"
Wait holy shit they have a noiseblocker eloop fan? I had two of those on my PC, those are seriously expensive fans!!
Ah 😀so there *are still* companies making quality designed and well-put-together pieces of equipment. Albeit in the somewhat high price-segment 😅 Beautiful instrument to see the internals and construction of, my compliments to the design-team of the R&S MXO4 🥇. And thank you for the cool video Dave ! 🌠
Thanks - I think most of the design team has already watched this video :)
@@pauldenisowski Paul, from one engineer to another. One of the many things that sets R&S apart from lots of others is your companies true desire to be a part of the maker and hobby community. Even though a fraction of a percent of us can even begin to afford your kit, you still are in the trenches and seeing what we all think. As you are very keenly aware, that the makers and hobbyists have technical day jobs and a portion of us use high end kit at work and/or have influence on the kit that our companies invest in. Its smart from a marketing and business POV, but it always seems far more genuine than business. It always seems driven by engineers and technicians, for engineers and technicians. I think that is what makes your company really special.
Siglent seems to also do this and it always seemed that's the way HP was run.
@@StreuB1 Thanks! R&S is very much an engineering-driven company and we have an awful lot of hobbyists and amateur radio operators (myself included - KO4LZ) working here. So yes, our interest is very genuine :)
@@pauldenisowski It shows!! You and your teams, keep it up. Please send my thanks to your guys! (can we still say that and mean men and women, together? lol) N0BPS 🙂
@@StreuB1 Will do!
wow, that FPGA is a pretty huge chunk of bare silicon right there
I'd love to know details on the process node and die size.
@@EEVblog it's a fairly old midrange part, still on TSMC 16nm
@@tommihommi1 geez, 16nm is considered old. Can I have 1M node elements FPGA for $5 now ? please ?
@@monad_tcp The chip was launched like 8 years ago. That's old.
7nm has been the leading FPGA node for 4 years already.
@@tommihommi1 yeah, its already been 8 years ? it seem like it was last 2 years . I'm still rocking a 14nm desktop here , wait, that thing survived for 6 years already ? wow.
Do the outlines that allow for rf cans provide rf shielding by themselves? With the tiny smd components being close to the pcb, they are almost in a box without a lid.
Green is the new black ! I love that matte green Board....cheers.
12:06 These 6 identical Micron memory chips (D9WFH) could form a single 12-bit word memory array for samplings storage.
20:22 There's another Micron 8Gb DRAM chip (D9TBK) near by which could be dedicated for MSO sampling. This is my guess. )
Could be right.
Those are (probably x8) DDR4 memory chips. 48bit memory bus, each transfer operation taking 4 cycles, 48 bytes total (32×12bit samples) per transfer.
Extra single DRAM chip - probably RAM for ARM cores
D9WFH is for the MT40A512M16LY-075:E which has a 512M16 configuration (4 pieces -> 64bit)
Who knows how they arranged the memory ... could be 12bits for the sample with 4bits input state -> 16bits x 4 channels -> 64bits
We can only have a pure guess at it.
But the max memory depth of the scope is 400Mpoints and those 4 memory chips are 512M16x4 which is certainly enough.
@ppdan im just amazed by the amount of knowledge you have.. what career did you study
Romanian electronics engineer here! Thanks Dave for pointing it out! I will research exactly where (at what plant in Romania) was made and make it public!
Maybe in the region of Sibiu. We have tow factories there. Seems to be an electronics cluster there.
Noiseblockers are awesome!
This oscilloscope is piece of art.
Cheers from Romania!
Beauty.😍
I haven't seen the main review and was trying to do the BOM price in my head, but in all the excitment I kindo lost track of myself...
Please can you do an update on the PSU, and what they're doing to that ground plane (I'm a little surprised they airn't using a cleaner internal one). If they're taking it straight from the wall they might as well pour vomit on the thing.
As for the DRAM it looks as though they're them for input buffers and RAM, with some switching to cope with the bandwidth (RAID0). It would be interesting to know how they're storing the input data, I'm guessing storing 2 values in a single 3-byte pointer (that would help with bandwidth), otherwise they would be wasting a lot of nibbles (Gen-Z probably wondering wtf is a nibble?😅). Little surprised the DRAM doesn't get a little privacy with its own can as a lot of laptops sheild the DRAM, amd ECC is probably a luxury at such speeds.
Also, I've never sceen those fans before. Noise Blocker is a fairly well known brand and PC enthusiates will pay a pretty penny for decent fans. I have 6x140mm cooling my CPU as I like silence (German too!)
Hello from Romania! I never expected this to be made here.
What do we have at 7:17 ?
Around each central pin of the BNC socket?
I can see some leftovers after... soldering paste?
Or is it just the dust?
Hi Dave,
Long time viewer. I really like your content.
I observed that on your two videos for the R&S MX04 Oscilloscope it was two different serial numbers.
Was the one on your first video a press sample(SN 200176-Cr) to be returned to R&S and the one on this second video is yours to keep/teardown(SN 200183-Bm - Show Room white tag on top)?
Thanks!
At 25:14, it's bothering me that the two differential pairs coming out of the DAC are different lengths before they go to the output amps. They could have placed the DAC in the middle of the two.
Hey Dave!
Seems you missed teardown as much as I did?
Exciting!
Mrs EEVblog had the night off.... Dave was utterly spent by the time he got home
Interestingly there are obviously plugged vias but did they not get a solder mask or did they get a separate silk screen for the white spots? Why?
Maybe the extra 'scope' for shielding (I'm here all week) is future-proofing for the line? They must be planning on using the same board for even higher bandwidths.
I get the feeling that Dave likes it!
At 13:43, ch 4 (from the left), does the pcb have leftover flux from some rework? I wasn't going to be too picky, but since you talked about contaminents on the pcb.. Don't get me wrong, the board layout is extremely beautiful.
Well you talked about it later on in the video. My guess would be a chippie needed to be replaced and the flux didn't get cleaned off.
I think it may just be roasted from a manual rework.
It feels like Christmas!
charming pcb 😍
WOW
Dude we need a warning about the power connectors. I shot off like a rocket.
That's what she said.
Good Videos,
Which Presenting app is that?
noiseblocker fan .. impressive !!
Those anti creepage slots - why is one of the three always slightly angled (not parallel) ?
You are really crazy! Greetings from Germany :)
The MXO 4 has at least 4 GB for temporary information storage for operating
system and instrument firmware and >= 7 GB for waveform data and measurement data.
Such information can be found in the "R&S®MXO 4 Series Oscilloscope Instrument Security Procedures"
21:22 That might be why there are 6 memory chips, to store 12 bits, you divide the address space in 2 and store 12 bits in 6 chips. But I still think they're storing 12 bits in 4 chips by dividing the address space in 3, and the other two are double-parity, You don't want the data getting corrupted do you ? or one of them is parity and the other is ASIC configuration RAM.
12:31 Usually when I see that, I think of parity, double parity ? woa !!!
Cool. I wonder if the ASIC is developed in Duisburg.
Just found a fan with a Db rating of 14. Gonna use 4 of 'em in a project.
13:47 Come on, where are the MELF fanboys?
All engineers that deign in small noisy fans must be permanently made unemployed for crimes against engineering!!!!!!!!!!!!!
4x SanAce 40x40x25 non-PWM go brrr
hello,
i am from Romania, but is the first time when i see a Made in Romania sticker on a oscilloscope, specially from R&S. That's a first
10:00 Well, I would have used a Noctua fan for this, might have been even quieter than the NB fan.
at 14cm size that NB is better than noctua 14cm offerings. until noctua releases their next gen 14cm, which have been in development for many years
This is probably a stretch, but would be a interesting to see an Infiniium UXR-series mmWave scope being torn down :)
@16:50 Isn't "1001" SMD resistor code for 1k resistor?
Pornographic nuthin', that board is Playmate of the Year material! Just plain gorgeous. I gotta say though I was pretty surprised at the price tag. I was expecting more like $20k than $7-8k. Of course I'm sure those options will jam the price right through the ceiling but still, that level of capability for a price like that is a damn good deal.
The one he's showing is close to $30k!
Beauty
@14:51 RuS likely is the abbreviation for Rode und Schwarz. Und just being German for and
The reason there are six ddr chips is because the adc's acquire width is 12bit,usually ddr chips used four chip in a group, It's corresponds to a 8bit adc.
So - how quiet is the scope when used?
I think Dave maybe feeling moist 😂
I feel like there should be a cover charge for some reason......
Was the power supply simply forgotten or did I miss something?
Also interesting to see that the parts they use also cost thousands of dollars a piece. But then again, they are not paying the retail price, but still.
The OnlyFans comment made me actually giggle.
Also, Noiseblocker. That's a name I haven't heard in a long time. Besides, kinda surprised to see a NB fan in an OEM product like that. Usually you have your Delta fans for that
Yeahp, the 15 year old Dave as surfaced......!!!! Great video.
You know how the razor / razorblade game goes for the "options" for these scopes. And how with the Siglent and Hantek and Rigol - we got the hacks.
The cost of this option stack = a nice new car. Hack?
I know. Anyone that's buying even the base model of this series wouldn't dare mess around.
Started to look for a helicopter pad on that lay-out.
VESA-mount!
I translates to Rudy&Czarny for those polonaise efficionados playing along at home.
Could you do a Teardown Tuesday or Teardown Thursdays?
is that the ill-fated dual link type b hdmi they used on the logic analyser? that port should count itself lucky 😂
I've been wondering as well. What company would they even source the plugs and sockets from?
What are all of the circles (like at 8:47)? Are they just vias?
Most of them look like tented vias yes, especially since some traces run to and terminate at them, and others have copper clearance around them (obviously the via is not connected to the same net that the copper fill is, and they wanted to avoid the extra cost and complexity of a blind/buried via where possible).
Question out of left field, is there a reason pcb traces cannot be made out of different materials that act the same as ic's, caps, transistors, inductors etc, so you don't actually need to put anything on top of the pcb?
PCB traces ARE resistors, inductors and capacitors. But usually only at higher frequencies. Look up "Distributed element filters". You can't do semiconductors like diodes and transistors with PCB traces.
I've certainly seen hybrid modules with resistors silk-screened directly onto the substrate. And of course in the microwave and millimeter wave spectrum, inductors and capacitors are often made with specially shaped pcb traces. As for active devices, I think what you just described is an IC
You can make resistors and caps out of PCB traces, and certainly high-frequency antennas. But transistors require doping tiny regions in certain ways with high precision. You get thousands or millions of transistors or ICs on an 8" or 14" radius wafer, but if you built a large PCB at that resolution, and only placed a few ICs on it, it would cost thousands or millions, rather than the cost being amortised over thousands of ICs
I do know that you shouldn’t run AC power line traces in parallel to serial communication lines.
Also: ua-cam.com/video/Zao8upe4phs/v-deo.html
Tear down with 64gsps ADCs when? Show us the AMD Versal or Intel Agilex!
What camera are you using?
Sony NX80
Ieeeei Romania!
That 12V is similar to how Intel wants it for the new ATX12V standard. Under that standard, the PSU only has to concern itself with 12V, it's up to the motherboard to provide all other power rails that a computer requires (5V, 3.3V, etc...). Which would make PSUs for computers much simpler.
Next do a teardown of a Keysight URX1104A. It has 110GHz of bandwidth.
That's more in the domain of The Signal Path, probably.
Calibration EVISCERATED
And what a joy it is!
0:22 I never knew we still produced scopes here in Romania! Much less this kind of scope
I'm sorry to say but the most beautiful osciloscope "teardown" was the Tektronix Mr. Carlson's Lab showed.
I won't argue.
I hope you found the toothed washer that fell into no mans land at 13:28
For all you metal nerds... I'm guessing the die cast is some high-aluminum alloy... since that's a better conductor of electricity and heat than zinc.
I bet the unpopulated cans are for a government/military specific sku.
Teardown Thursday lol
No leds on the fan 😢
Wash your mouth out!
Nice but ahhhhhh you didn't show the PSUuuuuu😮😂