Rigol MSO4000 and DS4000. Tests, bugs, firmware, questions, etc...

[hs3]

The cost reduction has to come from somewhere, and there is plenty to reduce in a DS4000 as I stated before (FPGAs, ADCs, memory). I suspect memory and GSPS are not upgradeable, but perhaps decoders and other functions. Obviously we would only be sure after opening one up and comparing IC markings.

Yes we have no idea if any re-design and cost reduction has actually been done.

But let's make a few calculations to see what might be possible with newer generation parts.

We can see in the earlier pictures that the DS4000 has 8 memory ICs per channel group. The part number is H5PS5162GFR which is a 32Mx16 DDR2 SDRAM.

So that is 64 MB per IC. So with 8 ICs total -> 512 MB memory. The spec for the scope is 140 MS so you would be able to fit that data over 3 times in the available memory. Not sure how Rigol is using the memory and if all of it is in use.

Total bandwidth needed to sample at 4 GS/s would be 4 GB/s. As there is 8 SDRAM ICs that is 512 MB/s per IC.

These are 16 bit wide so that would mean minimum 256 MHz memory clock + overhead. Let's say the memory interface would be running at DDR2-533 to have some headroom available. Not sure how fast memory interfaces are possible with Virtex 5.

Ok, googling "virtex 5 ddr2 bandwidth" helps:
http://www.xilinx.com/support/documentation/application_notes/xapp858.pdf
So looks like speed grade -1 Virtex 5 actually supports up to DDR2-533 for that reference design.

I think the current generation low cost FPGA from Xilinx would be Artix 7 probably.
http://www.xilinx.com/products/silicon-devices/fpga/artix-7.html
There it is mentioned that these support DDR3-1066 memory interfaces.

That would mean that you would be able to have similar bandwidth with the memory IC quantity reduced to half. That would allow reducing the FPGA I/O amount and package size needed.

That could be one way to reduce cost with newer components.

I don't think reducing the amount of memory would really offer much cost savings though. And if you need 4 memory ICs to get the bandwidth needed and consider the SDRAM IC sizes these days you might end up with quite a lot of memory anyway. So I think the sample memory size specification in a cost reduced design would be more about marketing and positioning the device in the overall product portfolio offered.

It seems the new DS4000E doesn't have much savings in the ADC performance area as if it supports 2 GS/s at the same time on all channels then the ADC capability and memory bandwidth needed would be similar to regular DS4000.

So who knows, maybe the E models are just the same hardware.

And if anyone notices errors in my calculations let me know.

[dr.diesel]

Do we know how much cheaper it even is?  TEq mentions it above, but it's not yet on their website.

[H.O]

Prices are listed on the Emona website and according to them, as I wrote earlier, you pay >80% of the price of the "standard" version (comparing the 100MHz models).

[alank2]

It doesn't make sense how they decide other pricing.

The Emona site shows a 18.5% (2779/3410) difference in price between DS4014 and DS4014E, but the rigol site shows a 46.2% (12800 CNY/23800 CNY) difference in price.

I don't know why anyone would choose the E model for 18%, but 46% is a different situation...

[seronday]

DS4000 series Option Codes.

I have updated the PDF document in my original post, ( https://www.eevblog.com/forum/testgear/sniffing-the-rigol%27s-internal-i2c-bus/msg523679/#msg523679 ) ,
to include the option selection codes for LIN and 1553B decode.

Also included are two possible future options,  Power Analysis and I2S decode.
The Power Analysis has been listed for some time, but the I2S decode is relatively recent.
The original un-documented "MA" option has become the 1553B decode option.
These codes have been tested with FW ver 00.02.03.00.03.

[3lite]

Hey guys

Recently I did purchase MS4014 and it's giving me a strange noise sound depending on the time-base. I mean it's a sound as if each time it refreshes the screen with given time-base it "beeps":



You might want to turn up the volume since the fan sound is a bit louder.

Shortly after I did receive it I did update it to DS4000 Firmware_00.02.03.00.03. Now I cannot recall whether it was like that before the firmware update.

[jjoonathan]

My 2011 DS4014 does that too, as does the DS4k in one of the youtube review videos. General consensus is that it's due to periodic trends in current consumption over the trigger cycle causing audible coil whine with the fundamental at the trigger frequency. Change the timebase, change the trigger frequency. As far as quirks go it's mostly harmless. At least next to other DS4k quirks like janky serial decode.

[3lite]

Thanks That's actually kind of a relief to hear since I was worried that it might be damaged, but you are right - it's definitely related to trigger. It's not that audible, but it wasn't there within DS2000 series which I did use before this one. It's kind of sad that they didn't fix it since 2011.

Anyway, I did check the scope itself and it's functioning properly. I suppose I can live with that.

[pascal_sweden]

I could not hear that sound, but maybe my laptop speakers don't have enough power

Anyhow, it's a remarkable that this issue is present on the high-end DS4000 series (overlooked?) and not on the low-end DS1000Z and DS2000A series (better design?)

So, whether it is present on all of their DS4000 scopes or not, and whether it is not annoying or just slightly annoying, it looks like a design issue!

But then again, the noisy fan as such is a design issue as well.

Anyhow, if they can make it work on the low-end series, they should be able to make it work on the high-end series. That is, the sound issue, which was reported in the thread here.

Courtesy of consistent design practices among the different design teams!

[mrpackethead]

Hi, just before i jump in and buy one, whats the status on hacking an entry level 4000 >

[jjoonathan]

Hacking is a go (500MHz, serial), but be warned: serial decode still has major bugs after all these years, and it's still done in software. Right now the big problem is that serial decode vanishes if you scale or shift the waveform (trigger still works though). A year ago the problem was that decode didn't work in segmented memory. Make sure the scope still makes sense to you with that caveat. When you include the price of a Saleae to compensate, the sticker price of a DS4k starts getting awfully close to that of an Agilent DSOX2014A + serial decode upgrade.

IMO the Rigol+Saleae still makes good sense -- more memory, more waveform/s, more bandwidth if you hack it, more decodes and more memory in the Saleae. But it's not the no-brainer of a $400 DS1054 vs a $1.5k Tek.

[rsjsouza]

Hacking is a go (500MHz, serial), but be warned: serial decode still has major bugs after all these years, and it's still done in software. Right now the big problem is that serial decode vanishes if you scale or shift the waveform (trigger still works though). A year ago the problem was that decode didn't work in segmented memory. Make sure the scope still makes sense to you with that caveat. When you include the price of a Saleae to compensate, the sticker price of a DS4k starts getting awfully close to that of an Agilent DSOX2014A + serial decode upgrade.

IMO the Rigol+Saleae still makes good sense -- more memory, more waveform/s, more bandwidth if you hack it, more decodes and more memory in the Saleae. But it's not the no-brainer of a $400 DS1054 vs a $1.5k Tek.

If you are looking for a brand new oscilloscope, if two channels are sufficient for you (especially if you plan to purchase a logic analyzer) and you are located in North America, Rigol's clearance bin has a good deal on a DS4012 - I bought my DS4014 there.

However, at the US$2k+ price level I agree with jjoonathan: the decision starts to become a bit more blurry. More so if you don't mind a used oscilloscope: depending on your location you may be tempted by a few used offers on eBay.

[Someone]

Hacking is a go (500MHz, serial), but be warned: serial decode still has major bugs after all these years, and it's still done in software. Right now the big problem is that serial decode vanishes if you scale or shift the waveform (trigger still works though). A year ago the problem was that decode didn't work in segmented memory. Make sure the scope still makes sense to you with that caveat. When you include the price of a Saleae to compensate, the sticker price of a DS4k starts getting awfully close to that of an Agilent DSOX2014A + serial decode upgrade.

IMO the Rigol+Saleae still makes good sense -- more memory, more waveform/s, more bandwidth if you hack it, more decodes and more memory in the Saleae. But it's not the no-brainer of a $400 DS1054 vs a $1.5k Tek.

Except other than at the extreme top end (10ns/div and faster when the synthetic 50,000 wfm/s limit is applied) and while only running a single channel the Agilent/Keysight 2000 still has a faster update rate, with dual channels or vectors the gap widens more as the Rigol slows down, 99% of use cases the Rigol update rate will be much slower by factors of x3, x10 or more. There are marketing specs which use corner cases to sell products, and then things like "the feel" of a responsive hardware driven capture system, compared to the Rigol display stuttering or failing to respond to knobs.

The longer memory of the Rigol products is their real selling point, its horses for courses.

[jjoonathan]

I'd be willing to believe it, the Keysights at work all have butter-smooth UI responsiveness while my Rigol DS4k doesn't. They're 1M wfm/s devices but regardless it shows that Keysight cares about UX more than Rigol. I mostly chalked up the UX difference to slow UI compositing or poorly optimized configuration routines rather than a wfm/s curve with a "hand crafted" maximum (EDIT: the waveform moves smoothly on the display, it's the 0V marker that's janky). Any chance you remember where the wfm/s tables live?

The Rigol has 5 hefty Virtex 5s, so they're not cutting corners on processing power, but that obviously doesn't prevent bottlenecking in the firmware or CPU.

[Someone]

Just running through the numbers:
https://www.eevblog.com/forum/testgear/rigol-mso4000-and-ds4000-tests-bugs-firmware-questions-etc/msg364748/#msg364748
https://www.eevblog.com/forum/testgear/first-impressions-and-review-of-the-rigol-ds2072-ds2000-series-dso/msg160064/#msg160064
http://cp.literature.agilent.com/litweb/pdf/5989-7885EN.pdf


And picking peak rates regardless of settings we can build the graph, noting that the additional memory depth of the DS4000 is slowing its updates at the lower timebases, but even when its capturing the same memory depths as the Agilent scopes it can't keep up. The DS2000 was set manually to minimum memory to get its peak rates, they drop as the memory is increased and similarly fall behind the Agilent scopes when set for the same memory depths, but you can always pick some area where each of the products offers some specific value, whats important depends on the application.

As far as I know the update rate at these high speeds is limited by memory bandwidth to the display buffer, whether that is synthetically limited or due to disabling memory on the 2000 X in part binning is unknown.

I've seen DS4000 units out in the field from before they were released, and right up to now. Usability I feel they're slower and harder to use than the benchmark Tektronix competitors which themselves are "adequate".

[jjoonathan]

Thanks for the graphs, much easier to read than tables!

Rigol wins on memory depth and hacked bandwidth, Keysight wins on decode and responsiveness (not just wfm/s, the differences in UI responsiveness are honestly more pronounced than the waveforms themselves, which are "pretty good" on both in my book, though that graph by Someone is telling). As for price, that's primarily a function of how the planets are aligning on ebay at the moment. If I had to buy right now I'd probably go for the 200MHz MSOX2024A for $1700. Six months ago I picked up my DS4014 for $1500. No regrets, but things might be different if I didn't already own a Saleae.

[TheSteve]

Hmm, how about a DSOX3014A for 1700 from Keysight? Easy to enable everything for the most part. If you wait for Keysight to list one on ebay you can get a deal.

[mrpackethead]

Hmm, how about a DSOX3014A for 1700 from Keysight? Easy to enable everything for the most part. If you wait for Keysight to list one on ebay you can get a deal.

100Mhz is'nt enough bw for me.. I need to do ethernet eye diagrams.. so minium 200Mhz. ideally more.

[TheSteve]

Hmm, how about a DSOX3014A for 1700 from Keysight? Easy to enable everything for the most part. If you wait for Keysight to list one on ebay you can get a deal.

100Mhz is'nt enough bw for me.. I need to do ethernet eye diagrams.. so minium 200Mhz. ideally more.

If you were to enable all of the software options you'd also get a bandwidth increase to 200 MHz.

[jjoonathan]

Because Real Men only need one and a half harmonic 

Seriously, though, it sounds like the DS4k is the appropriate choice here. 500MHz gets you 4+ harmonics rather than 1+ and deep memory isn't going to hurt for looking at packets and/or sticking them in segmented memory.

[mrpackethead]

and for this particular job, you can even get around having to have a really expensive diff probe.

[TheSteve]

Because Real Men only need one and a half harmonic 

Seriously, though, it sounds like the DS4k is the appropriate choice here. 500MHz gets you 4+ harmonics rather than 1+ and deep memory isn't going to hurt for looking at packets and/or sticking them in segmented memory.

I wouldn't argue at all. My post was more suggesting that if someone did consider the Keysight DSOX2xxx series over the Rigol they should seriously consider the DSOX3xxx instead as they can be had for very nice pricing with some patience/effort.

[Omikron]

Here's my silly "upgrade" to my MSO4014, inspired by another thread...



And if you can't tell what's different...



 

[jjoonathan]

Sweeet overmoulded knobs you got there  O0

Now where'd you get them? 

[Gen2ux]

Very nice!  I may have to do the same to mine.  I'll ask too, where did you get the knobs?