Rigol MSD2022A vs. Rigol DS4014 Scopes

[IDEngineer]

It's a gut feeling but people using the MSO seem to be less happy than DS users.

Arrrgh!  That puts me right back where I started. All I really want is a four channel 2000A series scope. I'm considering the MSO because its digital channels can (sometimes) substitute for the missing analog inputs. The alternative is to nearly double the price and halve the bandwidth to get four channels on a 4000A series. Grrrrrrr. 

[Pinkus]

Thats why you should add a Keysight 3000 series to your list

[JackP]

Or a DSO-X 2000A! £1400 for the barebones four channel model (or £800 for two), hack it to 200MHz, with ALL options (decoding, ARB...) and unlock the preinstalled digital channels?

[H.O]

The alternative is to nearly double the price and halve the bandwidth to get four channels on a 4000A series. Grrrrrrr.

It's probably not a fair comparison but if you pay twice the price for a DS4k (compared to a DS2k) you can, if you want, unlock the bandwidth to its full 500MHz. It's still twice the price  buy you get double the number of channels, 2.5 times the bandwidth (compared to a 200MHz DS2k), more sample memory, higher capture rate, larger screen and most likely something else I've missed.

You won't get the 500uV/div available on the DS2k though.

More food for thought.....

[rsjsouza]

JackP has a point, although you end up with less bandwidth, less sample rate and less memory than a Rigol DS4014 if you go for the DSOX2004A. For some the smaller memory is not a problem (discussed on some threads around), while others think the deep memory of DS4k is a joke without search and memory segmentation. The bandwidth can be improved on both models, but the DS4k has the edge - again, some think it is not worth upgrading the bandwidth but instead buy it from the get go.

To sweeten the DSOX deal, Keysight was running a promotion where you got all decoding options for free. This thread talks about the promotion, but the user there (dadler) went through a heck of problems to enable these options for free (just keep reading the thread).

[H.O]

while others think the deep memory of DS4k is a joke without search and memory segmentation

The DS4000 does have segmented memory.
It does not have a specific function/feature called search and I'll have to admit I don't know how such a function works on the scopes that has it (enlighten me please) but you can, for example, perform analysis on the frames (up to ~200k of them) in memory and when complete you can jump back and forth between the frames that failed the test - and of course you can scroll thru all the frames at "any" rate you want.

[hendorog]

while others think the deep memory of DS4k is a joke without search and memory segmentation

I think this is incorrect.

Memory is segmented on the ds4k based on each trigger event. How do other scopes do it?
Memory search is simplistic - based on variation from a template or variation from the average. I expect other brands have more features here.

This doc is hard to read, but gives the idea
http://beyondmeasure.rigoltech.com/acton/attachment/1579/f-02fa/1/-/-/-/-/file.pdf

It's probably not a fair comparison but if you pay twice the price for a DS4k (compared to a DS2k) you can, if you want, unlock the bandwidth to its full 500MHz. It's still twice the price  buy you get double the number of channels, 2.5 times the bandwidth (compared to a 200MHz DS2k), more sample memory, higher capture rate, larger screen and most likely something else I've missed.

^^ +1 that's what made the ds4k for me. There isn't anything else that can compete in that space. If you don't need the bandwidth/sample rate then you are lucky, as there are many more options!

[rsjsouza]

Thanks H.O and hendorog. I stand corrected on the segmentation.

The memory search on other brands is more flexible. I am familiar with WaveScan on LeCroy, which has more features, although some only applicable to their built-in LA.
 

[alank2]

Arrrgh!  That puts me right back where I started. All I really want is a four channel 2000A series scope. I'm considering the MSO because its digital channels can (sometimes) substitute for the missing analog inputs. The alternative is to nearly double the price and halve the bandwidth to get four channels on a 4000A series. Grrrrrrr. 

This is exactly where I was.  I even went through the trouble of trying two DS2072's side by side but was not happy with it.  I picked up a DS4014 and have not looked back.  It does all I ask of it.  You can run two channels (1 and 3) a 4GSa/s or all 4 channels at 2GSa/s.  It can be unlocked to higher bandwidth if you like as well.  My only concern besides the price was the noise people reported, but the noise doesn't bother me.

My opinion is that the scope based LA's are just not as easy to work with as something like a Saleae logic analyzer, in fact I use my older model Saleae 16 far more than I use the scope and it was 10% of the scope's price!  Saleae has USB3 versions out now that are much faster than the one I have.  I've tried to use the built in decoding on many Rigol's (DS2K, DS4K, older models) and it is not in the same class of usability.

[IDEngineer]

My opinion is that the scope based LA's are just not as easy to work with as something like a Saleae logic analyzer, in fact I use my older model Saleae 16 far more than I use the scope and it was 10% of the scope's price!  Saleae has USB3 versions out now that are much faster than the one I have.  I've tried to use the built in decoding on many Rigol's (DS2K, DS4K, older models) and it is not in the same class of usability.

Right, and agreed, thanks to others who chimed in the the same observations (especially from the perspective of protocol decode). I've removed protocol decoding as a primary criterion from the decision process. The only downside that I see is that captured analog and digital information isn't on a common display, but if you're willing to act like you're watching a ping pong match between two screens you can get past that. 

This is exactly where I was.  I even went through the trouble of trying two DS2072's side by side but was not happy with it.  I picked up a DS4014 and have not looked back.

I know this makes me look wishy washy and indecisive, but after a phone call with someone at Rigol this morning I'm reconsidering the 4000A series. Mostly because they confirmed my fears that the implementation of capture/playback on the 1000Z series is very limited, and that there's no magic special secret-moose-handshake four channel version in the 2000A series. If you want to capture four analog channels with a usable human interface, you're pretty much stuck with the 4000A.

...which, at that price, makes the KeySight 2000 series very competitive. You can get a 200MHz four channel DSOX2000 for basically the same money, AND they are tossing in all the options for free. I'm waiting for a call back from Rigol, we'll see what they can do, but that's some serious competition.

[atx]

Are you talking about only the LA digital channels here or does the decoding fail on the analog channels on the MSO2k too?

It's a gut feeling but people using the MSO seem to be less happy than DS users.

I do not believe that there is a difference between analog and digital decoding.  Would need to verify this.  Also the Rigol's statement that decoding only works with what is on the screen is kind of confusing.  For example, when decoding a 8MHz SPI signal using 1GHZ sampling and maximum memory, you can scroll through the data and everything is honky dory even though you only see a tiny bit on the screen.  In this setting the entire memory is decoded correctly.  But from a measurement point of view that is a total waste of memory.  You would scale down to something like 25Mhz or 50MHz to be on the safe side.  The Rigol has a weird sampling rate control.  you have to use the horizontal scale knob to change the sampling rate and not every turn changes it.  No direct control of the sampling rate is possible. The next rate down is 250MSa/s, then 25MSa/s with both analog channels off and maximum memory (28Mpts).  Here the decoding fails already at 250 MSa.

With a smaller memory let's say 7Mpts, you get different sampling options. Now you can sample at 1GSa, 500MSa, 125MSa, 25MSa, etc.  Decoding starts failing at 25MSa.  As you can see this gets rather complicated as it includes a variety of parameters.  Then, there are uncontrollable freezes in these operating modes which are annoying and render the device useless for up to a minute...It will eventually recover (the processor is probably overloaded)

The statement that users are less happy with the LA section seems to be correct.  At least it is true for me.  Many functions like the horizontal scale implementation in the analog section are fine but then the same implementation has been used on the digital LA side where it doesn't make a lot of sense IMHO.

atx

[hendorog]

I do not believe that there is a difference between analog and digital decoding.  Would need to verify this.

Yep, and please let us know which settings work/fail so we can replicate on the DS.

Also the Rigol's statement that decoding only works with what is on the screen is kind of confusing

Yes, but it makes sense to me when taken literally - nothing is decoded unless it's on the screen.
The triggering options on the various protocols are pretty good, so combined with the segmentation its fine.

Then, there are uncontrollable freezes in these operating modes which are annoying and render the device useless for up to a minute...It will eventually recover (the processor is probably overloaded)

I don't recall seeing anything like this on the DS. If you can work out the details I'll try and replicate when I get home.

[Muxr]

Arrrgh!  That puts me right back where I started. All I really want is a four channel 2000A series scope. I'm considering the MSO because its digital channels can (sometimes) substitute for the missing analog inputs. The alternative is to nearly double the price and halve the bandwidth to get four channels on a 4000A series. Grrrrrrr. 

This is exactly where I was.  I even went through the trouble of trying two DS2072's side by side but was not happy with it.  I picked up a DS4014 and have not looked back.  It does all I ask of it.  You can run two channels (1 and 3) a 4GSa/s or all 4 channels at 2GSa/s.  It can be unlocked to higher bandwidth if you like as well.  My only concern besides the price was the noise people reported, but the noise doesn't bother me.

My opinion is that the scope based LA's are just not as easy to work with as something like a Saleae logic analyzer, in fact I use my older model Saleae 16 far more than I use the scope and it was 10% of the scope's price!  Saleae has USB3 versions out now that are much faster than the one I have.  I've tried to use the built in decoding on many Rigol's (DS2K, DS4K, older models) and it is not in the same class of usability.

While it's really nice to be able to trigger an analog channel on your scope on a decode value, I have to agree, Saeleae software is really well done.

[IDEngineer]

While it's really nice to be able to trigger an analog channel on your scope on a decode value, I have to agree, Saeleae software is really well done.

Any chance those LA's can emit a trigger when they match a pattern? Plug that into your scope's external trigger input, and you'd have the best of both worlds.

[IDEngineer]

...at that price, makes the KeySight 2000 series very competitive. You can get a 200MHz four channel DSOX2000 for basically the same money, AND they are tossing in all the options for free. I'm waiting for a call back from Rigol, we'll see what they can do, but that's some serious competition.

A friend who works for KeySight called this afternoon, leaning on me HARD to go with the DSOX2024. He spelled it out: For about $2700, you get 4A @ 200MHz, plus they're enabling every option so you get all the serial decodes. Those options also include a function generator and onscreen DMM, features that the Rigol 4000A series doesn't even offer. (Rigol's integrated function generators stop at the 2000A series.)

KeySight or Rigol.

Rigol or KeySight.

Nice problem to have, but I can make an argument both ways.

[alank2]

A friend who works for KeySight called this afternoon, leaning on me HARD to go with the DSOX2024. He spelled it out: For about $2700, you get 4A @ 200MHz, plus they're enabling every option so you get all the serial decodes. Those options also include a function generator and onscreen DMM, features that the Rigol 4000A series doesn't even offer. (Rigol's integrated function generators stop at the 2000A series.)

I do love my DS4014, but that is a ton of stuff with the KeySight, hard to turn down those extras.  Good luck with your decision.

[atx]

I do not believe that there is a difference between analog and digital decoding.  Would need to verify this.

Yep, and please let us know which settings work/fail so we can replicate on the DS.

Decoding remains a mystery.  In the example attached I use a 16 bit data word that is incremented each frame defined by CS and a 8MHz clock signal.    I connected the analog channels to SCK and SDA.  The digital channels use the SCK, SDA and CS signals.  I use the first decoding on the digital channels (with CS) and the second decoding on the analog (with timeout).  I triggered on the digital channels with SDA=0x0.  At a sample rate of  200MSa (analog) or higher everything works fine.  When switching down to 100MSa both decoders fail but differently as you can see.

Yes, but it makes sense to me when taken literally - nothing is decoded unless it's on the screen.
The triggering options on the various protocols are pretty good, so combined with the segmentation its fine.

Not really as you can scroll through the table without changing the screen at all.  The table shows you all decoded values independent from what is on the screen.   Also, I posted scenarios where the data is clearly shown on the screen and decoding failed.  I think it is more a memory dependency.  Once the memory gets too large, decoding fails.  I had a theory that it works under any circumstances when the memory is set to AUTO.  But that is not true either.  Once you reach a certain memory size the decoding may fail.

Then, there are uncontrollable freezes in these operating modes which are annoying and render the device useless for up to a minute...It will eventually recover (the processor is probably overloaded)

I should have said spontaneous freezes.  I cannot reproduce those.  But they appear when a large memory is involved.  Sometimes more sometimes not at all.  I need to further investigate this.

Bottom line is that decoding is highly unreliable and thus more or less useless because you have to always verify it first.  Thus, paying for this option is not recommended IMHO.

atx

[Muxr]

While it's really nice to be able to trigger an analog channel on your scope on a decode value, I have to agree, Saeleae software is really well done.

Any chance those LA's can emit a trigger when they match a pattern? Plug that into your scope's external trigger input, and you'd have the best of both worlds.

Saleae doesn't have that option, and it would be difficult. Because you set the capture duration, and a non decoded trigger condition (usually just the logic level like a rising edge), the capture takes place and the client app does the decoding. So a real time ext. trigger out would not be real time.

This is my experience with the lower end 8 channel and 16 channel Saleae LA's. Not sure if the higher end ones are any different, but I would imagine not since the decoding is done in the computer at the application level, after the capture has taken place.

This approach has advantages, things like more advanced decoding possible and virtually unlimited memory, but external trigger out would not be possible.

[IDEngineer]

Saleae doesn't have that option, and it would be difficult. Because you set the capture duration, and a non decoded trigger condition (usually just the logic level like a rising edge), the capture takes place and the client app does the decoding. So a real time ext. trigger out would not be real time.

Yep, the LA hardware would have to directly handle the trigger detection. But if you did that, you could emit a trigger signal and still let decoding happen on the host machine. That would be rather nice in quite a few situations I can imagine.

[IDEngineer]

Wow, it's been a wild morning of trying to buy an oscilloscope. Get this:

I worked out a deal for a Rigol DS4024A. I know, I know... I reasoned the 4000A series was too expensive for our needs, etc. etc. I won't bore you with the details but in the end this was the sweet spot in the compromise curve. I literally had the company credit card out, reaching for the phone....

...and my buddy who works for Keysight calls and says "You bought that scope yet?" I had talked with him yesterday about the possibility of any discounts, brother-in-law "deals", and the like. They used to have such things, but after Agilent split away from HP they all went away and he reported nothing new on that horizon, so I dismissed any shot at a reasonably priced Keysight that could come anywhere close to the Rigol 4000A series.

But it apparently really bugged him that we were considering "leaving the family". So he shook the trees in his organization and they are now working on a deal to basically match our price for a Rigol 4024A with a DSOX3024T. Same bandwidth, same channel count, including all the enable options, etc. On a THREE thousand series scope, not the DSOX2000 series we had been considering. I don't care about the touchscreen - would rather not have it, actually - but it's the latest product and costs the same as the old A series and comes with a few more things standard.

Sheesh. It's a nightmare of wonderfulness. Rigol is really extending their hand here, I can't share the details but they really made an effort and I basically swore a blood oath that we'd take the deal. But then my almost-family buddy of 35 years, whose now-grown kids have called me Uncle since they were born, who has worked for HP and then Agilent and now Keysight the entire time, is also swinging the bat hard to arrange a killer deal on an awesome scope.

And then the products. I can argue for and against both. Keysight wins on support. But Rigol wins on many important technical specs. The 3000T series is relatively new so should get support for a good long time, while the 4000A series has been around a while so maybe it's nearing end of life. But then the 4000A series probably has most of its bugs worked out already, while the newness of the 3000T risks making us a beta site for new firmware.

The 3000T has 4Msamples of memory that can be divided into (IIRC) 1000 segments. The 4000A has 140Msamples that can be divided into 200K segments. Clear winner: Rigol. But the Keysight does a whole bunch of stuff in hardware that the Rigol handles in code, making things slower. Advantage: Keysight. Rigol has true hardware resolution down to 1mV/div, while Keysight only goes to a real 4mV/div and just optically multiplies from there. And so it goes.

Socially, I'm going to look like a jerk to someone. Technically, we gain some features with one, and some features with the other.

I know we have owners of both here. Opinions? Help a guy out. At the same price, which would you take?

[dr.diesel]

Unless you really need the memory depth, get the Agilent, hands down I'd say.

I've got a MSO4000, and although it's much better today, Rigol support totally blows, it's taken a long time to get the firmware straightened out.

[hendorog]

Given that choice I would go with DSOX3024T for sure Never used one though, but it just sounds like a fantastic offer to get a modern brand name scope and all of the options.
Perhaps someone who has used both recently can comment.

If only they would offer that bundle to everyone, then there might be some competition

[IDEngineer]

Unless you really need the memory depth, get the Agilent, hands down I'd say.

Yeah, that's the struggle I'm having too. The memory difference is just HUGE. I can always wait for Rigol's software to slowly do what the Keysight hardware does faster, but I can't smash more memory into the box. 35x or 70x the depth (depending on which spec sheet you read) is a LOT.

[dr.diesel]

Makes sure and review this thread before making your final decision:

https://www.eevblog.com/forum/testgear/rigol-mso4000-and-ds4000-tests-bugs-firmware-questions-etc/

[IDEngineer]

Yep, I've read that one (and many others) all the way through.

Lots of data out there. But some is poorly correlated, as with earlier in this thread where folks claimed a bug still existed when others here had documented its fix over a year ago.

Also, I wonder if Rigol has more bugs - or if those of Agilent/Keysight are just less documented. With Rigols being so much more affordable, it might be that they just get more exposure with the homebrew crowd who are perhaps more likely to post their findings. Not sure. But the lack of postings isn't proof of lack of bugs. Definitely Keysight gets the nod for "more likely to respond and fix a bug".

But the primary reason for needing this new scope - at least for now - is as an analog data capture system. Hence why memory looms so large as a consideration.

Just got a text that the Keysight people are having a meeting right now to discuss options. We'll see.