MSO, or DSO + LA, or MSO + LA ?

[Electro Fan]

This post is somewhat related to the question that has come up here recently and previously regarding "How many channels are needed on an oscilloscope?"  Obviously, the answer depends on the use case - but that in turn leads to the question by users who aren't sure what they might encounter such as "How do I know what I might need (or want) later?"

After working with both 4 channel (analog) oscilloscopes and a 2 channel (digital) oscilloscope I recently purchased a Logic Analyzer - an Intronix LogicPort.  The Intronix LogicPort has turned out to be an excellent LA - and it has influenced my thinking somewhat on how many channels are needed/desired, and it has even more so influenced my thinking on the question of "MSO, or DSO + LA?"

When I first got the EEVblog addiction a couple years ago one of my main goals was to be able to experiment with mixed signal types (analog and digital) in way that would enable me to see the relationships between the two.  At that time (when I first got started with EEV) I had pretty much nothing other than one analog scope and a very simple DMM.  When I got the bug for a digital scope I quickly became enamored with the idea of a MSO.

My theory was that a MSO would give more insight to the timing correlations between analog and digital signals than a DSO + a LA.  And I was sure that turning knobs and pushing buttons on a scope would be more satisfying than using a mouse with a computer display.  Unfortunately, I couldn't find a MSO that I could justify so I went with the Rigol DS2072.

I fell in love with the DS2072.  With the exception of a few fiddly small knobs the knob turning and button pushing all became second nature - but I kept wishing for the ability to examine/analyze digital signals in relation to analog signals (primarily out of curiosity). 

What I now think is that the software operation and viewing experience with the LogicPort is likely to be much more useful/easy/compelling than most of what could be done on a MSO with knobs and buttons plus a relatively small oscilloscope screen (compared to a computer monitor).  The LogicPort software UI is really good - not perfect - but really good.

Having said all that, and having modified the LogicPort to provide a trigger out function, and having actually played with the ability to trigger a scope from the LogicPort I now think the best solution would be a MSO and a LA. 

Here's why:  I think for doing lots of digital signal analysis using a good LA with a computer monitor would be preferred over just a MSO.  To be square, I can't say this with 100% certainty since I don't have a MSO, but with a standalone USB LA you get the benefit of the large screen and mouse functionality, and then with the trigger out to the scope you can get yourself in the realm of what causes what with mixed (digital and analog) signals.  Then, if you have a MSO, when you really want to zoom in on timing correlations you could put the digital and analog signals of most interest on the MSO.

So, I'd say if you can afford a 4 analog channel MSO, knock yourself out and go for it - but I think you might still want a PC-based LA.  On the LogicPort in just a few days I'm approaching 8 digital signals on the screen at one time and the system will go to 32 (plus 2 others for external state clocking).  I'm pretty sure that by the time you add 8 or more digital signals along with 1-4 analog signals on an oscilloscope screen it would become a challenging User Interface.  If you can only afford/justify or only need a 2 analog channel MSO (such as a Rigol MSO2072A) I think you will find that a 2 analog channel MSO plus a LA will also make a great combination. 

In summary, I started out thinking a MSO was the ultimate solution but after using the Intronix LogicPort I'm now thinking a MSO + a PC-based LA is the way to fly.  YMMV,

EF     

- and if you are on a tight budget a DS1054 + a LA would be A-OK too

[ez24]

Thanks for the insight.  Do you have any examples on what analog + digital that you are looking at?

[marshallh]

DSO+LA

8/16 channels is great until it isn't.

[Electro Fan]

Thanks for the insight.  Do you have any examples on what analog + digital that you are looking at?

Just Arduino stuff so far - not so interesting - but it's been a good DUT to help me figure out how to operate the LA.

[nctnico]

@Electro fan: I think that once you have a good MSO the logic port retires into a drawer. The biggest advantage of an MSO is that it can show digital signals quickly. A logic analyser is more like capture first and then analyse.

[Electro Fan]

@Electro fan: I think that once you have a good MSO the logic port retires into a drawer. The biggest advantage of an MSO is that it can show digital signals quickly. A logic analyser is more like capture first and then analyse.

nctnico - Please say more about this... Thanks

[pickle9000]

Serial decoding in a scope is super handy. I prefer LA's to be PC based.

[JoeB83]

In my experience, a DSO and a logic analyzer is the way to go.

I've had a couple MSO scopes in my past, and I still have one. (Sold Tek MSO2014, still have MSO4054)

If I were to do it all it again, I'd go with a regular DSO and buy a separate logic analyzer.

I found the MSO's logic analyzer (at least Tek's MSO 2014 & MSO4054) limited, awkward, & difficult to use.

My advice? Find a scope you love, DSO only, then buy a logic analyzer that fits your needs.

[Kilrah]

I have a scope + PC-based "record then watch" LA, but my next purchase in that department will likely be an MSO. There are times where being able to sync an analog waveform and associated logic is useful, as is the ability to watch digital signals change in real time (you straight away get a feeling of what happens like how often each byte of a frame changes to help determine their use, which takes ages on a PC where you have to manually look at hundreds of frames).

The PC LA is still preferable in some cases too, so I'd say ideally I'd recommend an MSO + a cheap PC-based LA. If that PC LA has an analog channel or 2 that can also come handy.

[Howardlong]

@Electro fan: I think that once you have a good MSO the logic port retires into a drawer. The biggest advantage of an MSO is that it can show digital signals quickly. A logic analyser is more like capture first and then analyse.

I'd agree with this. The only time that my Logicport comes out is when doing highly parallel busses (in today's terms), which happens less and less these days. I think it's been at least six months since I last used the LogicPort, but I use the logic channels on an MSO almost every day. I think it depends a lot on what you do, my work is predominantly in mixed signal. There's less and less call for very parallel busses in my line of work.

I realise that there's a lot of justification for a PC based LA too, it's just that I found I hardly ever use it nowadays, and although I am sure other people have different thoughts on this, I find it just as easy to set up an MSO as I do a PC. Yes, the MSO screen might not have the resolution of the PC, but with the triggering, decoding and search facilities available now on modern MSOs to me it's the path of least resistance.

I would make sure though that you investigate any limitations that there may be on the decoding part of any MSO you choose. The decoder on the DS/MSO1000Z for example will only decode what's on the screen and only up to a limited number of decodes before the software decoder gives up due to undersampling. Only this small decode of a few bytes can be written to your USB stick, so the save facility is almost worthless. It also doesn't work with waveform record/segmented memory. There is also no search facility. Sometimes details like these aren't at all transparent when you read the datasheets or user guides. These may or may not be a problem for you depending on how you use the scope.

Part of my setup is that the scope is always out on the bench, with probes plugged in at the ready. The scope sits permanently directly above the main PC monitor. If the scope were put away every day, or even just probes disconnected and hung up, I might use the LogicPort a bit more: it's just a matter of convenience. As I already stated, I can also completely accept and understand that others prefer the PC based LA approach.

[Electro Fan]

@Electro fan: I think that once you have a good MSO the logic port retires into a drawer. The biggest advantage of an MSO is that it can show digital signals quickly. A logic analyser is more like capture first and then analyse.

nctnico - Please say more about this... Thanks

- when you say a MSO can show digital signals quickly, is that a reference to an easier setup process, or an easier viewing process, or something else?  Seems like with the LogicPort the set up is no big deal and then you have signals scrolling across the screen - which you can set to stop according to all sorts of triggers.  And if you want you can setup decoding ("interpreting" in Intronix speak).  I'm not at all adverse to a MSO (it was my original preference)... just trying to understand the differences.  My only frame of reference is the serial decoding on the Rigol DS2072; on the 2072 you have to set about half a dozen or more variables to decode, then you watch waveforms and decodes fly by on the screen.  Then you generally stop the decoding and pan around and zoom in and out.  Not sure how that is different than the LogicPort (or what the equivalent process would be on a MSO)?

Thanks for any further insight...

[nctnico]

In some cases it helps to see live data. For example if you have an I2C device which is regulary updated. With an MSO it is easy to spot the bits flipping so you don't have to decode and go through the whole lot. An MSO gives more of an overview of the data.

[Electro Fan]

In some cases it helps to see live data. For example if you have an I2C device which is regulary updated. With an MSO it is easy to spot the bits flipping so you don't have to decode and go through the whole lot. An MSO gives more of an overview of the data.

Sorry for being so slow on the uptake, but what would you see when a bit flips on a MSO that you wouldn't see on the LogicPort? 

On the LogicPort you can adjust the sample rate and zoom in and out on the fly as the waveforms (bits) and any decodes (strictly optional, can be turned on or off, same as on a scope) are populating the screen.

I'm missing the distinction....  Thx

[Howardlong]

In some cases it helps to see live data. For example if you have an I2C device which is regulary updated. With an MSO it is easy to spot the bits flipping so you don't have to decode and go through the whole lot. An MSO gives more of an overview of the data.

Sorry for being so slow on the uptake, but what would you see when a bit flips on a MSO that you wouldn't see on the LogicPort? 

On the LogicPort you can adjust the sample rate and zoom in and out on the fly as the waveforms (bits) and any decodes (strictly optional, can be turned on or off, same as on a scope) are populating the screen.

I'm missing the distinction....  Thx

Speaking for myself, the distinction with the LogicPort is that actually I do find the MSO significantly easier and intuitive to set up, particularly when considering triggering. In fact triggering alone would be a good reason to go for an MSO rather than a LogicPort, there is no serial bus triggering on the LogicPort, unless I am suffering operator error.

However, completely coincidentally, I ended up debugging a 1-wire bus this evening... and the only way to decode it interactively here was with the LogicPort, so it came out of the drawer for its six monthly shake down and worked flawlessly as always. There are no fewer than six MSOs in my lab at the moment, none can do 1-wire decode directly. Sure, SPI, I2C, I2S, CAN, LIN, USB, Flexray, MIL something or other, but not 1-wire. Where the LogicPort let me down, as is so often the case, is that despite the compression, it doesn't take much for it to run out of memory.

[Electro Fan]

Hi Howard,

Thanks for sharing your experience with the LogicPort and MSOs.

I'd definitely like to have a MSO at some point - but I'm still trying to understand what the reluctance is regarding the LogicPort (other than the memory limitation - I get that it doesn't meet some needs in the memory department).

For example I've seen comments on the forum from other users saying the LogicPort is good for parallel decoding but maybe not serial.

Below is a screen shot from a serial decode of RS232.  It came directly from the pins on a PC's serial port.  It doesn't seem to require any more effort to set up or use than the Rigol DS2072.  In the first image the LogicPort cursor showed that when the trigger occurred the A cursor was on a "w".  You can place up to 6 cursors and at a glance the table will tell you where all 6 are when the trigger occurs.  With a couple clicks you can tell the LA display to jump to any of the cursors, or the trigger, or the first or last sample, or the next or last transition.  I don't think my 2072 has comparable controls.  In fact, I think the LogicPort will do a bunch of stuff the scope can't do.  (For navigation the big knob on the Rigol 2000 series is superior to the twiddly smaller knobs on the Rigol 1000 series but I have a hunch that on the Tektronix scopes their Wave Inspector has some useful nav functionality - so I know it can be done on a scope, it's just a matter of at what price do the human factors start to kick-in with good functionality.)

So I realize the LogicPort is not a MSO but it just seems to me that having the ability to put a lot of digital signals (up to 32) on a screen (whatever size screen you want) and click on samples rate changes, adjust triggers lots of ways, use cursors to measure, etc, etc. isn't all that bad a deal.  Not to mention the ability to do state clocking (although I have to admit I haven't had a need for it just yet).

The images don't do justice to the way the LogicPort renders on a computer monitor - they are significantly shrunk down.  On the computer monitor the Logicport is very attractive.  (I should have picked a more attractive set of colors - assigning signal colors is easy.  I just happened to use the white/orange LA lead and I've developed a habit of coloring the screen waveforms to match the lead colors.) 

Anyway, I've gone from thinking that a MSO is the way to fly to thinking that a LA + a DSO is an economical way to fly to thinking that the cat's meow would be a MSO plus a LA (for lots of signal viewing).   EF

[Kilrah]

Sorry for being so slow on the uptake, but what would you see when a bit flips on a MSO that you wouldn't see on the LogicPort? 

On the LogicPort you can adjust the sample rate and zoom in and out on the fly as the waveforms (bits) and any decodes (strictly optional, can be turned on or off, same as on a scope) are populating the screen.

I'm missing the distinction....  Thx

Unless I'm mistaken, but I don't see any info that would say te opposite on the LogicPort website, this device is like other common PC-based LAs (saleae etc) and work on a "record then analyze" principle. You start sampling, it fills a buffer without showing any activity, then stops and lets you look at the stored data.

An MSO displays stuff live. If you connect it to an interface (I2C or whatever) where a frame is being transmitted and a byte in that frame is a counter, you'll set up an appropriate trigger and see that frame in real time on the display and will see the counter update live on the screen. If a bit in that frame is supposed to change every half second you can look at it and easily see whether it does change every half second, or only every second because of a bug even if there are hundreds of frames sent. With the "record then analyze" device you'd have to closely look at the value of the bit over hundreds of frames to see when it changes, then put markers and look at timings.

[Howardlong]

The LogicPort does have triggering, but only fairly basic edge and pattern triggering with the option to add a second basic trigger event, although how that works has always eluded me. You can't, for example, trigger on an I2C address or SPI bit stream pattern. In addition the memory, although compressed, is still fairly small compared to the deep memory of a scope these days.

While I like the Analog Discovery and rate it highly, the sample rate is only 100MSa/s is good enough for a lot of things, but might be limiting in some circumstances. It also has tiny memory and it only has very basic triggering capabilities. The LA and scope on the Analog Discovery aren't correlated either as far as I've ever been able to figure out.

The Saleae on the other hand streams directly to the computer so is limited by the computer's memory, however on the lower end Saleaes this means sampling rate suffers as a result. Like the LogicPort, there is a rudimentary trigger based on levels and/or edges, but no sequenced events. Certainly the low end devices max out for me at 16Msa/s for me. The newer Pro models may well have been engineered around these limitations, I don't have one so can't comment.

What is common to the Saleae, Analog Discovery and LogicPort is that you can't trigger on serial data content. That is a key distinguishing feature between these and modern day MSOs.