MSO and USB LA owners

[Howardlong]

So there's a school of thought that says an MSO nowadays is of limited value compared to a DSO plus a USB LA like a Saleae or Logicport for example.

Typically MSOs have higher sample rates than USB LAs, have time correlation with analogue channles and offer much more comprehensive triggering options, whereas USB LAs offer a much larger screen and sometimes much deeper memory.

So which do you prefer?

[Sal Ammoniac]

I prefer an MSO for the reasons you state.

I have a Saleae Pro 8 that I use for debugging slow serial protocols, such as I2C, but it's really not much more than a toy. You can't set up anything other than a very simple trigger, and you can't group channels together in a bus like you can with a real analyzer. This last one makes it almost useless to me for any kind of hard-core debugging.The Saleae is big on eye candy but lacking in functionality.

[LabSpokane]

MSOs are obviously much better at analyzing signal integrity and timing issues.  USB Logic analyzers are much better at collecting long streams of decoded data for subsequent analysis. 

Both have a place.  Both have limitations that make one better than the other. 

[Berni]

I have two MSO scopes yet they rarely get any use. Thing is that almost everything is serial these days so usually the 4 analog channel are enough. It's always better to use those if you have them because it shows you any signal integrity issues, floating, two outputs fighting on a line etc.

When my saleae USB LA comes in then it is for looking trough large amounts of data where I can just record for seconds or even minutes and then ponder over it at my own pace. It's easier to look through it in a PC and I can export it in to other tools.

If I worked on old computers with parallel data buses going around that's where a real logic analyser comes in handy.

[D3f1ant]

Long captures of data I use a usb LA, when debugging a mixed analog digital, the mso is more useful. Seeing if there is psu glitches in time with bus errors a mso is ideal. But even the longest deep memory scope can't compare to streaming data in via usb for offline analysis, we're talking gigabyte captures. Tiny scope screens are not ideal for viewing 8+ channels...a pc just handles it better. I would buy a mso scope first, its more versatile and gets used more often. Basic LA like saleae is cheap enough that can pick one up if/when you need it.

[Performa01]

It obviously depends on the specific task and on the quality of both the MSO and LA.

Typically, an LA is more powerful than the digital part of an MSO, because it can have more channels and more sophisticated trigger options by providing several trigger levels to set up a trigger sequence. The biggest advantage is the huge choice of protocols available for bus decoding.

The MSO on the other hand, has the correlation between analog and digital channels and provides advanced trigger modes like pulse width, pulse timeout, runt and maybe some others, that are normally not available for digital channels, neither on the MSO nor on the LA. There are usually only few Protocols available for bus decoding.

A good USB-LA isn’t cheap and can easily exceed the price of a low cost MSO.

So it realy comes down to the task. When troubleshooting a mixed mode design, like e.g. a discrete precision ADC, the correlation between analog and digital matters and the MSO is the obvious choice.

When debugging a purely digital circuit, including the decoding of some exotic protocol, the LA is right tool.

With regard to screen space and UI when dealing with lots of channels, there are also USB-MSOs which can provide the best of both worlds in one instrument for certain tasks. There are more bus decoders than on the average MSO – but still not nearly as many as on a LA – and the number of digital channels is still limited to 16.

[TiN]

Where is option for real LA?
Very few USB LA's offer state and synchronous analysis mode. Tek page explains this bit further.

D3f1ant example is a great one, showing lack of this mode in both MSO and USB LA's, requiring gigabytes of useless samples to capture that short "event", while even 30 year old HP or Tek TLA714 can easily capture samples only when you need them, and wait for trigger on anything you don't need.

[jjoonathan]

Where is option for real LA?
Very few USB LA's offer state and synchronous analysis mode. Tek page explains this bit further.

D3f1ant example is a great one, showing lack of this mode in both MSO and USB LA's, requiring gigabytes of useless samples to capture that short "event", while even 30 year old HP or Tek TLA714 can easily capture samples only when you need them, and wait for trigger on anything you don't need.

If you need it you need it, but the price of a LA with state machine triggers buys a lot of DRAM.

[tautech]

And of course the memory depth of any MSO need be taken into account, they are each vastly different.

Siglent SDS2000X series MSO:

16 Ch
14Mpts/CH

Datasheet:
http://www.siglentamerica.com/USA_website_2014/Documents/DataSheet/SDS2000X_DataSheet.pdf

[Fungus]

What's the budget for the "MSO"?

Doesn't it depend an awful lot on how much the MSO costs? 

[Berni]

What's the budget for the "MSO"?

Doesn't it depend an awful lot on how much the MSO cost? 

Also a good point, my MSO9000 has 1Gpt of memory and can sample at up to 10GS/s on digital but I never used all of the avalable memory.

EDIT: sorry I mean 128 Mbit up to 2GS/s for digital channels

[jjoonathan]

What's the budget for the "MSO"?

Doesn't it depend an awful lot on how much the MSO costs? 

Wherever I look I look the price is always the same: "Call Us" 

EDIT: Well, it's either "Call Us" or "I got one for (statistically improbable price) on ebay!" 

[tautech]

What's the budget for the "MSO"?

Doesn't it depend an awful lot on how much the MSO cost? 

Also a good point, my MSO9000 has 1Gpt of memory and can sample at up to 10GS/s on digital but I never used all of the avalable memory.

Not quite, you're quoting Option 500 analog specs:

From the Datasheet:

Acquisition: digital channels
Maximum real time sample rate 2 GSa/s
Maximum memory depth per channel 128/ 64 Mpts with 2 GSa/s. 64/32 Mpts with sampling < 2 GSa/s (single/repetitive mode).

[Fungus]

What's the budget for the "MSO"?

Doesn't it depend an awful lot on how much the MSO cost? 

Also a good point, my MSO9000 has 1Gpt of memory and can sample at up to 10GS/s on digital but I never used all of the avalable memory.

So... don't keep us in suspense! How much do they cost?

[Berni]

Ah my mistake sorry. This shows how much use the MSO feature sees on my side. But I'm sure someone else out there couldn't live without the analog and digital correlation benefits of it.

[Berni]

Also a good point, my MSO9000 has 1Gpt of memory and can sample at up to 10GS/s on digital but I never used all of the avalable memory.

So... don't keep us in suspense! How much do they cost?

It goes in to the "I got it on eBay for A improbably low price" bin. I bought it from Keysights eBay store as a DSO9254H because they got discontinues for spot on 6 grand. Later on trough software turned it in to a MSO9404A with all options unlocked. It appears to be made in 2012 and had only been used for about 3 months all together by looking at windows logs.

Edit: Before anyone asks no it's not software upgradable to 4GHz, The 2.5GHz In the H version is the 4 GHz in the A version by hardware. The software downsampling used to get the 10 bit resolution limits it to 2,5GHz

[nctnico]

I have both an MSO and logic analyser (Tektronix TLA715 + 32Mpt 8Gs/s acquisition module). For debugging digital circuits I usually just use the TLA715. In the past I have used an MSO for debugging FPGAs but even though an MSO works for 99% of the logic analysis tasks the logic analyser still wins due to it's triggering and selective recording capabilities.

[aandrew]

I've got both; I use either Saleae or sigrok firmware on my USB LA depending on what I'm trying to do (being able to capture and manipulate over an ssh connection via remote is worth learning how to use sigrok). The logic analyzer in my MSO (MSO5074FG) is iffy at best, but it wasn't purchased for its LA; it was purchased because it's a cheap, hackable scope that just happens to have a kind of shitty LA and FG bolted on. If the LA and DSO sides were better synchronized and it had *some* triggering ability beyond the basics I'd use it a hell of a lot more, but as it is I like sigrok and the USB LA.

[rs20]

When using a borrowed USB LA for the first time, I was absolutely astonished to discover that the LA could not be configured to display a live stream of SPI data on my screen. I had an accelerometer taking measurements, and I just wanted to see them live, on screen, in response to me manipulating the sensor. I started a thread on this forum somewhere, and it was quickly pointed out to me that this "is not the point"/"is not the intended use" of USB LAs.

But I just can't get my head around this fact, given that it would be trivial for them to support this functionality. I accept that this is the way that LA's are, apparently, but still find it deeply difficult to understand. It's for this reason that I bought an USB MSO* in preference to a USB LA.

* Don't worry, I have a real display-and-knobs DSO as well, but a USB MSO seemed like a reasonable second purchase, especially as I have need of collecting lots of analogue data to disk.

[LabSpokane]

When using a borrowed USB LA for the first time, I was absolutely astonished to discover that the LA could not be configured to display a live stream of SPI data on my screen.

Someone correct me, but I believe low-end USB LAs of the Saleae variety (what I have) don't decode in hardware.  They actually are directly sampling and storing samples of the waveforms and can only store in computer RAM.  This is done for acquisition speed reasons.  I don't think changing this is trivial, at least not on a $150 USB LA. 

 Keysight (and others?) decode in hardware, and that makes live streams practical.

[Berni]

Yes the Saleae is very simple in terms of hardware. The older 8 bit ones contain only one chip inside. It is a MCU with built in USB that uses a DMA to barf the contents of an IO port register out of it as quickly as it can. The PC side software then does all the smarts of waveform compression, ram as sample memory and any decoding. Newer ones have a chip or two more to allow for faster sample rates and analog channels but that's about it. I love having essentially unlimited sample monody.

[Sal Ammoniac]

Yes the Saleae is very simple in terms of hardware. The older 8 bit ones contain only one chip inside. It is a MCU with built in USB that uses a DMA to barf the contents of an IO port register out of it as quickly as it can. The PC side software then does all the smarts of waveform compression, ram as sample memory and any decoding. Newer ones have a chip or two more to allow for faster sample rates and analog channels but that's about it. I love having essentially unlimited sample monody.

In the case of the Saleae it's not the limited hardware that bothers me. That's to be expected given the price. It's the Windows application that is lacking. It doesn't even support such basic logic analyzer functions such as grouping multiple signals together into a bus. If I have, say, four signals I'm using as state output, I'd like to be able to group them and read the state as a decimal number on the display rather than having to decode four separate lines as binary and combine them in my head.  I could go on and on, but it's clear that they have been very slow to add features LA veterans have come to expect in a LA.

[pascal_sweden]

That's interesting to hear about the software of the Saleae. I though that the software was their only selling point... and now I hear that their software is no good.

Some people buy a Chinese rip-off and use the Saleae software for all its goodness

So the software isn't something to write home about? =)

[rs20]

When using a borrowed USB LA for the first time, I was absolutely astonished to discover that the LA could not be configured to display a live stream of SPI data on my screen.

Someone correct me, but I believe low-end USB LAs of the Saleae variety (what I have) don't decode in hardware.  They actually are directly sampling and storing samples of the waveforms and can only store in computer RAM.  This is done for acquisition speed reasons.  I don't think changing this is trivial, at least not on a $150 USB LA. 

 Keysight (and others?) decode in hardware, and that makes live streams practical.

Anyone please correct me too!

The LA has an "infinite " (i.e. hard-drive-backed) memory capability, which means that the hardware is streaming data to the software. Since the computer software is taking the data and dumping it to the hard drive, there's nothing stopping them (apart from laziness) from doing SPI decoding in parallel at some clock speed. Maybe hardware is beneficial at extremely high clock speeds, but to say that the hardware is totally incapable is indeed wrong. It's could just be a software feature which works with the current hardware, at least for modest clock rates.

Alternatively, making the tiny change to the hardware that it can re-arm its own trigger, and thus send a stream of little blocks of data, makes it even easier for the software to handle. It could be much like the SPI decoding on Rigol scopes -- when the trigger rate gets too high, the SPI decoding can't translate every waveform that comes through. But Rigol (and others with no hardware decoding support) still provide the feature, since it's still hugely valuable even if it only displays a subset of SPI packets, and because it works perfectly at lower data rates too. Just blanket banning continuous triggering is really just cutting corners in software and firmware, and indefensible from a hardware feasibility standpoint AFAICT.

[nctnico]

When using a borrowed USB LA for the first time, I was absolutely astonished to discover that the LA could not be configured to display a live stream of SPI data on my screen.

Someone correct me, but I believe low-end USB LAs of the Saleae variety (what I have) don't decode in hardware.  They actually are directly sampling and storing samples of the waveforms and can only store in computer RAM.  This is done for acquisition speed reasons.  I don't think changing this is trivial, at least not on a $150 USB LA. 

 Keysight (and others?) decode in hardware, and that makes live streams practical.

AFAIK a logic analyser has never been about showing live streams and I don't know of any logic analyser which can show live streams at a decent speed. A logic analyser has always been about capturing signals and analysing them afterwards. At least that is how I have always been using them.