Digital 'scope with decoding or logic analyser?

[CJay]

Looking for opinions here.

As I mainly work on analogue/RF equipment I find there's not often a need for digital tools other than my trusty PICkits and my laptop but, occasionally, I need to dig into digital things and it'd be handy to be able to monitor and decode SPI, I2C, serial etc. transactions.

To that end, is a digital 'scope with decode options or a logic analyser 'better'?

I'm not looking to replace my analogue 'scopes or other predominantly analogue gear and (as I'm getting rather fed up of saying) money is tight so it won't be a $20K Keysight MXO or whatever, it will need to be a budget device and I don't mind buying cosmetically poor second hand if it works (or is repairable)

[nctnico]

I'd go for a digital scope with decoding because in many cases a problem is only truly visible in the analog signal.

[AndyP]

I use a mix of DSO (MicSig TO1104 @ home) to ensure the signal integrity is good then either the logic analyser on the scope (Rigol / Keysight / R&S @ work) OR USB logic analyser to decode I2C, SPI etc.

The big screen and mouse user interface on laptop/PC makes USB LA easy to use. (I use a DreamSourceLab DS Logic Pro - http://www.dreamsourcelab.com ), at $99 16 channel with a 16M buffer and up to 400MHz sample rate (4bit only) it has been fast enough for most of the boards I work on.
I've also heard recommended some £10 offerings that are 8 channel 25MHz which would do for the lighter applications I work with, that's cheap enough to try out and see if it does what you need.
 
I've not had the need to use a full blown stand alone logic analyser for quite a while and though the triggering options were useful, the speed and number of channels is really useful for looking a processor busses, but overkill for SPI / I2C traffic. (IMHO)

[CJay]

I'm leaning towards the USB LA because they can be a small piece of kit and I intend to keep my Tek/Hameg scopes

There won't be anything 'high speed' at the target end so one of the cheaper USB devices might be adequate, I'll order an eBay special if they're so cheap, recommendations?

[artag]

The only advantage of having the logic analyser in the scope is that it's sometimes useful to cross-trigger or view analog and digital waveforms at the same time. I've never met a digital scope/MSO that had anything near the capability of a USB logic analyser for unpicking serial bit streams. You can usually add further decoding to the USB analyser if you want to decode something it doesn't cover - see sigrok or the Saleae SDK, for instance.

[Gyro]

I'm leaning towards the USB LA because they can be a small piece of kit and I intend to keep my Tek/Hameg scopes

There won't be anything 'high speed' at the target end so one of the cheaper USB devices might be adequate, I'll order an eBay special if they're so cheap, recommendations?

It's pretty much a no-brainer from what you say. The 8-bit USB LAs are so cheap that even if you later come up with a higher speed need then you really haven't lost anything. If you want one from the UK then Hobby Components on ebay are pretty good (no association).


P.S. Some UK sellers indicate Saleae compatible and some aren't. It's a USB PID/VID difference in EEPROM. I'm sure you'll want to be ethical and use the open-source Sigrok Pulseview software... that works with both versions anyway.

[tggzzz]

As I mainly work on analogue/RF equipment I find there's not often a need for digital tools other than my trusty PICkits and my laptop but, occasionally, I need to dig into digital things and it'd be handy to be able to monitor and decode SPI, I2C, serial etc. transactions.

Have a look at a "bus pirate". I don't have one, but if/when I need to look at messages on serial busses, that will be my starting point.

[CJay]

I'm leaning towards the USB LA because they can be a small piece of kit and I intend to keep my Tek/Hameg scopes

There won't be anything 'high speed' at the target end so one of the cheaper USB devices might be adequate, I'll order an eBay special if they're so cheap, recommendations?

It's pretty much a no-brainer from what you say. The 8-bit USB LAs are so cheap that even if you later come up with a higher speed need then you really haven't lost anything. If you want one from the UK then Hobby Components on ebay are pretty good (no association).


P.S. Some UK sellers indicate Saleae compatible and some aren't. It's a USB PID/VID difference in EEPROM. I'm sure you'll want to be ethical and use the open-source Sigrok Pulseview software... that works with both versions anyway.

OK, I'll take a look at those, they're cheap enough that I can use it to stop a table wobbling if it's no good, do they really do 24MHz?

I will of course do the right thing as it's a current and supported product from Saleae.

Plus, I like the look of SIGROK.

T

[ralphrmartin]

Picoscope are clearing out some of their old models at bargain prices - check eBay. I just got a 200MHz MSO 2+16 channel USB scope for £300, and that includes all the decoding, math and whatnot you may want. Also see https://www.picotech.com/picoassured

Edit: Seems like all the bargain MSOs have gone.

[Gyro]

do they really do 24MHz?

Well yes... and no. Yes they will sample at 24MHz (PC and USB permitting), that means that their minimum time interval resolution is just under 42ns. You clearly need several clock samples to be able to make any sense of mark/space, setup and hold times etc. So in practice you are talking about significantly slower speeds for actual decoding capability. Off hand I don't know what maximum speeds people have actually managed for each different protocol. A data, clock and strobe type interface will make 'sense' at higher speeds than some others for instance. It should be good enough for most arduino type uses though.

To go much faster, you would need a device that has on board memory, triggering, etc. rather than just steaming it all over USB and letting the s/w sort it out. That involves a step change in cost of course. Some of the 16 bit streamers will manage faster speeds by cutting down on the number of chanels in use though.

[artag]

I will of course do the right thing as it's a current and supported product from Saleae.

Plus, I like the look of SIGROK.

T

It's a supported product from Saleae, but not current. Their current versions have analog acquisition too, and go to 500MS/s if you have the USB bandwidth for it. They're really very nice, though still not a replacement for a scope in most cases. I'd very much recommend them, especially if you can justify the more expensive models.

[CJay]

I could probably rationalise a new (to me) digital 'scope with logic features but spending the same on a dedicated LA (even though it'd have far better functionality) wouldn't make sense for the amount of digital 'stuff' I need to debug/repair/design etc.

I've dipped my toe in the water and have ordered a couple of the Cypress based devices just because they're dirt cheap and look like they may well do what I need with Sigrok.

I've still got the buidget to buy something like the Hantek 34 ch LA if the cheapies don't work for my needs.

[AndyP]

do they really do 24MHz?

To go much faster, you would need a device that has on board memory, triggering, etc. rather than just steaming it all over USB and letting the s/w sort it out. That involves a step change in cost of course. Some of the 16 bit streamers will manage faster speeds by cutting down on the number of chanels in use though.

The Dreamsource Labs Ds Logic Pro I settled for is 'only' $99, 256Mbit of memory (16MSamples 100MHz 16bit, or 8 bit @200MHz, or 4 bit @400Mhz) - complex triggers including serial trigger, optional external clock, trigger in and trigger out to sync with 'scope, FPGA  is open source for those that want to tinker. It can also stream continuously at USB speeds if needed.   

http://dreamsourcelab.com/wiki/index.php?title=DSLogic_Core_Board
There is also a guide somewhere for putting your own FPGA image on there!

I found the speed / buffering useful for higher speed serial links when the cheap $10 USB logic analyser ran out steam.
The point was to give an idea of what a $100 USB logic analyser can provide, I expect there are similar offering from other manufacturers, of course if the $10 LA does the trick then stick with it.

[JPortici]

I'd go for a digital scope with decoding because in many cases a problem is only truly visible in the analog signal.

i can't agree more.
I just had a problem with a preproduction unit and a level translator. Data was being transmitted but no responce back. probing with a scope revealed a problem in the level shifter low voltage side and a shutdown mcu, possibly fried from it.
a logic analyzer would have displayed the value as if it was fine

i suggest a pico2000 if you're on a budget

[tggzzz]

I'd go for a digital scope with decoding because in many cases a problem is only truly visible in the analog signal.

That has its problems. In particular you are presuming that the scope "interprets" the analogue signal in the same way that the digital receiver will.

A very good technique is:

• use a scope to ensure that the (analogue) signal is a well-behaved digital signal, in terms of voltages, undershoot/overshoot, transition monotonicity and speed
• then, once you know you have a valid "digital" signal, flip to debugging in the digital domain with whatever tool is best, e.g. logic analyser, printf, bus pirate etc

That first step, assuring signal integrity, can be done with any scope that has sufficient bandwidth. Bandwidth is everything in this context, and 100MHz is barely sufficient.

For example, driving low loads Fairchild NC7SV74 flip flops reportedly have 77ps risetimes, which is so insane as to be suspicious!
https://dl.dropboxusercontent.com/u/53724080/Parts/Logic/NC7SV74_2.JPG
https://dl.dropboxusercontent.com/u/53724080/Parts/Logic/NC7SV74/NC7SV74_test.JPG
However sub-ns risetimes are common, even when driving loads.

[nctnico]

IMHO you are overreacting when it comes to risetimes and bandwidth when in case of protocol decoding. Most communication busses transport data over a certain distance so you don't want sharp edges (high frequencies) to prevent excess radiation and inputs are usually designed to provide some filtering. 100MHz bandwidth really is more than enough for 99.9% of the UART, SPI, I2C or CAN busses you find in the field.

[Fungus]

i suggest a pico2000 if you're on a budget

Apart from the fact that a reasonable spec. Pico2000 costs more than the equivalent oscilloscope.

[JPortici]

i suggest a pico2000 if you're on a budget

Apart from the fact that a reasonable spec. Pico2000 costs more than the equivalent oscilloscope.

and has a lot more support for decoding than the equivalent oscilloscope

[tggzzz]

IMHO you are overreacting when it comes to risetimes and bandwidth when in case of protocol decoding. Most communication busses transport data over a certain distance so you don't want sharp edges (high frequencies) to prevent excess radiation and inputs are usually designed to provide some filtering. 100MHz bandwidth really is more than enough for 99.9% of the UART, SPI, I2C or CAN busses you find in the field.

In many cases I agree.
But sometimes there are "interesting" effects which should not be there - and if they are there then you need to find them.
And if you are going to get a scope, you might as well get one that can accurately observe all the signals on/between your boards.

Besides, you are only considering one definition of "protocol". Even register-to-register transfers have a very simple "protocol" which involves setup and hold times.

[ralphrmartin]

i suggest a pico2000 if you're on a budget

Apart from the fact that a reasonable spec. Pico2000 costs more than the equivalent oscilloscope.

Not when you factor in the cost of decoding being a pay for option on the scope, but is included with the pico.

[JPortici]

and some of the standars decoded are on other scopes only when they have one more zero on the baseprice