Choosing Logic Analyzer advise needed

[sairfan1]

Hi I'm looking for logic analyzer, I mostly work on UART, SPI and I2C basic protocols in near future would like to work on LIN and CAN bus.  I don't like PC/USB based logic analyzers because we need to switch again and again between logic info and code etc.

I searched ebay there are some cheap professional logic analyzers like TLA614, TLA704 and HP 16600A.  I'm good to spend  around $300 to $500. But I'm not sure how much they can full fill my requirements does they support such basic protocols? in case of TLA614/704 what modules are enough to install?  And does they display byte info in a way like Saleae does like each byte can be shown in hex, bin or dec format, and we can separately configure each pin for logic as on.

[rstofer]

I can't help much with the logic analyzer question.  I built mine with an FPGA board (www.sump.org) and it does decoding.  But it's pretty elementary.

I can help with the switching back and forth!  Get a second monitor.  I'm thinking that 3 monitors would be better.  Datasheet on the left screen, code on the center screen and logic analyzer on the right screen.  Two of my workstations have multiple monitors and they are a joy to use when developing just about anything.

[ade]

At that price point you may also want to consider a digital storage oscilloscope like the popular Rigol 1054Z which can decode rs232, i2c, and SPI.  (I believe the more expensive DS2000 series is required for CAN bus decoding.)

While not as good as a true logic analyzer for certain things, a DSO can diagnose many serial/i2c/SPI signaling problems which can't be "seen" by a logic analyzer.

[sairfan1]

@ade thanks for reply, any idea about logic analyzers i mentioned? are they workable in required situations? and what model would you suggest for DSO?

[suicidaleggroll]

I know you said no USB-based, but all you need is a second monitor or a second workspace (native on Linux and OSX, there are 3rd party tools on Windows) and the code/logic switching is a non-issue.  You're hunting for a solution to a problem that doesn't really exist IMO.

The Saleae analyzers are great pieces of equipment and fit right in your budget.  You can get the Logic Pro 8 (8 ch, 500 MS/s digital, 50 MS/s analog (USB bandwidth permitting)) for $480 and it'll decode anything you can throw at it, you can even write your own protocol analyzers if you need to.  The analog section isn't nearly as good as what you'd with with an MSO, but the huge recording buffer and cross-platform UI software more than makes up for it IMO.  The analog is perfectly usable for "slow" <5 MHz signals (UART, I2C, some SPI), and for the rare occasion when you need to do analog debugging on a link running faster than that, you can hook up a scope.

[ade]

Those older logic analyzers are highly capable + complex and I generally don't recommend getting them unless you have very specialized needs (i.e., the need to analyze dozens or hundreds of high speed digital channels simultaneously).  Otherwise I think you are actually better off with lower cost USB-based logic analyzers.

Also the logic analyzers of that era were basically PCs with hard drives on them -- if you do get one you'll probably want to replace the internal hard drive and/or make a good backup, as the old software may no longer be available from the manufacturer.  They tend to run ancient OS like Windows 98 and Windows 2000.

If you don't have any oscilloscope yet then the Rigol DS1054Z 4-channel DSO previously mentioned is very popular and is within your budget range.  There is a "hack" (key generator) people use to "unlock" all of the scope's options for free, including serial/I2C/SPI decoding (I'll leave it up to you on whether or not to do so -- seems like individuals almost universally hack theirs).

The DS1054Z won't decode LIN/CAN but you can still easily debug most LIN/CAN issues with (any) scope.

Having an oscilloscope allows you to see for yourself what the actual signals look like, rather than what the logic analyzer thinks the signals look like.

[sairfan1]

Though I like Saleae interface, but it doesn't make sense at all to spend this much and still depend on my computer. If it was around $100 I would definitely go for that.

In current scenario it's better to spend little more to have more, I guess Rigol is better choice and I should consider its model with logic analyzer?  But then question comes does it shows values in bin, hex and dec formate for each signal line? And does it support to configure input line as per my own logic as well. (That is because some time we have to debug different logic like TFT interfacing etc)

[ade]

Depending on a computer isn't a bad thing.  E.g., for debugging CAN bus in the field, you can easily hook up a USB analyzer to a laptop and take it with you anywhere.   Moving an old big bench-top logic analyzer even to the garage is a big PITA -- plus you always need to be mains connected through long extension cords, etc.

DSOs like the the 1054Z can decode and show bin/hex/dec even without a logic analyzer module; The decoding is all done in the 1054Z's firmware instead of via hardware.

The ones with a logic analyzer module (Mixed Signal Oscilloscopes -- MSOs) are very nice, can handle more digital channels (16 or more), but usually are also quite a bit more expensive... usually US$600 and up... nice models like the Rigol MSO1074Z are US$850 or so.

[danadak]

I have a couple of these (just search on ebay for logic analyzer) -

http://www.ebay.com/itm/24MHz-8-Channel-USB-Logic-Analyzer-8-CH-Logic-Analyzer-for-Arduino-MCU-/191685084604?hash=item2ca15245bc:g:KFYAAOSwyQtV1vOO

They use the Saleae software. I don't think they really do 24 Mhz, but great for I2C, low speed SPI, UART.

Note inputs are not protected, some guys making up simple breadboard with R and diode protection
networks. If you are only working in LV environment you do not need to do this.

Regards, Dana.

[tszaboo]

I dont think the LAs you mentioned are a good fit. They are old, they dont have much memory. They are designed to work with massive amount of signals, like 100+ while you are only going to connect something like 5 signals to it. They are designed for state analysis ei for a microcontroller system, and while they are maybe going to do the job, they are not the right size hammer.
CAN is not handled correctly on many expensive LAs, or they lack features which are only in the dedicated CAN loggers. Like a message might be ok for CAN, but not for the perticular system and such, bad timing etc...
Debugging the other busses are very simple. But if your system is complicated, you are going to need all the memory you can get. Like a bit error, happening 0.01% of the cases, communication every 100us will break your system very often, but unless you can trigger on the error, you will not be able to catch it if you dont have the memory.
The USB steaming solutions are somewhat OK (saleaeaea or whatever), depending on the speed. Best is to get a few Mb of memory, and online compressing, so the memory is filled with data, not zeros.

[onlooker]

Something like the one in the link below may be of interest to the OP. It supports many protocals.  This one sales for $189.  I had someone bought one for me from China for about $150 a while back (not from the link below). The software got updates every several months. It worked OK, though I used it only for serial analysis so far.

http://www.aliexpress.com/item/Kingst-LA5016-LA5016-PC-USB-Logic-Analyzer-500M-max-sample-rate-16CH-5B-samples-english-software/32566563077.html?spm=2114.01020208.3.1.XNMDiW&ws_ab_test=searchweb201556_1,searchweb201602_1_10036_10035_301_10034_507_10032_10020_10017_10005_10006_10021_10022_401_10018_10019,searchweb201603_9&btsid=b31beca7-ca94-40d3-b467-a7344fef50b8

[sairfan1]

I'm convinced to go for Rigol I recently watched review by Dave and liked it.  What is difference between DSO and MSO models? in term of functionality and ability?

[suicidaleggroll]

Though I like Saleae interface, but it doesn't make sense at all to spend this much and still depend on my computer. If it was around $100 I would definitely go for that.

I don't really see what the problem is though.  If you're using a logic analyzer on a serial line then 99.9% of the time it means you're debugging firmware on some kind of MCU, which means you almost certainly have a computer there anyway.  What's the harm in using it to run the LA as well?

I think you're looking at it the wrong way, you're not spending $500 and still depending on your computer, you're only spending $500 because you're taking advantage of using the computer you already have there anyway, that otherwise would just be sitting idle with an IDE or text editor open.  Without the computer, you'd need to spend an order of magnitude more money on a completely self-contained device with the same functionality.  Much of the time a logic analyzer needs a big buffer and powerful searching functionality to do its job properly, the 12Mpts on the Rigol just won't cut it for any relatively high speed signal.  That's a 1.5 MB buffer per channel (0.75MB if you're using >8 channels)...a USB analyzer can take advantage of the GBs you have in your machine.  Nearly every time I need a logic analyzer, I need to record at least 5-10 seconds of data at at least a few MHz (usually closer to 25-50 MHz since the serial line will be running at 1-4 MHz) on a handful of lines.  That already excludes the $1100 Rigol as a viable option, right out of the gate.

[rstofer]

MSO vs DSO -

DSO -> Digital Storage Oscilloscope - the ordinary type of digital scope

The MSO (Mixed Signal Oscilloscope) will have some number of digital channels in addition to the analog channels that may either be displayed or the scope may be used as a logic analyzer.  Look at the User Manual for the Rigol MSO 1104Z  DSO -> Digital Storage Oscilloscope

http://www.rigolna.com/products/digital-oscilloscopes/mso1000z/mso1104z-s/

I haven't read the manual very well and I haven't read the manuals for any other Rigol MSOs but here's my concern (may not be correct):  The digital samples to be displayed are clocked by the scope sample clock which is in no way related to the incoming data.  As a result, you have to sample at some multiple of the data rate.  Even then, signals that are acceptably aligned for the system may not show up properly aligned on the scope.  The one important feature I need is a 'state' clock input.  The data on the bus represents the state of some Finite State Machine and I have a state clock that I want to use as the capture clock.  I don't care what Rigol's sample clock thinks, I want the data aligned to the state clock.  Make sure you have this capability in any LA you buy.  Most will have it, look for 'external clock input'.

Dave Jones has a video where he pretty much pans the MSOs as making very poor LAs.  If you need an LA, buy an LA.  A scope, even an MSO, is not a LA.

Here is V2 of a very cool piece of equipment:  http://store.digilentinc.com/analog-discovery-2-100msps-usb-oscilloscope-logic-analyzer-and-variable-power-supply/

I have V1 and I use it quite a bit.  It works as a dual power supply (V2 is variable voltage, I believe), dual channel analog scope, digital scope (16 channels), logic analyzer, arbitrary waveform generator and so on.  This thing is a lab in a box.

No, it doesn't have the state clock input.  But it does decode RS232, SPI and I2C.  It doesn't  have all of the display options the OP wants but it's still a pretty cool gadget!

[matkar]

I have a DS1074Z which I used for serial decoding in the past. Later I bought a USB based cheap 16ch 100MHz LA on ebay and I don't use my DSO for decoding anymore. I just check if the communication is not distorted and the levels are OK before starting the LA.
It is faster and in general much nicer to view messages on a bigger screen, where you can zoom messages in and out quickly on a very long record. For me there is no way back to clumsy button pressing and menu searching on a small 7" display.

[nomadd]

I have a DS1074Z which I used for serial decoding in the past. Later I bought a USB based cheap 16ch 100MHz LA on ebay and I don't use my DSO for decoding anymore. I just check if the communication is not distorted and the levels are OK before starting the LA.
It is faster and in general much nicer to view messages on a bigger screen, where you can zoom messages in and out quickly on a very long record. For me there is no way back to clumsy button pressing and menu searching on a small 7" display.

This^

I have a "hacked" DS1050Z and all the options plus speed-up. The SPI and I2C decode are fussy, slow and a bugger to get to trigger properly at times. The scope controls are finicky for this kind of work and the "decode" is very hit-n-miss, even at relatively slow serial data rates. The data display is also poor: small text, and most of it disappears into "blobs" with no attached data the minute you zoom out to see more than 3 or 4 bytes of data; you then switch to the "table" view of decoded data which is even smaller and more difficult to track...

I bought a cheap Saleae clone for £10 and use it with the latest Beta of their software (v1.2.5) It's a dream to use in comparison.

Love the little Rigol, but not for Serial decode. Especially when a cheap clone LA is a fraction of the price (albeit with Saleae's excellent software.)

[matkar]

I also had this unusual experience decoding with DS1074Z: https://www.eevblog.com/forum/testgear/ds1074z-decode-rs232-ascii-problem/msg774455/#msg774455
It took me some time to figure out it was the trusty scope who pulled my leg. All is fine after an update.

[nomadd]

I also had this unusual experience decoding with DS1074Z: https://www.eevblog.com/forum/testgear/ds1074z-decode-rs232-ascii-problem/msg774455/#msg774455
It took me some time to figure out it was the trusty scope who pulled my leg. All is fine after an update.

Thanks for that. I waited a few months for the firmware to stabilise before I bought the Rigol, then updated the firmware the moment it arrived. Fortunately I never suffered the problem you discovered.

In a tight spot, the Rigol is passable as an LA for Serial Decode - I can usually come up with some bizarre arrangement of triggering to make it work - but it's just not worth the hassle when a cheap LA is so much better, as we've realised.

Guess you can't have everything for the low price.

[ez24]

I'm convinced to go for Rigol I recently watched review by Dave and liked it.  What is difference between DSO and MSO models? in term of functionality and ability?

Download and read the manual before you buy to see if it will fit your needs.  They have good manuals.

[Howardlong]

I have the MSO1074Z-S, and have used it fairly extensively for serial decoding.

I also have a number of USB based LAs like Saleae, Logicport, Bus Pirate, Digilent Analog Discovery.

Be careful about what is read into Dave's video about whether MSOs are worth it or not. Yes, you can get a DSO and a USB LA for less than an MSO, but read on...

While I know I'm swimming upstream on this one I almost always prefer to use a scope/MSO for serial decode. Why? Because the triggering facilities are far, far better than any of the LAs I mentioned, and it's common for me to want to time correlate between analogue and digital channels. The only exception is with very parallel busses which are becoming less and less common nowadays. Also note the limited sampling rate with some of these LAs, depending on your application that might be quite a drawback. A further benefit of the MSO1074Z is that it will sample up to 1GSa/s (0+8 ch, with trigger off digital channel) which beats any of those LAs I mentioned.

Be aware that to hack the MSO variant, unlike the DS1000Z, you'll need to physically disassemble them and use a JTAG emulator. It's not hard, but put aside at least the best part of half a day if you've never attempted something like this before.

Regarding the comment earlier about having a state clock, yes, of course you're not going to see a beautiful 50% square wave sampling a 100MHz signal at 250MSa/s with a 1 bit ADC, the same applies to most USB LAs in that respect, although the LogicPort supports that, although at a lower samplig rate. In practice the high sampling rate of the MSO1074Z means it's never been a particular concern of mine. I've successfully decoded 50MHz SPI on it, and I2C works reasonably well on both analogue and digital channels, but be aware that you can only decode a limited number of frames at a time. I did a YouTube video about I2C decoding here...



Take note that the MSO1000Z is either 4+0, 3+8 or 2+16 channels: an analogue channel is lost for each block of 8 digital channels. In practice thise is rarely a problem, the screen is already busy as it is.

The LA portion of the MSO1074Z does have a few quirks when you start using it with much gusto, I was decoding and triggering off a DDR bus a few months back out on site and found a couple of "features". They were obvious, but still it took my attention from the work I was doing. This is documented here but might well be fixed now, I haven't retested since perfoming a firmware upgrade in the interim. https://www.eevblog.com/forum/testgear/rigol-ds1000z-series-(ds1054z-ds1074z-ds1104z-and-s-models)-bugswish-list/msg727413/#msg727413

So in short, some people prefer USB based LAs, and I can understand that, the amount of screen real estate on your average PC far outweighs that of your average scope. The added benefit with the Saleae is the almost infinite buffer memory, but on the lower models the sample rate is lacking and the trigger options on all of them are very rudimentary, but you tend to use the Saleae in a different way to more traditional LAs, it's more like looking at a post dump report with the Saleae than looking at things in real time.

If you're happy with setting up more complex triggers you might find an MSO a better option for you, certainly that's my preferred option after being a regular LogicPort user for about a decade. I just got frustrated with the lack of trigger options and limited memory on the LogicPort.

Just keep in mind that the screen is limited on the MSO1074Z. It's pretty small for the resolution, so if you're at the bench for long hours, day after day, you may find it frustrating. There is also no post-capture search as you get on the Saleae, so getting your trigger right is more important on an MSO, it's just a different way of working. I use ithe MSO1074Z as my field scope, and for that purpose I really can't complain. Sure I'd like more bandwidth but as a compromise, in terms of features and size it works very well.

[rstofer]

Regarding the comment earlier about having a state clock, yes, of course you're not going to see a beautiful 50% square wave sampling a 100MHz signal at 250MSa/s with a 1 bit ADC, the same applies to most USB LAs in that respect, although the LogicPort supports that, although at a lower samplig rate. In practice the high sampling rate of the MSO1074Z means it's never been a particular concern of mine. I've successfully decoded 50MHz SPI on it, and I2C works reasonably well on both analogue and digital channels, but be aware that you can only decode a limited number of frames at a time. I did a YouTube video about I2C decoding here...

Great writeup!

In my case, only, I really need that state clock.  Whatever logic I am displaying is synchronous with the state clock and if the signals make the setup time for an external FPGA, they very likely make the setup time internally.  Clearly, there will be some variable amount of logic going on between state clocks, some signals taking longer than others to settle, and getting them displayed synchronous with the clock seems important.

But my application may be a very small subset of uses for a LA and, truth be known, the home-built sump.org LA is a bit cumbersome to use.  It works extremely well but I really should package it up.  As is, it is just an FPGA board laying alongside another FPGA board with wires running between.  BTW, 32 channels is not excessive and the next time around, I want 64 channels.

In fact, I saw an FPGA project where the creator simply built a LA into the project FPGA.  No external wiring required!  OTOH, there is a recompile of the FPGA project every time the LA inputs need to change.  But that's almost always the case anyway.  There's always 'just one more thing to look at!'  I'll think about this next time around.

The problem of width comes about because, left to its defaults, Xilinx ISE implements finite state machines with 'one hot' encoding - one bit for each enumerated state.  Got 100 states?  Fine, the state word is 100 bits wide!  Yes, I know I can force a different implementation but I also know that Xilinx takes that approach for a reason - notably less gates used for state decoding back to 'one hot' before implementing the state logic.  Coming up with a subset of state bits to send diagnostics to the LA can be a challenge.

That Saleae Pro 16 seems like a decent unit.  I don't know anything about it but the price is reasonable and the number of channels, while below my desired count, seem workable for many projects.

I don't have an opinion on MSO versus LA because I have never used an MSO.  As mentioned, triggering is everything.  In some cases, I have the project generate a trigger.  This is really easy to do with an FPGA project.  Kind of like a 'start debugging here' signal.  Just what you would do with C code - print an "I am here!" message.

[Howardlong]

You can trigger on state and one edge on the MSO1074Z but I am pretty sure the edge is sync'd to the sample clock, and thus you probably need a minimum pulse width too.