EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: MathWizard on September 03, 2022, 03:36:24 pm
-
I don't need a $400-500 LA, but I don't want a $20 one either. And I don't want a $20 one sold for $100 either.
What do people recommend these days ? And what makes the good ones so expensive ?
-
And what makes the good ones so expensive ?
Sampling rate, width, competent signal integrity. Plus, of course triggering and capturing filters so you don't even see all the irrelevant data that is inhaled by the cheap LAs. Optionally protocol analysis, but that can often be done with computers and printf() statements.
Without the filters and signal integrity, all else is in vain.
-
I don't need a $400-500 LA, but I don't want a $20 one either.
What specifically do you need and what do you want to do with it? Are you looking for a stand-alone or PC-based LA? How many channels? What protocols or busses do want to look at (and at what speeds)? Do you want analog capability on some/all channels? Do you need to capture very long sequences (which may require either deep memory or high-bandwidth USB streaming)?
-
Do you need to capture very long sequences (which may require either deep memory or high-bandwidth USB streaming)?
Very long sequences can usually be avoided by decent filtering in realtime, so that irrelevant stuff is neither captured nor stored.
Examples:
- check setup and hold times are valid, by storing nothing until an invalid timing is found
- only capture state changes (plus time interval), not all the crap that occurs between successive clocks
- verify maximum latency between event and action is observed by only capturing failures
Decide whether you need protocol analysis, i.e. the semantic content of sequences of messages and responses. If so, a conventional LA is probably the wrong tool.
-
a) What do you need to analyze?
would be a good start....
-
a) What do you need to analyze?
would be a good start....
Agreed. Nowadays an oscilloscope with digital inputs is a good place to start. The lower end logic analysers don't bring much extra to the table.
-
Do you need to capture very long sequences (which may require either deep memory or high-bandwidth USB streaming)?
Very long sequences can usually be avoided by decent filtering in realtime, so that irrelevant stuff is neither captured nor stored.
The problem is that the low cost logic analysers don't have the trigger abilities to support such a workflow. And it is better to capture large amounts of data so you can look back what has happened before a problem occured. In the end the fact a problem occured isn't that interesting, the cause is what you are after. Even better if you can record several good 'transactions' together with the 'failed transaction' to make a 1 on 1 comparison based on having the system in the exact same hardware / software configuration.
-
Do you need to capture very long sequences (which may require either deep memory or high-bandwidth USB streaming)?
Very long sequences can usually be avoided by decent filtering in realtime, so that irrelevant stuff is neither captured nor stored.
The problem is that the low cost logic analysers don't have the trigger abilities to support such a workflow.
My contribution is centred on what makes good LAs expensive.
And it is better to capture large amounts of data so you can look back what has happened before a problem occured. In the end the fact a problem occured isn't that interesting, the cause is what you are after. Even better if you can record several good 'transactions' together with the 'failed transaction' to make a 1 on 1 comparison based on having the system in the exact same hardware / software configuration.
Any LA with decent triggering and filtering will have exactly that capability. Digitising scopes nicked the "trigger point is on the right hand side of the screen" from LAs.
-
a) What do you need to analyze?
would be a good start....
Agreed. Nowadays an oscilloscope with digital inputs is a good place to start. The lower end logic analysers don't bring much extra to the table.
LAs enable state to be captured, i.e. the value at the instant the clock occurred with the relevant chip select and address signals, without the irrelevant crap between clocks and w.r.t. other devices
The real competition for LAs is protocol analysers and application specific printf() statements
-
Again, the low end LAs typically don't have the trigger & storage abilities to do what you suggest.
Additionally, MSOs do 99% of the tasks you typically use a logic analyser for and most modern MSOs support rather complicated trigger conditions. I have a high end logic analyser but I hardly ever use it (it has probably been sitting unused for 2 years already) because the MSO that sits on my desk can fullfill most of the logic analysis tasks AND show digital signals change in realtime.
Bottom line: if you don't have a specific need that can only be fullfilled by a logic analyser, the better option is to spend a bit extra on an oscilloscope which has digital channels. For most oscilloscopes the only thing you need extra are the digital probe pods.
-
Again, the low end LAs typically don't have the trigger & storage abilities to do what you suggest.
You know better than to snip context and invent strawman arguments. Your doing that is surprising and unimpressive.
Which bit of this did you not understand....
My contribution is centred on what makes good LAs expensive.
-
And thus your remark is not applicable to what the OP is looking for; you should have made clear in your earlier replies that what the OP is looking for might not exist in her/his budget range. But it all depends on the requirements we still know nothing about.
Fact is that in today's world a logic analyser is a very niche piece of equipment. Keysight and Tektronix have stopped making logic analysers a long time ago. In 20 years from now it may not even be possible to perform measurements where a logic analyser would be ideal because the equipment is no longer in working order.
-
Fact is that in today's world a logic analyser is a very niche piece of equipment. Keysight and Tektronix have stopped making logic analysers a long time ago. In 20 years from now it may not even be possible to perform measurements where a logic analyser would be ideal because the equipment is no longer in running order.
I am pretty sure Keysight still make "traditional" logic analyzers, targeted mostly at debugging and validating DDR memory systems. It is a small niche but I don't think it is going away anytime soon.
I was assuming that the OP is most likely looking for a Saleae-style USB pod. Although I realize that "PC compatible" might mean (or used to mean) that it runs PC software.
-
Fact is that in today's world a logic analyser is a very niche piece of equipment. Keysight and Tektronix have stopped making logic analysers a long time ago. In 20 years from now it may not even be possible to perform measurements where a logic analyser would be ideal because the equipment is no longer in running order.
I am pretty sure Keysight still make "traditional" logic analyzers, targeted mostly at debugging and validating DDR memory systems. It is a small niche but I don't think it is going away anytime soon.
You are right. I was under the impression that Keysight also had left the logic analyser market. Tektronix has for sure.
-
And thus your remark is not applicable to what the OP is looking for;
Congratulations, you've now got the chance to see the limitation of my replies, as given in my reply #2.
you should have made clear in your earlier replies that what the OP is looking for might not exist in her/his budget range. But it all depends on the requirements we still know nothing about.
Congratulations. That second sentence demonstrates why your first sentence is nonsensical.
Fact is that in today's world a logic analyser is a very niche piece of equipment. Keysight and Tektronix have stopped making logic analysers a long time ago. In 20 years from now it may not even be possible to perform measurements where a logic analyser would be ideal because the equipment is no longer in working order.
As you have subsequently acknowleged, that isn't quite right.
You seem to think the sun shines out of MSOs, and that they don't have limitations. MSOs are just as limited as LAs, but in different ways.
Neither MSOs nor LAs are suitable for "higher up the stack" debugging, but if I had the option I would choose a decent two channel scope, a decent LA, and printf() statements.
-
Ok what I want it for is general purpose hobby repair/tinkering, like reading what a MCU/eeprom are saying. I have a Siglent scope that can decode I2C, but so far, if I'm doing it right, I find the time base is way too short to capture what's going on, using single shot. Like on a MP3 player or monitor I was working on. For figuring out what a MCU is doing on a mobo, the more channels the better.
I bought a $20 LA ages ago, but it got lost in the mail.
So yeah more than just a logic analyzer, I'd want software or an MCU to do the common protocol analyzing. And yeah I don't want a standalone thing, I'd want to use it with PC.
I see there's people with Arduino LA projects. That's interesting too, but in a week or 2, I'm going to order something, even if it is just a $20 LA again, I could use it for now on big MCU's like on a mobo.
What about Hantek models, they seem to have cheap hobby level stuff that's PC compatible, and not total garbage.
-
Ok what I want it for is general purpose hobby repair/tinkering, like reading what a MCU/eeprom are saying. I have a Siglent scope that can decode I2C, but so far, if I'm doing it right, I find the time base is way too short to capture what's going on, using single shot. Like on a MP3 player or monitor I was working on. For figuring out what a MCU is doing on a mobo, the more channels the better.
You likely need to use zoom mode to force a longer capture on your Siglent scope. In normal mode, it doesn't use all the memory even if you manually set it; just enough to fill the screen (unlike oscilloscopes from other brands).
-
Yeah I should hookup my ardiunio and make it blink an output a know # of times at a known rate, etc, and figure out what the scope does.
I see the Hantek4032L, something like that is probably a good fit for me. I've been learning a lot more about 4-bit breadboard computers, and for when I'm learning MCU's, it would also be nice to have a LA/PA show me what's happening, on my computer screen.
-
What about Hantek models, they seem to have cheap hobby level stuff that's PC compatible, and not total garbage.
When you buy a USB logic analyzer, you are also paying for the software, which can make big difference in practice. Hantek does not exactly have a great reputation in that regard, although I do not know about the logic analyzers specifically. But you can usually download a demo version that you can explore without hardware. It is highly recommended to do that before buying and compare. Other brands to potentially look at are DSLogic, Zeroplus, Ikalogic, Saleae Logic, as well as Digilent's Digital Discovery.
For the non-brand 10$ gadgets, you will likely be using the open-source software PulseView. It has support for a lot of other devices but some work better than others (even if they are technically listed as "supported").
-
Agreed. The software is what makes or breaks a good device. For most modern day test equipment the hardware is pretty simple; it is the software that makes it worth the money (or not).
-
Thoughts on the Digilent Analog and Digital Discovery devices for this purpose?
-
I have a Siglent scope that can decode I2C, but so far, if I'm doing it right, I find the time base is way too short to capture what's going on, using single shot.
Like Nico says, are you using Zoom mode ?
You should be using something like 50ms/div then press Zoom where the timebase control switches to the zoomed window where you can zoom into each packet and pan around the record.
However it's best you get rock solid triggering sorted first before using Zoom as for most protocols requires a falling edge trigger, signal thresholds set correctly with the packet vertical amplitude and most importantly Holdoff applied to prohibit retriggering within a packet.
Show us some screenshots with appropriate menus visible.
-
Thoughts on the Digilent Analog and Digital Discovery devices for this purpose?
I've bought digital discovery and so far I'm happy with it. I used it mostly to configure SPI devices, so my LA usage is so far limited. No issue with sw so far, seems well made. The ability to make a quick script to configure your SPI devices is super nice.
-
I see there's people with Arduino LA projects. That's interesting too, but in a week or 2, I'm going to order something, even if it is just a $20 LA again, I could use it for now on big MCU's like on a mobo.
Yep. Get one of those cheapo 24Mhz, 8bit analyzers and see how it goes. They're about $6 on Aliexpress. Use "Pulseview"software.
https://www.aliexpress.com/wholesale?SearchText=logic+analyzer (https://www.aliexpress.com/wholesale?SearchText=logic+analyzer)
If it doesn't work out then you only lost $6.
-
The OP might like to consider one of the "bus pirate" family of devices.
Downloadable "personalities" are available, and they can also generate signals.
For the avoidance of doubt, I haven't used them.
-
The OP might like to consider one of the "bus pirate" family of devices.
Downloadable "personalities" are available, and they can also generate signals.
For the avoidance of doubt, I haven't used them.
I have used bus pirate, Bus blaster and the logic sniffer from DP all are recommended products for signal debugging, sniffing and jtag access.
I would recommend them as starting device's for any home lab.
Of the 3 the bus pirate has the most uses as it can act as 4 channel LA, general purpose uart (i2s/spi/serial etc) , jtag adapter and more.
darkspr1te
-
MathWizard
Your Siglent scope can do more than you imagine and you just need learn how.
Below SPI decode with timebase set to max packets...it warns you and then just press the timebase control to enter Zoom mode and again set the timebase and/or H Pos to examine any packet.
Ch1 yellow is Clk set to Timeout for SPI decoding with just 2 channels. For simple decodes do set the Trigger edge correctly and add some Timeout for a little more than a packet.
Decode table also shown that you can scroll through which displays some 2600 bytes are decoded that you can examine the result in 2 ways, Pan the decode bar with the H Pos encoder or select Scroll and examine the results list.
Always keep an eye on both the blue V arrows to know where you are within the capture and BTW this is NOT a capture but live continuous decoding that has been captured by pressing the Print button.
-
MathWizard
Your Siglent scope can do more than you imagine and you just need learn how.
Below SPI decode with timebase set to max packets...it warns you and then just press the timebase control to enter Zoom mode and again set the timebase and/or H Pos to examine any packet.
Ch1 yellow is Clk set to Timeout for SPI decoding with just 2 channels. For simple decodes do set the Trigger edge correctly and add some Timeout for a little more than a packet.
Decode table also shown that you can scroll through which displays some 2600 bytes are decoded that you can examine the result in 2 ways, Pan the decode bar with the H Pos encoder or select Scroll and examine the results list.
Always keep an eye on both the blue V arrows to know where you are within the capture and BTW this is NOT a capture but live continuous decoding that has been captured by pressing the Print button.
I don't know whether the following are of interest to the OP, but when debugging at that kind of level I've always needed:
- running silently until "Aargh" (or some other message) is encountered? Then displaying the messages that lead up to it?
- ignoring everything except messages "Vpos xxxx" interleaved with other uninteresting messages?
No tool can do everything, and there are good reasons why a tool shouldn't attempt to do something poorly. Basically "eierlegende Wollmilchsau" are a pain.
The key is to know which tool to use for which job.
-
The key is to know which tool to use for which job.
And read the #15 post that clearly indicates MW is unclear about his scopes capabilities of which I hope I've demonstrated are adequate for his reported needs.
We could also use Search or dedicated protocol triggering to analyze more complex requirements.