Products > Test Equipment
What to buy - Dedicated Logic Analyzer vs "The new Scopes" aka. SDS824/DHO924(S)
Kean:
--- Quote from: tggzzz on November 29, 2024, 08:24:06 pm ---The buspirate 5 is an interesting device that does more than capture and process data. It cannot easily be characterised in conventional terms, since it is in a different niche.
There is a lot of information on the website and forum https://forum.buspirate.com/ , but not a conventional manual. The principal creator is active there, and the natives seem friendly.
It is open source, so in theory you could create addons - but that wouldn't be trivial.
Overall it is useful to know the tool exists, so that you can use it when your needs fall into its niche.
--- End quote ---
The Bus Pirate 5 can operate as a basic logic analyser, but also so much more. It can also act as a device programmer, and supports many protocols like 1-Wire/SPI/I2C/UART/IR/analog for debug and reverse engineering. Think of it a bit like a swiss-army knife for embedded developers.
Usage does require some practice, and there is a learning curve compared to some other device - but it is incredibly versatile and extensible. There are also various adapters and accessories available to help get started.
The better doc site for exploring the capabilities of bp v5 is https://firmware.buspirate.com/
Traceless:
--- Quote from: btest on November 30, 2024, 02:21:38 am ---Traceless,
Sorry, I did not mean in any way to imply that the software was 'less refined'. I think it being based off of Sigrok is a good thing. I was just saying to not expect it to have the triggering capability of a scope, but I do not know of any PC based logic analyzers that do. The fact that it has multi-stage triggering puts it above almost all of the others, although that's a feature I have maybe used twice over the years. The strength of a PC based LA is the ability to do long captures for post analysis. You do not need fancy triggering to do that. I've only had the software crash a couple of times over the years and cannot remember the last time it did so.
--- End quote ---
Hi btest, no worries you didn't imply that the software was less refined, it was ebastler who actually mentioned this:
--- Quote from: ebastler on November 29, 2024, 07:06:06 am ---... There are B-brand alternatives with better hardware performance, e.g. the DSLogic series from DreamSourceLabs, although with less refined software.
--- End quote ---
Since I don't have hands on experience with either the Saleae nor the DSLogic I thought I'd double check with you and ebastler, who based on the comment also seems to have some experience either with the device or at least the sofware, how to interpret "less refined". I just wanted to make sure that I'm not running into buggy banana-ware situations as was reported in context of the Rigol scopes on multiple occasions in the forum.
--- Quote from: btest on November 30, 2024, 02:21:38 am ---Just for grins I got my Logic Pro 8 out and downloaded their latest software. It's come a long way since the last time I used it. But at over 3x the price of the DSLogic I don't see how anyone can justify it. Turning the Analog channels on puts such a limit on the sample rate that for me it becomes pointless. As tggzzz said: "My preference is to have two tools each of which is good at one job, rather than one tool that is poor at two jobs". I think we have all learned that lesson the hard way. Worst case you can always run the trigger out of the scope into an LA input to time correlate domains.
--- End quote ---
I'm actually very grateful for you helping me out with your very first post here on the forum, and ebastler commenting on the same device. Together with all the input from other users here on the downsides of the scope based solutions I decided to go with the DSLogic U3Pro32. I could still get the SDS824HD without the SLA1016 as standalone scope, albeit I'm currently considering to just stick with my current Hanmatek DOS1102 for now.
--- Quote from: btest on November 30, 2024, 02:21:38 am ---I also want to point out you need to buy for your specific needs. I write embedded control software on many different platforms but at a top level its all the same: gather data from multiple sensors, analyze it and generate a control output. My last job one of the inputs was IMU data at the rate of 4000 messages per second over a UART. My current job is primarily a combination of CAN and UART. In the digital domain the DSLogic has easily met all my needs. I am not sure what you are trying to do but take a good look at what type of signals you are trying to analyze and buy the appropriate tools. You can download the DSView app and run it in demo mode before buying the hardware. Same with the Saleae.
As much as I do in the digital domain there are times where you have to be in the analog domain. If you get noise coupled onto your signal lines the LA will show you the invalid transitions but to really see what is going on you need a scope. It's the only way to see how much your square wave has rounded off when sent over long cables. Or, like in a case I had a couple of years ago, where the EE team tried to get away with just using the RC clock internal to a STMicro instead of putting an external crystal oscillator on it. Worked great until it got warm. Then the RC clock drifted so much that it corrupted the CAN bus timing. Not obvious on the LA but clearly shown on the scope doing an eye diagram.
Hope this helps. Again, I can only relate to my needs which I am sure are not the same as yours and these are the tools that make me most efficient.
--- End quote ---
--- Quote from: tggzzz on November 30, 2024, 09:48:34 am ---
--- Quote from: btest on November 30, 2024, 02:21:38 am ---Hope this helps. Again, I can only relate to my needs which I am sure are not the same as yours and these are the tools that make me most efficient.
--- End quote ---
Just so, hence my reluctance to make A Pronouncement on this topic. (Doesn't stop me on other topics, though :) )
--- End quote ---
Currently I'm mostly interested in reverse engineering finished designs. Thus I don't expect to have to deal with analog design/noise issues because the designers of the DUT probably have figured this out already. As mentioned in my post above currently I'm interested into investigating the communication between Memory, CPU/Microcontroller and TPMs. I've also worked with CAN-bus communication in the past, however currently there is nothing CAN-related in the pipeline. I'm sure once I have the analyzer I'll quickly come up with more use cases. So buying a device with some headroom might just save me a redundant upgrade purchase later down the road.
One sophisticated use case in context of the TPM scenario would be differential power analysis. A stand-alone logic analyzer is not the right tool for that and the Scope/Logic combos might be more useful here. The basic idea here is: Cryptographic operations involve a key, depending on the key bits power consumption changes. As a simple example the Square-And-Multiply algorithm squares in each step but if there is a one bit you also have an additional multiplication operation which increases the power consumption. Of course you have to correlate your power consumption with key-based instructions somehow. If this is observable the issue is called "side-channel". Apart from the power consumption there is also often an observable timing difference (due to the different number of instructions in key-bit dependent code paths). Proper implementations will be "constant-time", i.e. perform "unnecessary" dummy operations so that both code paths for 1-bits and 0-bits in the key use the same amount of time, and power. This prevents leaking information about the secret key through timing and/or power consumption.
--- Quote from: btest on November 30, 2024, 02:21:38 am ---Just realized I missed your comment on 16 vs 32 channels. The issue with any PC based LA is USB bandwidth. The more channels you have active the lower the max sample rate. For me personally I rarely have more than 7 or 8 active channels and typically its 2 or 3 so 32 channels is way way overkill so I did not spend the extra $100. Again, it's what meets your needs. It's the same software for both, and with 3 channels or less they both can sample at 1GHz. For 6 channels it's 500MHz, etc. I have no experience with Siglent scopes so cannot make any comments there.
--- End quote ---
32 channels are indeed a lot. But double the channels for 100$ extra seemed like a no brainer. Even if I don't need the additional 16 channels, DSLogic sells the fly-wires for 25$, so the additional wires alone are worth the price difference. Plus currently they have a sale so the 32 channel version is sold at the 16 channel version price ;)
@Kean: Thanks for the bus pirate link I'll take closer look later. At its price point it might be worth to just get one and see what it can do. Currently I'm mostly using the el-cheapo USB-Uart adapters, and a RT809-F, T48 or CH341A for programming.
tggzzz:
--- Quote from: Traceless on November 30, 2024, 01:22:08 pm ---So buying a device with some headroom might just save me a redundant upgrade purchase later down the road.
--- End quote ---
Alternative strategy: buy something cheap that will allow you to understand what you really need to spend money on.
--- Quote ---One sophisticated use case in context of the TPM scenario would be differential power analysis. A stand-alone logic analyzer is not the right tool for that and the Scope/Logic combos might be more useful here. The basic idea here is: Cryptographic operations involve a key, depending on the key bits power consumption changes. As a simple example the Square-And-Multiply algorithm squares in each step but if there is a one bit you also have an additional multiplication operation which increases the power consumption. Of course you have to correlate your power consumption with key-based instructions somehow. If this is observable the issue is called "side-channel". Apart from the power consumption there is also often an observable timing difference (due to the different number of instructions in key-bit dependent code paths). Proper implementations will be "constant-time", i.e. perform "unnecessary" dummy operations so that both code paths for 1-bits and 0-bits in the key use the same amount of time, and power. This prevents leaking information about the secret key through timing and/or power consumption.
--- End quote ---
You should understand a scope's ENOB as a function of frequencies. That's particularly important when you are subtracting two nearly equal numbers. Z=X-Y, where X and Y are +-1% does not mean Z is +-1%!
I suspect you would benefit from an analogue tool dedicated to generating an analogue signal that is related to the power, and capturing that signal on a scope.
ebastler:
--- Quote from: Traceless on November 30, 2024, 01:22:08 pm ---Since I don't have hands on experience with either the Saleae nor the DSLogic I thought I'd double check with you and ebastler, who based on the comment also seems to have some experience either with the device or at least the sofware, how to interpret "less refined". I just wanted to make sure that I'm not running into buggy banana-ware situations as was reported in context of the Rigol scopes on multiple occasions in the forum.
--- End quote ---
I'd rather defer to people who have used the Saleae or DSLogic in earnest, or ideally both. My own experience with the software is only based on trying a cheap 8-channel "logic analyser" module with both, Saleae's software and Sigrok (on which DSLogic have based their software).
I just never quite "got" Sigrok, while Saleae's software seemed more logical and consistent to me. I have had similar experiences with other open-source projects, e.g. picture, video, audio editors: While they are powerful, they often can't hide their "design by committee" and "I'll implement my favorite feature here!" background.
A while ago I downloaded the DSLogic software since I was contemplating to buy one of their U3pro units. My impression was that while they had re-skinned Sigrok to make it look cooler (dark theme!), the UI logic was still largely the same. Usable, but not very satisfying (for me, personally). On the other hand, I would not expect bugs to be a major issue, since the underyling Sigrok base is pretty mature.
I never pulled the trigger on actually buying a better logic analyser -- thinking it over, I did not really have a need for more than the 4 channels of my scope so far. So again, someone who has used the actual DSLogic or Saleae units can hopefully provide better insight.
ebastler:
--- Quote from: tggzzz on November 30, 2024, 01:38:34 pm ---
--- Quote ---One sophisticated use case in context of the TPM scenario would be differential power analysis. A stand-alone logic analyzer is not the right tool for that and the Scope/Logic combos might be more useful here. The basic idea here is: Cryptographic operations involve a key, depending on the key bits power consumption changes. As a simple example the Square-And-Multiply algorithm squares in each step but if there is a one bit you also have an additional multiplication operation which increases the power consumption. Of course you have to correlate your power consumption with key-based instructions somehow. If this is observable the issue is called "side-channel". Apart from the power consumption there is also often an observable timing difference (due to the different number of instructions in key-bit dependent code paths). Proper implementations will be "constant-time", i.e. perform "unnecessary" dummy operations so that both code paths for 1-bits and 0-bits in the key use the same amount of time, and power. This prevents leaking information about the secret key through timing and/or power consumption.
--- End quote ---
You should understand a scope's ENOB as a function of frequencies. That's particularly important when you are subtracting two nearly equal numbers. Z=X-Y, where X and Y are +-1% does not mean Z is +-1%!
I suspect you would benefit from an analogue tool dedicated to generating an analogue signal that is related to the power, and capturing that signal on a scope.
--- End quote ---
I agree that this may be a challenging project -- both on the analog side (picking up the potentially small power variations, which may sit on an uncorrelated background of other fluctuations), and on the digital/logic analysis side (getting from the fluctuation patterns to the actual key).
But if the planned approach involves physically opening the device and tapping into as many digital signals as you can access (outside of the chip under investigation), then it would certainly be a use case which benefits from a mixed-signal scope which captures and displays the analog and digital channels in sync.
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version